3g Rf Engineering Guidelines

Flexent®/AUTOPLEX® Wireless Networks CDMA 3G-1X RF Engineering Guidelines

401-614-040 Issue 2 February 2003 Lucent Technologies - Proprietary This document contains proprietary information of Lucent Technologies and is not to be disclosed or used except in accordance with applicable agreements Copyright © 2003 Lucent Technologies Unpublished and Not for Publication All Rights Reserved

Copyright © 2002 Lucent Technologies. All Rights Reserved.

This material is protected by the copyright and trade secret laws of the United States and other countries. It may not be reproduced, distributed, or altered in any fashion by any entity (either internal or external to Lucent Technologies), except in accordance with applicable agreements, contracts, or licensing, without the express written consent of Lucent Technologies and the business management owner of the material.

Notice

Every effort was made to ensure that this document was complete and accurate at the time of printing. However, information is subject to change. Mandatory Customer Information Interference information: Part 15 of FCC rules

NOTE: This equipment has been tested and found to comply with the limits for a Class A digital device, pursuant to Part 15 of the FCC Rules. These limits are designed to provide reasonable protection against harmful interference when the equipment is operated in a commercial environment. This equipment generates, uses, and can radiate radio frequency energy, and if not installed and used in accordance with the documentation, may cause harmful interference to radio communications. Operation of this equipment in a residential area is likely to cause harmful interference, in which case the user will be required to correct the interference at his/her own expense.

Security statement

In rare instances, unauthorized individuals make connections to the telecommunications network through the use of remote access features. In such event, applicable tariffs require the customer to pay all network charges for traffic. Lucent Technologies cannot be responsible for such charges and will not make any allowance or give any credit for charges that result from unauthorized access. Trademarks

5ESS is a registered trademark of Lucent Technologies. AUTOPLEX is a registered trademark of Lucent Technologies. Flexent is a registered trademark of Lucent Technologies. Solaris, Sun, and SunOS are trademarks or registered trademarks of Sun Microsystems, Inc. UNIX is a registered trademark in the United States and other countries, licensed exclusively through X/Open Company Ltd. X Window System is a trademark of the Massachusetts Institute of Technology.

Ordering information

The ordering number for this document is 401-614-040. To order this or other Lucent Technologies information products, see “To obtain documentation, training, and technical support or submit feedback” on the 401-010-001 Flexent®/AUTOPLEX® Wireless Networks System Documentation CD-ROM or the documentation web site at https://wireless.support.lucent.com/amps/rls_info/rls_doc/cd_docs/.

Technical support

For technical support, see “To obtain documentation, training, and technical support or submit feedback” on the 401-610-010 Flexent®/AUTOPLEX® Wireless Networks System Documentation CD-ROM or documentation web site at https://wireless.support.lucent.com/amps/rls_info/rls_doc/cd_docs/.

Developed by Lucent Learning.

Lucent Technologies - Proprietary See Notice on first page

Contents

............................................................................................................................................................................................................................................................

About Purpose Reason for reissue Related information products Related training To obtain technical support, documentation, and training or send feedback Notations used

i ii iii iii iv iv

............................................................................................................................................................................................................................................................

1

Discussion of CDMA 3G-1X RF engineering

1-1

Introduction

1-2

Capacity and coverage for voice applications

1-4

Spectrum requirements Link budget Voice capacity RF engineering for data

1-4 1-4 1-5 1-6

Introduction Overview of traffic theory Data link budget Resource management Deployment

1-6 1-7 1-8 1-9 1-9

............................................................................................................................................................................................................................................................

2

Voice coverage, capacity and link budget

2-1

Introduction

2-2

Analysis

2-4 Reverse link Forward link

2-4 2-20

............................................................................................................................................................................................................................................................

3

RF engineering for data

3-1

Introduction

3-3

Traffic theory

3-4

Introduction

3-4

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

C O N T E N T S

i

General Erlang model Special cases: Erlang B and Erlang C Applications of Erlang C to 3G-1X data Data capacity

3-5 3-7 3-10 3-13

Introduction Data link budgets

3-13 3-19

Reverse link Forward link Resource management: RF scheduling

3-19 3-22 3-36

Introduction Scheduling algorithm Conclusions

3-36 3-36 3-43

............................................................................................................................................................................................................................................................

4

System deployment

4-1

Introduction

4-2

Spectrum use: Carrier assignments and guard band

4-4

Cellular band PCS band Preferred channels 2G/3G-1X spatial and frequency design

4-4 4-8 4-10 4-11

Coverage (spatial) design: overlay and greenfield Frequency design Mixed 3G-1X voice/data capacity and coverage

4-11 4-13 4-19

............................................................................................................................................................................................................................................................

5

Handoff

5-1

Introduction

5-3

Soft handoff definition Procedure IS-95B soft handoff algorithm Signal combining Coverage contour Discussion

5-3 5-3 5-6 5-8 5-8 5-12

Soft handoff costs on channel elements and packet pipe Soft handoff cost on forward link Soft handoff advantages Qualitative description of forward link soft handoff benefit IS-95B parameters SOFT_SLOPE, DROP_INTERCEPT, ADD_INTERCEPT SCH anchor transfer vs. SHO Hard handoffs

5-12 5-12 5-13 5-22 5-25 5-30 5-31 5-36

References

5-37

............................................................................................................................................................................................................................................................

6

Power control

6-1

Introduction

6-2

........................................................................................................................................................................................................................................................... C O N T E N T S

ii

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Reverse power control

6-4

Reverse power control for voice traffic RPC for packet data traffic Reverse SARA for 3G-1X packet data calls Forward power control Forward power control for voice traffic Forward power control for packet data traffic

6-5 6-8 6-9 6-11 6-12 6-15

............................................................................................................................................................................................................................................................

7

Extended carrier

7-1

Introduction

7-3

Single extended carrier

7-6

Reverse link Forward link Forward Data Capacity Growth strategies Applications Concentric carriers

7-6 7-8 7-14 7-15 7-18 7-19

Core carrier reverse link Core carrier forward link Traffic density Determining mobile location Growth strategies Applications Amplifier sharing - Quasi omni

7-20 7-23 7-25 7-25 7-26 7-26 7-28

Growth strategies Amplifier sharing - Asymmetric cell

7-29 7-31

Growth strategies

7-32 7-33

Summary

............................................................................................................................................................................................................................................................

8

Fixed wireless voice networks

8-1

Introduction

8-2

Parameters for fixed wireless analysis

8-3

Reverse link interference ratio (br) Required reverse link Eb/Nt for 3G Walsh code overhead Recommended loading factor Channel activity factor Reverse link coverage

8-3 8-4 8-6 8-8 8-8 8-9

System capacity calculation

8-10

Capacity calculation methodology Reverse link based capacity calculations Power requirements of forward link

8-10 8-11 8-17

3G-1X RC3 3G-1X RC4

8-17 8-21

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

C O N T E N T S

iii

3G-1X with SMV Conclusions

8-21 8-23

References

8-24

........................................................................................................................................................................................................................................................... C O N T E N T S

iv

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

About this information product

............................................................................................................................................................................................................................................................

Purpose

This document, CDMA 3G-1X RF Engineering Guidelines, addresses selected radio frequency engineering topics for the Lucent implementation of CDMA2000-1X, also known as CDMA 3G-1X, or simply 3G-1X. 3G-1X is a first-phase implementation of an IS-95 based, third generation CDMA network that complies with the recommendations for third generation wireless systems advanced by the ITU. In particular, 3G-1X offers both voice and data capabilities that are significantly improved with respect to IS-95 (second generation or 2G) offerings. Voice capacity is increased, offering up to twice the Erlang capacity per Hz achieved by IS-95. Features allowing burst speeds of up to 153.6 kbps for packet-switched data are also provided, in contrast to the maximum 14.4 kbps circuit-switched capability provided in IS-95. Furthermore, voice and data users can coexist within the same wideband carrier. In spite of the differences, many RF engineering principles of 3G-1X remain comparable to those of IS-95, particularly for voice applications. For example, the frequency reuse remains at 1. Voice link budget and voice capacity analyses are similar. Management of cochannel interference remains key for both voice and data users, and

............................................................................................................................................................................................................................................................ 401-614-040 i Lucent Technologies - Proprietary Issue 2, February 2003 See Notice on first page

is accomplished through application of familiar IS-95 principles such as fast power control, variable-rate voice coding, and careful network optimization. Accordingly, this document does not offer extensive discussions of topics with strong IS-95 counterparts; rather, in such cases, the differences relative to 3G-1X implementation are emphasized. More detailed information on IS-95 can be found in Lucent documents 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines, 401-703-201, PCS CDMA RF Engineering Guidelines, as well as TIA/EIA/IS-2000A standards. In contrast, much attention is devoted to topics without clear IS-95 analogues, such as the RF engineering issues associated with the advent of wireless packet data. These include packet data coverage, coexistence of voice and data users within the same carrier, and allocation of communication resources such as power amongst competing data users. Reason for reissue

Intended audience

How to use this information product

Starting from Issue 2, the information in this document is divided into two parts. Part I includes the updated information for Issue 1 of this document. Part 2 introduces the following new chapters. •

Chapter 5: Handoff

•

Chapter 6: Power Control

•

Chapter 7: Extended Carrier

•

Chapter 8: Fixed Wireless Voice Networks

This document is intended for engineers who will be responsible for system design and performance analysis of a Lucent Technologies 3G-1X system. This document is organized as follows: •

Part I, which consists of Chapters 1 through 4, provides a systemlevel picture of 3G-1X RF engineering. – –

–

Chapter 1, “Overview,” provides a brief overview of Part I Chapter 2, “Voice Coverage/Capacity/Link Budget,” discusses the essential coverage and capacity issues for voice applications Chapter 3, “RF Engineering for Data,” offers a discussion of RF data issues for 3G-1X, including a contrast between the Erlang B (voice) and Erlang C (data) models, analysis of capacity and coverage, and an examination of resource management

...........................................................................................................................................................................................................................................................

ii

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

– •

Related information products

Chapter 4, “System Deployment,” describes deployment issues, with focus on transition from 2G to 3G-1X Part II, which consists Chapters 5 through 8, provides more specialized discussions on individual topics such as power control and soft handoff. –

Chapter 5, “Handoff,” discusses the soft handoff procedures, algorithms, coverage, cost and benefit for the CDMA 3G-1X voice and packet data calls

–

Chapter 6, “Power Control,” describes the power control functions for both the forward link and reverse links for the CDMA 3G-1X voice and packet data calls

–

Chapter 7, “Extended Carrier,” provides guidelines for RF planning for “extended” carrier deployment

–

Chapter 8, “Fixed Wireless Voice Networks,” provides a detailed analysis of the system performance of 2G and 3G1X CDMA fixed wireless voice networks.

The Lucent document 401-610-000, Flexent®/AUTOPLEX® Wireless Networks Documentation Guide, provides a brief overview of each information product that supports Flexent®/AUTOPLEX® wireless networks systems, products, and features. The following Flexent®/AUTOPLEX® wireless networks information products are either referenced in this information product or provide additional information that relates to the Prepaid Services feature:

Related training

•

401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines

•

401-703-201, PCS CDMA RF Engineering Guidelines

•

TIA/EIA/IS-2000A, family of standards for “CDMA2000 Standards for spread Spectrum Systems” Global Engineering Documents 1-800-854-7179 1-303-397-7956

Lucent Technologies offers the following training products that relate to CDMA RF design and operation: •

CL3715, Understanding CDMA

•

CL8301, CDMA IS-95 and 3G-1X RF Design and Growth Engineering for Cellular System

•

CL8302, CDMA IS-95 and 3G-1X RF Design and Growth Engineering for PCS System

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

iii

To obtain technical support, documentation, and training or send feedback

•

CL8303, CDMA IS-95 and 3G-1X Base Station Call Processing

•

CL8304, 3G-1X RF Design Engineering and Base Station Call Processing.

The current release of the Flexent®/AUTOPLEX® wireless networks documentation is provided on the Lucent Technologies wireless networks customer technical support web site to all customers free of charge. To access the site, please visit: https://wireless.support.lucent.com To provide the most current, complete, and technically accurate documentation to customers as quickly as possible, revisions and updates of information products on the current release of the 401-010-001 Flexent®/AUTOPLEX® Wireless Networks Electronic Documentation CD-ROM are also provided on the site for all customers free of charge. For details on obtaining technical support, documentation, and training or sending feedback, refer to document “To Obtain Technical Support, Documentation, and Training or Send Feedback.”

Notations used

Notations used in this document are listed below. AG = Cell site antenna gain in dBi BL/VL = Building or vehicle penetration loss in dB, whichever is applicable CL = Cell site cable loss in dB d = The Eb/Nt required for acceptable quality Eb/Nt = The ratio of channel bit energy to spectral density of total channel impairment F = The receiver noise figure Fmobile = The mobile receiver noise figure Fcell = The base station receiver noise figure Fade = Fade (in dB) at mobile location g = The spread spectrum processing gain gnet = The net gain consisting of the product of mobile antenna gains, body (head) loss, building/vehicle penetration loss, cell site antenna gain, and cell site cable loss HL = Head (body) loss in dB int = The dB path loss at a 1 km reference point k = The multiplier used in a Gaussian distribution to achieve a certain percentile; for example, k=1.3 corresponds to a 1.3×σ choice which yields a 90th percentile M = The length of queue for the general Erlang model N = The number of active channels

...........................................................................................................................................................................................................................................................

iv

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Nmax = “pole” capacity No = Thermal noise density Nsect = The total number of sectors Nk = The total number of mobiles in sector k Ntotal = The total number of mobiles within the network Nlinks = The number of links per sector Nsuppl = The number of supplemental links Nfund = The number of fundamental links Phost = The mobile received power from its host or serving sector Pother = The mobile received power from surrounding non-serving sectors PL = Point to point (average) path loss in dB between mobile antenna and cell site antenna Qtotal = The current (steady-state) average power radiated at the J4 port Qmax = The maximum average power allowed at the J4 port before overload (blocking) occurs Qover = The constant overhead power ri = The random position of the ith mobile within the cell R = The cell radius Ri = The channel bit rate of the ith mobile Si = The base station received power from the ith mobile Smin = The minimum receiver sensitivity sij = The distance from the jth surrounding cell to the ith mobile u = Loading factor W = The carrier bandwidth wmax = The maximum mobile power into the mobile antenna xij = Link (traffic channel) power as measured at the J4 port for the jth mobile in the ith sector Xmax = Maximum mobile transmit power (in dBm) out of mobile antenna xi = A sample drawn from a Gaussian (0,8) distribution, thus corresponding to a dB fade drawn from lognormal fading statistics with a 0 dB mean and 8 dB standard deviation Y = A random number defined in Equation 2-19 αi = The channel activity factor for the ith mobile

αij = The channel activity of the jth mobile in the ith sector ak;ij = The attenuation from the kth sector to the jth mobile in the ith sector

β, βreverse = The ratio of other cell interference to serving cell interference for the reverse link βi = The ratio of other cell interference to serving cell interference plus receiver noise floor for the forward link βomni = The ratio of other cell interference to serving cell interference for the forward link and omni antenna configuration ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

v

δi = The fraction of the mobile received host power dedicated to the ith traffic channel ε = r/R

ε d = d i / ηd i

ε g = gi / η g i

ξ = The orthogonality factor λ = The average arrival rate for the general Erlang model ηa = The mean of α ηb = The mean of β σa = The standard deviation of α σb = The standard deviation of β γ = The fixed fraction of the maximum average power dedicated to the overhead channels χ = s/R

µ = The average server completion (of service) rate for the general Erlang model

...........................................................................................................................................................................................................................................................

vi

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

1

Discussion of CDMA 3G-1X RF engineering

Overview ............................................................................................................................................................................................................................................................

Objectives

Contents

This chapter provides a brief overview of Part I of this document, which consists of Chapters 1 through 4. Introduction

1-2

Capacity and coverage for voice applications

1-5

Spectrum requirements Link budget Voice capacity RF engineering for data

1-5 1-5 1-6 1-7

Introduction Overview of traffic theory Data link budget Resource management Deployment

1-7 1-8 1-9 1-10 1-10

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

1 - 1

Discussion of CDMA 3G-1X RF engineering

Introduction

Introduction ............................................................................................................................................................................................................................................................

The ITU-2000 recommendation calls for third generation wireless communication systems with a number of features. These include enhanced voice capacity as well as wireless packet data features, with the latter offering rates of up to 144 kbps for outdoor mobile subscribers. CDMA2000 (also known as CDMA3G) is an IS-95 based standard that satisfies ITU recommendations. This standard allows for a phased implementation of 3G capabilities. CDMA 3G-1X RF Engineering Guidelines summarizes the radio frequency engineering aspects of the Lucent implementation of the first phase, known as CDMA2000-1X, or CDMA3G-1X. This implementation offers enhanced voice capacity as well as wireless packet data at burst speeds of up to 153.6 kbps. Voice and data users can coexist within the same 3G-1X carrier. The Lucent implementation of 3G-1X will support existing IS-95 (2G) services of voice and circuit-switched data as well as 3G-1X voice and packet switched data. The 3G-1X voice will provide improved capacity, expected to be greater by up to a factor of two in terms of supported Erlangs. The 3G-1X packet data service supports access to the Internet via the IP protocol. The 3G-1X and IOS (Inter-Operability Specification) Packet Data services feature(s) provides a subscriber the ability to transmit and receive data with raw rates of up to 153.6 kbps over a packet data network via the 3G-1X IS-2000 air interface. The 3G-1X Packet Data feature(s) enable mobile users with laptop computers or other data devices conforming to the IS-2000 and IS-707A1 standards to access various data applications, such as Internet access, Intranet access, Database access, e-mail, and file transfer at higher speed. The 3G-1X physical layer incorporates a number of major enhancements that provide for higher data rates and better spectral efficiencies compared to second generation CDMA systems. A burstmode capability is defined to allow better interference management and capacity utilization. An active high-speed packet data mobile always has a traffic channel using a Fundamental Code. This channel is called the Fundamental Channel (FCH). An active Packet Data call with the need for higher bandwidth, either in the forward or reverse direction, ...........................................................................................................................................................................................................................................................

1 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Discussion of CDMA 3G-1X RF engineering

Introduction

could be allocated an additional channel for the duration of a data burst, whose duration can be up to a few seconds. The additional channel during this state is called the Supplemental Channel (SCH). A wide range of data rates (raw data rates of 9.6 to 153.6 kbps) is supported over each SCH. One SCH is assigned per data service. An SCH with a data rate of 19.2 kbps or higher is equivalent to multiple voice calls from the consideration of air interface capacity. The assignment of the SCH, along with its data rate, is controlled by the infrastructure based on system load and interference conditions. Static allocation of multiple codes to a small number of users can result in inefficient use of CDMA air interface capacity. Dynamic infrastructure-controlled burst allocation makes it possible to efficiently share the bandwidth among several high-speed packet data mobiles. Efficient algorithms to support dynamic burst allocation have been developed by Lucent. The burst allocation scheme is designed to maximize utilization of CDMA channel bandwidth and system resources. As has been determined during the extensive design process for Lucent Technologies’ HSPD (High Speed Packet Data) Service, the potential risks and issues that arise in designing the packet data service (especially risks of voice quality impact) are minimal, and are easily manageable with minimal impact on voice or data capacity. The data rate and duration of the burst (i.e., the supplemental channel) will be dynamically determined by the infrastructure, depending on load, interference, and resource availability conditions. Therefore, the supplemental channel does not offer any guaranteed bit rate. However, the data rate offered by the fundamental channel with raw data rate of 9.6 kbps is always guaranteed to the 3G-1X data user. For the forward direction, the burst allocation is triggered when data gets backlogged in the network side of the system. For the reverse direction, data builds up at the mobile, which in turn sends a supplemental channel request message to the system, triggering the burst allocation procedure. The new service can be asymmetric, i.e., the high speed packet data mobile, at any given instant, may be assigned different bandwidths on the forward and reverse links. This helps to maximize the efficient use of bandwidth in both directions, still meeting the bandwidth demand of the end-user in each direction. The 3G-1X CDMA HSPD product is built on the 2G/3G CDMA Low Speed Packet Data (LSPD) software since the operation of the fundamental channel and packet data call setup and tear-down procedures are almost identical to the LSPD service when there is no data burst in progress. To end users, the most visible advantage of HSPD over LSPD releases is speed. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

1 - 3

Capacity and coverage for voice applications

Discussion of CDMA 3G-1X RF engineering

Capacity and coverage for voice applications ............................................................................................................................................................................................................................................................

Spectrum requirements

Spectrum requirements for 3G-1X are modest and identical to those for an IS-95 carrier. 3G-1X requires a 1.23 MHz carrier in the cellular band or a 1.25 MHz carrier in the PCS band, with a recommended guard band of 270 kHz between the CDMA and AMPS carriers in the cellular band, and a guard band of 625 kHz (~ ½ carrier) on either side of the PCS block. The guard band recommended is typical, and may be relaxed or expanded depending upon the specific wireless applications in contiguous spectrum. In most cases, it is anticipated that the 270 kHz for the cellular band or 625 kHz for the PCS band should be sufficient. As in IS-95 engineering, no guard band is required between contiguous 3G-1X carriers. Additionally, no guard band is required between an IS95 carrier and an adjacent 3G-1X carrier. 3G-1X and IS-95 subscribers may, in fact, share the same carrier frequency with concomitant effects on each technology's capacity. This strategy is discussed further in Chapter 4, "System deployment".

Link budget

3G-1X voice coverage is essentially determined via link budget analysis, which follows a strategy comparable to that pursued in IS-95 applications. The cell footprint is first sized using the reverse link, which properly takes into account the impact of limited mobile transmit power. Forward link budget analysis focuses on ensuring that sufficient forward power is available to support operations within the footprint dictated by the reverse link. The link budgets used for voice coverage follow a format similar to that for IS-95; however, key parameters differ in value and meaning. For example, the receiver Eb/Nt requirement used to determine cell site receiver sensitivity is based on the total mobile transmit power, rather than the fraction of mobile power dedicated to the traffic channel (unlike IS-95, the uplink consists of both a traffic channel and a pilot channel). In addition, a more aggressive loading with respect to the pole point is allowed due to the inherently greater number of users within a single carrier. These topics are discussed in greater detail in Chapter 2, "Voice coverage, capacity and link budget", which derives both forward and reverse link budgets. A comparison is also drawn between 3G-1X and IS-95 coverage. The slight improvement offered by 3G coverage is key to a 2G to 3G (i.e., IS-95 to 3G-1X) migration strategy, as discussed in Chapter 4, "System deployment".

...........................................................................................................................................................................................................................................................

1 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Capacity and coverage for voice applications Voice capacity

Discussion of CDMA 3G-1X RF engineering

The analysis of 3G-1X voice capacity is also similar to that of IS-95, albeit with different values of key parameters. In particular, relaxed Eb/ Nt requirements on both links drive Erlang capacity/Hz up to twice the value available for IS-95, i.e., up to 26.4 Erlangs per 1.23 MHz carrier for an 8 kbps vocoder. The improved Eb/Nt requirements derive from a number of air interface features, such as enhanced convolutional coding, faster power control, and a reverse link pilot channel that provides a reference signal to aid in signal demodulation. The analysis of 3G-1X capacity is coupled to that of 3G-1X coverage, since both are ultimately driven by Eb/Nt requirements on each link. This analysis is presented in some detail in Chapter 2, "Voice coverage, capacity and link budget".

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

1 - 5

Discussion of CDMA 3G-1X RF engineering

RF engineering for data

RF engineering for data ............................................................................................................................................................................................................................................................

Introduction

Unlike voice applications, the analysis of RF engineering issues for wireless packet data has no ready analogue in IS-95. This discussion therefore occupies a major portion of these guidelines. Key differences include the use of packet-switched rather than circuit-switched data principles, subscriber time-sharing of the same data channel, and new measures of capacity that vary widely with subscriber usage statistics. From a simple overall perspective, a collection of 3G-1X data users within the cell footprint are subscribers engaging in data sessions (e.g., web-browsing) that are inherently bursty in nature. Each user maintains a constant low-rate data connection (fundamental channel) to the cell in order to maintain the call, provide infrequent signaling frames, and occasionally aid in data transmission. For example, in an 8 kbps system, the fundamental channel operates at 1/8 rate on a 9.6 kbps channel, powering to full-rate when signaling information is present. In addition, each subscriber intermittently transmits bursts of data at a much higher rate. This rate is negotiated for each burst between the mobile and base station in a process that takes into account a number of factors including the current interference background, the mobile’s RF conditions, the amount of data that needs to be sent, and the history of the data session (i.e., when the user was last served). These bursts take place over supplemental channels that are set up and torn down as necessary, with raw data rates ranging up to 153.6 kbps. Since the system can simultaneously support only a limited number of supplemental channels due to the higher data rate, this dynamic process of allocating and removing supplemental channels to each user can be viewed as time-sharing a small number of high-speed data pipes amongst the users. In this model, the user transmissions “queue up” for service until one of the high-speed pipes is available. Since the traffic is bursty in nature (i.e., user need for the supplemental channels is brief and not simultaneous across users), the time-sharing of resources is not readily apparent to the end user. For example, wait time in the queue is modest. In this sense, the air interface is packet-switched rather than circuitswitched, since channels are time-shared throughout the user session (packet-switched) rather than completely dedicated to a user (circuitswitched) for this time. Accordingly, performance criteria distinct from those employed in voice networks (circuit-switched) must be used.

...........................................................................................................................................................................................................................................................

1 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Discussion of CDMA 3G-1X RF engineering

RF engineering for data

These include data throughput, average wait time, probability of being delayed, and average length of the queue. The performance may also vary considerably with user statistics, which are necessarily a function of the data applications employed (e.g., e-mail, web-browsing, etc.) and of user behavior (e.g., ‘think’ or idle time between the download of each web page). Accordingly, performance predictions obtained by employing user statistics that are significantly different from those observed in commercial systems may not match the commercial performance. Overview of traffic theory

IS-95 has typically employed a circuit-switched analysis of traffic, since this body of theory is based on a dedicated resource (channel) per user. The resource is held exclusively by the user for the duration of service (i.e., for the duration of the call) and released upon call completion. The performance of this system approximates that of an Erlang B model, which dictates the probability of blocking for a traffic load incident upon a fixed number of servers. The probability of blocking represents the probability that a user will be turned away because all channels are occupied. Although IS-95 principles deviate in some important ways from Erlang B assumptions, the use of circuit-switched principles is correct in that each user occupies a channel resource that is dedicated to its application for the duration of the user session. In contrast, the packet-switched data feature of 3G-1X is not as readily captured by Erlang B principles, since subscriber transmissions (data messages waiting to be burst) can wait or queue up for service rather than be blocked when all resources are busy. In packet-switched data, high-speed data users are serviced by a small number of supplemental channels capable of supporting a high data rate. These channels are time-shared by a fairly large number of data users that transmit bursts of data in turn when cued to do so by the network. This situation is better (although still not precisely) described by an Erlang C model, which relates the probability of delay and average wait time for an incident traffic load funneled through an infinite queue to a fixed number of servers. Since the queue is infinite, no blocking can occur; however, arrivals wait in the queue for service when all channels are busy. In this model, the supplemental channels are viewed as the fixed number of servers. The arrivals are message bursts that are either immediately transmitted (if a channel is idle), or wait in memory at the mobile (reverse link) or cell site (forward link) for their chance at transmission.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

1 - 7

Discussion of CDMA 3G-1X RF engineering

RF engineering for data

The Erlang C model can be readily applied to estimate such performance parameters such as wait time and throughput provided that the number and data transmission rate of the servers are known. For the 3G-1X air interface, these values must be determined from RF analysis. This determination is complicated by the fact that the Erlang C model requires a fixed number of servers each with a fixed data rate; however, the number and rate of supplemental channels within the air interface vary dynamically with factors such as user number, speed, multipath, fade, and transmission history. Accordingly, numerical analysis must be employed to obtain probability distributions of the number and type of supplemental channels available within the cell footprint. This information is then employed to drive an Erlang C model in a manner that reflects the varying, statistical nature of the servers. The process described is computationally intensive, and must be repeated for every design scenario where key input aspects such as performance requirements (e.g., average delay, minimal data rate supported at cell edge) are changed. Some baseline results (see Chapter 3, "RF engineering for data") have been established for a Lucent traffic model, and may be used in planning in the absence of more specific information regarding subscriber behavior and performance requirements. If baseline results are employed, design scenarios can be addressed by using link budget analysis to verify that the air interface can support the total number of fundamental and average number of supplemental channels required within the cell footprint. Data link budget

The data link budget serves two primary purposes. First, the analysis dictates coverage by establishing a minimum data rate available at the cell edge. Second, the analysis verifies that the system has sufficient power to support the mix of fundamental and supplemental channels that are required within this design footprint in order to achieve performance (e.g., data throughput). The reverse link budget for data applications is relatively straightforward in that only the coverage of the supplemental channel need be considered to establish a footprint. This strategy follows from the fact that the high data rate of the supplemental channel renders its coverage the limiting factor. The necessary coverage requirements are typically expressed by requiring a minimum data rate at the cell edge with a specified level of probability (e.g., 90%). For a high data rate, the coverage is naturally limited, and is usually less than that of the lower rate 3G-1X voice or fundamental channel. In these instances, service providers may choose to locally or globally augment cell count

...........................................................................................................................................................................................................................................................

1 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Discussion of CDMA 3G-1X RF engineering

RF engineering for data

in order to achieve a ubiquitous coverage for high-rate users, or may allow the network to naturally restrict the higher data rates to users within the interior of the cell. The forward link budget analysis is more complex, in the sense that forward power must be appropriately shared between fundamental (voice and low-speed data) and supplemental (high-speed data) channels in order to provide coverage within the footprint. In addition, the forward link supplemental channel does not enter soft handoff at the boundary. This design strategy limits forward link interference by ensuring that only one high-rate burst is simultaneously active to the mobile. Supplemental channel performance at the cell edge is enhanced by anchor transfer (essentially a fast hard handoff to the best serving cell), which exploits the fact that the supplemental channel is bursty rather than continuous in nature. The anchor transfer allows the mobile to be served by the best cell for the burst duration. Resource management

The complexity of capacity analysis is a natural consequence of resource allocation or resource management across data subscribers. This strategy dictates the optimal use of available RF resources such as power and data rate in light of the demands being made upon the network. For example, each subscriber’s data rate can be adjusted during the course of a call, and is a function of the subscriber’s reported RF condition (e.g., interference, fading, multipath) as well as the amount of data waiting (queued) for transmission. Although resource management might not be properly regarded as an RF engineering issue per se, the subject is so fundamental to overall performance that it is discussed in detail. The throttling down of the rate of a high-speed data call as it moves from the interior to the exterior of the cell, or as it moves from benign RF conditions to poor RF conditions, is a straightforward consequence of resource management.

Deployment

Deployment of a 3G-1X system entails considerations such as carrier spectrum assignment, overlay ratios, and 3G-1X channel element provisioning. These issues are discussed in detail in Chapter 4, "System deployment". 3G-1X may be deployed in a separate wideband carrier or within an existing IS-95 carrier. The latter may be preferable in areas where spectrum resources are constrained or a gentle migration from IS-95 to 3G-1X is desired; however, the former will result in somewhat greater capacity per Hz within the 3G-1X carrier. A dedicated 3G-1X carrier is

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

1 - 9

Discussion of CDMA 3G-1X RF engineering

RF engineering for data

engineered simply by restricting carrier access to 3G mobiles only. Restricted access is achieved via messages that can be read by 3G mobiles only; i.e., the 3G carrier is ‘invisible’ to 2G mobiles. The overlay ratio for upgrade of an existing IS-95 system to 3G-1X is recommended to be at least 1:1 (i.e., one 3G-1X cell for every existing IS-95 cell) since the 3G-1X voice coverage is slightly better than the IS-95 voice coverage. The improvement is not enough to recommend an overlay consisting of fewer 3G-1X cells than IS-95 cells, such as 1:1.5. Overlays that exceed 1:1 (e.g., such as two 3G-1X cells for every IS-95 cell, or 2:1) are not generally recommended unless the service provider desires to obtain a high-speed data coverage that entirely matches the underlying (low-speed) voice coverage. A 1:1 overlay will supply a low-rate data channel across the entire voice coverage area, while confining higher-rate users to the interior of the cell. Channel element provisioning, i.e., the determination of the number of channel elements required at the cell site to support a traffic load that can consist of 3G-1X voice users, 3G-1X data users, and IS-95 voice users, is not straightforward, but facilitated by the fact that the dualmode 3G-1X channel element can support both 3G and IS-95 (2G) calls. This feature reduces the problem difficulty somewhat, as the exact proportion of 2G and 3G users need not be known in order to produce a channel element number that is operationally sufficient.

...........................................................................................................................................................................................................................................................

1 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

2

Voice coverage, capacity and link budget

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter describes the essential coverage and capacity issues for voice applications. Introduction

2-2

Analysis

2-4

Reverse link Solution--Exact Solution--Approximate Link budget Forward link Solution--Exact Solution--Approximate

2-4 2-6 2-9 2-14 2-20 2-25 2-29

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 1

Voice coverage, capacity and link budget

Introduction

Introduction ............................................................................................................................................................................................................................................................

The 3G-1X principles of voice coverage and capacity are similar to those of IS-95. This similarity is to be expected, as 3G-1X is a spread spectrum system based upon IS-95. In the following sections, we briefly review these principles. A more detailed discussion can be found in Lucent documents 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines, and 401-703-201, PCS CDMA RF Engineering Guidelines, as well as TIA/EIA/IS-2000A standards. In 3G-1X, all users share the same wideband carrier; i.e., the frequency reuse is 1. Transmissions within this channel are distinguished by coding. This approach stands in contrast to other approaches such as frequency division multiple access (each user occupies a distinct narrowband channel) or time division multiple access (each user occupies a distinct time slot). The simultaneous use of the same wideband carrier means that all users interfere with one another. On both forward and reverse links, this interference is tolerated but mitigated through means such as processing gain, fast power control, variable-rate coding, and soft handoff. Interference from other users is suppressed by the processing gain (typically about 20 dB), which derives from the manner in which each traffic channel is uniquely coded to allow ready identification. Power control dynamically adjusts each traffic channel power to the minimum required to maintain performance. Variable-rate coding further suppresses the background interference level by powering down the link (i.e., reducing the voice coding rate) whenever the user is not speaking. Finally, soft handoff reduces overall interference levels by allowing the call to be simultaneously supported by multiple base stations, thereby introducing a diversity gain that lowers the net traffic power required per mobile. Soft handoff is also important in mitigating interference to the forward link receiver (mobile) from a nearby base station that is not supporting the call. Once the mobile enters into a soft handoff state with this base station, this cell becomes a source of signal rather than of interference. This effect of soft handoff is important in real-time applications such as voice, but is less significant in data applications where real-time decoding is not as critical since messages received in error are retransmitted. ...........................................................................................................................................................................................................................................................

2 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Introduction

The improvements to the air interface in 3G-1X have improved capacity performance to the point where the limiting resource in some cases will be the number of available Walsh codes, as opposed to air interface resources.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 3

Voice coverage, capacity and link budget

Analysis

Analysis ............................................................................................................................................................................................................................................................

In the following sections, we outline the coverage and capacity analysis for both the reverse and forward links. In each case, the governing performance equations are presented and discussed. Although exact solution of the equations via numerical simulation is outlined, the focus of the discussion is directed towards simplifications or approximations that can be used in planning processes such as design. The exact solution is discussed only in order to illustrate the complexity underlying accurate performance predictions, and to aid in understanding some of the simulation results presented throughout the document. The latter include offline simulation values employed as line items in planning approximations such as the link budget, as well as key performance results that are based upon numerical analysis outside the scope of this document. Note that in all cases, warrantable performance predictions must be obtained via a mixture of numerical simulation as well as trial (field) results. Reverse link

The key to reverse link analysis lies in assessing the receiver sensitivity; i.e., the minimum power (usually expressed in dBm) required per receive diversity branch at the cell site receiver input. This input (the J4 port) lies at the end of the cable connecting receiver to antenna; i.e., at the point where the incoming signal has already suffered cable loss. Consider a collection of mobiles within a sector. For the moment, we presume a steady-state condition; i.e., one where all mobile positions are fixed and the mobile conditions of voice activity factor, multipath, and fade are unchanging. At the J4 port, each mobile must satisfy its particular Eb/Nt (ratio of channel bit energy to spectral density of total channel impairment) requirement, which is a function of mobile speed, multipath, and required channel Frame Erasure Rate (FER). For all mobiles within the sector:  Eb   Nt

  ≥ d i i

Equation 2-1: Eb/Nt requirement ...........................................................................................................................................................................................................................................................

2 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

In this equation, the letter i is the index of the mobile in question. The left-hand side is the achieved Eb/Nt at the cell site receiver (J4 port); the right-hand side is the required median Eb/Nt corresponding to the particular mobile’s condition (speed, multipath) at the design FER (e.g., 1%). For the sake of simplicity, we presume an isolated sector with N mobiles. Expanding the above, we obtain:

 Eb  Nt

α i 

(W / R )α i S i  gα i S i α i S i / Ri  = = = = αidi N N N  i FN + 1 ∑ α S FN oW + ∑ α j S j FN oW + ∑ α j S j o W j =1 j j j =1 j =1 J ≠i

J ≠i

J ≠i

Equation 2-2: Expanded Eb/Nt definition

This expression is the heart of system analysis for reverse link coverage and capacity; accordingly, we consider it in some detail. In the above, the energy per bit (numerator) is determined by the ratio of received power Si to channel bit rate Ri. The spectral density of receiver interference plus receiver noise (denominator) is determined by the sum of receiver noise density (the thermal noise density No scaled by the receiver noise figure F) and the sum of power received from the other N-1 mobiles. In voice applications, the channel bit rate is constant for all users provided that a single vocoder, either 8 or 13 kbps speech, is employed within the mobile population; hence Ri=R. This is not the case in data applications, where the channel bit rate can vary per user. Additionally, a voice network may contain a mixed population of 8 and 13 kbps vocoders. These points are explored later on in this document. W is the channel (carrier) bandwidth. The quantity W/R = g is the spread spectrum processing gain. Equation 2-2 shows that the ratio of signal power to impairment (noise plus interference) power, when multiplied by the processing gain, must equal or exceed the Eb/Nt requirement.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 5

Voice coverage, capacity and link budget

Analysis

The variable α is the mobile voice or channel activity factor with possible values ranging from 0 to 1 in discrete steps of 1/8, ¼, ½, and 1.0.1 The last value applies when the user is speaking; the first applies when the user is listening. Intermediate values are transitional rates inserted to avoid a clipped sound to speech when the channel is changing between the speak/listen states. The probability (relative dwell time) of each value has been determined from analysis of vocoder speech and is known. The statistics of alpha are therefore completely characterized. The di is the median full-rate (i.e., α =1) Eb/Nt requirement. In the above, we have explicitly made the assumption that the Eb/Nt requirement is scaled by the voice activity; e.g., the Eb/Nt requirement for a user in the 1/8 state (listen) is 1/8 of the full-rate Eb/Nt requirement. The Eb/Nt requirement as a function of multipath, speed, and Frame Erasure Rate (FER) is determined via a combination of link level simulations and receiver tests. Equation 2-2 represents a set of linear equations in the variables S1, S2,…SN. These equations express the coupling between mobiles; i.e., the fact that each user’s signal is interference to all other users. Solution--Exact

We presume an ideal power control, which would without error ensure that all mobiles just achieve (rather than exceed) their Eb/Nt requirement. Accordingly, we change the inequality in Equation 2-2 to an equality. This expression can then be expanded to the matrix equation:  g / d1  −α 1   .  .   − α1

− α2 g / d2 . . − α2

− αN − αN . .

 S1  1    S    1 2   .  = FN oW  .     .    .  1 . . g / d N  S N   

. . . .

. . . .

Equation 2-3: Reverse link Eb/Nt matrix

...........................................................................................................................

1

The ½ and ¼ are transitional rates (from speak to listen), and are not always employed.

...........................................................................................................................................................................................................................................................

2 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

Each value of Si can be replaced by aixi, where ai is the total attenuation (loss) from the transmit antenna of the ith mobile to the J4 port and xi is the transmit power out of the ith mobile. Note that the former includes total loss, and therefore could be computed by the dB sum of body (head) loss, building/vehicle loss, (random) fade, point-to-point (distance-dependent) path loss, receiver antenna gain, and receiver cable loss. The latter constitutes the total mobile transmit power, including the 3G-1X pilot signal that accompanies traffic power in order to aid demodulation at the cell site receiver. The Equation 2-3 becomes:

 g / d1  −α 1   .  .   − α1

− α2 g / d2 . . − α2

−αN −αN . .

      . . g / d N 

. . . .

. . . .

a1      

a2 . .

      a N 

 x1   1      x2   1  .  = FN W  .  o     .   . x   1    N

Equation 2-4: Reverse link expanded matrix form

Note that the matrix containing the attenuations (a) is diagonal, with 0’s in all nondiagonal entries. The importance of Equation 2-4 cannot be overemphasized, since it represents the key to analysis of system performance via numerical simulation. In this Monte Carlo process, the performance limits of capacity and coverage are established by computing performance for a range of possible values of sector coverage and capacity. In this process, a sector perimeter (footprint) and number of mobiles N are first selected. A trial is conducted by randomly placing N mobiles within the footprint, and assigning them random values of voice activity, fade, and multipath. The multipath value and 0 velocity (fixed position) dictate the full-rate requirement d for each mobile. The expression Equation 2-4 is then solved for the transmit powers x. This process is repeated over many trials until the statistics of the mobile transmit powers can be determined for the selected perimeter and capacity. One or both of these values (perimeter, capacity) is then altered. The process of determining mobile transmit power distributions by conducting multiple trials is then repeated, thereby characterizing performance for this new selection. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 7

Voice coverage, capacity and link budget

Analysis

This process is repeated for a number of selections (perimeter, capacity). Given a target probability of outage for the mobiles, e.g., no more than 5% of the transmit powers observed can exceed the mobile maximum transmit power, this analysis can determine the best values of coverage and capacity that can be supported. For example, the maximum value of N that can be supported within a given fixed footprint at a 5% outage can be determined by computing the probability distribution of mobile transmit powers for each value of N. At a small value of N, the probability distribution is unlikely to exceed the mobile maximum transmit power xmax at all; at a larger value of N, a significant portion of observed values may be above xmax. The desired value of N is that which yields a probability distribution that displays the value xmax for its 95th percentile (i.e., the 95th percentile of the mobile transmit power distribution can be no greater than the maximum mobile transmit power). Although the analysis outlined by Equation 2-4 has been pursued, the results are generally not applicable to network performance unless the model is expanded in two ways: The incorporation of the impact of moving (non-fixed) mobiles, and the incorporation of the effects of other sectors. For completeness, these are described below. In the above, we have presumed that the mobiles are fixed. This concept lends itself readily to the steady-state assumption, where position, fade, multipath, and voice activity do not change with time. In each trial, the required Eb/Nt, di, for each mobile was obtained solely as a function of the random choice made for multipath since the speed was fixed at 0. The situation for moving mobiles is assessed by using a randomly assigned value of speed as well as multipath to determine the required Eb/Nt, (di) in Equation 2-4. The performance of a system with moving mobiles is thus determined by applying mobile Eb/Nt requirements to an otherwise static situation. This approach, which approximates the more complex situation where the mobile positions are changing from instant to instant, is sometimes referred to as analysis via a series of static snapshots. The analysis embodied in Equation 2-2 and Equation 2-4 considered only an isolated sector; in contrast, an embedded sector, i.e., a sector surrounded by a sea of cells, is clearly a better model of real-world conditions. The effects of other sectors can be included by expanding the denominator of Equation 2-2 to include the interference at the sector receiver from mobiles transmitting in other surrounding sectors. These expressions expand and alter the matrix in Equation 2-4; in ...........................................................................................................................................................................................................................................................

2 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

particular, the size of the matrix increases from N×N to Ntotal × Ntotal, where Ntotal is the number of mobiles in all sectors. The analysis proceeds similarly but with considerably more computational complexity, since for each trial the Ntotal × 1 vector of transmit strengths, representing the transmit strengths of all mobiles within the network, must be solved for. The techniques described have been used to simulate the performance of IS-95 (2G) systems, achieving results that are supported by field data. For example, the capacity of a fully mobile system within a nominal cell footprint (i.e., a footprint dictated by the reverse link budget analysis outlined in "Solution--Approximate" section on Page 2-9 and "Link budget" section on Page 2-14) is the equivalent of 13 channels (7.4 Erlangs at 2% block) and the equivalent of 20 channels (13.2 Erlangs at 2% block) for 13 kbps and 8 kbps coding, respectively. These values apply to the early version of the ASIC receiver chip (1.0), and rise to 9.0 Erlangs and 16.6 Erlangs, respectively, with use of the ASIC 1.1 chip in the cell site receiver. The same techniques have been employed in predicting 3G-1X capacity for nominal (link budget) footprint, indicating 26.4 Erlangs at 2% block (35 channels) for 8 kbps coding. This value is as yet unsupported by extensive field data, since no 3G-1X commercial systems have been deployed. Table 2-1

Air interface

Air interface capacity

IS-95 at 13 kbps

Capacity @ 2% block 7.4 Erlangs

IS-95 at 8 kbps

3G/1X at 8 kbps

13.2 Erlangs

26.4 Erlangs

Solution--Approximate

We now consider means of obtaining solutions to Equation 2-2 that are approximate. Although any final performance prediction should rely upon a mixture of exact solution (see "Solution--Exact" section above) as well as trial results, the approximate solutions are useful for planning as well as lending insight into performance trends. We seek an approximate solution to Equation 2-2, repeated here for convenience.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 9

Voice coverage, capacity and link budget

Analysis

(W / R )α i Si  Eb  gα i S i α i Si / Ri   = = = = α i di N N N  N o i FN + 1 ∑α S FN oW + ∑α j S j FN oW + ∑α j S j o W j =1 j j j =1 j =1 J ≠i

J ≠i

J ≠i

As discussed above, real-world conditions are better modeled by incorporating the effects of other sectors. This can be done by altering the summation term in the denominator appropriately. The interference from other (outside) sectors can be viewed as the outer interference. The interference from mobiles within the sector is the inner interference, represented by the summation term over N-1 users in the denominator of Equation 2-2. Simulations employing the techniques described in the "Solution--Exact" section have shown that the ratio of outer to inner interference can be approximated by a constant β, for an embedded sector in a sea of cells with uniform sector loading. The impact of outer cell interference is therefore captured by altering Equation 2-2 to:  Eb  gS i   = = di N  N t  i FN W + (1 + β )∑ α S o j j j =1 J ≠i

Equation 2-5: Reverse link Eb/Nt with interference ratio

The di represents the per-path median Eb/Nt requirement of the ith mobile, which is dictated by conditions of multipath, speed, and FER. At 1% FER, the range of possible values is not large; moreover, the existence of at least two paths is guaranteed in the presence of twobranch spatial diversity2. The analysis can therefore be considerably simplified by making the conservative assumption that all mobiles achieve the Eb/Nt for the worst-case (maximum) of the 2-path multipath cases (di=dmax). The condition that all mobiles achieve the same d= dmax introduces a symmetry into the above expression that requires all received powers be equal as well; i.e., Si=S:

...........................................................................................................................

2 The two-path existence is also guaranteed in some other diversity schemes such as slant polarized diversity branches; however, these schemes are not applicable at all frequencies.

...........................................................................................................................................................................................................................................................

2 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

gS FN oW + (1 + β ) S ∑ α j N

= d max

j =1 J ≠i

Equation 2-6: Reverse link Eb/Nt assuming worst case required Eb/Nt

This expression is readily solved for the single value of the key parameter S, the required signal strength per diversity branch at the J4 port of the cell site receiver:

S=

FN oW g d max

− (1 + β )∑α j N

j =1 J ≠i

Equation 2-7: Required received reverse link power

The value S is a random variable, since the summation in the denominator is a sum of the independent but identically distributed values of channel (voice) activity. For planning purposes, we seek the expected value of S for use as the minimum receiver sensitivity Smin. This value is conveniently expressed as:

S min

    FN oW = E {S }=    g − (1 + β )( N − 1)η  α  d max 

   g − (1 + β )( N − 1)η  α d  E  max  N  g − (1 + β )∑ α  j  d max  j =1 J ≠i  

Equation 2-8: Reverse link receiver sensitivity

Here, E denotes the expectation operator; also, the ηα represents the expected value of the channel activity. The expectation on the far right of Equation 2-8 can be computed analytically since the distribution of the random value of voice activity is known. This value is close to 1. For large g/dmax, this result can be obtained by inspection; moreover, regardless of the value of g/dmax, the expected value is always 1 for very large N since the sum over (N-1) voice activity values is equal to (N-1) times the mean voice activity. We therefore approximate the mean receiver sensitivity as: ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 11

Voice coverage, capacity and link budget

Analysis

    FN oW E{S }=    g − (1 + β )( N − 1)η  α  d max  Equation 2-9: Reverse link receiver sensitivity - approximation

This expression shows that the receiver sensitivity is a monotonically increasing function of N, the sector loading. Since increased sensitivity clearly requires decreased cell radius, this expression illustrates the fundamental trade-off between coverage and capacity that can be pursued in CDMA systems. In addition, there is clearly a hard limit to the loading N, since the denominator must be greater than zero. The limit Nmax can be obtained by setting the denominator equal to zero, obtaining the reference or “pole” point at which the required receiver sensitivity grows without bound:

N max =

g 1 +1 ηα d max (1 + β )

Equation 2-10: Pole capacity definition

The receiver sensitivity can be recast using Nmax as:

E{S}=

1  FNoW  1 1  FNoW  1   1   dFN R 1 − = =   o  ηα (1+ β )  Nmax − N  ηα (1+ β ) Nmax  1− u  1− u  Nmax  Equation 2-11: Reverse link receiver sensitivity in terms of loading

where u = N/ Nmax is the loading with respect to the pole point. Equation 2-11 can be used to determine the receiver sensitivity for use in a link budget, as discussed below. For:

g d max

>> ηα (1 + β )

Equation 2-12: Receiver sensitivity simplifying assumption ...........................................................................................................................................................................................................................................................

2 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

Equation 2-11 can be simplified to:

E{} S =

dmax  1  FN W g 1−u o

Equation 2-13: Simplified reverse link receiver sensitivity

This approximation allows the receiver sensitivity to be determined by the dB sum of Eb/Nt requirement, data rate, interference margin 1/(1-u) and receiver noise floor. The reverse link budget format (see "Link budget" section below) is based upon this approximation; however, the underlying calculations rely upon Equation 2-11 since the approximation Equation 2-12 may not always be satisfied. Equation 2-11 suggests that any integer value of N less than Nmax (i.e., any value of u less than 1) is permissible provided that one is willing to pay the penalty of reduced cell coverage associated with very high pole loadings (e.g., u=0.95). In practice, loadings approaching u=1 are avoided due to the possibility of associated instabilities. Such instabilities exist regardless of the nature or form of power control, as can be demonstrated by a sensitivity analysis that relates relative changes in loading u to relative changes in required receiver sensitivity. Differentiating Equation 2-11 with respect to u, we obtain:  dS min   S min

  u   =    1 − u 

 du     u 

Equation 2-14: Sensitivity of receiver sensitivity to loading

This expression indicates that relative changes in required receiver sensitivity are related to relative changes in loading u by the sensitivity factor u/(1-u). This factor indicates to what degree relative changes in u are suppressed or amplified into relative changes in Smin. The sensitivity in Equation 2-14 increases with loading, rising from values <<1 (where relative changes are suppressed) to 1 when u reaches u=0.5 (50% loading). Loadings greater than 50% yield sensitivity factors greater than 1, indicating that required relative changes in S are amplified relative to changes in u; in particular, the sensitivity factor is greater than 3 for loadings exceeding u=0.75 and rises rapidly thereafter.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 13

Voice coverage, capacity and link budget

Analysis

Large sensitivities indicate that minor changes in loading can require large changes in Smin. In this region, the finite time and finite accuracy associated with any power control loop can result in large overshoots (instability) as the system tries to make the necessary large adjustments in response to small, fast changes in loading. This effect constrains the maximum loading that can be tolerated. These concepts are illustrated in the curve in Figure 2-1 below.

Sensitivity Factor

mu/(1-mu)

20 15 10 5 0 0

0.2

0.4

0.6

0.8

1

mu

Figure 2-1

The sensitivity factor maps relative changes in loading into relative changes in receiver sensitivity. The factor is a function of the design loading µ. For large values of µ, small relative changes in loading are “amplified” into large relative changes in receiver sensitivity. The choice of design loading factor µ must avoid this region of the curve.

Simulation and field results suggest that the maximum tolerable loading falls within the range of u=0.5 to u=0.75. The allowed loading improves with better power control and with lower dmin (i.e., higher pole point). The latter effect arises since a larger number of users associated with any value of u tends to stabilize that value; i.e., the relative change of u per the addition or deletion of a single user is less. Link budget

The required receiver sensitivity Smin in Equation 2-11 can be used to obtain a reverse link budget. This budget dictates the maximum allowable path loss between mobile transmit antenna and cell site receive antenna. Provided further analysis indicates that the forward ...........................................................................................................................................................................................................................................................

2 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

link can support performance at the same loss (See "Forward link" section on Page 2-20), the loss can be used on a market-by-market basis to perform RF design. This process employs algorithms that map loss into cell radii via consideration of local variables such as tower height, terrain, and clutter. The allowed point-to-point path loss is determined by considering the terms that dictate net loss from mobile to cell. Components of the net loss are indicated in the following figure (head loss and fading are not shown in the figure, but are included in the link budget equations).

CDMA Mobile

Max. Path Mobile EIRP

Building Penetration Loss

Figure 2-2

Antenna Gain

Cable Loss

CDMA Base Station

Receiver Sensitivity

Components of net path loss from mobile to base station

The terms characterizing the net loss are captured in the following relation, which requires that at maximum mobile transmit power the signal power achieved at the J4 port must equal or exceed 10log(Smin): X max - HL - fade - BL/VL - PL + AG - CL ≥ 10 log(S min ) Equation 2-15: Reverse link budget equation

where: Xmax= Maximum mobile EIRP (Effective Isotropic Radiated Power) (in dBm) HL = dB head (body) loss Fade = dB fade at mobile location BL/VL = dB building or dB vehicle penetration loss, whichever is applicable PL = dB point to point (average) path loss between mobile antenna and cell site antenna AG = dBi cell site antenna gain CL = dB cell site cable loss.

This expression is readily rewritten for the allowed maximum dB path loss. This value dictates the edge (boundary) of the cell coverage. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 15

Voice coverage, capacity and link budget

Analysis

PL ≤ X max - HL - fade - BL/VL + AG - CL - 10log(S min ) = PLmax Equation 2-16: Reverse link budget equation

Equation 2-16 can be viewed as constructing the allowed maximum path loss as a dB sum of credits (e.g., mobile transmit power) and deficits (e.g., cable loss). This dB process is captured in the reverse link voice budget. Several examples are shown in Table 2-2 below. The link budgets serve as examples only and will vary from market to market per the service provider’s requirements. For instance, the cell site antenna gain, cable loss, fade margin, and building penetration margin could be modified, substantially altering the (bottom line) allowed path loss to be used in design. Table 2-2 shows PCS link budgets for second generation (2G) voice coded at 13 kbps (total rate with overhead bits is 14,400) and at 8 kbps (total rate with overhead bits is 9600). These are included for reference. The 3G-1X budget for 8 kbps is shown in the right-hand column. The 2G budgets are created from parameters (e.g., noise figure) applicable to the IS-95 Minicell and the ASIC 1.0 chip. The 3G-1X budget uses parameters appropriate to the Flexent® Modular Cell. In all cases, the format of the link budget is essentially obtained from Equation 2-16, with Equation 2-13 used to create the value of Smin. As discussed above, the approximation Equation 2-13 is not always valid; hence, in spite of the format the spreadsheet uses an embedded form of Equation 2-11 to obtain the Smin.

...........................................................................................................................................................................................................................................................

2 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis Table 2-2

Reverse PCS link budget for IS-95 9.6 kbps, IS-95 14.4 kbps, and 3G-1X 9.6 kbps voice, mobility application

Item

Units

2G Voice 14.4kbps

2G Voice 9.6kbps

3G-1X Voice 9.6kbps

21

21

21

Comments

(a) Maximum Transmitted power per channel

dBm

(b) Transmit Cable, connector, combiner, and body losses

dB

2

2

2

(c) Transmitter Antenna Gain

dBi

2

2

2

(d) Transmitter EIRP per channel (a - b + c)

dBm

21

21

21

(e) Receiver Antenna Gain

dBi

18

18

18

(f) Receiver Cable and Connector Losses

dB

3

3

3

(g) Receiver Noise Figure

dB

5

5

4

(h) Receiver Noise Density

dBm/Hz

-174

-174

-174

(i) Receiver Interference Margin

dB

3.4

3.6

5.5

(j) Total Effective Noise plus Interference Density = (g + h + i)

dBm/Hz

-165.6

-165.4

-164.5

(k1) Information Rate (10log(Rb))

dB

41.6

39.8

39.8

(l1) Required Eb/Nt

dB

7

7

4

(m) Receiver sensitivity (j + k +l)

dBm

-117.2

-118.7

-120.7

(n) Hand-off Gain

dB

4

4

4

For 90% cell edge coverage and 8 dB log-normal standard deviation

(o) Explicit Diversity Gain

dB

0

0

0

Diversity gain has been included in required Eb/ Nt

(p) Log-normal Fade Margin

dB

10.3

10.3

10.3

(p') Building/Vehicle Penetration Loss

dB

0.0

0.0

0.0

(q) Maximum Path loss {d-m+ef+o+n-p-p'}

dB

146.9

148.4

150.4

Body loss

PCS Minicell for 2G and Modcell for 3G-1X

72% loading for 3G-1X

Considering 2 spatial receive diversity branches

For 90% edge coverage with 8dB log-normal standard deviation For outdoor coverage

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 17

Voice coverage, capacity and link budget

Analysis

Note that the link budget creates the fade margin in Equation 2-16 as a sum of two terms: The single-link (simplex) fade margin and the soft handoff gain. The simplex fade margin is obtained conventionally by selecting a dB value from a normal distribution of possible values. For a simplex connection, the path loss at the cell edge therefore accommodates all values of fade up to and including this value. For example, a selection of 10.3 dB means that at cell edge all fades up to and including 10.3 dB can be tolerated without requiring that the mobile exceed its maximum transmit power. Since the 10.3 dB is the 90th percentile within the distribution of fades3, this choice corresponds to a 90% probability of cell edge coverage. The probability of area coverage is greater, since inside the cell boundary the path loss is less and the mobile has more transmit margin to overcome deeper fades. The fade margin required for 90% edge coverage is actually less than the simplex value, since a CDMA mobile at the cell edge is in a soft handoff state with at least two legs. The full simplex margin would only be required if both legs faded simultaneously and equally. Since the leg-to-leg fading is at least partly uncorrelated, the net fade margin required to achieve a given probability of coverage is less. The soft handoff gain is the difference between the simplex and actual fade margin. The exact value is a weak function of probability of edge coverage and is determined by offline calculations that are supported by field data. Recommended values are tabulated, below. These values correspond to a 60% correlation between soft handoff legs in a lognormal fading environment (8 dB standard deviation). Table 2-3

Reverse link soft handoff gains

Probability of Edge Coverage

Soft Handoff Gain (dB)

75%

3

90%

4

...........................................................................................................................

3 This is true for lognormally distributed fades with 8 dB standard deviation, this distribution is common and often observed in path loss measurements

...........................................................................................................................................................................................................................................................

2 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

The differences between the 3G-1X and the IS-95 voice reverse link budgets must be emphasized. They include the following: •

For 2G, the mobile transmitted power consists solely of traffic channel power; however, for 3G-1X, the mobile transmitted the power includes the traffic channel and reverse pilot channel power. The analysis described in the "Solution--Exact" and "Solution--Approximate" sections applies in either case since the Eb/Nt requirements di are adjusted appropriately. It is simply a matter of interpretation of the transmit power x.

•

The required Eb/Nt d (i.e., the traffic channel Eb/Nt requirements for the 2G, and the total traffic plus pilot Eb/Nt requirement for the 3G-1X) to achieve 1% target Frame Erasure Rate (FER) differ. For the 9.6 kbps voice, mobility application and 1% FER target, the requirement for the 3G-1X is 4 dB, less than the 7 dB required for the IS-95 system (ASIC 1.0 chip).

•

The pole loading factor for 3G-1X is higher than the pole loading factor for IS-95, due to a larger user base and slightly improved power control (see the "Solution--Approximate" section). This difference is reflected within the interference margin. The example budgets employ the maximum loading recommended for the scenarios chosen. Lower loadings are allowed, increasing coverage at the expense of reducing capacity.

•

The air interface capacity of the 3G-1X 8kbps voice application is 26.4 Erlangs per sector per carrier (corresponding to 35 channels at 2% blocking) while that of the IS-95 8kbps voice is 13.2 Erlangs per sector per carrier (corresponding to 20 channels at 2% blocking). This difference arises due to the 3G-1X reduced Eb/Nt requirement, as well as the increased 3G-1X maximum pole loading.

•

The base receiver noise floor of the PCS CDMA Minicell is 5 dB while that of the PCS CDMA Flexent Modular Cell is 4 dB. The former has been extensively deployed within the field, and was therefore used as a 2G reference in Table 2-2.

The examples above indicate that 3G-1X can tolerate more path loss than IS-95 under identical (“normalized”) conditions; i.e., equal values of antenna gain, fade margin, building penetration loss, etc. This difference allows an IS-95 system to be upgraded to 3G-1X on a 1:1 basis without loss of coverage performance. Overlay strategies are discussed in more detail in Chapter 4, "System deployment".

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 19

Voice coverage, capacity and link budget

Analysis

Forward link

Reverse link analysis is used to establish the cell footprint. This analysis can be viewed as driven by the limit on mobile transmit power. This limit is a key constraining factor in cell size, driven by market demands for more compact subscriber units and longer battery life. The objective of forward link analysis is to ensure that the forward link has sufficient power to support performance within the footprint dictated by the reverse link. Accordingly, the dB design path loss determined by reverse link analysis is an input to the forward link analysis process, which assesses whether the forward link has sufficient resources to deliver adequate power to each mobile receiver within the design path loss. This analysis differs in three important ways from that of the reverse link: •

First, the link transmitter power (forward link amplifier power) considered in analysis is shared amongst multiple users. In contrast, the transmit power employed in reverse link analysis (the mobile transmit power) is dedicated to a single subscriber.

•

Second, the effect of other sectors at the receiver is more important, as a mobile receiver near the cell boundary can be subjected to a significant amount of interference broadcast by nearby neighbor sectors. In contrast, the other-cell interference considered in reverse link analysis consisted of power from modest transmitters at a greater distance from the cell site receiver.

•

Third, the available link level information does not consist (directly) of receiver Eb/Nt requirements; rather, the fractional forward link power (“Ec/Ior”) as a function of mobile geometry is used in analysis. The ‘geometry’ is defined as the ratio of the total power within the active set to the sum of receiver noise and total power received from all sectors not within the active set. A sector is in the mobile’s active set when it is supporting the mobile call; i.e., providing a signal or leg that the mobile is demodulating.

In order to ensure clarity, we provide a few examples of the last point. The fractional forward link power requirement Ec/Ior (or x = Ec/Ior, used here for convenience) is a pure (dimensionless) number and a function of the mobile geometry G:

...........................................................................................................................................................................................................................................................

2 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

x = Ec / Ior; x < 1 x = f (G) Equation 2-17

For example, x may be 0.05, indicating that 5% of the total forward link power broadcast by a sector is required to maintain forward link FER. Note that this relationship says nothing about the total power broadcast by the sector, but simply indicates what fraction of the power being broadcast is required by the mobile in question. The geometry must be defined with care. For a mobile not in soft handoff, the numerator of the geometry consists only of the power received from its host sector. For a mobile in soft handoff, the numerator consists of the power received from the host as well as all other sectors supporting the call. In each case, the denominator consists of the sum of receiver noise and the received power from all other sectors not supporting the call. As a specific example, we consider the following. Without loss of generality, we may denote the received host sector power P1 for a mobile not in handoff. For a mobile in soft handoff with sectors 1 and 2, we denote the received host sectors’ power P1 and P2. Then: Let

I i = Pi / W ; or received sec tor power density I1

G= FN t +

all sec tors

∑I

; mobile not in soft handoff ; host sec tor is 1

i

i=2

G=

I1 + I 2 FN t +

all sec tors

∑I

; mobile in soft handoff with sec tors 1 and 2

i

i =3

Equation 2-18: Geometry calculation example

Note the contrast between the first (no soft handoff) and last (soft handoff) definition of geometry. In the former case, only sector 1 supports the call; accordingly, only the received power from sector 1 is in the numerator. In the latter case, both sectors 1 and 2 support the call. In this case, the received power from sector 2 is removed from the sum in the denominator and placed within the numerator.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 21

Voice coverage, capacity and link budget

Analysis

In all cases, the I represents the received spectral density from all multipath reaching the mobile receiver from the sector in question. For clarity, some sample mappings of x = Ec/Ior vs. geometry G are shown below. The mapping is from an early study examining the impact of power control, which, as expected, improves the link performance by lowering the x required. The study per se will not be discussed further here; the chart is used only to demonstrate the general shape of the curve x = f(G).

Figure 2-3

Required fractional power versus geometry example

Note that, in general, the required x shrinks as the number of multipath from the sector to the mobile increase. We now consider the system-level analysis of forward link employing this information. Consider a collection of mobiles within a sector. We again presume a steady-state condition; i.e., one where all mobile positions are fixed and the mobile conditions of voice activity factor, multipath, and fade are unchanging. The sector is embedded, surrounded by a sea of other sectors containing mobiles. At the sector J4 port, the fractional transmit power allocated to each mobile must be sufficient to reach or exceed the mobile receiver’s Ec/Ior requirement, which is dependent upon its speed and geometry. The geometry is dependent upon the host and other sector powers received at the mobile; these, in turn, are dependent upon the mobile ........................................................................................................................................................................................................................................................... 2 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

position and fading state. In order to support N links (including primary, soft, and softer handoff links), the total fractional transmit power required must not exceed the fraction of amplifier power that is available for traffic:

Y = ∑α j x j ≤ N

j =1

Qmax − Oover Qmax

Equation 2-19: Sum of powers less than available traffic power

In the above, αj represents the channel activity of the jth forward link, and xj represents the fractional power required to support this link. The values Qmax and Qover are the maximum average available power and the average power assigned to overhead channels (e.g., pilot), respectively. This inequality must be satisfied for each sector within the system. For clarity, we write the value xj more completely to show its functional dependence: x j = x j ( speed , multipath, geometry[ all sec tor powers , fading , location, handoff state]) Equation 2-20: Required fractional power dependencies

The fractional transmit power requirement is a function of speed, multipath, and mobile geometry. Geometry, in turn, is a function of mobile location, powers broadcast by all surrounding sectors, fades between the mobile and all sectors, and the mobile handoff state. Given the randomness of location, speed, multipath, and fading, it is clear that for fixed N, the sum Y is a random variable with an associated probability distribution. As N varies, the distribution retains (approximately) its shape but shifts to the right or left; see Figure 2-4. For a given coverage footprint and given number of links, i.e., given number of users, the computation of the associated probability distribution provides the probability that the sum Y satisfies the inequality (Equation 2-19). In particular, for a given footprint, the forward link capacity limit can be obtained by finding the highest value of N such that the sum Y still satisfies the inequality (Equation 2-19) with acceptable probability. This probability should be high, e.g., 90%, 95%, in order to be consistent with the high probability of coverage generally provided by the reverse link.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 23

Voice coverage, capacity and link budget

Analysis

Increasing links

N links

N +1 links

N +2 links

Nmax links

probability

..... ....

Max allowed

Total fractional power used Figure 2-4

Various probability distributions as a function of N (number of users)

For fixed footprint, as N increases, the probability distribution associated with the sum Y (total fractional transmit power) retains its shape (approximately) but shifts to the right. A maximum amount of fractional transmit power (“max allowed”) is available for traffic. The maximum capacity Nmax can be found by locating the highest value of N for which the probability distribution still has an acceptably small probability of violating the maximum allowed fractional transmit power.

In order to conduct the above analysis, Equation 2-19 must be solved for all sectors. This analysis is not straightforward; in particular, the method of solution is not analogous to that employed in the reverse link (linear algebra). The additional complexity arises from several factors, including: •

The nonlinear mapping x = f(G) in Equation 2-17, which can be tabulated but not readily expressed in analytical form

•

The dependence of fractional transmit powers x on a number of factors (Equation 2-20), including geometry

•

The fact that the computation of geometry for a given location depends upon knowledge of the radiated power from all sectors…but the radiated power from all sectors cannot be computed unless the fractional transmit powers x are known. These, in turn, depend upon geometry. This circularity prevents a straightforward solution; rather, an iterative approach that eventually results in an answer with self-consistent sector powers, fractional transmit powers, and geometries must be employed.

...........................................................................................................................................................................................................................................................

2 - 24

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

This complexity has resulted in a number of forward link analysis techniques, which vary depending upon the speed, accuracy, and extent of information desired. We describe two examples below. Solution--Exact

The exact solution of Equation 2-19 is complex, but nevertheless employed in system performance simulations. We outline the method of solution here. The analysis can be done by collecting data from a number of simulated snapshots, where within each snapshot a predetermined number Ntotal of mobiles are randomly distributed throughout the sectors comprising the network. The snapshot is static in the sense that the mobiles do not move; however, motion can be modeled in a limited sense by randomly selecting a velocity for each mobile, and using the geometry curves for that velocity to assign fractional transmit powers to the (stationary) mobile. Given a sufficiently large number of snapshots, the probability distribution for the sum Y (Equation 2-19) can be obtained. This curve then allows specification of the probability that the total fractional transmit power remains below an allowable level. As discussed above, the geometry associated with a given location depends upon the total powers broadcast by the sectors. These total powers cannot be determined unless the fractional transmit powers x are known; however, these cannot be specified without knowledge of the geometry. This interrelationship dictates an iterative approach to the problem, which can be generally pursued as follows: •

Construct a nominal (i.e., hexagonal) arrangement of cells, with per-sector footprint dictated by an allowable path loss and a path loss law (e.g., Hata model). The former is usually dictated by the reverse link budget

•

Choose a value of Ntotal users within the network, which corresponds to a desired average value of users per sector.

•

Choose an initial value (e.g., Qmax) of power broadcast by all sectors

•

Create and solve a single static snapshot via the following steps: 1.

Randomly place the Ntotal users within the network

2.

For each user, randomly select a multipath and velocity value

3.

Specify a value of Qmax power broadcast by all sectors

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 25

Voice coverage, capacity and link budget

Analysis

4.

Using these values, compute each user’s soft handoff state

5.

Compute each user’s geometry

6.

Using the appropriate curve for each user’s multipath and velocity, map user geometry into required fractional transmit power from its host sector

7.

Compute actual user (link) power from host sector

8.

Compute total power broadcast by all sectors, and compare with values assumed in Step 3, above. If they do not match, reassign sector broadcast powers in Step 3 to the new values computed in this Step 8.

9.

Repeat Steps 3 through 8 until convergence; i.e., until the assigned sector powers in Step 3 match the computed sector powers in Step 8.

10. Store the sector power for this value of Ntotal as a single point within the probability distribution of required power (see Figure 2-4). This point corresponds to a single static snapshot for Ntotal users. 11. Repeat Steps 1 through 10 (i.e., run additional static snapshots) until a sufficient number of points is obtained to characterize the probability distribution of required power for Ntotal users (see Figure 2-4) •

Assess this distribution to ascertain whether the available power is sufficient to support the capacity within the coverage footprint.

The list above summarizes the general steps to be taken in forward link analysis. This process can be implemented in several ways; in particular, it is possible to obtain a set of solutions for a normalized network (e.g., unity power, unity coverage, etc.) and then scale these solutions in a simple way to address a wide variety of design scenarios. This strategy obviates the difficulty of running a computationally intensive model for every design scenario; rather, solutions for a normalized scenario can be scaled to a variety of other scenarios through straightforward adjustments of antenna gain, cell radius, fade margin, etc. For this approach, the normalized results are captured in a set of coefficients (Figure 2-5) that are used to reconstruct values relevant to the design scenario at hand by using scenario-specific parameters such as uplink coverage footprint, antenna gain, and available forward sector power. The coefficients capturing baseline performance are sometimes termed “Hong Yang” coefficients, after their author. ...........................................................................................................................................................................................................................................................

2 - 26

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

Distribution of mobile position

Distribution of shadow fading

CDMA Forward Link Solver

Distribution of soft handoffs

µ(xN) µ(xI) σ(xN) σ(xI) E(xNxI)

Ec/Ior vs. Geometry curves

Figure 2-5

“Hong Yong” coefficients

A set of coefficients that captures forward link performance for a normalized case (e.g., unity coverage, unity power, etc.) can be obtained via a computationally complex model. These coefficients can then be used to scale results to a variety of design scenarios, using design-specific parameters such as antenna gain and forward power. Design scenarios that cannot be scaled from a normalized result include those in which such underlying assumptions as fading and voice statistics, velocity distribution, and path loss laws differ from those employed in the normalized result. In these cases, a different set of normalized results employing the new assumptions is required. An example of a spreadsheet that functions as a link budget in that it extrapolates normalized results is shown in Table 2-4.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 27

Voice coverage, capacity and link budget

Analysis Table 2-4

Forward link budget

Forward Link Budget for 3G PCS 3 Sector 8 kbps CDMA with ASIC 1.1 and Mobility Applications (on Street)

Line #

B

C

D

E

F

G

Description

Power

W

Power

10.5

W

40.2

dBm

Max power available

Comments

Transmit Power calculations 5

Nominal available power at J4 point

6

Pilot Channel Power

1.575

W

32.0

dBm

15% of max. power

7

Sync Channel Power

0.2

W

22.0

dBm

10% pilot power

8

Paging Channel Power

0.6

W

27.4

dBm

35.1% pilot power

9

Power available for the traffic Channel

8.2

W

39.1

dBm

78.2% total power

%

10

Total Overhead

21.8

C10 = 100*(1 - (c9/c5))

11

Overhead factor to convert from mobiles to the number of active power channels

1.75

2.4

dB

12

Cell site Cable Loss and combiner loss

2.0

3.00

dB

13

Cell site Transmit Antenna Gain

63.1

18.0

dBi

14

Propagation loss

15

Max. mean Propagation Path Loss

1.06E+15

150.2

dB

16

Mobile RX Signal power Calculations

17

Mobile Receive Antenna Gain

1.6

2.0

dBi

18

Mobile Body/Cable/Building Losses

1.6

2.0

dB

19

Thermal Noise Calculations

20

Mobile Noise Figure (F)

7.9

9.0

dB

21

Thermal Noise Density (No = KT)

3.98E-21

-174.0

dBm/Hz

22

Total thermal Noise power per Hz (NoF)

3.16228E20

-165.0

dBm/Hz

23

Spreading bandwidth (W)

24

1.23E+06

Hz

60.9

Total thermal noise power (NoWF)

3.88581E14

W

-104.1

dBm

25

External (intermod/spectrum clearance) interference

1.58489E15

W

-118.0

dBm

26

Number of Mobiles per Sector

27

Power Outage Probability

IS-95B new handoff

dB

36 0.040

...........................................................................................................................................................................................................................................................

2 - 28

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Voice coverage, capacity and link budget

Analysis

28

Pilot Ec/(No+Io) at cell edge

29

Voice Activity Factor

30

Mean of VAF

31

Variance of VAF

32

Hong Yang's Coefficients

0.04

-13.7

dB

0.48 0.122725

µ(xN)

0.0107

µ(xΙ)

0.0200

E(xNxI)

0.0003

σ(xN)

0.0368

σ(xΙ)

0.0064

µ(Ξ)

0.01025894

σ2(Ξ)

0.00008474

µ(Ψ)

0.64631291

σ2(Ψ)

0.00533881

A “Forward Link Budget” that extrapolates normalized results. Since the assessment of forward link performance can be computationally intensive, a few normalized scenarios (e.g., unity coverage, unity power, etc.) are assessed and can be later extrapolated in a straightforward way to design scenarios of specific interest. The normalized results are captured in the Hong Yang coefficients. The extrapolation uses design-specific values such as antenna gain, sector power, and uplink coverage footprint. New coefficients must be generated if the design scenario of interest has different fundamental assumptions (e.g., fading statistics, path loss laws) from those employed in generating the normalized results. Solution--Approximate

Additional means may be used for forward link analysis. These are simpler but approximate. Since the time required to obtain a solution by extrapolating from a normalized baseline (see the "Solution-Approximate" section) is usually comparable to the time required to obtain an approximate solution, the former is preferred. Nevertheless, approximate methods can provide useful insight. For completeness, we briefly outline several methods below. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

2 - 29

Voice coverage, capacity and link budget

Analysis

The method outlined in the "Solution--Approximate" section can be simplified by placing all users at cell edge in an identical multipath and handoff state. Given the specification of user number, cell radius (i.e., allowed path loss, usually from the reverse link budget), forward sector power, and fading statistics, a geometry value for each user can be computed. The distribution of these values is then examined to determine whether it is appropriate to deliver performance; for example, the range of user velocity that can be supported could be assessed. This approach has the value of reasonable simplicity, particularly since the fades of the links from the host cell are assumed independent. Its disadvantages are the inaccuracies stemming from several sources, including the following: •

No fading can be assigned to the interference background experienced by each mobile at cell edge…for simplicity, this background is presumed constant

•

The presumption of all users at the cell edge is very conservative

•

The presumption of identical multipath for each user is not correct

Although these limitations introduce error, the result is usually conservative, particularly if a high value of surrounding interference background is used in computing user geometry. Accordingly, this approximate method remains useful. A simpler approximation may be obtained by analyzing a single mobile at the cell edge, and assessing its performance when assigned the maximum allowed value of single link traffic channel power. This approach renders the forward link very similar to the reverse link, since the fundamental issue of power-sharing amongst multiple mobiles is removed. Although this approach also provides insight, it is less often used since it completely decouples coverage from capacity; i.e., it is difficult to extrapolate from the single-link result how many additional mobiles may be served within the footprint.

...........................................................................................................................................................................................................................................................

2 - 30

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

3

RF engineering for data

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter offers a discussion of RF data issues for 3G-1X, including a contrast between the Erlang B (voice) and Erlang C (data) models, analysis of capacity and coverage, and an examination of resource management. Introduction

3-3

Traffic theory

3-4

Introduction General Erlang model Special cases: Erlang B and Erlang C Applications of Erlang C to 3G-1X data Data capacity

3-4 3-5 3-7 3-10 3-13

Introduction Data link budgets

3-13 3-19

Reverse link Forward link Symmetric forward data link analysis Example forward link budget Monte carlo forward link analysis Resource management: RF scheduling

3-19 3-22 3-23 3-31 3-35 3-36

Introduction Scheduling algorithm Fundamental Channel (FCH) assignment and release Forward link supplemental channel (F-SCH)

3-36 3-36 3-36

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 1

RF engineering for data

assignment and release Reverse link supplemental channel (R-SCH) assignment and release Load Balancing Conclusions

3-38 3-41 3-42 3-43

...........................................................................................................................................................................................................................................................

3 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Introduction

Introduction ............................................................................................................................................................................................................................................................

The availability of packet data features introduces additional complexity into the air interface. In voice-only applications, channel rates are fixed and known; moreover, network influence on the channel is limited to handoff decisions and power control. In contrast, packet data features allow variable channel rates dictated by the network. In addition, the network exerts further influence on the use of air interface resources through instructing channels when to transmit (burst) and when to wait. The role of the network in managing resources is discussed in some detail in "Resource management: RF scheduling" section on Page 3-36. The performance impact of these differences must be added to the already-present random effects of mobile speed, position, and fade. The additional parameters of channel rate and network control increase the complexity of performance prediction to the point where the situation is best analyzed via detailed end-to-end simulations of the 3G network. Simulators providing this level of detail have been developed to assess the 3G performance. Approximate analyses via other methods, e.g., link budget, have also been developed, but are of less utility for packet data than for voice. The use and limitations of approximate methods are discussed below. In the following sections, insight into 3G-1X performance is offered via several discussions. In "Traffic theory" section, we overview the differences between circuit-switched (voice) and packet-switched (data) transmissions. This information provides a brief but necessary framework for the performance discussions that follow. such as the identification and computation of performance metrics applicable to a data network. These are shown to depend upon the number and data rates of channels available, which are obtained from RF analyses that employ numerical modeling within an RF footprint. The design of this footprint is addressed in the "Data link budgets" section, which presents the 3G-1X data link budget for various data rates.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 3

RF engineering for data

Traffic theory

Traffic theory ............................................................................................................................................................................................................................................................

Introduction

The 3G-1X air interface is packet-switched in the sense that a limited number of high-speed RF data channels are shared across many users. A packet-switched network may share channels across users for the duration of the calls, or user sessions. In contrast, a circuit-switched network dedicates a channel exclusively to the user for the duration of the session. The latter model is frequently used in voice applications, where a dedicated channel is allocated and held for the duration of the voice call. The former model is used in packet data applications, where multiple data sources transmit data intermittently over a group of shared channels. The “sharing” on the 3G-1X air interface does not entail the sharing of a physically tangible resource; rather, the sharing concept derives from the fact that users transmit high-speed data bursts only when cued to do so by the network. Since channels at higher data rates produce more interference, the network manages these bursts in a way that ensures only a limited number of high-speed data bursts are simultaneously active. This process prevents the interference background from rising above acceptable levels while still allowing users to experience high data rates. This resource management (see "Resource management: RF scheduling" section on Page 3-36) can be viewed as time-sharing a limited number of high-speed data channels amongst the users, and is thereby characterized as a packet-switching process. The restricted availability of the data channel for the duration of the user session would be unacceptable for a real-time application such as voice, but is an efficient means for data support since the user need for transmission is at most intermittent for many data applications (e.g., web-browsing). For our purposes, the packet-switched nature of the air interface can be captured through use of the Erlang model. This model can also be used to illustrate the differences between the more familiar circuit-switched (voice) and packet-switched approaches, as well as to develop the performance metrics that are relevant to a data network. Although this model is well documented in a number of references,4 it is overviewed below in order to establish a framework for the performance discussions that follow. A variation of this model shall be used to estimate network performance, below (see "Data capacity" section on Page 3-13). ...........................................................................................................................

4

See for example Mischa Schwartz; “Telecommunication Networks”

...........................................................................................................................................................................................................................................................

3 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Traffic theory

General Erlang model

The Erlang model applies to the following general scenario, applicable to either voice or data:

Completions µ N servers Arrivals λ

Queue (length M)

Completions µ Figure 3-1

General Erlang Model

The assumptions and results associated with this model are absolutely essential in characterizing data performance. We review these briefly, below. The model shows service requests (arrivals) entering a waiting area (queue). From the queue, the arrivals are vectored out into one of N possible servers. Each server can serve only one arrival at a time. An arrival has immediate access to any non-busy server. If all servers are busy, the arrivals wait in the queue. The number of arrivals waiting in the queue is therefore variable. The maximum size accommodated by the queue, or queue length, is M arrivals. If the queue reaches a size of M, further arrivals are turned away or blocked until at least one arrival can exit the queue and enter a non-busy server. The definition of arrivals is very general. For wireless purposes, we view the arrivals as either voice calls requiring service (voice network) or message bursts requiring transmission (data network). In either case, the server is a transmission channel. In the former case, the server is a ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 5

RF engineering for data

Traffic theory

dedicated channel that holds the voice call for its entire duration; in the latter case, the server is a transmission channel that transmits the burst. In either case, once service for the arrival is complete, the resource is then freed up for the next arrival. The rate of arrivals is characterized by a process in which the time of arrivals is random and independent; i.e., the probability of an arrival in any one instant is identical to and independent of the probability of an arrival in any other instant. The average rate of arrivals is usually characterized by λ (e.g., calls/minute, messages/hour). The service process is similarly characterized. For a busy server, the time of service completions is also random and independent. The average rate of completions for each server is usually characterized by µ (e.g., completed calls/minute, messages transmitted/hour). The perserver completion rate is of course distinct from the system completion rate, since the system rate is dependent upon the number of busy servers. For example, the system average completion rate when N servers are busy is N×µ. Knowledge of the parameter µ can be used to compute the probability distribution of the inter-completion time. For voice calls, the intercompletion time is clearly the hold time. Its random distribution reflects the random duration of voice calls. For messages, the intercompletion time is simply the time required for the channel to transmit the message. Given a fixed channel data rate (e.g., 64 kbps), this random distribution reflects the random length of arriving messages. In both cases, the average inter-completion time is computed to be 1/µ. Within these very general assumptions, the model in Figure 3-1 can be solved analytically for the probability of all possible states, where the state is determined by the total number of arrivals within the system; i.e., the sum of all arrivals being served as well as any arrivals waiting in the queue. The possible states, therefore, range from 0 to (M+N). The last state is the blocking state, since in this state no more arrivals can enter the system. The probability of the state (M+N) is therefore the probability of blocking. The probability states are found to depend upon the ratio of λ /µ, rather than the value of either alone. This ratio is a system load parameter measured in Erlangs. Since 1/µ is the average inter-completion time, this load measure may be viewed as the arrival rate weighted by the average “stress” (average hold or average transmit time) each arrival places on the system. ...........................................................................................................................................................................................................................................................

3 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Traffic theory

The computations described above produce an analytical relation between the Erlang load, number of channels, queue length, and state probabilities. This relation has been extensively tabulated for the special case where the queue length is zero. This Erlang B table demonstrates the relation between the Erlang load, the number of channels, and the probability of state N. The latter is the probability of blocking in this case: since there is no queue, the probability of new arrivals being turned away or blocked is simply the probability that all N servers are occupied. The Erlang B model is discussed in more detail below, and contrasted to an alternate special case of infinite-length queue (Erlang C). The above concepts are summarized in the table below. Table 3-1

Summary of Erlang model

Average arrival rate

λ

Average server completion (of service) rate

µ

Average server inter-completion time

1/µ

System load (Erlangs)

λ /µ

Number of servers

N

Length of queue

M

Maximum system occupancy Probability of block Average system completion (of service) rate

N+M = Probability of state (M+N) nµ, where n = current system state

Special case

Erlang B (M = 0 or no queue)

Special case

Erlang C (M → infinity or infinite queue)

Special cases: Erlang B and Erlang C

In any situation, the collection of state probabilities depends upon the Erlang load as well as the values of M (queue length) and N (servers). Two limiting cases are of especial interest. In the first case, the queue length is set to 0. Arrivals are therefore blocked as soon as all servers are busy; i.e., the probability of blocking is probability of the state N. The blocking probability is entirely determined by the Erlang load and the number of servers N. The relation between the three is captured in an Erlang B table, which allows computation of any one (e.g., Erlangs) from specification of any other two (e.g., probability of block, number of servers).

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 7

RF engineering for data

Traffic theory

The Erlang B table is widely used for voice calls, since its underlying model captures the scenario where voice users make a single call attempt (a single arrival) and are either immediately served or blocked. In the latter case, the user may try again at a later time, but the elapsed time is sufficient to ensure that the next attempt resembles a new, independent arrival to the system. In the second case, the queue length is presumed infinite. Arrivals are therefore never blocked; however, there is a probability of delay. The probability of being delayed (i.e., of waiting some nonzero time in the queue for service) is the probability of the state N, where all servers are busy. The delay probability is entirely determined by the Erlang load and the number of servers N. The relation between the three is captured in an Erlang C table, which allows computation of any one value (e.g., Erlangs) from specification of any other two (e.g., probability of delay, number of servers). Since the average wait time in the queue can be determined from the probability of delay (and vice-versa), an alternate 3-way relation of Erlangs, average wait, and number of servers may be tabulated. The Erlang C table may also be used for voice calls, since its underlying model captures the scenario where voice users make repeated, multiple attempts as necessary to be served. In this interpretation, each arrival is a single voice user attempting to access the system. The user is either served on the first attempt, or not; in the latter case, the user continues to attempt to access the system until served. These continuous reattempts place the user in the system queue, “waiting” for service. Note that the queue in this application is conceptual, representing the collection of users attempting but not yet achieving access. The use of this model for voice calls is less prevalent than that of Erlang B, since it requires that each user continuously attempt access until finally served; in contrast, Erlang B requires a single access attempt per user. Neither model can therefore accommodate scenarios where each user may execute 1 or 2 immediate re-access attempts before being served or blocked, but Erlang B is frequently considered to be a better approximation of this situation than Erlang C. The Erlang C model is more typically used for packet data, since the presence of a queue lends itself to modeling the management of data resources over shared channels. In this application, a large number of data sources time-share a modest number of transmission channels N. The data sources send brief data transmissions (bursts) as permitted by ...........................................................................................................................................................................................................................................................

3 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Traffic theory

the network, which attempts to share the transmission channel resources in some efficient manner. Bursts (messages) undergoing transmission are being served by one of the N servers in the Erlang C model (see Figure 3-1). Messages waiting their turn for transmission are in the queue, regardless of whether these messages are stored at each data source (a conceptual queue) or stored in a single intermediate physical buffer between the data sources and the servers (physical queue). The random nature of the arrival process into the queue is driven by the fact that the data source does not require the sending of continuous messages; rather, the arrival of messages is randomized by the bursty, interactive nature of the data application (e.g., web-browsing). Indeed, this randomness is exploited in order to efficiently serve the data users with a number of servers that is less than the total number of active data sessions. The random nature of the service process at each server derives from the random variations in message length. For a server of fixed transmission rate, these random variations in length randomize the service or hold time for each arrival. In Erlang C, the net rate at which arrivals exit the system after being served depends upon the number of busy servers. For n busy servers, the net rate is n×µ, where n can vary from 0 to N. Since the probability of all states is known, the average net rate at which arrivals exit the system (the throughput) can be computed as:

throughput =

N −1

∑ n ⋅ µ ⋅ p (n) + N ⋅ µ ⋅ p n=0

delay

Equation 3-1: Throughput equation

The units of throughput are messages/second or more conventionally bits/sec. In the above, the average rate is computed by weighing the service rate at each state by its state probability. For n less than N, the service rate associated with each state is n×µ, since for these states, the queue is empty and n servers are busy. For n greater than or equal to N, all servers are busy and the service rate becomes fixed at N×µ. This rate is therefore weighted by the probability of delay, since the probability of all servers busy is simply the probability that an arrival will be delayed (i.e., will need to wait in the queue). The concept of throughput is directly applicable to 3G-1X traffic planning, as described below.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 9

RF engineering for data

Traffic theory

Applications of Erlang C to 3G-1X data

We now consider specific use of the Erlang C model for 3G-1X data. Identity of the model components, as well as specific measures of performance for planning and analysis, are discussed below. In 3G-1X data, high-speed or supplemental channels are dynamically set up for the users when a burst is cued, and torn down when the burst is complete (see "Resource management: RF scheduling" section on Page 3-36). In order to control interference, only a limited number of supplemental channels may be simultaneously active; accordingly, we may view the process as simply time-sharing a fixed number N of highspeed servers. These high-speed channels are the servers within the Erlang C model. (The number of servers is determined via RF analysis in "Data link budgets" section on Page 3-19). Message bursts requiring transmission are either immediately transmitted, or stored awaiting transmission. At the reverse link, storage occurs within the mobile data device. At the forward link, storage occurs at a buffer in the cell site. In both cases, this storage corresponds to the queue in the Erlang C model where arrivals are waiting for access to the servers. The forward link queue is physical in that a single buffer can be identified where messages arrive and await service. The reverse link queue is more conceptual in that it consists of the collection of stored messages across the mobile data devices. In both cases a large number of arrivals can be stored; hence, the queue length is approximated as infinite. For this queue and these servers, a three-way relation between Erlang load, number of servers, and average wait time in the queue is readily determined from the Erlang C model. Once these values are determined, the throughput can also be calculated (see Equation 3-1). Accordingly, the load that can be accommodated by a sector can be obtained by specifying the number of servers and the average wait time. In 3G-1X, the determination of the number of servers per sector is a constraint dictated by the RF interface. The average wait time (e.g., 5 seconds) is specified as a requirement and corresponds to the average time between actual message transmission and the time at which data enters the buffer (mobile or cell) to await service. Given these values, the Erlang load that the sector can accommodate follows directly from the Erlang C model. This load can then be compared to the load offered from the subscriber population to assess how many sectors are required.

...........................................................................................................................................................................................................................................................

3 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Traffic theory

This process is followed in 3G-1X traffic planning with a slight modification: the offered load is assessed in throughput rather than in Erlangs. This approach differs from that employed in Erlang B, since in the Erlang B model the subscriber load in Erlangs can be determined in a manner independent of the network; i.e., the load depends only upon the characteristics of the subscriber population. In Erlang C, the subscriber load in Erlangs depends upon transmission properties of the network as well as upon characteristics of the subscriber population. This difference, which drives the use of throughput as an alternate load measure in planning data networks, is described below. In a voice network modeled by Erlang B, the load in Erlangs is the product of arrival rate and of average hold time (see "General Erlang model" section on Page 3-5 and "Special cases: Erlang B and Erlang C" section on Page 3-7). Arrival rates (e.g., calls/minute) are clearly a function of the subscriber characteristics alone. Hold times (e.g., minutes/call) are also a function of the subscriber characteristics alone provided that any additional network processing time is negligible by comparison (a very good assumption for hold times that are typically measured in minutes). The Erlang load offered to a sector can therefore be determined from subscriber characteristics alone; indeed, since Erlangs from different sources add together to yield net Erlang loads5, the total load offered to an unspecified voice system can be determined by multiplying the estimated average subscriber contribution (typically expressed as milliErlangs per subscriber) by the estimated subscriber population. In planning, this offered load is readily compared to the accommodated load per sector in order to determine the number of sectors needed within any geographic area. Accordingly, we seek an alternative measure of load that is a function of subscriber characteristics alone. Since Erlang loads from multiple data sources add, we view the subscriber population as a large collection of data sources, each contributing a modest Erlang load to the data network. The arrival rate (e.g., messages/second) for the ith data user is λi. The average hold time per message is identical at 1/µ, which can be expressed as the average message length in bits divided by the server capacity of C bits/second:

...........................................................................................................................

5 This addition property for multiple sources holds provided that each source has identical hold time statistics. This assumption holds well for voice users.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 11

RF engineering for data

Traffic theory

Total Erlangs =

L 1 λtotal λi = = λi = ∑ ∑ ∑λ L C C data subscriberi s i µ data subscribers i µ data subscribers i Equation 3-2: Total Erlang calculation

In the above equation, the expression within the final sum has the units of messages/sec * bits/message or bits/sec. This measure represents the total data transmission load (usually expressed in kbits/sec) or throughput offered by the subscriber population. This measure of load is proportional to the Erlang load via the transmission capacity C as shown above, and is a property of the subscriber population alone. Since the throughput accommodated by the network can be readily computed from the Erlang C model (see Equation 3-1), this value can be compared to the throughput offered by the subscriber population in much the same way that the Erlang load possible on a voice network can be compared to the Erlangs offered by the subscribers. Traffic planning for 3G-1X data networks can therefore be done as follows: 1. 2. 3. 4.

5.

Establish the number and rate of servers available on the air interface per sector via RF analysis Specify average wait time required Compute the throughput that can be accommodated by the sector by using the Erlang C model Estimate the throughput offered by the subscriber population through estimating the messages/sec and the average message length (see Equation 3-2) Compare the accommodated throughput to the offered throughput to determine how many sectors are required.

For example, if a sector can accommodate 100 kbps satisfying the wait time constraint specified, 10 sectors are needed to address an area offering a total 1000 kbps. More sectors might be needed to address other requirements, such as RF coverage throughout the area. Steps 1 through 3 are addressed in the Lucent modeling described below. This information establishes a throughput per sector for an average wait time of 5 seconds and various other assumptions on the data traffic encountered. The value obtained varies as wait time and data statistics are altered.

...........................................................................................................................................................................................................................................................

3 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data capacity

Data capacity ............................................................................................................................................................................................................................................................

Introduction

In this section, we determine the capacity offered by a 3G-1X sector. Although this capacity is best determined by detailed time-dependent simulations, considerable insight can be gained by less computationally intensive modeling that exploits the Erlang C model. The essential details of this modeling are overviewed below. In summary, information obtained from standards is used to develop link level simulations. This data is input to a system simulation. This simulation determines the number and rate of channels available from RF considerations. This information, coupled with end-user traffic models and system performance constraints, is used to determine capacity via an Erlang C model (see Figure 3-1). We focus our attention on the capacity calculation. An average wait time for a message is specified as a requirement. The throughput for the sector can then be calculated from the Erlang C model provided that the other components within the model are specified. These include hold time per server, rate of server, and number of servers. To determine this information, we presume a forward-link limited situation within a cell coverage area. This presumption divorces the analysis from specific output powers and specific cell radii. The power required to balance the links is discussed separately in "Data link budgets" section on Page 3-19. Within the coverage area, one of the possible 3G-1X supplemental data rates (e.g., 19.2 kbps) is selected and fixed. Given a presumed average message size, the average supplemental channel hold time required per message is computed. The number of channels accommodated by the air interface at this rate is a random function depending upon a number of variables including mobile position, speed, multipath, and fade. Monte Carlo analysis is therefore employed to produce a probability distribution function of the number of channels at this rate. For each possible number of channels within the distribution, a throughput is calculated from the Erlang C model. The probability of this throughput is the probability of the associated number of channels. An overall average throughput for this data rate is computed by weighing each value of throughput by its probability.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 13

RF engineering for data

Data capacity

The process is repeated for each 3G-1X data rate of interest, yielding an average throughput for each rate. Using estimated average throughput per subscriber (where the averaging interval includes delays between transmissions required to read or think about a downloaded message), the number of subscribers accommodated at each rate is calculated. The results for each data rate are tabulated and then combined in a weighted fashion that reflects the anticipated mix of users at different data rates. The thoughputs calculated are then compared to offered subscriber loads to determine how many sectors are required. This process is summarized in Figure 3-2, and described in greater detail below.

IS-2000 STANDARD

User Mobility

Physical Layer Specs.

Channel Structure

Data Protocols

LINK LEVEL SIMULATION Power Requirements per Channel

SYSTEM LEVEL SIMULATION Number of Channels End User Traffic Model

QUEUING MODEL QoS Requirements

CAPACITY Figure 3-2

Estimation of data capacity

Overview of computation process for capacity

From an air interface perspective, we anticipate that the 3G-1X packet data capacity and throughput will be governed by several interlocking factors including: •

The number of users (fundamental and supplemental channels) that can be supported

•

The number of users that share the supplemental channels

...........................................................................................................................................................................................................................................................

3 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data capacity

•

The relative position of the users within the cell site coverage area. This information is key since the maximum data rate supported by the link can clearly increase when the user is closer to cell center.

•

The average throughput per data user

•

The associated FER target

•

The Automatic Repeat Request (ARQ)

•

The link channel activity and packet call size.

Several studies have been done to examine aspects of these elements. For example, analysis on the air interface limit of supplemental channels has been done for each supplemental channel data rate by conducting link level and system level simulations. The air-interface limit of supplement channels derived in this analysis is the distribution of the simultaneous active channel number that depends on the target Frame Error Rate (FER), mobile locations, mobile speeds, propagation environments, other user interference and the base station power allocated to each traffic channel. Once the distribution of the number of supplemental channels is determined, the M/M/m queuing model (Poisson arrival, exponential distribution of service time and m servers) is used to compute the average total throughput and data user capacity that can be supported for a given data traffic model including average packet call size, target queuing delay, and supplemental channel rate. More specifically, the link level simulations are performed to obtain the required base station power fraction for a traffic channel versus the geometry that is a function of the mobile location and propagation environments. The geometry is defined as the ratio of the mobile received serving sector power to the mobile received other interfering sector power plus noise power. In the link level simulation, we consider the following parameters: •

Radio Configuration 3 (RC3)

•

9.6kbps at 1% FER, 19.2kbps at 2% FER, 38.4kbps at 2% FER, 76.8kbps at 3% FER, and 153.6kbps at 5% FER

•

No handoff on the forward link for supplemental channels.

In order to obtain the Probability Density Function (PDF) of supportable supplemental channels from the system level simulator, the following assumptions are made: •

Snapshot simulation technique

•

3-sector configuration, 19 cells and 57 sectors

•

Randomly generate data user location within center cell

•

ITU vehicular propagation model

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 15

RF engineering for data

Data capacity

•

8 dB log-normal standard deviation and 0.5 site-to-site correlation

•

Mobile speed distribution: 50% for Additive White Gaussian Noise (AWGN) and 50% for 1 path Rayleigh fading at 3 kmph (pedestrian speed)

•

5% outage probability

•

Turbo code gain is considered for data rates greater than or equal to 19.2 kbps.

Having determined the distribution of the supplemental channel number, we employ the M/M/m queuing theory to derive the average throughput and data user capacity based on the data traffic model for web browsing (illustrated in Figure 3-3). In the data traffic model, a session is defined as the interval between the time instant when a data user logs in the web site and the time instant when the user logs off the web. A session consists of a number of packet calls (web pages for the web browsing application), each of which is comprised of several packets. For the web browsing application, the total delay per page is defined as the time interval between a mouse click and the completion of a web page download. In other words, the total delay per page is the sum of the access time, network delay, queuing delay and download time. The average packet call inter-arrival time between two adjacent mouse clicks equals the total delay per page plus the think time. Think time is the duration between the time instant when starting reading a web page and the time instant when clicking a mouse for the next page. Therefore, the average throughput is obtained by dividing the average packet call size by the average packet call inter-arrival time. In the following, we will provide the average throughput and data user capacity in terms of simultaneous data sessions for the case where 5 sec is selected as the queuing delay per packet. To characterize the data session fully, some additional assumptions are made: •

Exponentially distributed packet call (web page) size with a mean of 41.1 kBytes

•

Exponentially distributed “think” time between packet calls (page downloads) with a mean of 40 sec

•

Packet Call Inter-Arrival Time = Access & Network Delay (3 sec) + Target Queuing Delay (5 sec) + Download Time (dependent on the channel rate) + Think Time

•

Average number of packet calls per session = 20

•

“Equal user throughput” scheduling policy

...........................................................................................................................................................................................................................................................

3 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data capacity

Web Browsing Session Active

Dormant

Supplemental Channel Bursts

First Data Arrived at IWF 38.4

Active

Dormant

153.6 SCH

153.6 SCH 76.8 SCH

76.8 SCH

76.8 SCH

9.6 kbps FCH

9.6 FCH

Start Reading Web Page

Mouse Click

Access Time

Download Network Time Delay Queuing Delay ( incl. SCH Setup Delay)

Figure 3-3

153.6 SCH

Mouse Click

Fundamental Channels

Dormancy Timer Duration "Think" Time

Data traffic model for web browsing application with the 3G-1X packet data users

Feeding the system level simulation results into the queuing model with the data traffic scenario, we obtain that average throughput per user for 3G-1X packet data. The packet data capacity is shown in Table 3-2 as a function of supplemental channel data rate. In the table, the average total delay per page is defined as the sum of the access time, network delay, queuing delay, and download time. The average number of simultaneous data sessions can be calculated by dividing the average sector throughput by the average user throughput. It is observed that as the supplemental channel data rate decreases, the average total delay increases but the average sector throughput and the average number of simultaneous data sessions are comparable. The average data user capacity is determined by several dominant factors: average packet call size, supplemental channel rate, think time, and Quality of Service (QoS) including the target FER and queuing delay.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 17

RF engineering for data

Data capacity Table 3-2

Average data user capacity and throughput with an average packet call size of 41.1 kBytes and a target queuing delay of 5 seconds

Channel Total Delay Average User Throughput per Page Rate (kbps) (sec) (kbps) 19.2 25.1 38.4 16.6 76.8 12.4 153.6 10.3 Average 16.1

4.9 5.7 6.1 6.4 5.8

Throughput per Number of Simultaneous Data sector per carrier(kbps) Sessions 23 111 20 112 18 111 17 109 20 110.8

Note that the above values indicate that the throughput per sector is robust with respect to the data rate. This result indicates that an anticipated throughput of 109 to 112 kbps is robust with respect to whatever weighting is employed to combine the results at individual data rates into an overall estimate. This observation is useful in planning, but could change as components of the underlying traffic model or requirements (e.g., think time and the average wait in queue) are altered.

...........................................................................................................................................................................................................................................................

3 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

Data link budgets ............................................................................................................................................................................................................................................................

Reverse link

The reverse link data budget for 3G-1X data is readily obtained by recognizing that coverage is dictated by the data rate desired at the edge of the data coverage; i.e., the edge of the coverage of the reverse link supplemental channel used to transmit high-speed data bursts. This edge may or may not correspond to the physical edge of the cell, which could for example be designed to support voice rate at its perimeter and higher data rates only within its interior. With the data rate desired at the edge of data coverage specified, the data coverage footprint is determined by presuming that all users within this footprint operate at this data rate when transmitting on a supplemental channel. The analysis outlined in Chapter 2, "Reverse link" section on Page 2-4, for voice therefore applies directly with only a few simple modifications. These are: •

The voice activity for the supplemental channel is presumed to be 1. This high usage reflects the assumption that the few supplemental channels supported by the air interface will be almost continuously busy as they are shared from user to user.

•

The information rate is higher, corresponding to the data rate (e.g., 19.2 kbps) selected for cell edge

•

A voice user certainly requires a body (head) loss; e.g., 2 dB. A data user employing a data device may encounter little or no loss. In the below, a 0 dB loss for data users is assumed.

•

The Eb/Nt requirement is lower for the data application due to the relaxed target FER. The relaxed FER is permissible since the data application is not real time; i.e., frames received in error can be retransmitted.

The PCS Modcell reverse link budget examples for 3G-1X 19.2 kbps to 153.6 kbps packet data, mobility applications and 9.6 kbps voice application are shown in Table 3-3. Note that the target FER relaxation for the data application is used to increase the maximum path loss (or cell coverage). Simulation results indicate that the increased FER does not cause significant TCP/IP throughput degradation.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 19

RF engineering for data

Data link budgets Table 3-3

Reverse link budget for PCS 3G-1X 9.6 kbps voice, 19.2 kbps to 153.6 kbps packet data, mobility applications

Item

(a) Maximum Transmitted power per traffic channel

Units

dBm

3G-1X Data 19.2kbps

3G-1X Voice 9.6kbps

3G-1X Data 3G-1X 76.8kbps Data 38.4kbps

3G-1X Data 153.6kbps

Comments

21

21

21

21

(b) Transmit Cable, connec- dB tor, combiner, and body losses

2

0

0

0

0 No body loss for data user

(c) Transmitter Antenna Gain

dBi

2

2

2

2

2

(d) Transmitter EIRP per traffic channel (a-b+c)

dBm

21

23

23

23

23

(e) Receiver Antenna Gain

dBi

18

18

18

18

18

(f) Receiver Cable and Con- dB nector Losses

3

3

3

3

3

(g) Receiver Noise Figure

dB

4

4

4

4

4 For Modcell

(h) Receiver Noise Density

dBm/Hz

-174

-174

-174

-174

(i) Receiver Interference Margin

dB

5.5

5.5

5.5

5.5

-164.5

-164.5

-164.5

-164.5

-164.5

51.9

(j) Total Effective Noise plus dBm/Hz Interference Density = (g + h + i)

21

-174 5.5 72% loading for 3G-1X

(k1) Information Rate (10log(Rb))

dB

39.8

42.8

45.8

48.9

(l1) Required Eb/Nt

dB

4

3.4

2.6

1.8

(m) Receiver sensitivity (j + k + l)

dBm

-120.7

-118.5

-116.5

-114.5

(n) Hand-off Gain

dB

4

4

4

4

4 90% cell edge coverage

(o) Explicit Diversity Gain

dB

0

0

0

0

0 Diversity gain has been included in required Eb/Nt

(p) Log-normal Fade Margin dB

10.3

10.3

10.3

10.3

10.3 For 90% edge coverage with 8dB log-normal standard deviation

(p') Building/Vehicle Penetration Loss

dB

0.0

0.0

0.0

0.0

(q) Maximum Path loss {d - m + e + o + n - p - p'}

dB

150.4

150.2

148.2

146.2

1 With Turbo code gain for data; 1% target FER for 9.6kbps, 2% for 19.2kbps, 2% for 38.4 kbps, 3% for 76.8kbps and 5% for 153.6kbps; considering 2 spatial receive diversity branches -112.7

0.0 For outdoor coverage 144.4

...........................................................................................................................................................................................................................................................

3 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

If the design goal of a newly deployed 3G-1X system is to provide a ubiquitous coverage for a high-rate data service, then the link budget based on the supplemental channel rate should be used for RF design. In this case, the physical edge of the cell is determined by the edge of data coverage. In contrast, if the voice link budget is used, then the high-rate data service will be available with the same probability of coverage as voice coverage in an inner circle of the cell coverage. In this case, the supportable packet data rate, or alternatively the probability of achieving a higher data rate, will reduce when the mobile moves close to the cell edge. The maximum allowable path loss for the packet data can be extended by employing the data terminals with higher antenna gain and transmitted power and increasing the base station transmit power. In the above examples, the interference margin is retained at a constant 5.5 dB in spite of the fact that the number of supplemental channels available at each data rate decreases as the data rate increases. A reduced number of supplemental channels could force a reduction in loading in order to ensure system stability; however, the interference background is stabilized by the constant (1) value of voice activity for the few channels present (see Chapter 2, "Solution--Approximate" section on Page 2-9). The link budgets shown above can be applied to the situation of ubiquitous coverage at a given data rate. For example, if 76.8 kbps is desired throughout the coverage area, the cell footprint would be designed by employing the 76.8 kbps budget: since this cell spacing extends the 76.8 kbps to the cell edge, this rate is by extension available throughout the interior of the cell. Since each data rate has equal interference margin, the budgets shown can also be used to map out the relative coverage areas for a mix of supplemental channels within a larger footprint. For example, the outer physical perimeter of the cell could be established using the 9.6 kbps link budget. Within this perimeter, the restricted dB losses shown for the link budgets at higher rates establish the inner coverage areas where the higher rates are available as shown in the following figure:

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 21

RF engineering for data

Data link budgets

- 9.6 kbps Fundamental Voice or Data Coverage - 19.2 kbps Supplemental Coverage - 38.4 kbps Supplemental Coverage - 76.8 kbps Supplemental Coverage - 153.6 kbps Supplemental Coverage

alpha gamma Cell Cite

Figure 3-4

Forward link

beta

Inner coverage areas for higher supplemental channel data rates

Analysis of the forward link is best conducted by numerical simulation; however, the computational load associated with such calculations drives the need for simpler planning tools. In the following, we briefly consider several alternatives for simplified forward link analysis. The relative merits of each are discussed. All techniques can be employed in two basic planning configurations. The first or embedded configuration is determined by a single data rate, e.g., 76.8 kbps. This data rate corresponds to the physical edge of the cell and thus determines the physical cell footprint. The second or concentric configuration is determined by two data rates, e.g., 76.8 kbps and 9.6 kbps. The lower data rate determines the physical edge of the cell, i.e., the outer boundary. The higher data rate corresponds to the boundary of an inner footprint within the cell where the higher data rate can be supported. The inner footprint is thus a subset of the overall cell coverage. This concept is illustrated in Figure 3-5. The concentric configuration is more common, since upgrades to 3G-1X frequently involve overlay on existing networks where the outer physical boundary is determined by the IS-95 voice data rate.

...........................................................................................................................................................................................................................................................

3 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

Outer boundary; e.g., 9.6 kbps

Inner boundary; e.g., 76.8 kbps

Figure 3-5

Concentric configuration: inner and outer boundaries dictated by two data rates

In the following, we consider the application of symmetric link budget analysis and Monte Carlo link analysis to the forward link. Note that the former is essentially identical in form to a spreadsheet analysis in which all terms (including symmetric terms common to each link) are employed; however, the discussion below focuses on the symmetric approach in order to more compactly indicate the concepts involved. In each case, application to embedded and concentric configuration is discussed. Symmetric forward data link analysis

The purpose of symmetric forward link analysis is to establish the path loss within which a given data rate can be supported. This analysis might be done to assess whether a footprint established by reverse link can be supported, or to establish limits on path loss imposed by forward link considerations alone. The latter is useful in situations where, by design, the forward link is expected to carry higher data-rate traffic than the reverse link. In this case, support of the footprint established by the lower-rate reverse link would not be relevant; rather, the design footprint would be established solely by forward link limitations. The data rate to be supported at the cell edge is chosen. This rate is the rate desired for the supplemental channel. A path loss to the cell boundary is computed (e.g., from reverse link considerations), or simply presumed as a starting point for analysis. All forward links are presumed to burst at this data rate, and the mobile receivers are symmetrically arranged in a worst-case situation at the cell edge. The analysis then determines whether the available forward link power is ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 23

RF engineering for data

Data link budgets

sufficient to achieve the required forward link Eb/Nt at the mobile receiver in light of fading phenomena across the links. The symmetric arrangement of the mobiles ensures that the Eb/Nt requirement for each link is identical, and renders the problem soluble without extensive numerical computation. Although the approach is conceptually similar to voice analysis, important differences exist. Unlike voice, the rates of all links are not identical. The analysis must consider mobiles employing both the lowrate fundamental and the high-rate supplemental channels. The former are mobiles transmitting at low levels while waiting to burst, i.e., waiting for a supplemental channel. The latter are mobiles bursting, i.e., transmitting on a supplemental channel. In addition, soft handoff is only available for the fundamental channel. No soft handoff exists on the forward link supplemental channel. Furthermore, the definition of “cell edge” in this analysis depends upon the configuration employed. In the embedded configuration, there is no ambiguity: the cell edge corresponds to the physical outer perimeter of the cell. In the concentric configuration, the cell edge in analysis corresponds to the physical edge of an inner coverage circle where the data rate of interest can be supported. The outer coverage boundary of the cell is dictated by a lower data rate and corresponds to the physical edge of the actual cell footprint (see Figure 3-5). To avoid confusion, we will refer to the edge corresponding to the high data rate of interest as the data cell edge. The data cell edge is the boundary of cell footprint in an embedded configuration, but is the edge of the inner boundary in the concentric configuration. With these definitions, we proceed as follows for both configurations. We first establish the path loss corresponding to the data cell edge by noting the sensitivity required for the uplink at the data rate of interest (e.g., 76.8 kbps):   FN tW S min = E{S }=   g − (1 + β )( N − 1)η α d  max S min a= wmax g net ( fade)

     

Equation 3-3 ...........................................................................................................................................................................................................................................................

3 - 24

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

The above expression essentially corresponds to reverse link analysis for the supplemental channel data rate of interest (see "Reverse link" section on Page 3-19). Note that the processing gain g must equal the ratio of bandwidth to data rate, R, as in voice. However, the processing gain may well be modest (in comparison to voice) since the supplemental channel data rate can be as high as 153.6 kbps. The receiver sensitivity is used to solve for the attenuation (path loss) a, which will serve as a starting point for forward link analysis. Alternatively, a value of a could be assumed. We compute the value in this way for this example under the presumption that it is desired that both links achieve the same supplemental channel data rate at the data cell edge, regardless of whether concentric or embedded configurations are employed. We now proceed by placing all mobiles at the data cell edge. Mobile receiver Eb/Nt requirements and total forward power constraints are then used to produce the governing relationship that must be satisfied with high probability.

N links

αidi βi

i =1

gi

∑

[1 + η (1 / β i − d i / g i ) ] ≤ (1 − γ )Q max Q total

Equation 3-4

This form of Equation 3-4 is essentially identical to that employed for voice. In spite of this similarity, the value and meaning of many of the underlying parameters are different. The summation is over both supplemental and fundamental links. Accordingly, the values of channel activity α, forward link Eb/Nt requirement d, interference ratio β, and processing gain g extend across both supplemental and fundamental channels. This extension is the reason behind the subscript i on the processing gain. Unlike voice, this value is not constant per link, but varies in accordance with whether the link is fundamental or supplemental. Accordingly, we alter the form as follows:

ηd

N links



∑ α i =1



links

i

i

( )

N εd  β [1 + ξ (1 / β − η / η )]= η ∑ α new β [1 + ξ (1 / β − η / η )]≤ η (1 − γ ) i d g d i i i d g g ε gi  i i =1

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 25

RF engineering for data

Data link budgets

where:

di = ε d E d = ε d ηd i

i

gi = ε g E g = ε g η g i

i

and

(α )= α ε new

i



i

  ε gi  di

Equation 3-5

Presuming that the total number of links Nlinks is composed of Nsuppl and Nfund links, the means of d and g are readily computed by using constant values of d and g for supplemental links and constant values of d and g for fundamental links:

ηd =

N sup pl N d sup pl + fund d fund N total N total

ηg =

N sup pl N g sup pl + fund g fund N total N total

where N total = N fund + N sup pl Equation 3-6 Composite means over fundamental and supplemental channels for processing gain and required Eb/Nt

The newly defined channel activity has statistics defined by the fundamental and supplemental channel activities, weighted by the deviations of Eb/Nt, d, and processing gain g from their means. This random variable is independent of others in the sum. The solution for satisfying this relation with high probability, provided that it is understood that the channel activity variable in this solution is the newly defined channel activity above is:  1 + ηg N linksηα η β   (1 − γ )

k N links

σ α2σ β2 σ β2 σ α2   + + ηα2η β2 η β2 ηα2  

( −1)

[1 + ξ (1 / η

β

]

− η d / η g ) −1 ≥ η d

Equation 3-7 Forward link budget statement

Satisfaction of this inequality indicates that the supplemental and fundamental channels can indeed be supported at the data cell edge. The forward link budget essentially evaluates the left-hand side ...........................................................................................................................................................................................................................................................

3 - 26

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

expression, and ensures that it is greater than or equal to the right hand side. This analysis, encapsulated in the spreadsheet, can be used to solve for alternate values, given others. For example, the current format uses an assumed path loss and total J4 power to assess the Eb/Nt that can be achieved with high probability in order to compare this value to the right hand side requirement in Equation 3-7. Alternatively, the required Eb/Nt and J4 power can be used as inputs to solve for the path loss that can be tolerated while still satisfying Equation 3-7 with high probability. Equation 3-7 can also be evaluated using a more detailed calculation that simply includes symmetric terms like antenna gain. As expected, the results are identical since the precise value of the symmetric terms has no effect in establishing forward link viability. Nevertheless, the more detailed spreadsheet (see Table 3-4) is sometimes preferred since it lists such terms explicitly. With the exception of terms related to the interference ratio β and newly defined channel activity α, all other factors required to evaluate Equation 3-7 can be found in Equation 3-6. We now consider the evaluation of these channel activity and interference ratio terms. The computation of the mean and variance of the newly defined channel activity can be done in a conventional way provided that the statistics of this random variable are known. These are readily developed as follows. We presume that the shared supplemental channel(s) are continuously employed; accordingly, their activity is 1. In contrast, the fundamental channels operate at 1/8 of the fundamental channel rate of 9.6 kbps, i.e., a low rate sufficient to maintain the call between bursts. Considering the definition in Equation 3-5, the statistics underlying the newly defined channel activity are: With probability Nsuppl/Ntotal:

α new = (1)

dsup pl /ηd

gsup pl /ηg

With probability Nfund/Ntotal:

α new = (1/ 8)

d fund / ηd

g fund / ηg

Equation 3-8 Voice activity for data link budget ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 27

RF engineering for data

Data link budgets

Note that the former is simply the value of new channel activity for a supplemental channel. The latter is the value of new channel activity for a fundamental channel. The Equation 3-8 completely characterizes the statistics of the newly defined channel activity; accordingly, mean and variance can be computed in the standard fashion. We assume that the values and statistics of interference ratio are identical for both fundamental and supplemental channels. This assumption generally follows from the symmetric placement of all mobiles at data cell edge, and is either accurate or simply conservative depending upon the configuration employed (see below). In the embedded configuration, the value of Pother/Phost for the supplemental channel is larger than in voice applications, since the forward link supplemental channel does not enter into soft handoff at the data cell edge. Accordingly, the value of Pother is raised since the broadcasts from the neighbor cell(s) can no longer be treated as a source of signal rather than of interference. (In voice, the power from neighbor cell[s] does not contribute to Pother since these cells contribute a soft handoff link). The value for supplemental channel thus increases from –4 dB (voice) to nearly +2 dB. The latter value can also be used as a very conservative estimate for the fundamental channels, which do enter into soft handoff in a manner similar to a voice channel. In the concentric configuration, the value of Pother/Phost for both fundamental and supplemental channels varies depending upon the data rates that establish the inner boundary (data cell edge) and outer boundary (physical perimeter) of the cell. Generally, the outer boundary is well removed from the inner; accordingly, neither fundamental nor supplemental channels are in soft handoff and the properties of Pother/Phost are identical for both. In the forward link budget, the constant value of Pother/Phost is determined in an offline fashion by simply computing the sum of received interference from neighbor cells. The relevant path loss in this computation is not the loss from neighbor cell to host cell boundary, but the path loss from neighbor cell to the data cell edge (i.e., inner boundary). This value is determined by presuming a simple path loss law; e.g., 38 dB/decade. The distance between inner boundary (data cell edge) and outer boundary (physical perimeter) is essentially determined by examining difference in high and low data rates (e.g., 76.8 kbps, 9.6 kbps) characterizing the two. These considerations are illustrated in Figure 3-6 and Figure 3-7 below. ...........................................................................................................................................................................................................................................................

3 - 28

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

Other Cell

X

X Host

Figure 3-6

mob

X Other

Embedded configuration cell boundaries

In an embedded configuration, the physical boundary of the cell corresponds to the data cell edge. A mobile at cell edge receives power from its host and power (interference) from other surrounding cells. The interference from all surrounding cells must be considered, as the forward link supplemental channel is not in soft handoff with any cell. In contrast, for a voice channel, surrounding cells supporting the mobile in soft handoff would be sources of signal, and not contribute interference. The value of beta for the supplemental channel is therefore greater than the values employed for voice.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 29

RF engineering for data

Data link budgets

Other Cell

mobile

Host Cell

Figure 3-7

Other Cell

Concentric configuration cell boundaries

In a concentric configuration, the data cell edge corresponds to an inner boundary where the higher data rate is available. The outer (physical) boundary of the cell corresponds to a lower data rate. This configuration is common for upgrades/overlays of 3G-1X data on a 2G-voice footprint, since the 2G voice data dictates the outer (physical) boundary of the cell. A mobile at the data cell edge receives power from its host and power from surrounding cells; however, the interference from surrounding cells is proportionately less than the host power since the mobile is no longer equidistant to all cell sites. The value of beta for the supplemental channel varies according to the inner and outer data rates. The computation of Pother/Phost for the embedded configuration ignores the effect of supplemental channel anchor transfer. This transfer is essentially a very fast hard handoff in which the forward data link is dynamically switched to the best serving cell. This switching is facilitated by information provided by the mobile’s forward link fundamental channel, which can enter soft handoff with surrounding cells as appropriate. The presence of anchor transfer should reduce the effects of other cell interference by ensuring that the strongest cell is

...........................................................................................................................................................................................................................................................

3 - 30

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

always a source of signal rather than of interference; nevertheless, in link budget analysis, we make the conservative assumption of ignoring any anchor transfer gain. An example forward link spreadsheet for the 3G-1X fixed 153.6 kbps data application is shown below. The spreadsheet uses the approaches described above, but with symmetric terms (e.g., antenna gain) added for example completeness. The example is conservative in that: •

The embedded configuration is employed. As described above, the embedded configuration establishes the data cell edge as the physical cell boundary. The symmetric arrangement of mobiles at this cell edge maximizes the interference from surrounding cells while minimizing host signal power.

•

The chosen rate of 153.6 kbps minimizes the spread spectrum channel processing gain (W/R). The ability of the supplemental channel to reject interference from surrounding cells is therefore reduced.

•

No gains are allowed for anchor transfer.

In spite of these restrictions, the spreadsheet indicates 11 fundamental channels (i.e., 11 mobiles) can be supported at data cell edge. In addition, a single supplemental channel can be simultaneously supported. This channel essentially operates at a channel activity of 1, downloading high-speed (153.6 kbps) bursts of data to each mobile in turn. Note that the specific mobile served by the supplemental channel at any time is not relevant to the link budget calculation due to the symmetric arrangement of the mobiles. Example forward link budget

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 31

RF engineering for data

Data link budgets Table 3-4

Example forward link budget for PCS 3G-1X 3-sector Melodizes with fixed 153.6 kbps packet data application

Line # Description Transmit Power calculations 5 Nominal available power at J4 point 6 Pilot Channel Power 7 Sync Channel Power 8 Paging Channel Power 9 Power available for the traffic Channel 10 Total Overhead 11 SCH rate 12 FCH rate 13 Required SCH Eb/Nt 14 Required FCH Eb/Nt 15 Number of FCHs per sector

Power

W

16 2.4 0.2 0.8 12.5 21.8 153.6 9.6 1.8 2.5 11

16 Number of SCHs per sector

W W W W W % kbps kbps

Power

Comments

42.0 33.8 23.8 29.3 41.0

dBm dBm dBm dBm dBm

51.9 39.8 2.5 4.0

dB dB dB dB

1

17 Overhead factor for FCH

1.75

18 Total number of active FCH power channels

19.3

19 Overhead factor for SCH 20 Total number of active SCH power channels

1 1.0

21 Mean Voice Activity Factor (VAF) for Fundamental 0.125 Channel 22 Mean Voice Activity Factor (VAF) for Supplemental 1 Channel 23 Average Traffic Channel Power for Fundamental Chan0.14 nel 24 Average Traffic Channel Power for Supplemental 9.9 Channel 25 Peak Traffic Channel Power per Fundamental Channel 1.1 26 Peak Traffic Channel Power per Supplemental Channel 9.9 27 Cell site Cable Loss 2.0 28 Cell site Transmit Antenna Gain 44.7 29 Fundamental Traffic Channel EIRP per user at full rate 24.6 30 Supplemental Traffic Channel EIRP per user at full rate 221.1 31 Total EIRP 358.2 32 Propagation loss 33 Maximum Path Loss 6.46E+12 34 Lognormal Fade Standard Deviation 6.3 35 Multiplier for fading standard deviation (e.g., 1.3 for 90th percentile) 36 Mobile RX Signal power Calculations 37 Mobile Receive Antenna Gain 38 Mobile Body Loss & Building Penetration Loss and Cable Loss

2.4 dB

0.0 dB

W

21.4 dBm

W

39.9 dBm

W W

30.4 39.9 3.0 16.5 43.9 53.4 55.5

W W W

Maximum power available Set at 15% of max. power Set at 10% of pilot power Set at 35.1% of pilot power 78.2% of total power C10 = 100*(1 - (c9/c5))

No SHO on SCH Total number of simultaneously active data users Total number of simultaneously active supplemental channels users Due to users being in 2 way and 3 way hand-off; from IS95B new handoff algorithm # of Fundamental channels supported by the transmitter # of Supplemental channels supported by the transmitter

dBm dBm dB dBi dBm dBm dBm

128.1 dB 8.0 dB

90% edge coverage; Assume no SHO for SCH

1.3

1.6 31.6

2.0 dBi 15.0 dB

...........................................................................................................................................................................................................................................................

3 - 32

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

39 Mobile Rx Fundamental Channel Signal power at full rate 40 Mobile Rx Supplemental Channel Signal power at full rate 41 Mobile Rx Total pwr from the Serving cell 42 Interference Power Calculations 43 Orthogonality Factor for Other users in Serving Cell

1.91E-13 W

-97.2 dBm

1.72E-12 W

-87.7 dBm

2.78E-12 W

-85.6 dBm

0.16

44 Standard Deviation of SCH Activity Factor 0.0 45 Standard Deviation of FCH Activity Factor 0.20 46 Ratio of mean other sector interference to same sector 1.8 power at cell edge 47 Other Cells Interference Power 5.06E-12 W 48 Thermal Noise Calculations 49 Mobile Noise Figure (F) 8 50 Thermal Noise Density (No = KT) 3.98E-21 51 Total thermal Noise power per Hz (NoF) 4.07E-20 52 Spreading bandwidth (W) 1.23E+06 Hz 53 Total thermal noise power (NoWF) 5.01E-14 W 54 External (intermod/spectrum clearance) interfer1.58E-15 W ence 55 Noise Floor and Other Cell Interference to the 5.33E-12 W Fundamental traffic channel 56 Noise Floor and Other Cell Interference to the Funda- 4.34E-18 W/HZ mental traffic channel per Hz 57 Noise Floor and Other Cell Interference to the Supple- 5.33E-12 W mental traffic channel 58 Noise Floor and Other Cell Interference to the Supple- 4.34E-18 W/Hz mental traffic channel per Hz 59 Mean of Other Cell to Serving Cell Interference Ratio 1.92 for SCH 60 Mean of Other Cell to Serving Cell Interference Ratio 0.49 for FCH 61 Standard Deviation of Other Cell to Serving Cell Inter0.53 ference Ratio for SCH 62 Standard Deviation of Other Cell to Serving Cell Inter0.53 ference Ratio for FCH 63 Aggregate Margin for SCH Interference Ratio and 1.32 Activity Factor 64 Aggregate Margin for FCH Interference Ratio and 1.32 Activity Factor 65 Adjustment for SCH due to Serving Cell Interference 1.05 and Orthogonality Factor 66 Adjustment for FCH due to Serving Cell Interference 1.32 and Orthogonality Factor 67 Bit Energy to Interference calculations 68 Fundamental Traffic Channel Bit Rate 9600 bps 69 Fundamental Channel Energy per bit at full rate 1.99E-17 W/Hz 70 Fundamental Traffic Channel Eb/(No+Io) 2.6 71 Supplemental Traffic Channel Bit Rate 153600.0 bps 72 Supplemental Channel Energy per bit at full rate 1.12E-17 W/Hz 73 Supplemental Traffic Channel Eb/(No+Io) 1.859258

-8.0 dB

2.6 dB

From same sector's other Walsh channels

for SCH

-83.0 dBm 9 -174.0 -163.9 60.9 -103.0 -118.0

dB dBm/Hz dBm/Hz dB dBm dBm

-82.7 dBm -143.6 dBm/Hz -82.7 dBm -143.6 dBm/Hz 2.8 dB -3.1 dB -2.8 dB -2.8 dB 1.2 dB 1.2 dB 0.2 dB 1.2 dB

39.8 -137.0 4.19 51.9 -139.5 2.69

dB dBm/Hz dB dBm/Hz dB

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 33

RF engineering for data

Data link budgets

In this example, a 128 dB path loss (line 33) has been analyzed. This path loss corresponds to the reverse link budget for a 153.6 kbps reverse supplemental link under the conditions of: •

0 dB head loss

•

16.5 dBi antenna gain

•

3.0 dB cable loss

•

4.0 dB noise figure

•

0.8 dB Eb/Nt requirement

•

10.3 dB fade margin

•

15 dB building penetration margin

•

5.5 dB interference margin.

Accordingly, the spreadsheet indicates that the fundamental and supplemental channel numbers listed above can be supported within this footprint since the average forward link Eb/Nt requirement is met. (This requirement is the right-hand side of the condition Equation 3-7). Since the analysis employs very conservative assumptions (see above), the actual number could be larger. The average forward link Eb/Nt requirements for all data rates are tabulated below for reference. These values can be used in symmetric forward link analysis (embedded or concentric) for any rate chosen. Note that the values employed can be used for mobile or fixed applications; in the latter case; improvements in power control relative to 3G suggest that any Eb/Nt advantage for a fixed user may be negligible. Table 3-5

Average Eb/Nt requirements for forward link budget data channel analysis (8 kbps RC3)

Channel Data Rate (kbps) Average Eb/Nt Requirement (dB)

TBD

TBD

TBD

TBD

TBD

The symmetric link budget analysis has the advantages of computational simplicity; however, the approach may be overly conservative. This conservatism is especially evident in the embedded configuration, since all high-rate supplemental channels must be supported at the physical cell boundary without allowance for anchor transfer gain. In contrast, the Monte Carlo analysis allows for random distribution of the data subscribers but adds complexity to the planning analysis. This approach is further discussed, below. ...........................................................................................................................................................................................................................................................

3 - 34

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Data link budgets

Monte carlo forward link analysis

This work is in progress.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 35

Resource management: RF scheduling

RF engineering for data

Resource management: RF scheduling ............................................................................................................................................................................................................................................................

Introduction

In the above sections, the essential strategy of the network in managing RF resources has been used in order to assess performance. Such strategies include the sharing of high-rate supplemental channels and the exploitation of “bursty” subscriber behavior (e.g., idle think time between web page downloads) to maximize the number of data sessions served. In the followings, we consider the aspects of resource management in greater detail. Efficient radio resource management is critical for the success of 3G wireless data in the multi-user environment. The CDMA2000-1X standard defines physical channels with transmission rates of up to 153.6 kbps, more than a 10-fold increase compared with IS-95A. However, the ultimate end-user data experience depends to a large degree not only on data rate capability, but also on transmission latency, resource availability, and service coverage. Complex interactions are expected within the radio resource management function due to competition between multiple user demands and due to the selfregulating delay-sensitive nature of upper layer data protocols such as TCP. Additionally, resource management has to support both voice and data services on the same frequency carrier without compromising voice quality achieved in 2G systems.

Scheduling algorithm

Fundamental Channel (FCH) assignment and release

A fundamental channel (FCH) is mandatory for the data call and is needed for carrying signaling and control information. This channel must be established for each user before a high-rate connection can start. The FCH is set up in both directions, forward and reverse, and it uses the same modulation and coding for data and voice. Lucent 3G-1X implementation supports data FCH using Radio Configurations 3 and 4 (RC3 and RC4) on the Forward link and RC3 on the Reverse link. As for voice service, the data FCH reduces its rate according to the data source activity in order to reduce co-channel interference to other users. In other words, the FCH reverts to the 1/8 rate when there is no data or signaling to transmit. Figure 3-8 (reproduced here for convenience) depicts an example of how data (e.g., web pages) can be transmitted over the 1X air interface in the Lucent implementation. In the following sections, the details of channel management are discussed in greater detail. ...........................................................................................................................................................................................................................................................

3 - 36

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Resource management: RF scheduling

Web Browsing Session Active

Dormant

Active

Dormant

Supplemental Channel Bursts

First Data Arrived at IWF

153.6 SCH

153.6 SCH 76.8 SCH

76.8 SCH

38.4

9.6 kbps FCH

9.6 FCH

Start Reading Web Page

Mouse Click

Access Time

Download Network Time Delay Queuing Delay ( incl. SCH Setup Delay) Figure 3-8

76.8 SCH

Mouse Click

Fundamental Channels

Dormancy Timer Duration "Think" Time

Data traffic model for web browsing application with 3G1X packet data

FCH assignment

The Fundamental channel is set up every time a data call enters an Active state. This occurs in the beginning of the call, or when the user returns from a Dormant state. The Dormant-to-Active transition may happen due to both mobile origination and termination. The data FCH is established in the same way as the voice traffic channel after exchanging signaling messages on Paging and Access channels. FCHs are assigned to users on a first come, first served basis. User admission algorithms are designed to maximize the number of simultaneous active users while protecting the system from overload. To achieve this goal, the system continuously monitors performance and resource availability and takes appropriate corrective measures when resource utilization becomes sub-optimal. A set of admission thresholds is designed to provide acceptable level of service to all existing and incoming users. A decision whether to establish the FCH is based, among other things, on current power (forward) and interference (reverse) loading, frame error rate performance,

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 37

Resource management: RF scheduling

RF engineering for data

availability of base-station and back-haul hardware resources, etc. In the first implementation of Lucent 3G-1X, the system treats FCH assignment for voice and data calls in the same way. FCH assignment and resource allocation for both voice and data calls takes precedence over the Supplemental Channel allocation to ensure coverage for signaling and minimum rate data traffic over the same cell area as voice. If resources needed for setting up an FCH are unavailable due to their utilization by an existing SCH, the system releases the SCH to make room for the incoming FCH (early SCH release is discussed in "Early F-SCH termination" section on Page 3-40). The probability of this scenario can be made low by providing sufficient margins when allocating SCH resources such that in the majority of cases, there are enough resources to admit a new FCH during SCH operation without triggering the early SCH termination. The FCH is used for transmitting signaling information and may also be used for transmitting data traffic. For example, if the Supplemental channel is not active, the data traffic is transmitted on the Fundamental channel. On the forward link, the system prefers transmission of new traffic data on the high-rate Supplemental channel. However, the retransmit data may be sent over the Fundamental channel even if the Supplemental channel is active. On the reverse link, it is left up to the mobile station to decide whether to send data over FCH and SCH simultaneously. Active-to-dormant transition and FCH release

Data users go into a Dormant state after a period of inactivity. When Active-to-Dormant transition occurs, the user loses any air-interface connection with the base station. However, the PPP connection is maintained throughout the transition. The transition is triggered by the expiration of the Dormancy timer. The value of the timer setting is the same for all users and can be adjusted by the operator via translation. Forward link supplemental channel (F-SCH) assignment and release

Forward link SCH (F-SCH) can be established using the following physical transmission rates: 19.2 kbps, 38.4 kbps, 76.8 kbps, and 153.6 kbps of Radio Configurations 3 and 4. The duration of F-SCH allocation can span multiple 20-ms frames, depending on the amount of the data buffered for transmission. F-SCH transmission may be continued beyond this initial duration if more data is buffered, and if resource availability conditions permit. ...........................................................................................................................................................................................................................................................

3 - 38

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Resource management: RF scheduling F-SCH rate and duration allocation

There are a number of factors used by the scheduling algorithm when choosing transmission rate and duration of the F-SCH assignment. Some of the most important factors considered by the algorithm are listed below: •

Fraction of amplifier power required by the supplemental channel. This metric is determined as a result of RF measurements, service negotiations and translation settings in the following areas: •

•

Mobile RF environment including fast and shadow fading effects • Propagation path loss between the mobile and the base station • Interference level experienced by the mobile • Radio Configuration used for the supplemental channel • Turbo coding support • Target Frame Error Rate for each data rate Fraction of amplifier power consumed by other voice and data users and a corresponding power fraction available for establishing the supplemental channel (power computation provides sufficient margins for ensuring a low probability of overload during SCH operation and therefore a low probability of early SCH termination)

•

Channel element, back-haul, Walsh function and other hardware and system resource availability at the serving base station that can be assigned to the supplemental channel

•

Amount of data buffered for transmission to the mobile

•

Scheduling policy that prevents monopolizing system resources by one or a small group of users for an extended period of time.

The system makes the best effort to satisfy above constraints/ requirements when assigning a forward SCH rate. The duration of SCH assignment is determined by the amount of data in the transmit buffer, the handoff state of the user, and the transmission rate resulting from the algorithm described above. F-SCH continuation

If, at the end of current F-SCH transmission, the user still has data in the transmit buffer, the F-SCH burst may be continued using the same rate. This happens only if the system determines that the available ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 39

Resource management: RF scheduling

RF engineering for data

resources are sufficient to proceed with such continuation. The duration of continuation burst is determined using the same calculation as for initial burst transmission. The number of consecutive continuations is limited to prevent one, or a small number of users, from monopolizing system resources for a long period of time. The number of allowed continuations is a translation parameter. This limit takes effect if there is a contention for SCH resources from other users. Otherwise, continuation beyond the limit is allowed. Transmission rate of the supplemental channel may be increased if, after certain number of continuations, the system determines that a significantly larger amount of resources have become available for the data user. Normal F-SCH termination

The Supplemental channel is terminated normally if transmission on this channel spans exactly the duration assigned to the mobile in the Extended Supplemental Channel Assignment Message (E-SCAM) before the start of the F-SCH burst. Early F-SCH termination

F-SCH may be terminated earlier than specified in the E-SCAM. Early termination may be caused by power overload reached during F-SCH operation due to power control operation or due to the increased loading and/or interference. Early termination may also occur due to a need to free up base station resources to admit new of handoff F-FCH channel (data and voice). SCH resource management provides sufficient margins to make the probability of early termination low. In the event that early termination does occur, the user may be assigned a new SCH at a lower rate based on the re-evaluation of resources available at the time of termination. Note that F-SCH operation will not be terminated early if soft handoff adds or soft handoff drops occur, or some combination of them. This is because the F-SCH resides on only one leg and is not effected by changes in the mobile’s active set unless the serving cell itself drops.

...........................................................................................................................................................................................................................................................

3 - 40

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

RF engineering for data

Resource management: RF scheduling

Reverse link supplemental channel (R-SCH) assignment and release

Reverse link SCH (R-SCH) may be established using the following physical transmission rates: 19.2 kbps, 38.4 kbps, 76.8 kbps, and 153.6 kbps of Radio Configurations 3. The duration of R-SCH allocation can span one or more 20-ms frames, depending on the amount of data in the mobile’s transmit buffer. R-SCH rate Allocation

Unlike the single-leg F-SCH operation, the operation of the R-SCH will be on all legs of the call. Therefore, the rate of the burst is determined by the minimum rate that can be supported among all legs. Some of the most important factors considered by the reverse supplemental channel rate allocation algorithm are listed below: •

Maximum mobile transmit power available for R-SCH

•

Additional loading that would be produced by the supplemental channel after it is assigned. This projected loading increase is determined as a result of RF measurements, service negotiations and translation settings in the following areas: –

Mobile RF environment including fast and shadow fading effects

–

Propagation path loss between the mobile and all sectors in the Active set

–

Target Frame Error Rate for each data rate

–

Turbo coding support

•

Current loading and interference levels on all handoff sectors (Active set) of the call and corresponding loading and interference budgets available to support a new SCH (interference budget computation provides sufficient margins for ensuring a low probability of overload during SCH operation and therefore a low probability of early SCH termination)

•

Channel element and back-haul availability at all handoff sectors of the call

•

Amount of data buffered for transmission by the mobile.

System makes the best effort to satisfy the above constraints/requirements when assigning a Reverse SCH rate.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 41

Resource management: RF scheduling

RF engineering for data

R-SCH normal termination

Normal termination is triggered when the mobile station completes transmission of data in its buffer. Mobile station requests the R-SCH release by sending a Supplemental Channel Request Message (SCRM), specifying that zero amount of data needs to be transmitted. R-SCH early termination

R-SCH could be terminated by the system prior to mobile requesting such termination. This early termination may be caused by reverse link interference overload reached during R-SCH operation due to power control operation or due to the increased loading. Early termination may also occur due to a need also to free up base station resources to admit new or handoff R-FCH channels (data and voice). SCH resource management provides sufficient margins to make the probability of early termination low. In the event when the early termination does occur, the user may be assigned a new SCH at a lower rate based on the re-evaluation of resources available at the time of termination. Unlike in the single-leg only F-SCH case, early R-SCH termination may also happen as a result of handoff activity, e.g., adding new handoff legs. Load Balancing Overview

System supports carrier load balancing on origination. With this algorithm, the system directs originating users to a different carrier if the load of initial carrier is larger than the load of that new carrier by more than a translation-defined delta load. Handoff calls are admitted on the carrier independently of carrier load. Dual 3G/2G deployments

There is also a mechanism providing load balancing in dual 3G/2G deployments. Specifically, the system may be configured with the translation to give preference to 3G carriers for originating 3G mobiles, and to 2G carriers for originating 2G mobiles. A degree of allowed load imbalance between 2G and 3G carriers resulting from this approach is limited through the translation parameter that can be adjusted by system operator. This parameter specifies the maximum load imbalance between carriers beyond which 2G mobiles are directed to a 3G carrier, or 3G mobiles capable of 2G are directed to a 2G carrier, to mitigate the imbalance. ...........................................................................................................................................................................................................................................................

3 - 42

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Resource management: RF scheduling

RF engineering for data

Additional load delta parameter is provided for handling load balancing for data calls. This translation parameter specifies the additional load imbalance between carriers beyond which 3G data mobiles are directed to a 2G carrier. Conclusions

RF resource scheduling is an essential part of the Lucent CDMA2000-1X data system. It consists of a set of call admission, load balancing, channel and rate assignment algorithms designed to optimize resource utilization, system capacity and performance. Future enhancements will provide even greater flexibility in carrier scheduling, non-assured QoS support, and throughput optimization.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

3 - 43

Resource management: RF scheduling

RF engineering for data

...........................................................................................................................................................................................................................................................

3 - 44

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

4

System deployment

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter describes the deployment issues, with focus on transition from 2G to 3G-1X. Introduction

4-2

Spectrum use: Carrier assignments and guard band

4-4

Cellular band General considerations Frequency planning for systems with 3G-1X and AMPS PCS band Preferred channels 2G/3G-1X spatial and frequency design

4-4 4-4 4-7 4-8 4-10 4-11

Coverage (spatial) design: Overlay and greenfield Frequency design Estimating capacity: Mix of 3G-1X voice and 2G voice Planning: Mixed vs. dedicated carriers for 3G-1X Mixed 3G-1X voice/data capacity and coverage

4-11 4-13 4-13 4-15 4-19

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 1

System deployment

Introduction

Introduction ............................................................................................................................................................................................................................................................

Implementation of 3G-1X is straightforward. The key points include the following: •

Most existing Lucent 2G systems may be upgraded to 3G via addition of the 3G channel card and the appropriate software release

•

The 3G-1X channel card (CCU-32) is dual-mode, supporting both 3G-1X and 2G calls. The card automatically makes this decision, depending upon the nature of the mobile trying to originate a call.

•

Since the 3G-1X and 2G voice footprints are comparable, a 1:1 upgrade provides 3G-1X coverage that is better than or equal to that of 2G voice coverage

•

3G-1X may be implemented in spectrum cleared for this purpose (dedicated 3G carrier), or within spectrum already serving 2G subscribers (shared 2G/3G carrier). In the latter case, the net voice Erlang capacity is intermediate between that of a 3G-only carrier and a 2G-only carrier.

•

Mobile standards specify that the 3G-1X mobile be dual-mode, supporting both 2G and 3G-1X calls

•

3G-1X need not be deployed ubiquitously. The 3G-1X infrastructure supports 3G to 2G handoffs for mobiles exiting a 2G/3G area into a 2G-only area.

Given the above, an existing 2G system can be gradually upgraded to 3G-1X simply by implementing the appropriate software release, seeding the subscriber population with 3G-1X mobiles, and deploying the appropriate channel cards. The last may be done selectively (limited 3G/2G area) or ubiquitously (3G throughout). The 3G-1X mobiles may operate in the same spectrum as an existing 2G carrier. Since both mobiles and cards are dual-mode, exact knowledge of the proportion of 3G-1X voice users is not required in order to properly provision the cell site. Alternatively, the 3G-1X mobiles could be deployed within a carrier dedicated to that purpose. 3G-1X can also be implemented as a greenfield design, i.e., within an area that does not already possess 2G service. In this case, the processes of spectrum clearance, system design, and cell provisioning are analogous to that of 2G. Furthermore, if designed for full 3G voice capacity and ubiquitous 3G voice (as opposed to high-speed data) ...........................................................................................................................................................................................................................................................

4 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

Introduction

coverage, the cell count should be virtually identical to that of a 2G design. This comparison is useful for service providers that desire to specify the coverage design prior to deciding whether to initially implement 2G or 3G service. In the following sections, we provide further detail on deployment issues, such as carrier assignments, underlay/overlay considerations, and the mix of 2G and 3G within available radio spectrum. The mix of 3G voice and data is also considered.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 3

Spectrum use: Carrier assignments and guard band

System deployment

Spectrum use: Carrier assignments and guard band ............................................................................................................................................................................................................................................................

Carrier assignments and guard band remain the same as in 2G. The recommendations for carrier assignments are provided for two band classes: Band Class 0 (i.e., the cellular band) and Band Class 1 (i.e., the PCS band) defined by the IS-2000. For detailed information, please refer to Lucent documents 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines, and 401-703-201, PCS CDMA RF Engineering Guidelines. Cellular band

This section will address frequency planning considerations in dual mode systems, which support AMPS and IS-95 standards as well as 3G-1X. This section assumes that the reader is familiar with the frequency planning considerations and techniques used in AMPS, and is not intended to be a tutorial in basic frequency planning. It will address only those frequency-planning issues that are the result of dual mode system operations. General considerations

Table 4-1 lists the five bands of 832 channels available to the A- and BBand service providers. Valid channels for 3G-1X assignments are designated by “CDMA” in the “Valid CDMA Frequency Assignments” column, and invalid assignments by “//////////”. This information is taken from the IS-2000 and provided here for convenience. Note that Band Class 0 is also referred to as the cellular band in North America.

...........................................................................................................................................................................................................................................................

4 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

Spectrum use: Carrier assignments and guard band Table 4-1

System Band

Valid CDMA Frequency Assignments

/////////////////

AMPS and 3G-1X channel numbers and corresponding frequencies for Band Class 0

Number of Analog Channels

22

A'' (1 MHz) A'' (1 MHz)

CDMA

11

A (10 MHz)

CDMA

311

A (10 MHz) B (10 MHz)

/////////////

//////////////

22

22

B (10 MHz)

CDMA

289

B (10 MHz)

//////////////

22

A' (1.5 MHz)

/////////////

22

A' (1.5 MHz)

CDMA

6

A' (1.5 MHz)

/////////////

22

B' (2.5 MHz) B' (2.5 MHz) B' (2.5 MHz)

/////////////

CDMA

/////////////

22

39

22

AMPS/CDMA Channel Number

Transmitter Frequency Assignment (MHz) Mobile

Base

991

824.040

869.040

1012

824.670

869.670

1013

824.700

869.700

1023

825.000

870.000

1

825.030

870.030

311

834.330

879.330

312

834.360

879.360

333

834.990

879.990

334

835.020

880.020

355

835.650

880.650

356

835.680

880.680

644

844.320

889.320

645

844.350

889.350

666

844.980

889.980

667

845.010

890.010

688

845.640

890.640

689

845.670

890.670

694

845.820

890.820

695

845.850

890.850

716

846.480

891.480

717

846.510

891.510

738

847.140

892.140

739

847.170

892.170

777

848.310

893.310

778

848.340

893.340

799

848.970

893.970

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 5

Spectrum use: Carrier assignments and guard band

System deployment

The non-allowed bands of channels are 22 AMPS channels in width and are dictated primarily by the 1.23 MHz bandwidth (41 AMPS channels) of the 3G-1X channel. These valid 3G-1X assignments do not take into account practical considerations such as guard-band needs and/or the channel needs for AMPS in dual mode systems. The subsections below discuss the channel needs for AMPS and 3G-1X that should also be considered when allocating the spectrum in dual mode systems. Because of the need for guard bands and/or channels in dual mode systems, it should be understood that allocations of spectrum channels to a specific standard should be done as much as possible in terms of contiguous channels/bands for each (AMPS/3G-1X) technology. By using contiguous channels/bands for one standard, there is only a single guard band penalty for the overall spectrum allocation given to the standard in question. For example, if an A-Band, dual mode, 3G-1X application required two 3G-1X channels, a good first 3G-1X channel selection would be channel 283. In the case of a dual mode (AMPS/3G1X) system, this is the highest available channel in the 10 MHz A-Band that could be selected without concern for interference in the A-Band setup channels (313-333). This channel selection already provides a 0.27 MHz guard band of channels between the nominal 1.23 MHz 3G-1X channel band and the AMPS setup channels (313-333) required for the A-Band service provider. The logical choice for the second 3G-1X channel would be channel 242, which that is 41 channels away from 283 for a carrier frequency separation of 1.23 MHz. Any selection resulting in a carrier frequency separation of less than 41 channels would result in the two 3G-1X carriers being separated by less than the nominal 1.23 MHz 3G-1X channel bandwidth and would cause excessive interference between the two carrier bands. Using a separation of greater than 41 channels results in inefficient use of the spectrum. Two preferred channel assignments specified in the IS-2000 are: •

Primary Setup Channel - Channels 283 and 384 for A- and B-Band, respectively

•

Secondary Setup Channel - Channels 691 and 777 for A- and B-Band, respectively.

If the 3G-1X mobile supports the preferred roaming list feature defined by the IS-683, then any valid channel assignment can be used by the mobile station for initial acquisition. Otherwise, an operational CDMA system must use at least one of the two channels, primary and/or ...........................................................................................................................................................................................................................................................

4 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

Spectrum use: Carrier assignments and guard band

secondary in every CDMA cell, and therefore, the selection for CDMA frequencies in any start up system requiring only one CDMA channel per cell is quickly narrowed to one of these two preferred channels. Frequency planning for systems with 3G-1X and AMPS

It is recommended that for the 3G-1X and AMPS operating in the same cellular band (A Band or B Band), the guard band of 270 kHz be implemented on both sides of the consecutive 3G-1X carriers and no guard band between the 3G-1X carriers be required. For the derivation of the 270 kHz guard band, please refer to Lucent document 401-614012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines. Table 4-2 and Table 4-3 below show frequency assignments for dual mode AMPS and 3G-1X operations in the A- and B-Band spectrums. These assignments are given for various numbers of 1.23 MHz bandwidth 3G-1X channels. As highlighted by the asterisks (*) in the AMPS columns, the frequency assignments and number of available channels includes the 21 setup channels. Table 4-2

Recommended A-Band 3G-1X center frequency assignments for Band Class 0

Number of CDMA Channels

CDMA Center Frequency Assignments

Number of AMPS Channels*

AMPS Channel Assignments*

1

283

356

1-252, 313-333, 667-716, 991-1023

2

242, 283

315

1-211, 313-333, 667-716, 991-1023

3

201, 242, 283

274

1-170, 313-333, 667-716, 991-1023

4

160, 201, 242, 283

233

1-129, 313-333, 667-716, 991-1023

5

119, 160, 210, 242, 283

192

1-88, 313-333, 667-716, 991-1023

6

78, 119, 160, 201, 242, 283

151

1-47, 313-333, 667-716, 991-1023

7

37, 78, 119, 160, 201, 242, 283

110

1-6, 313-333, 667-716, 991-1023

8

691, 37, 78, 119, 160, 201, 242, 283

60

1-6, 313-333, 991-1023

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 7

System deployment

Spectrum use: Carrier assignments and guard band Table 4-3

Recommended B-Band 3G-1X center frequency assignments for Band Class 0

Number of CDMA Channels

CDMA Center Frequency Assignments

Number of AMPS Channels*

AMPS Channel Assignments*

1

384

356

334-354, 415-666, 717-799

2

384, 425

315

334-354, 456-666, 717-799

3

384, 425, 466

274

334-354, 497-666, 717-999

4

384, 425, 466, 507

233

334-354, 538-666, 717-999

5

384, 425, 466, 507, 548

192

334-354, 579-666, 717-799

6

384, 425, 466, 507, 548, 589

151

334-354, 620-666, 717-799

7

384, 425, 466, 507, 548, 589, 630

110

334-354, 661-666, 717-799

8

384, 425, 466, 507, 548, 589, 630, 777

57

334-354, 661-666, 717-746

In both the A- and B-Band cases, the Secondary Setup Channel was the last 3G-1X channel added. The reason for this is that this channel incurs the greatest AMPS channel loss because it requires its own guard band penalty in addition to the 0.54 MHz guard band penalty for the other 7 CDMA channels. If added setup channel capacity is needed, this channel may have to be implemented sooner than assumed here. PCS band

Although the 3G-1X channel numbering algorithm with 50 KHz channel spacing implies the availability of 1200 of 50 kHz for 3G-1X carriers, not all 1200 are actually usable. Table 4-4 indicates the availability of the channels by classifying them as valid (usable) channels, conditionally valid, or not valid. The designation of channels 0-24 and 1176-1199 as being not valid eliminates the possibility of interference between PCS systems and the services allocated to the spectrum above, below, and between the two 60 MHz spectrum allocations comprising the PCS spectrum. The channels specified as conditionally valid are the 25 lowest (except for Block A) and the 25 highest (except for Block C) channels in each block. These channels are valid only under the condition that the service provider also owns the adjacent block of spectrum. Looking at it another way, all channels are valid for use as 3G-1X carriers except for the 25 lowest channels and the 25 highest channels in each block. Thus, there are 251 channels unconditionally available

...........................................................................................................................................................................................................................................................

4 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

Spectrum use: Carrier assignments and guard band

(i.e., “Valid”) for designation as carrier frequencies in each of Frequency Blocks A, B, and C, and there are 51 unconditionally available channels for each of Blocks D, E, and F. If a service provider were to obtain licenses in two adjacent blocks, then an additional 50 channels would become available from the conditionally available channels. Note that Band Class 1 is also referred to as the PCS band in North America. Table 4-4

3G-1X channel allocation availability for Band Class 1

Frequency

CDMA Frequency

CDMA

Transmit Frequency (MHz)

Block

Assignment Validity

Channel Number

Personal Station

Base station

A

Not Valid

0-24

1850.000-1851.200

1930.000-1931.200

(15 MHz)

Valid

25-275

1851.250-1863.750

1931.250-1943.750

Conditionally Valid

276-299

1863.800-1864.950

1943.800-1944.950

D

Conditionally Valid

300-324

1865.000-1866.200

1945.000-1946.200

(5 MHz)

Valid

325-375

1866.250-1868.750

1945.600-1948.750

Conditionally Valid

376-399

1868.800-1869.950

1948.800-1949.950

B

Conditionally Valid

400-424

1870.000-1871.200

1950.000-1951.200

(15 MHz)

Valid

425-675

1871.250-1883.750

1951.250-1963.750

Conditionally Valid

676-699

1883.800-1884.950

1963.800-1964.950

E

Conditionally Valid

700-724

1885.000-1886.200

1965.000-1966.200

(5 MHz)

Valid

725-775

1886.250-1888.750

1966.250-1968.750

Conditionally Valid

776-799

1888.800-1889.950

1968.800-1969.950

F

Conditionally Valid

800-824

1890.000-1891.200

1970.000-1971.200

(5 MHz)

Valid

825-875

1891.250-1893.750

1971.250-1973.750

Conditionally Valid

876-899

1893.800-1894.950

1973.800-1974.950

C

Conditionally Valid

900-924

1895.000-1896.200

1975.000-1976.200

(15 MHz)

Valid

925-1175

1896.250-1908.750

1976.250-1988.750

Not Valid

1176-1199

1908.800-1909.950

1988.800-1989.950

Not all of the valid and conditionally valid channels can be used simultaneously as carriers in a given system. Once a channel number has been specified for use as the first carrier in a system, there are minimum spacing rules for carriers in use, which limit how close the new carrier can be above or below the previously existing carrier(s). While the classification of channels as valid and conditionally valid is ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 9

System deployment

Spectrum use: Carrier assignments and guard band

by FCC decree, the minimum spacing between active carriers is determined by 3G-1X technology considerations. Generally, the channels are specified as dictated by the minimum carrier spacing of 25 3G-1X channels, which is consistent with the nominal 1.25 MHz bandwidth for 3G-1X. Preferred channels

The preceding subsection specified the channels which are valid, or at least conditionally valid, carrier frequencies that the service provider can specify for use in the system's frequency plan. The selection of these frequencies might be dictated by issues dealing with inter-system or intra-system interference. If these issues are not significant factors in the system performance, the number of channels that the service provider might consider for carrier frequencies can be reduced significantly to the list of “preferred channels” in the table below. These are the channel numbers that a personal station will “scan” when looking for service. Thus a system must use at least one (or more) of these carriers at each site in the system if the sites are to be capable or providing (CDMA) access to the system. Table 4-5

Preferred CDMA channels for Band Class 1

Frequency Block

Preferred Channel Numbers

A

25, 50, 75, 100, 125, 150, 175, 200, 225, 250, 275

D

325, 350, 375

B

425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675

E

725, 750, 775

F

825, 850, 875

C

925, 950, 975, 1000, 1025, 1050, 1075, 1100, 1125, 1150, 1175

Conditionally valid channels 300, 400, 700, 800, and 900 are excluded from the above list because they can only be used if the service provider has licenses for both the frequency block containing the channel and the immediately adjacent frequency block, e.g., Channel 300 is a likely carrier channel if the service provider has licenses for both Blocks A and D. If conditionally valid channels are used, they should be used for traffic only and not access. For details about intra-system and inter-system frequency planning, please refer to Lucent document 401-703-201, PCS CDMA RF Engineering Guidelines.

...........................................................................................................................................................................................................................................................

4 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

2G/3G-1X spatial and frequency design

System deployment

2G/3G-1X spatial and frequency design ............................................................................................................................................................................................................................................................

Coverage (spatial) design: overlay and greenfield

A PCS Modcell reverse link budget example for 3G-1X 9.6 kbps voice application was presented in Chapter 2, "Link budget" section on Page 2-14. This example indicates a fundamental governing principle in deployment planning for 3G-1X, that the 3G-1X voice coverage (footprint) is (slightly) better than or equal to the 2G footprint. Accordingly, a new or “greenfield” 3G deployment will have essentially the same cell count as a greenfield 2G deployment. In addition, upgrade or migration of a 2G network to a 3G network can be accomplished through a 1:1 overlay of 3G on 2G, i.e., 3G voice coverage is obtained by upgrading each 2G cell to 3G functionality. The resulting 3G coverage will match or slightly better that of the underlying 2G network. This comparison applies to the situation where 3G-1X and 2G are each fully loaded. A lighter design loading on 3G-1X will expand the voice coverage at the expense of cell capacity. This design trade-off is identical to the coverage-capacity trade-off that exists in 2G systems. Since full 3G-1X loading is required to reach the full 3G-1X voice capacity (see Table 2-1, "Air interface capacity" on Page 2-9), we presume a fully loaded system in the following discussions. Link budgets for the 19.2 kbps - 153.6 kbps packet data applications have also been presented in Chapter 3, "Data link budgets" section on Page 3-19. These examples show that the radio coverage (footprint) for 3G high-rate packet data can be considerably less than that of (2G or 3G) voice. This difference is fundamental, and a straightforward consequence of the higher rates at which the supplemental channel must operate. The coverage difference between data and voice is a key issue in design. We consider two scenarios, overlay and greenfield deployment, below. Consider a 2G system upgraded to (overlaid by) 3G-1X. The physical outer perimeter of the cell is determined by the existing 2G design. The 3G-1X voice coverage extends to this perimeter. Since the link budget comparison indicates that the voice system supports a greater maximum path loss than the 3G-1X high-rate packet data, the high-rate data service will be available only within an inner circle of cell coverage. In this case, the supportable packet data rate for a call originated within the inner circle will dynamically reduce when the

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 11

System deployment

2G/3G-1X spatial and frequency design

mobile moves closer to the cell edge. This reduction will be controlled by the radio resource management algorithms (see Chapter 3, "Resource management: RF scheduling" section on Page 3-36), which assign data rates based on reported RF conditions as well as other factors, e.g., mobile history. Similarly, the data rate will dynamically increase as the mobile moves closer to the cell center. Now consider the scenario where the overlaid 3G-1X system must provide an ubiquitous coverage for the high-rate 3G-1X data. In this case, the link budget based on the high-rate supplemental channel is used for 3G cell layout since the design coverage of the high-rate data channel must extend out to the 3G cell edge. The footprint of these cells is modest compared to a voice footprint. A 1:1 overlay would not be feasible since the high data rate could not be supported at all locations between the cells. Additional 3G cells must be added to obtain ubiquitous high-rate data coverage. The overlay would increase from 1:1 to N:1 (i.e., N 3G cells required for each 2G cell). The N:1 restriction could be relaxed under several conditions. These include scenarios where: •

High-rate data subscribers possess an additional advantage, such as a directional antenna, to compensate for the lack of coverage. This advantage must be symmetric, i.e., applicable to both link directions; for example, the provision of higher mobile transmit power to the high-rate subscribers would not be effective since this change would not provide a forward link benefit as well.

•

The underlying 2G system is not coverage but capacity-driven. In urban or dense urban areas where the cell count is driven by the capacity, the actual path loss between a 2G mobile and the serving base station in the existing 2G network could be less than the maximum allowable value dictated by the 3G link budget. Under such a circumstance, the 1-to-1 overlay of CDMA2000 on the IS95 may still be a feasible migration.

•

The design restriction of extending high-rate data coverage to the 3G cell edge is removed. If voice rate coverage to the 3G cell edge is acceptable, then a 1:1 overlay becomes feasible (see above). In this case, mobile data rates would be dynamically adjusted, depending upon their location within the coverage area.

Further means for extending data coverage relative to voice are discussed in “Mixed 3G-1X voice/data capacity and coverage” section below.

...........................................................................................................................................................................................................................................................

4 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

2G/3G-1X spatial and frequency design

The issue of achieving comparable coverage at high data rates is obviated in greenfield deployments, since there is no underlying network for comparison. In these scenarios, the cell design is driven by selecting the data rate required out to the cell edge and then using the appropriate link budget in design. Selection of a very high data rate decreases the cell footprint and increases the design cell count considerably. Design alternatives employing a more modest data rate at cell edge may be more cost-effective, especially if the coverage of anticipated high-rate users is enhanced by the use of subscriber directional antennas. Frequency design

The implementation of 3G-1X within available radio spectrum offers a rich array of possibilities. 3G-1X can be deployed as 1.23 MHz wideband carriers within spectrum cleared for this purpose. Alternatively, 3G-1X can be deployed within an existing 2G carrier, yielding a net per-carrier capacity that lies between that achieved by 3G-1X alone and that achieved by 2G alone. Decisions regarding specific implementation paths depend upon several factors, including the availability of radio spectrum, the prediction (and accuracy) of voice and data demands, and the priority placed upon obtaining maximum air interface capacities and maximum channel element efficiency. Some insight into these factors is supplied in the discussions below. Estimating capacity: Mix of 3G-1X voice and 2G voice

We consider a scenario where the anticipated demand is a known mix of 3G-1X traffic and 2G traffic. For simplicity, we examine the case where all 3G-1X traffic is voice only; the extension of concepts to include data traffic as well is straightforward, although computationally more difficult. In scenarios where 2G and 3G are implemented as distinct carriers, the total Erlang capacity per sector is readily computed as an appropriately weighted sum of the two. For example, let: E3G = Total voice Erlangs per carrier per sector for 3G-1X E2G = Total voice Erlangs per carrier per sector for 2G N2G = Total number of 2G carriers per sector N3G = Total number of 3G carriers per sector. Then, the total Erlangs per sector is readily computed as:

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 13

System deployment

2G/3G-1X spatial and frequency design

Etotal = N 2G E2G + N 3G E3G Equation 4-1: Total Erlang calculation for 2G-3G mix

Equation 4-1 can be used in planning situations where the demand of 3G and 2G traffic is known. In these situations, the best fit of the integer numbers N2G and N3G are obtained to ensure that the demand is met. In scenarios where 2G and 3G are mixed within each carrier, the total Erlangs are best determined by simulation but can be approximated in the following manner. The total number of 2G voice Erlangs is an upper bound that cannot be exceeded when the subscriber population consists of 2G users alone. Trivially: Etotal ≤ E2G Equation 4-2

Let fraction x of the total Erlangs be 2G, and fraction (1-x) of the total Erlangs be 3G. Assume that the Erlang values E2G and E3G that can be achieved by each population alone are proportional to the total interference that can be tolerated. The equivalent 2G Erlangs generated by each 3G user is therefore the 3G Erlangs scaled by the ratio E2G/E3G. For example, a 3G user generates about half the interference as a 2G user; accordingly, the 3G usage must be scaled by a factor of ½ in totaling equivalent 2G usage. The total 2G Erlangs can therefore be computed and limited by the upper bound E2G:

xEtotal + (1 − x) Etotal (

E2G ) ≤ E2G E3G

Equation 4-3

In Equation 4-3, the first term on the left hand side is the number of 2G Erlangs, which is a fraction x of the total. The second term is the number of 3G Erlangs (a fraction 1-x of the total) scaled to an equivalent number of 2G Erlangs. The sum of (equivalent) 2G Erlangs is then limited to the same upper bound as a population consisting entirely of 2G users. Equation 4-3 can be solved for Etotal:

...........................................................................................................................................................................................................................................................

4 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

System deployment

2G/3G-1X spatial and frequency design

Etotal =

1 x / E2G + (1 − x) / E3G

Equation 4-4: Total Erlang capacity based on 2G & 3G capacities

The result Equation 4-4 is a planning approximation, and requires a modification in the interpretation of E3G and E2G in order to be employed. Specifically, in order to convert Erlangs from one population into equivalent Erlangs of another, the values of E3G and E2G employed must each correspond to the same interference margin, i.e., the same loading with respect to pole. This requirement ensures that each population see the same interference rise under full load. Since the 2G population tolerates a lower (~55%) loading than the 3G population (~72%), the 2G loading must be used. Use of a higher loading could be tolerated by the 3G population, but would require the 2G population to operate within a background interference that is too high, thereby compromising 2G performance. The E3G employed within the above calculation therefore must be the 3G Erlangs that can be achieved when a 3G carrier is loaded to the lower (55%) point, rather than its maximum of 72%. The restriction of 3G to a lower loading in a mixed carrier scenario influences the decision of deploying 3G in a mixed or dedicated mode, as described below. Planning: Mixed vs. dedicated carriers for 3G-1X

The decision regarding deployment via mixed or dedicated carriers is driven by several factors. The relative importance of each of these factors ultimately drives the decision. These factors are discussed below. Accurate mixing proportions

Employment of dedicated carriers naturally restricts the possible mixes of subscribers. For instance, if only two carriers are available for two populations, the only possible mix is to dedicate one carrier to the first population and the other carrier to the second population. This combination cannot reflect all possible target mixes of the two population. In contrast, mixing populations within the same carrier allows tailoring to a much larger set of possibilities.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 15

2G/3G-1X spatial and frequency design

System deployment

For example, suppose that only two carriers are available and the principle of dedicated carriers is employed. Further suppose that the anticipated subscriber Erlangs are 2/3 3G voice and 1/3 2G voice. This mixture of total Erlangs is readily addressed by devoting one carrier to 2G voice and one carrier to 3G voice, since the 3G voice carrier handles twice as many Erlangs as the 2G voice carrier. However, this solution would not be adequate if the anticipated mix was 60% 2G voice and 40% 3G voice. Within the two-carrier limit, there is no combination using dedicated carriers that would support these proportions when both carriers were fully loaded. In contrast, this mix could be achieved within each of the two carriers individually if non-dedicated carriers are allowed (see Equation 4-4). The sum across the two carriers would then match the design 60/40 target. Accordingly, the decision of dedicated vs. mixed carriers can be influenced by the desire to accurately achieve a design or target mix of subscribers. Mixing carriers allows more degrees of freedom in achieving specific values. If only approximate values relative to a target need be achieved, this distinction becomes less important. Further, if many carriers rather than few are available, the ability to achieve a specific mixture improves since more design degrees of freedom are available. Maximum total capacity

The computation of capacity for any combination of dedicated carriers is straightforward (See “Estimating capacity: Mix of 3G-1X voice and 2G voice” section). The total capacity achieved is simply the linear combination of the capacities offered by each carrier. The computation of total capacity for mixed carriers is more complex, since a truly accurate result for mixed subscribers within the carrier must account for nonlinearities. The impact of these nonlinearities depends upon the differences between the subscriber populations; for example, if two populations are distinguished solely by a small difference in Eb/Nt requirement, the impact of nonlinearities becomes negligible. In contrast, these effects can become important for large Eb/ Nt differences. Presuming that the achievement of an exact mix of subscribers (see above) is unimportant, the presence of nonlinear effects means that the total Erlang capacity in a mixed subscriber situation is always less than or equal to the total Erlang capacity that can be achieved with ...........................................................................................................................................................................................................................................................

4 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

2G/3G-1X spatial and frequency design

System deployment

dedicated carriers. The latter is actually an upper bound on the former, and often used as a reasonable approximation in planning scenarios. The validity of this approximation decreases as the differences between the populations become more distinct. These concepts were outlined in “Estimating capacity: Mix of 3G-1X voice and 2G voice” section and are expanded upon here. Consider the reverse link. A 13 kbps 2G population can tolerate about 3 dB interference rise over the noise floor. An 8 kbps 3G-1X population can tolerate about 5 dB interference over the noise floor. In a dedicated carrier scenario with all carriers full, each population experiences and tolerates its maximum rise. In contrast, in a mixed carrier scenario, the 3G population must be limited to (roughly) a 3 dB interference rise, since a larger value would result in a background interference that compromises the ability of the accompanying 2G population to reach the cell site. More 3G users could be added without loss of 3G performance since each 3G user can tolerate a higher interference level, but this higher level would degrade the performance of the 2G population. Since the 3G population is constrained by the presence of the 2G users, the total capacity (2G plus 3G) must be less than what could be achieved by dedicating carriers to each group. For 55% loading (the typical 2G value) the 3G capacity is reduced from 26.4 Erlangs (see Table 2-1, "Air interface capacity" on Page 2-9) to 18.4 Erlangs. The effects described above are mitigated somewhat by other factors (e.g., the 2G population benefits somewhat by the statistical benefits of more total users within a single carrier), but in all cases the dedicated carrier scenario remains an upper bound on achievable capacity. Since these effects depend upon the extent and nature of the differences in properties between the populations mixed, they are best assessed on a case-by-case basis. In the situation of small differences, the mixed carrier scenario may indeed approach the upper bound of performance. Efficient use of channel elements

A 2G channel element can accommodate only 2G calls. In contrast, a 3G channel element is dual-mode, accommodating both 2G and 3G calls. In addition, the 3G channel elements are offered in packs with higher density than 2G elements, e.g., 32 per pack vs. 20 per pack. From a provisioning point of view, the choice of mixed or dedicated carriers has little impact for fully loaded carriers. In the dedicated case, a requisite number is employed per carrier. In a mixed case, the total number required can be calculated from the anticipated subscriber

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 17

2G/3G-1X spatial and frequency design

System deployment

mixture. The calculation need not be that accurate, since 3G CE will support both 2G and 3G calls. A conservative estimate of 3G requirements, i.e., overprovision rather than underprovision, minimizes the impact of any inaccuracy. Extra 2G cards in this process can be removed, thereby using all elements to best advantage. In contrast, the mixed/dedicated choice becomes important in a growth scenario where the 3G population is slowly becoming a sizable percentage of the total traffic. In the dedicated case, this small 3G population is placed on a dedicated carrier, requiring 3G channel elements. Since the density of each 3G pack is large, a sizable fraction of the CE present may be idle, particularly when the nascent 3G traffic is low. In contrast, CE are used to better efficiency if the emerging 3G traffic is mixed into a 2G carrier. Within this scenario, a 3G pack is added to accommodate the 3G traffic. The existence of any extra idle CE can be balanced by removing the corresponding number of 2G CE, since the 3G CEs are dual-mode. The use of mixed carriers therefore uses hardware resources more efficiently in the traffic growth stages. This distinction may not be important if growth on a dedicated 3G carrier is expected to be rapid. Conclusions

The decision regarding dedicated vs. mixed carriers is therefore driven by several factors. These must be weighted in overall importance since all do not indicate the same decision. As an example, a possible deployment scenario could entail mixing 3G users into 2G carriers in the early stages of 3G growth. As 3G traffic becomes significant, 3G users could be migrated to a dedicated carrier(s). This scenario could apply in a situation where there is need to accurately meet in early growth a forecasted target demand (target mixture of total capacity) within the constraint of available spectrum, and to use CE as efficiently as possible. This scenario also provides for the maximum possible capacity in later phases of growth, as dedicated carriers are then employed. These advantages must be weighed against the disadvantages of not providing the maximum possible capacity in early phases, and against the difficulty of migrating 3G users to a separate carrier later on.

...........................................................................................................................................................................................................................................................

4 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Mixed 3G-1X voice/data capacity and coverage

System deployment

Mixed 3G-1X voice/data capacity and coverage ............................................................................................................................................................................................................................................................

In the following discussion, we presume that 3G-1X is deployed within a dedicated carrier, and consider the impact of mixing voice and data within the carrier. We further assume that the 3G-1X is an upgrade (1:1 overlay) of an existing 2G network, where the cell spacing is dictated by the 2G voice footprint. This scenario is of particular interest since many service providers desire to upgrade their current 2G networks to 3G. When the 3G-1X packet data service is introduced into the voice network, the high speed data will have an impact on the voice capacity and coverage. Analysis of the 3G technology indicates that the requirement of ubiquitous high rate packet data coverage is generally more stringent than that of voice coverage for comparable assumptions on RF parameters. This difference mainly comes from the decrease in processing gain. As mentioned in Chapter 3, “Data link budgets” and “Coverage (spatial) design: overlay and greenfield” sections of this chapter, if the design goal of a 3G-1X system is to provide an ubiquitous coverage for a high-rate data service, then the link budget based on the supplemental channel rate should be used for cell layout. If the voice link budget is used, then the high-rate data service will be available in an inner circle of the cell coverage. In this case, the supportable data rate will reduce when the mobile moves close to the cell edge. In order to extend the data coverage, the following methods may be employed: •

Relaxing the target FER for the data application without causing significant TCP/IP throughput degradation

•

Considering less body loss when using a data terminal

•

Using higher gain antenna at the data terminal

•

Increasing the base station transmit power and data terminal transmit power

•

Implementing a scheduling policy to provide fair access to data users on the cell edge

•

Increasing the number of cell site to provide additional coverage. This could also require re-design of the network and re-location of some of the existing sites and addition of new cell sites.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 19

Mixed 3G-1X voice/data capacity and coverage

System deployment

The total capacity for the mixed voice and data system is expressed as two numbers, the data throughput capacity and voice Erlang capacity. The higher data subscriber percentage, the more data throughput and the fewer voice erlangs. Data throughput and voice Erlang capacity will clearly vary depending upon the mix of voice and data users. Based on the voice and data traffic projection for a service area, a service provider can calculate the percentages of voice Erlang and data sessions, and then determine the trade-off between voice capacity and data throughput. Calculating the capacity values for the mixed voice and data capacity is a somewhat involved process. In both cases, capacities are calculated from maximum number of channels from the traffic (Erlang) model. The general model is described in Chapter 3, “General Erlang model” section, and is characterized by random arrivals at a system with finite queues and fixed number of channels (servers). For voice, the Erlang B (a.k.a, blocked calls cleared) version of the traffic model is typically used. The Erlang B version is the General Erlang model with no (zero length) queue. No queue implies no waiting. When a call arrives it is either served or turned away (blocked). The carried load on N channels is measured in Erlangs (average active channels). The associated performance is measured by probability of blocking, i.e., all channels busy. For data, the Erlang C (a.k.a., blocked calls delayed) version of the traffic model is typically used. The Erlang C version is the General Erlang model with infinite length queue. An infinite length queue implies that all arrivals are (eventually) served: hence, there is no blocking in the Erlang C model. The carried load on N channels (N data pipes) is measured by throughput (kbits/sec). The associated performance is measured by average wait in queue. The specification of throughput and Erlangs for a particular mix therefore depends upon the performance requirements (blocking percentage for voice, average delay for data) imposed on each population within the mix.

...........................................................................................................................................................................................................................................................

4 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Mixed 3G-1X voice/data capacity and coverage

System deployment

The methodology used to estimate the capacity for mixed voice-data networks uses the following steps: •

Select mix (e.g., 70% voice, 30% data)

•

Treat mix as forward power constraints, i.e., 70% is used for voice and 30% is used for data

•

For data portion: • • • •

For each assigned channel rate: Obtain probability distribution of number of supplemental channels over the coverage area from system level simulation For each possibility, compute throughput under average wait time constraint using Erlang C Obtain throughput via weighted sum

• •

Then obtain overall average throughput by using probability of seeing each rate. For voice portion: •

•

Determine the number of RF channels the forward link power can support • Compute carried Erlangs at required probability of block for the number of RF channels. Repeat for different mixes.

Following this methodology yields a collection of (Erlang, throughput) points for the range of mixes. A different curve can be generated by varying the performance specifications, either wait time constraint for data or blocking for voice. Typically, voice blocking is held constant across all curves and average data delay time is varied to produce a family of curves. However, there is no inherent reason that the performance specifications must be the same across the family of curves. Varying the performance specifications will change the shape of the curve. Figure 4-1 shows two curves. The straight line is obtained by varying the wait time specification for the data services for the different mix ratios, i.e., longer wait times for lower percentage of data versus voice. The straight line is an approximate upper bound and is recommended for provisioning purposes (i.e., packet pipe and CEs). The lower curve represents a constant wait time specification (5 seconds) and is recommended for capacity planning purposes.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

4 - 21

Mixed 3G-1X voice/data capacity and coverage

System deployment

Figure 4-1

Voice capacity versus packet data throughput in a mixed carrier

For example, if the demand for a typical sector was expected to be 18 voice Erlangs and 70 kilobits, the RF engineer would start by plotting this point on the figure above. Clearly, the point lies above the capacity curve and cannot be supported by a single carrier. The RF engineer would then divide both demand numbers by N until he got a point that fell below the curve. N is then the number of carriers required to support the capacity demand. In this N=2 gives 9 voice erlangs and 35 kilobits per carrier, which falls below the capacity curve, and hence, can expect to be supported.

...........................................................................................................................................................................................................................................................

4 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

5

Handoff

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter discusses the soft handoff procedures, algorithms, coverage, cost and benefits for the CDMA 3G-1X voice and packet data calls. Introduction Soft handoff definition Procedure IS-95B soft handoff algorithm Signal combining Forward link Reverse link Coverage contour Discussion Soft handoff costs on channel elements and packet pipe Soft handoff cost on forward link Soft handoff advantages Qualitative description of reverse link soft handoff gain Quantitative description of reverse link soft handoff gain Qualitative description of forward link soft handoff benefit Quantitative description of forward link soft handoff benefit IS-95B parameters

5-3 5-3 5-3 5-6 5-8 5-8 5-8 5-8 5-12 5-12 5-12 5-13 5-14 5-19 5-21 5-24 5-24

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 1

Handoff

T_ADD, T_DROP T_TDROP T_COMP SOFT_SLOPE, DROP_INTERCEPT, ADD_INTERCEPT SCH anchor transfer vs. SHO Fundamental Channel (FCH) – voice and data Data Supplemental Channel (SCH) Hard handoffs

5-26 5-28 5-29 5-29 5-30 5-30 5-31 5-35

References

5-36

...........................................................................................................................................................................................................................................................

5 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Introduction

Introduction ............................................................................................................................................................................................................................................................

Soft handoff definition

In soft handoff, multiple cells simultaneously support the mobile's call. In softer handoff, the mobile's call is simultaneously supported by multiple sectors of the same cell. The mobile continuously scans for the pilot signals transmitted by each cell/sector (site), and establishes communication with any site/sector (up to 6 6) whose pilot power exceeds a given threshold. Communication with the site/sector is terminated when the pilot power drops below a threshold for a time period. These types of handoff do not require an interruption of the communication link as a new link (“leg”) is added before an old leg is dropped. In contrast, a hard handoff (e.g., AMPS) requires a brief interruption of the link as the single supporting link is switched from one cell to another. Hard handoffs can also occur in CDMA when the mobile executes a handoff between carriers.

Procedure

The soft and softer handoff procedures dictate the way in which a call is maintained as a mobile crosses boundaries between CDMA cells. In soft handoff, multiple cells simultaneously support the mobile's call; in softer handoff, multiple sectors of the same cell simultaneously support the mobile's call. The distinction between soft and softer handoff is important since the same Channel Element (CE) is shared to support the handoff legs in the softer handoff case, but a separate CE is required to support each handoff leg in the soft handoff case. Each sector transmits a pilot signal of sufficient power to be detected by mobiles within its vicinity. The mobile continuously scans for pilots, and establishes communication with any sector (up to six) whose pilot exceeds a given threshold. Similarly, communication with sectors whose pilot drops below a threshold is terminated. The identification of distinct pilot signals by the mobile relies on the fact that each pilot exhibits a different time offset within the same PN code. ...........................................................................................................................

Typically, mobiles only have three “fingers” that demodulate three different signals (soft handoff legs or multipaths of a single leg). In six-way soft handoff, signals are transmitted from six different sectors. The mobile chooses the best three to demodulate, so not all signals are used by the mobile. Previously, only three-way soft handoff hand been supported. Even in three way soft handoff the mobile’s three fingers might demodulate different multipaths of the same transmission and not use a signal from one of the transmitting sectors. The six-way handoff feature is useful in pilot pollution areas. The feature needs to be carefully optimized so as to not compromise system capacity - see CDMA Translation Application

6

Note #4: Handoff. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 3

Handoff

Introduction

The mobile's search for pilots is facilitated by the fact that these offsets are in integer multiples of a known time delay. The pilots identified by the mobile, as well as other pilots specified by the serving sector(s), are categorized by the mobile as follows: •

The Active Set consists of those pilots whose sites are currently supporting the mobile's call

•

The Candidate Set consists of those pilots whose sites, based on the received strength of their pilots, could also support the mobile's call

•

The Neighbor Set consists of those pilots whose sites are not in the active set or the candidate set, but are nevertheless likely candidates for soft handoff; for example, these sites may be in known geographic proximity. Each sector in the network has an associated “neighbor list” provisioned. As sectors are added to the active set the network sends a Neighbor List Update message with the “best” 20 neighbors from the combined neighbor lists of all active set participants. The mobile uses the information from the network, as well as the normal movement of pilots (i.e., pilots in the candidate for longer than T_TDROP seconds), to populate the neighbor set.

•

The Remaining Set consists of those pilots within the CDMA system but not within the other three sets. The mobile may move pilots from the remaining set to the candidate set. However, the mobile typically uses more resources on the neighbor set than the remaining set; hence, it is less likely for pilots in the remaining set to move into the candidate set, than it is for the pilots in the neighbor set. Furthermore, because of the possible confusion about the unique identification of a sector by PN offset, the network does not add pilots from the remaining set to the active set that do not appear on the neighbor list. The undeclared neighbor list feature can be used to track these occurrences so that neighbor lists can be optimized. Note that provisioning of neighbor lists is one of the most important optimization activities to assure system performance.

Movement of pilots among the sets is determined by the mobile's assessment of pilot signal strength and a set of (adjustable) thresholds. This movement is coordinated with the serving sector. The mobile assesses pilots by comparing pilot strengths to one another, and by comparing each pilot's power to the total received forward link power. The latter comparison (normalized pilot strength) is the ratio of the ...........................................................................................................................................................................................................................................................

5 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Introduction

pilot energy in a time chip to the spectral density of total received forward link power. This ratio is called pilot channel Ec/Io and is defined as:

 Ec  µ ⋅ Pi   =  Io i FNoW + ∑ Pj all _ j

Equation 5-1: Pilot channel Ec/Io definition

where: µ = Fraction of sector power allocated to the pilot channel Pi = The power received from the ith sector F = Mobile receiver noise figure No = Thermal noise density W = The carrier bandwidth. Pilots in the neighbor and/or remaining set whose Ec/Io exceeds a threshold are associated with sites that can support the call; accordingly, these pilots are moved to the active or candidate set. The threshold is a fixed number (T_ADD) in IS-95A and a dynamic number in IS-95B that depends on the quality of the pilots in the active set. Similarly, pilots in the active and/or candidate set whose Ec/Io drops below a threshold (T_DROP for IS-95A and dynamic for IS-95B) for a period of time exceeding the parameter T_T_DROP are moved to the neighbor or remaining set. Finally, a candidate set pilot whose strength exceeds an active set pilot by at least T_COMP (and an additional dynamic criteria for IS-95B) will be moved to the active set, possibly displacing that pilot, as shown in Figure 5-1.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 5

Handoff

Introduction

Active Active set not full and Pilot exceeds T_ADD Or Active set full but swap criteria met (see text)

Pilot replaced by Candidate pilot

Candidate

Pilot is below T_DROP for T_TDROP seconds

T_TDROP expires Pilot exceeds T_ADD

Neighbor

Remaining

Figure 5-1

Simplified Pilot Set transactions (diagram does not show all possible transitions)

Figure 5-1 is a simplified diagram showing the movement of pilots between sets. Rather than attempting to show every possible event, we focus the diagram on those events most influenced by the translatable handoff parameters. IS-95B soft handoff algorithm

The field data shows that under some conditions there may be more soft handoffs occurring than are necessary when using the current IS95A handoff algorithm. Such handoff overheads may also overuse system resources, thereby degrading total system capacity. An improved soft handoff algorithm was defined in IS-95B and will be used for 3G-1X. The new soft handoff algorithm is intended to improve these situations by introducing the dynamic handoff threshold determined by combining the pilot strengths from all pilots in the active set. IS-95B added the following three new parameters to the soft handoff algorithm: •

SOFT_SLOPE

•

ADD_INTERCEPT

•

DROP_INTERCEPT.

...........................................................................................................................................................................................................................................................

5 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Introduction

These parameters lead to a variable threshold for adding and dropping pilots as opposed to the fixed threshold in IS-95A (i.e., T_ADD and T_DROP). The threshold is a function of the mobile's measure of the strength of the pilot's in the active set. The stronger the sum of the pilots strength the less likely a mobile is to add a pilot to the active set and more likely the mobile is to drop a pilot from the active set. The equations for the thresholds are:

  ADD _ THRESH = max SOFT _ SLOPE × 10 × log ∑ PSi + ADD _ INT , T _ ADD    i∈A   DROP _ THRESH = max  SOFT _ SLOPE × 10 × log ∑ PS i + DROP _ INT , T _ DROP    i∈A where PSi is the mobile's measure of pilot Ec/Io and the sum is performed over all pilots in the active set. The threshold is plotted as function of combined active set pilot strength below.

Add Threshold

IS-95B

Pilots not added in IS-95B that would have been added in IS-95A

T_ADD IS-95A

Combined Active Set Pilot Strength

Figure 5-2

IS-95B dynamic add/drop thresholds

Under this algorithm, the mobile will send out a PSMM message to request the base station to add a pilot into the active set only when the pilot is worthy of being added. This benefit can be seen in the figure as the gray area of pilot strengths that are not added in IS-95B that would have been added to the active set in IS-95A. The better the pilots the mobile is currently using (further to the right on combined active set pilot strength axis), the less likely is that a pilot will be added to the ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 7

Handoff

Introduction

active set (higher add threshold). A similar figure can be drawn for the drop threshold. The mobile will request the base station to drop a pilot from the active set if the pilot contributes little. These improvements will reduce the time a call is in soft handoff and also filter out unnecessary handoffs from each call; therefore, the average number of legs for each call is reduced and the forward link capacity is increased. Intuitively this makes sense since additional base station power should not be spent on a mobile that is receiving strong signals elsewhere. Improving forward link power utilization efficiency will lead to increased system capacity. Simulations have shown a range of improvements for the soft handoff power overhead factor used in forward link budgets. For forward link budget planning purposes, a reduction from the typical value of 1.85 for IS-95A to 1.75 for IS-95B is recommended. Signal combining

Forward link

On the forward link, all of the signals from the sectors in soft and softer handoff are combined in the mobile in a Maximum Ratio Combining (MRC) technique (see CMDA Systems Engineering Handbook, Jhong Sam Lee & Leonard E. Miller). In MRC, each of the soft handoff legs, in addition to any discernible multipaths, are added together with a weighting for the channel quality, which for IS-95 based systems is the pilot channel Ec/Io. Reverse link

For sectors involved in softer handoff the signals from the mobile are combined in the Channel Element in a MRC fashion as described for the forward link. For cells involved in soft handoff, the signals from the mobile are not actually combined, but a “frame selector” at the MSC chooses the “best” signal. The CRCs for the physical layer frames are examined, and the frame without an error is chosen as the best. If neither packet has an error, the decision is made randomly. Coverage contour

Mobiles evaluate base stations' suitability for providing a serving traffic channel by measuring the base stations' pilot signal strengths relative to total forward link power, or Ec/Io, as described above. One criteria for determining a coverage contour is that the mobile have at least one pilot with Ec/Io that is equal to the value of T_ADD: Values of Ec/Io within the contour will be greater than T_ADD; values outside the contour will be less. Accordingly, a mobile crossing the boundary

...........................................................................................................................................................................................................................................................

5 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Introduction

into the cell will add that cell's pilot to its active set. (A mobile crossing the boundary out of the cell will not necessarily drop the pilot, as this function depends on the values of T_DROP and T_T_DROP.) Coverage areas also change with varying T_ADD.

X Cell 1

Mobile in soft T_DROP - Cell 2

Figure 5-3

T_ADD - Both

X Cell 2

T_DROP - Cell 1

T_ADD coverage contour

Consider the sequence shown in Figure 5-3 (note the figure is drawn so that the T_ADD boundary for both cells exactly coincides. In practice the boundaries overlap given the geometry of the cell layout). As a mobile moves from Cell 1 to Cell 2, it will go through the following sequence: 1. 2. 3.

As the mobile moves past the T_DROP boundary for Cell 2, nothing happens When the mobile reaches the T_ADD boundary, it will add Cell 2 to its active set and will be in soft handoff with Cell 1 and Cell 2 When the mobile moves past the T_DROP boundary for Cell 1, it will drop Cell 1 from its active set and leave the soft handoff state.

A mobile moving in the opposite direction, from Cell 2 to Cell 1, goes through the following sequence, as shown in Figure 5-4.

Mobile in soft handoff

X Cell 1 T_DROP - Cell 2

Figure 5-4

1. 2. 3.

T_ADD - Both

X Cell 2 T_DROP - Cell 1

T_ADD contour mobile moves opposite direction

As the mobile moves past the T_DROP boundary for Cell 1, nothing happens When the mobile reaches the T_ADD boundary, it will add Cell 1 to its active set and will be in soft handoff with Cell 1 and Cell 2 When the mobile moves past the T_DROP boundary for Cell 2, it will drop Cell 1 from its active set and leave the soft handoff state.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 9

Handoff

Introduction

Note that designing coverage contours for pilot channel Ec/Io values of T_DROP will lead to coverage holes. Consider the following Figure 5-5:

No pilot!!!

X Cell 1 T_ADD - Cell 1

Figure 5-5

T_DROP - Both

X Cell 2

T_ADD - Cell 2

T_DROP coverage contour leads to coverage holes

As a mobile moves from Cell 1 to Cell 2, it will go through the following sequence: 1. 2.

As the mobile moves past the T_ADD boundary for Cell 1, nothing happens When the mobile reaches the T_DROP boundary, it will drop Cell 1 from its active set. The pilot signal from Cell 2 will still be below T_ADD, and hence, should not be in the active set. The mobile will not drop the pilot since it is its only active pilot, but because the pilot is weak, the mobiles performance (FER) will degrade and the call may drop.

Of course given the geometry of cell site coverage, it is impossible to have T_ADD contours matching exactly between cells. Therefore, it is important that in designing a network that all areas receive at least one pilot that is above T_ADD. This design approach will lead to most areas having overlapping pilots above T_ADD. In these overlapping areas, the mobile will be expected to be in soft handoff. The mobile will also be expected to be in soft handoff outside these overlapping T_ADD contours, but the specifics of the soft handoff locations depend on the mobile direction of travel. For networks with fixed subscribers, the soft handoff areas will be solely the areas of overlapping pilot strength above T_ADD. The areas of soft handoff in a mobile with one pilot below T_ADD (but above T_DROP) and another above T_ADD will not be soft handoff areas in a fixed network. Note that this discussion uses IS-95A terminology (i.e., T_ADD and T_DROP) but is applicable to IS-95B as well. As discussed earlier (“IS-95B soft handoff algorithm” section), the add and drop thresholds in IS-95B are a function of aggregate pilot Ec/Io. However, the IS-95B ...........................................................................................................................................................................................................................................................

5 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Introduction

thresholds will equal the IS-95A fixed thresholds for areas without strong pilot coverage, i.e., low aggregate pilot Ec/Io. The cell edge is expected to fall into this category of low aggregate pilot Ec/Io, and hence, the thresholds for an IS-95B network at the cell edge are expected to be T_ADD and T_DROP.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 11

Handoff

Discussion

Discussion ............................................................................................................................................................................................................................................................

Soft handoff costs on channel elements and packet pipe

The cost of soft-handoff is two-fold: 1.

2.

Increased number of channel elements (CE). A CE is required at every cell that supports a soft handoff leg. In the case of softer handoff Increase in backhaul network capacity required. Since multiple cells support the call and the frame selector that chooses the best soft handoff leg resides at the MSC, backhaul capacity will be required from all cells supporting soft handoff legs.

However, the benefit of soft handoff is increased coverage. Therefore, fewer base stations are required to cover the same area. The reduction in base station count outweighs the increase in CE and backhaul facility count. For example, for 95% probability of area coverage, the reverse link soft handoff gain is 4.0 dB. For a typical path loss slope of 38.5 dB/decade, the increase in cell radius is 27%, which equates to an increase of 61% in cell area. The same area can be covered with 38% fewer cells. This reduction in cell count typically outweighs the cost associated with the extra CEs and backhaul facilities for those cells. Softer handoffs require fewer resources than soft handoff in terms of channel elements and packet pipe bandwidth, since the signals are combined in a single channel element. The differentiation is important for provisioning required channel element and packet pipe resources. Soft handoff cost on forward link

The cost of soft handoff is forward link capacity in that the soft handoff legs on the forward link require power that cannot be used to support other users. This cost is captured in the forward link budget with the line item “Overhead factor to convert from mobiles to the number of active power channels”, commonly referred to as the “power overhead factor”. The value used is a function of soft handoff algorithm (IS-95A vs. IS-95B), terminal mobility, and cell site antenna configuration. The following table captures values typically used for planning purposes, which are rounded to nearest 5/100ths.

...........................................................................................................................................................................................................................................................

5 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion Table 5-1

Soft handoff overhead factors for voice link budgets

Cell Antenna

IS-95A

IS-95B

Terminal Mobility Mobile

Indoor Fixed

Outdoor Fixed

Omni

1.50

1.35

1.0

3-sector

1.85

1.60

1.25

Omni

1.45

1.30

1.0

3-sector

1.75

1.55

1.25

The soft handoff legs also consume Walsh codes. The power overhead factor is slightly different than the overhead for Walsh codes. However, the Walsh code overhead is only an issue when there are insufficient Walsh codes. Typically, Walsh codes are not a limiting factor. Networks with fixed subscribers that have high capacities are cases where the Walsh codes may be limiting. Note also that the power overhead factor is not the same as the Channel Element (CE) overhead factor. Since softer handoff does not require extra CEs, the CE overhead factor is less than the power overhead factor. Soft handoff advantages

Further insight into soft handoff operation can be gained by contrasting this process with the hard handoff process used in an analog system. In an analog system, each cell is assigned a set of narrowband channels for use in communication links. Co-channel interference is controlled by not reusing the same channels in adjacent cells. A mobile proceeding out of one cell into another must switch to an available channel in the new cell. This switch requires a brief interruption of the communication link. In a CDMA system, the same wideband channel is reused in every cell. Co-channel interference is accepted but controlled so as to achieve greater capacity. Accordingly, soft/softer handoff does not require channel switching and its associated link interruption. Moreover, with proper threshold settings, the acquisition of new sites is accomplished before the old (serving) sites are too far away to be useful. The soft handoff procedure is more robust because the connection with the new host(s) is made before the connection with the old is broken. This process is often referred to as a make-before-break connection, as opposed to the analog break-before-make. The makebefore-break handoff is more robust and leads to fewer dropped calls at handoff boundaries.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 13

Handoff

Discussion

Soft handoff also provides for advantages in terms of coverage and capacity. These advantages appear on both the forward and reverse links. Soft handoff provides a diversity gain without which some areas at the cell boundary (the locations furthest from base stations) would be regions of poor link quality because of shadow fading. The mobiles in these fringe areas would also be more susceptible to base station interference (see Chapter 2). Furthermore, the soft handoff state assures that the mobile is in a two-path channel. Two-path channels generally have lower Eb/Nt requirements relative to own-path channels. These effects increase the probability that a call will be dropped, since a hard handoff procedure would typically not be initiated until a mobile reached this area; that is, until the mobile noted a drop in signal strength from its host cell. In addition, the use of power control without soft handoff could create a situation where a mobile generates considerable amounts of interference to neighbor cells. Such interference would reduce capacity. The last situation arises because the mobile would detect a drop in received signal strength before it requested a handoff. Since cell boundaries overlap, this reporting point could be well into the boundary of the neighbor cell. Within this area, power control would boost the mobile's transmit signal strength in an attempt to maintain the link with the (distant) serving cell. This call-dragging phenomenon reduces the capacity of the neighbor cell because the mobile 's transmissions increase the level of interference at the neighbor cell. In contrast, if the mobile were in soft handoff, power control commands from both cells would ensure that the mobile did not produce undue interference; in fact, the reverse link could be maintained at a lower level of mobile transmit power due to the gain involved in combining the signals received at the two base stations. Qualitative description of reverse link soft handoff gain

The effect of soft handoff gain can be understood by considering a simple case of a mobile driving from one base station to another base station, as shown in the following figure.

...........................................................................................................................................................................................................................................................

5 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

Base Station A Figure 5-6

Base Station B Mobile traveling between two base stations

The mobile must generate enough signal (Smob) to overcome the path loss (Lp) and provide the required signal (Sreq which accounts for interference from other users) at the base station. This can be expressed mathematically as: Smob = Sreq + Lp

Clearly, as the path loss increases, the required power from the mobile will increase. In an ideal case, the path loss profiles would look something like the following. Pathloss to Base Station B

Pathloss to Base Station A

Base Station A

Base Station B Figure 5-7

Mobile required power with and without soft handoff

The following are shown in the above figure: •

The dashed line shows the path loss to base station A

•

The solid line shows the path loss to base station B

•

The heavy dashed line shows the required mobile power for a system without soft handoff

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 15

Handoff

Discussion

•

The heavy solid line shows the required mobile power for a system with soft handoff.

The system with soft handoff decides frame by frame which path is the better one, allowing the switch from base station A to base station B to happen immediately. The switch can be immediate since both base stations, A and B, are receiving and processing the signal from the mobile and the MSC is deciding which signal is best - hence the “soft” part of the handoff. In a system without soft handoff, the switch will be made later since it needs to have some hysteresis and has delay associated with signaling, etc. The difference in power during the delay in switching is not the gain claimed for soft handoff, but demonstrates the critical factor of a soft handoff: the fact that the decision of the best path is done frame by frame allowing the best path to always be chosen. The specific gain for soft handoff is shown in the following example that shows the effect of a fade.

Fade to Base Station A

Pathloss to Base Station B

Pathloss to Base Station A

Base Station B

Base Station A Figure 5-8

Mobile required power during fade with soft handoff

The following are shown in the above figure: •

The dashed line shows the path loss to base station A

•

The solid line shows the path loss to base station B

•

The heavy solid line shows the required mobile power for a system with soft handoff.

The figure demonstrates the benefit of soft handoff. As the mobile goes into a fade to base station A, it does not have to increase its power to the level of the fade, even for a short period. The mobile only has to increase its power to the level to reach base station B, which is unlikely to be also faded with respect to the mobile. The difference between

...........................................................................................................................................................................................................................................................

5 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

these two levels is what leads to “soft handoff gain”, which is a function of the extent of dissimilarity (decorrelation) between the fading processes with respect to A and B. It is difficult to definitively measure soft handoff gain in the same way that other link budget parameters such as antenna gain can be measured. This difficulty arises from the fact that the specific gain is variable. The gain depends upon the decorrelation of the fading processes, which can vary by market/morphology, and even by drive routes within a market. However, for planning purposes, a gain based upon a conservative decorrelation can be used in link budget analysis (see "Link budget" section on Page 2-14). Furthermore, this gain can be shown to map directly into reduction in mobile transmit strength, thus enhancing coverage as the link budget dictates. This demonstration is discussed further below. Lucent’s lab environment allows us to set up a specific path loss for a mobile. Utilizing this capability, the following path loss profile was created.

Figure 5-9

Path loss profile

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 17

Handoff

Discussion

The mobile was then put through this simulated path loss environment and the mobile transmit power was measured. The following figure focuses in on the area called “Jump #6” in Figure 5-9 (note that the time scales are different).

Figure 5-10

Observed mobile power without soft handoff

As one would expect, the mobile power increases by 12 dB, the same magnitude as the fade (increase in path loss). A second path loss profile for a different sector was also created as shown in the following figure.

...........................................................................................................................................................................................................................................................

5 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

Figure 5-11

Additional path loss profile for second soft handoff leg

The test was rerun with the mobile receiving from and transmitting to the two separate base stations through the two path loss profiles shown above. The mobile power transmit power was measured and is plotted in the figure below.

Figure 5-12

Measured mobile power with soft handoff

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 19

Handoff

Discussion

In this case with soft handoff, the mobile power only increases by 6 dB. As can be seen from the two plots, the soft handoff case reduces the mobile power by the difference between the fade magnitude and the difference in path loss to the two base stations. Quantitative description of reverse link soft handoff gain

In a CDMA system, there is an advantage due to soft hand-off gain that results in effectively lowering the fade margin required to obtain a specific probability of edge coverage, as compared to other technologies. The soft handoff gain calculation methodology sketched out below follows the development in Reference [1] of this chapter. For a CDMA system that admits soft handoff, for any given frame, the better, or alternatively, stronger of two or more base stations’ reception will be utilized at the switching center. For simplicity, consider that the decision will depend only on the attenuation, and that the base station with lesser of the two or more attenuations will control the AT. The attenuation of an AT to base station i is given by

10 log(α (di ,ζ i ) = 10γ log(di ) + ζ i Equation 5-2

Where: α(di,ζi) represents a function of d and ζ di is the distance to the ith base station ζ is the corresponding lognormal shadowing µ is the path loss exponent.

One problem is that the random component of the attenuation to the different base stations [the various ζs (i=0,1,2,...)] could be correlated with one another. To get around that problem, theζs are alternately expressed in terms of two independent random variables. Following along the same lines as the development in Reference [2] of this chapter, we define ζi = aΣ+ bΣi , where, a2+b2 = 1. The idea here is that by using different values for a and b, we can vary the correlation between the ζ’s. a = 1, b = 0 is the completely correlated case, while a = 0, b = 1 represents the completely uncorrelated case. For numerical calculations, values of a =b=1/√2, a partially (50%) correlated case, will be considered. Next, we evaluate the excess link margin required in this

...........................................................................................................................................................................................................................................................

5 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

case. Consider the scenario when a AT is in two way soft handoff. Link outage will occur in this case only if attenuation to both soft handoff sectors is greater than the margin γ. Hence,

Pout = Pr{Min[10µ log d 0 + ζ 0 ,10µ log d1 + ζ 1 ] > γ } Equation 5-3

Even before we evaluate the above expression, a review of the equation gives us an idea of why we have gain due to soft handoff. Instead of a single random variable, ξ, being greater than a fixed value resulting in an outage, we now need two partially independent random variables, each of which has to be greater than the fixed value to have an outage. The probability of the later event occurring is certainly less than the former, or alternatively, for the same outage probability, we need less margin in the later case. This is the advantage of the soft handoff capability in reducing the margin required, which effectively translates into soft handoff gain. We do not go into the details of evaluating Equation 5-3. The interested reader is referred to Reference [2] of this chapter. For a = b = 1/√2, path loss exponent of 4, and fading standard deviation of 8 dB, and probability of edge coverage the soft handoff gain numerically works out to 4 dB. For probability of edge coverage of 75%, the handoff gain is less and a value of 3 dB is used in the link budget. Due to the soft handoff feature, excess link margin requirement has dropped by 4 dB, from 10.3 dB to 6.3 dB. The soft handoff gain for the case of fading standard deviation equal to 8 dB, but probability of edge coverage of 75% (probability of area coverage of 90%) works out to approximately 3 dB. Due to the soft handoff feature, the excess link margin requirement has dropped by 3 dB from 5.4 dB to 2.4 dB. This reduction in link margin is the advantage due to soft handoff that results in increased coverage. Reverse link budgets typically contain the fade margin entered for the no-soft handoff case. Then, a separate line called soft handoff gain is included to capture the effect of soft handoff. The values typically used in the reverse link budget are conservatively rounded down from the values calculated by the methodology above, since the precise correlation is not known. The values used in the reverse link budget for fading standard deviation of 8 dB are shown in the following table.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 21

Handoff

Discussion Table 5-2 Probability of Edge Coverage

Reverse link soft handoff gain Reverse Link Fade Margin

Reverse Link Soft Handoff Gain

75

5.4

3.0

80

6.7

3.3

85

8.3

3.5

90

10.3

4.0

For standard deviations other than 8 dB, the required margin to achieve a specified outage (probability of edge coverage) criteria can be numerically determined using the methodology outlined in Reference [1] of this chapter. The soft handoff gain to be entered in the link budget is just the difference between the computed required margin and fade margin. Qualitative description of forward link soft handoff benefit

The co-channel nature of CDMA makes soft handoff critical for the forward link. A mobile at the cell edge will see equal strength signals from at least two base stations. In a non-CDMA system, the adjacent base station would not be using the same frequency channel. In CDMA the adjacent will be using the same frequency channel. Soft handoff allows for these co-channel signals that would be interferers to contribute to supporting the call. Consider this simplified case of a mobile receiving equal signals from two different base stations, no thermal noise and perfect orthogonality.

Base Station 1

Figure 5-13

Base Station 2

Mobile in soft handoff

...........................................................................................................................................................................................................................................................

5 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

In the no-soft handoff case, the Eb/No at the mobile is:

Eb µ ⋅ g ⋅ S1 = No S2 where g is the processing gain and µ is the fraction of power from base station 1 required to support the traffic channel for the mobile. If we assume that the signals from the two base stations of are equal strength, we can solve for µ as follows:

µno _ sho =

1 Eb g No

In the soft handoff case, we know from the theory of maximum ratio combining that the achieved Eb/No is the sum of the Eb/No's (linear) from the different soft handoff legs. Therefore, the combined Eb/No at the mobile is:

Eb µ1 ⋅ g ⋅ S1 µ2 ⋅ g ⋅ S2 (µ1 ⋅ S12 + µ2 ⋅ S22 )⋅ g = + = No S2 S1 S1 ⋅ S2 If we assume that the total power (Si) from each base station is the same and the power fraction for the two base stations are equal, i.e., µ1 equals µ2, µ can be solved as:

µsho =

1 Eb 2 ⋅ g No

Therefore, the power required from each base station in the soft handoff case is half of what would have been required without soft handoff. Of course, Base Station 2 is now transmitting power (utilizing its forward link capacity) to support the call, which it was not doing in the no soft handoff case; however, the net power received by other mobiles in the vicinity is unchanged since each base station is transmitting half of the original power. The benefit of soft handoff on the forward link comes from the fact that the mobile receives signals from different base stations that provide a diversity gain against fading. When a mobile enters a fade with respect to one base station, it is unlikely that it will be also be in the same fade with respect to the other base stations in its active set. Hence, the base

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 23

Handoff

Discussion

station will not require power to overcome the deepest fade a mobile at the edge may enter, since during that fading period the mobile can rely on signals from other base stations in its active set. We can consider the same example with the mobile entering a fade (F) to Base Station 1, the benefit of soft handoff becomes very clear.

µ1 ⋅ g ⋅ Eb = No 1 S2

S1

F

and

Eb µ ⋅ g ⋅ S2 = 2 S1 No 2 F

The combined Eb/No is then:

  S2  µ1 ⋅ 1 2 + µ2⋅ ⋅ S22  F Eb   = S1 No S F⋅ 2 If we assume equal power from both base stations, the equation simplifies to:

 µ1 + µ  ⋅ g 2⋅  Eb  F 2  = g µ1 + F ⋅ µ  =  2 1 No F   F If we assume that F is large, the µ1 term can be expected to be much less than the µ2 term. If then compare to the previous (non-faded case) Eb/No and require that the Eb/No be maintained at the same level:

F ⋅ g ⋅ µ2 = 2 ⋅ g ⋅ µSHO−nofade we see that the traffic fraction must be:

µ2 =

2 ⋅ µSHO−nofade F

The traffic fraction will not increase if the fade is greater than 2 (3 dB) and actually decreases for deeper fades. This analysis makes it appear that fading is beneficial, due to the assumptions of perfect orthogonality and no thermal noise leading to the single cell being the only interference term for the given leg. Hence, the deeper the fade the interferer is in, the better. If that perfect orthogonality assumption is removed, the self-interference (from the same cell) will become the predominant interference term for the non-faded leg during a fade to ...........................................................................................................................................................................................................................................................

5 - 24

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

other soft handoff legs. Of course, the faded leg will benefit from having the self-interference reduced by the same fade amount as the traffic signal. Quantitative description of forward link soft handoff benefit

The impact of soft handoff is captured in the Monte Carlo simulations used to derive parameter values for the forward link budget. These simulations capture the benefit of lower power per handoff leg due both to the diversity gain against fading and maximum ratio combining at the mobile. The forward link budget contains a term for the soft handoff power overhead factor. This term increases the number of traffic channels the forward link is supporting. For IS-95A, empirical data shows that a value of 1.85 to be a good estimate. No empirical data exists for IS95B. However, simulations suggest a reduction in this value to 1.75 as a good estimate for planning purposes. IS-95B parameters

IS-95B added the following three new parameters to the soft handoff algorithm. Lucent's 3G-1X system supports the IS-95B handoff algorithm and hence has these parameters. •

SOFT_SLOPE

•

ADD_INTERCEPT

•

DROP_INTERCEPT.

These parameters lead to a variable threshold for adding and dropping pilots as opposed to the fixed threshold in IS-95A, i.e., T_ADD and T_DROP. The threshold is a function of the mobile's measure of the strength of the pilot's in the active set. The stronger the sum of the pilots strength, the less likely a mobile is to add a pilot to the active set and more likely the mobile is to drop a pilot from the active set. Intuitively this makes sense since additional base station power should not be spent on a mobile that is receiving strong signals elsewhere. Improving forward link power utilization efficiency will lead to increased system capacity.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 25

Handoff

Discussion

The equations for the thresholds are as follows:   ADD _ THRESH = max SOFT _ SLOPE × 10 × log ∑ PSi + ADD _ INT , T _ ADD    i∈A   DROP _ THRESH = max  SOFT _ SLOPE × 10 × log ∑ PS i + DROP _ INT , T _ DROP    i∈A

where PSi is the mobile's measure of pilot Ec/Io , and the sum is performed over all pilots in the active set.

Add Threshold

IS-95B

T_ADD

Pilots not added in IS-95B that would have been added in IS-95A IS-95A

Combined Active Set Pilot Strength

Figure 5-14

IS-95B dynamic threshold

These thresholds are also applied when applying the T_COMP (see "Procedure" section on Page 5-3) criteria. T_ADD, T_DROP

Lower T_ADD and T_DROP thresholds lead to the mobile having more pilots in its active set. More pilots mean that mobile will have more forward link legs to support it. More forward links can help a mobile in disadvantageous RF conditions. However, this must be traded off against the cost of supporting those forward links. The power required to support those soft handoff legs will not be available to support other calls, thereby possibly lowering capacity. There are two factors that mitigate those costs.

...........................................................................................................................................................................................................................................................

5 - 26

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

First, the IS-95B dynamic thresholds, as described above, reduce the number of soft handoff legs by not assigning legs to mobiles that have high good pilots already. The quality of the pilots the mobile is seeing is determined by the aggregate Ec/Io term. Second, the faster forward link power control in 3G-1X allows the sectors involved in soft handoff to realize a greater gain from soft handoff. When a soft handoff leg is added in 3G, the mobile will see an immediate improvement in Eb/No and ask for less power from all base stations involved in the handoff. The base stations can quickly reduce the power for those links. In IS-95, the impact of adding a soft handoff leg was realized much more slowly as the power control is EIB based. The power is reduced slowly while no errors are reported from the mobile. The impact of faster power control is illustrated in the following simulations. Figure 5-15 and Figure 5-16 show time series plots of the 2G EIB based power control and the 3G-1X Eb/Nt (800 Hz) based power control. The top sub-plot of each figure shows mobile received Ec/Io from various pilots. Bolded lines indicate the pilots in the active set. As shown in the figures, the mobile gets into 3-way hand-off around the 82nd second. Hence, the geometry increases dramatically from simplex to 3-way handoff. However, the EIB based power control method cannot track the geometry changing very efficiently. This results in transmitting excessive power. On the other hand, the 3G Eb/ Nt (800 Hz) based power control can fully take advantage of tracking capability and results in saving transmit power.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 27

Handoff

Discussion

Figure 5-15

2G EIB based power control

...........................................................................................................................................................................................................................................................

5 - 28

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

Figure 5-16

3G-1X Eb/Nt based power control

T_TDROP

A timer is started when the strength Ec/Io of an active or candidate set pilot falls below T_DROP (or dynamic threshold for IS-95B). An active set pilot that falls below T_DROP for a period exceeding T_TDROP is moved to either the candidate or neighbor set (the decision is based on the serving site direction). A candidate set pilot that falls below T_DROP for a period exceeding T_TDROP is moved to the neighbor set. It is expected that the settings for T_TDROP for both the IS-95B and IS-95A implementations will be similar. ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 29

Handoff

Discussion

T_COMP

The parameter T_COMP controls movement of pilots from the candidate set to the active set. A candidate set pilot with strength Ec/Io exceeding that of an active set pilot by T_COMPx0.5dB is moved to the active set, replacing that pilot. T_COMP is measured in units of 0.5 dB. The SOFT_SLOPE, DROP_INTERCEPT, and ADD_INTERCEPT terms determine the dynamic portion of the add/drop threshold. Higher values will lead to fewer pilots in the active set, while lower values will lead to more pilots in the active set. More SHO legs can benefit a call and lead to fewer dropped calls and possibly lower error rates. However, more SHO legs will reduce base station capacity as more forward link power is used for SHO legs. These parameters need to be optimized to find the correct trade-off. Such optimization can be done, for example, in pre-commercial drive test.

SOFT_SLOPE, DROP_INTERCEPT, ADD_INTERCEPT

Insight into the initial settings for the new IS-95B parameters can be gained by plotting the improvement in aggregate pilot channel Ec/Io (i.e., the linear sum of Ec/Io's of pilots in the active set) for a given initial aggregate pilot channel Ec/Io and additional leg Ec/Io. Pilot Strength (dB)

10.0

Improvement (dB)

9.0 8.0

-6

7.0

-7

6.0

-8 -9

5.0

-10

4.0

-11

3.0

-12

2.0

-13

1.0 0.0 -18 -16 -14 -12 -10

-8

-6

-4

-2

0

Aggregate Ec/Io (dB)

Figure 5-17

Improvement in aggregate pilot strength

...........................................................................................................................................................................................................................................................

5 - 30

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion

Following any one specific line representing an additional pilot at a fixed Ec/Io, the plot shows the benefit of adding that pilot decreases as the aggregate Ec/Io increases. The knee of the curve defines a logical point for deciding whether to add that pilot or not. The knee is not precisely defined, but an approximate inflection point can be determined where the benefit of adding the additional pilot diminishes. SCH anchor transfer vs. SHO

While soft handoff is clearly beneficial for voice communications, the cost benefit trade-off is not as clear for bursty data transmissions on the SCH (see Chapter 3, "RF engineering for data"). For this reason, Lucent has chosen not to implement soft handoff for the SCH forward link. Lucent instead has implemented an optimized fast switching algorithm (i.e., anchor transfer) that provides similar performance to soft handoff without the drawbacks. In contrast, soft handoff is provided for the fundamental channel that serves voice and provides a continuous support link for the supplemental channel bursts. More detail is provided below. Fundamental Channel (FCH) – Voice and data

A fundamental channel is defined as a circuit-switched 9.6 kbps channel, supporting either voice or data. An FCH for voice is required to maintain a target Quality of Service (QoS) in terms of FER over the duration of a call. Call holding times can be several seconds to tens of minutes. During the call, the user most likely moves through a variety of RF conditions, crosses multiple cell boundaries, changes speed, etc. Soft handoff is designed to reliably maintain the call without speech quality degradation during any part of the call through these changing conditions. An FCH for data provides underlying support for data bursts on the supplemental channel (SCH), as well as to transmit low speed data. Similarly to voice calls, the FCH for data may stay active for durations of several seconds to durations of hundreds of minutes. The FCH is used to reliably deliver signaling and to guarantee minimum rate data services throughout the coverage area. Therefore, Lucent has implemented soft handoff for the FCH for both voice and data.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 31

Handoff

Discussion

Data Supplemental Channel (SCH) Data service requirements

High rate packet data transmissions are bursty in nature. SCHs are set up for durations expected to be much shorter than the typical voice call, in the range from hundreds of milliseconds to a few seconds. Additional reliability for SCH is provided by the RLP protocol that automatically retransmits physical layer frames in error. Therefore, the SCH does not have the same requirement for a continuous, low errorrate channel. Soft handoff cost

Soft handoff has a built-in cost in terms of both backhaul facilities between the base station and the Internet infrastructure and channel elements. Facilities, which are one of the highest operating expenses for network providers, would be required between every cell in the soft handoff and the MSC. This cost of facilities and channel elements is worthwhile for voice that requires a continuous, low latency channel. Qualitative performance impact of soft handoff

Supporting forward link soft handoff would increase the setup time for the channel, and hence the latency any given transmission would see. TCP flow control is very sensitive to round trip delay. At high data rates, even if the pipe is large (i.e., high bandwidth channel), it will not be fully utilized unless the end-to-end latency is minimized. Providing higher rate channels provides no advantage unless latency is controlled. But setting up a data burst in soft handoff would necessarily take longer and introduce more delay. Soft handoff requires coordination among the different base stations for the following: •

Channel element availability

•

Backhaul facility availability

•

RF resource availability

•

Time synchronization of the transmission of the burst.

Furthermore, to support soft handoff requires that all base stations providing a forward link as a soft handoff leg have sufficient power. Data channels are expected to require, on average, more power than voice channels. Therefore, it is more likely in data, as opposed to voice, that sufficient power will not be available to support the desired forward link rate. The channel rate would have to be reduced to support the weakest leg with the least available power.

...........................................................................................................................................................................................................................................................

5 - 32

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion Lucent anchor transfer solution

Lucent has implemented an anchor monitoring and transferring solution for the forward link SCH. “Anchor” means the sector that is determined to provide the best server for a given mobile. The mobile monitors the pilot Ec/Ios of the nearby base stations, and reports these measurements to the network on the reverse link. The network can then determine which base station will provide the best forward link performance. In this manner, most of the diversity advantage of soft handoff is maintained. Currently, the mobile reports its measurements up to every 2 seconds. Enhancements to the standard provide for mobile reporting when significant changes in pilot strengths are observed. Quantitative comparison of capacity and coverage impact of anchor transfer

Lucent has performed performance simulations to study the performance of ideal anchor transfer compared to soft handoff. The simulation had the following assumptions: •

Cell layout is based on 3G-1X voice link budget

•

Lognormal shadow fading with standard deviation of 8dB and 50% site-to-site correlation

•

Maximum supplemental channel transmission power fraction is -3dB with respect to full power.

•

Due to load variation (voice and/or data), half of the time the maximum Supplemental channel transmission power fraction for calls in handoff is restricted to -6 dB in at least one of the legs. The rest of the time, all handoff legs have up to -3dB available. This assumption is the most critical for the performance comparison. Different distributions for available power among the proposed handoff legs will yield different results.

•

No transmission diversity

•

Turbo codes for SCH

•

Mobile environments: AWGN, 3kmph one-path Rician (K=2, K=5)

•

IS-95B handoff algorithm.

The first plot shows the average SCH power, relative to total power, as a function of RF environment and SCH channel rate. Lower values are clearly better, as less power per user means that more users can be ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 33

Handoff

Discussion

supported. From the plot, it is clear that handoff does not provide a capacity advantage, and in many cases, provides a capacity disadvantage. The second plot shows the SCH coverage area, relative to FCH coverage, as a function of RF environment and SCH channel rate. In all cases, the no handoff case provides equal or better coverage compared to the handoff case. Average SC H Power as a Function of RF Enviro nm ent and R ate K= 2 19.2

K=5 19.2

AWGN 19.2

K=2 38.4

K=5 38.4

AWGN 38.4

K=2 76.8

K=5 76.8

AWGN K= 2 76.8 153.6

K=5 AWGN 153.6 153.6

-2.0

-18.0

-6.4

-9.5

-8. 0

-8.1 -9.6 -11.7

-8.9

-9.9

-12.4

-11.2

-11.7

-14.3

-15.1

Handoff No Handoff

-16. 5

-13.9

-16.0

-14.2

-14.0

-11.5

-12.0

-10.0

-10.0

-9.6

-8.0

-8.0

-8.2

-6.3

-6.0

-6. 9

-4.0

-12.2

Average Supplemental Channel Ec/Ior(dB

0.0

Figure 5-18

Simulation results 1 - soft handoff impact on data performance

SCH Area Coverage as a Function of RF Environment dR t

SCH Coverage (% of FCH Coverage)

K=2 19.2 100% 90%

K=5 19.2

K=2 AWGN 19.2 38.4

10 10 10 10 10 10 0% 0% 0% 0% 0% 0%

99 99 % %

K=5 38.4

AWGN

10 10 0% 0%

38.4

K=2 76.8

10 10 0% 0% 94 96 % %

80% 70% 60%

K=5 76.8 98 99 % %

AWG 76.8

K=5

AWGN K=5 153.6 153.6 153.6

10 10 0% 0% 85 81 % %

92 87 % %

99 91 % %

Handof No

50% 40% 30% 20% 10% 0%

Figure 5-19

Simulation results 2 - soft handoff impact on data performance

...........................................................................................................................................................................................................................................................

5 - 34

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

Discussion R-SCH

Lucent’s equipment does support soft handoff on the reverse SCH. Reverse link soft handoff has no cost from an air capacity point of view, and does provide benefit in reducing the mobile required power. The mobile is only broadcasting a single channel that is received by multiple base station receivers. Compare this to the forward link where the different base stations in soft handoff are transmitting separate signals, consuming part of their power and hence capacity. Supporting reverse link soft handoff does require extra channel elements and backhaul facilities, but given the expected asymmetrical nature of data, the cost is expected significantly less than if forward link soft handoff were supported.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 35

Handoff

Hard handoffs

Hard handoffs ............................................................................................................................................................................................................................................................

Although the focus of this chapter is on soft handoff, it should be noted that, hard handoffs also occur in a 3G network. Briefly, hard handoffs would occur in the following cases: •

From 3G-1X to 3G-1X on a different carrier

•

From 3G-1X to 2G on the same carrier

•

From 3G-1X to 2G on a different carrier

•

From 3G-1X to AMPS on a different carrier.

Note that hard handoffs from 2G to 3G and AMPS to 3G are not currently supported. Of course, hard handoffs from 3G-1X to either 2G or AMPS require the mobile to support the other technology being handed to (i.e., dual mode mobile). The reliability of hard handoffs is enhanced by carrying forward all of the improvements in hard handoff that Lucent has made for 2G IS-95. These improvements include: •

CDMA Inter-frequency Handoff Trigger Improvement (IFHOTI)

•

Pilot-Only Carriers

•

CDMA Multiple Pilots Interfrequency Handoff (CMPIFHO).

The combination of these features has led to extremely robust interfrequency handoff performance. Further information can be found in the Reference [2] of this chapter.

...........................................................................................................................................................................................................................................................

5 - 36

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Handoff

References

References ............................................................................................................................................................................................................................................................

[1]. “Soft Handoff Extends CDMA Cell Coverage and Increases Reverse Link Capacity,” Andrew Viterbi, Audrey Viterbi, Klein Gilhousen, Ephraim Zehavi, IEEE Journal On Selected Areas in Communications, Vol. 12, No. 8, October 1994. [2]. “CDMA Multi-Carrier Performance Enhancements,” Neil Berstein, October 1998.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

5 - 37

Handoff

References

...........................................................................................................................................................................................................................................................

5 - 38

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

6

Power control

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter describes the power control functions for both the forward link and the reverse link for the CDMA 3G-1X voice and packet data calls. Introduction

6-2

Reverse power control

6-4

Reverse power control for voice traffic RPC open loop for voice traffic RPC closed loop for voice traffic RPC for packet data traffic Reverse SARA for 3G-1X packet data calls Forward power control Forward power control for voice traffic FPC inner loop for voice FPC Outer Loop for Voice Forward power control for packet data traffic F-FCH power control for packet data Forward SARA for 3G-1X packet data calls

6-5 6-6 6-6 6-8 6-9 6-11 6-12 6-13 6-14 6-15 6-16 6-18

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 1

Power control

Introduction

Introduction ............................................................................................................................................................................................................................................................

This chapter describes the power control functions for both the forward link (base station transmitting signal to mobile) and the reverse link (mobile to base station) for the CDMA 3G-1X voice and packet data calls. The primary objective of power control mechanism is to maintain satisfactory traffic channel quality and reliability with minimal required power while maximizing system capacity within the design coverage area. The quality of each channel depends strongly on the ratio of signal power to the interference power, or Eb/Nt, Eb being the energy per signal bit and Nt the spectral density of the interference and noise. The required Eb/Nt is a function of vehicle speed and channel conditions. In addition, the forward link Eb/Nt requirement can also be affected by the mobile location with respect to the serving cell and other mobiles. This varying signal-to-noise ratio influences the frame error rate (FER) and the measured FER values can best characterize the voice quality for the CDMA system providing voice services. The power control algorithm is formulated based on tracking the measured FER values and comparison against the FER design target. The reverse power control (RPC) is more complex than that of the forward link. The RPC consists of an open loop and a closed loop. The latter consists of an inner loop and an outer loop. The open loop power control algorithm primarily resides in the mobile. This serves to adjust the mobile transmit power level to compensate for larger scaled, slow varying effects such as propagation loss and shadow fading. The closed loop algorithm involves both the base station and the mobile, and mainly serves to compensate for fast power fluctuation such as Rayleigh fading. The outer loop algorithm continuously updates the appropriate target Eb/Nt value required to maintain a desired average reverse FER for signals received at the serving cell. The inner loop then compares the measured Eb/Nt value with the target value. As the base station examines each reverse traffic frame reported by the mobile via the inner loop with each frame subdivided into 16 power control groups (PCG) having 1.25 msec time duration. The reported FER value is used as a reference in the outer loop to determine a new Eb/Nt target value. Both 2G and 3G RPC algorithms support the same basic open and closed loop functions, although the 3G algorithm offers significant enhancement over the 2G. The 2G average reverse link output power ...........................................................................................................................................................................................................................................................

6 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Introduction

for sub-rate frames (non-voice frames) is reduced by gating off PCG’s in the reverse traffic channel while maintaining the same power level per PCG. This reduces the reverse link power control speed for sub-rate frames. For instance, for 1/8-rate frames, the power control speed is reduced from 800 Hz (for full–rate frames) down to 100 Hz. The 3G reverse power control allows for continuous transmission rather than the gated transmission for the sub-rate frames, thus maintaining the 800 Hz power control speed regardless of the frame rate. In addition, the 3G fundamental and supplemental channels are adjusted using a simple integrated scheme, established by introducing the reverse link pilot channel (R-PICH), which serves as a reference in the inner closed loop for measuring the mobile Ec /Io level and for scaling. The forward link power control (FPC) algorithm is less complex than that of the reverse link. The mobile measures the FER statistics over a time frame and reports that to the base station. The measured FER is then compared with the FER target value. Upon comparison, the base station increases the forward link output power level if the measured FER is higher than the target, and vice versa. Unlike the 2G FPC, which was not designed to effectively mitigate fading, the 3G-1X FPC algorithm adopts a faster FPC scheme operating at a higher rate up to 800 Hz. The 3G power control mechanism facilitates a faster tracking of RF fades and provides a tighter gain adjustment to satisfy the minimum required Eb/Nt per call, thereby enhancing forward link capacity. The 3G FPC algorithm for voice calls operates at 800 Hz. For packet data service, the forward fundamental channel (F-FCH) power control operates at 800 Hz when the power control function for forward supplemental channel (F-SCH) is off. The F-FCH power control rate reduces to 400 Hz during the F-SCH bursts while F-SCH power control is on, also at a rate of 400 Hz.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 3

Power control

Reverse power control

Reverse power control ............................................................................................................................................................................................................................................................

The primary objective of reverse link power control is to resolve the near-far issue, where a mobile that is near the serving cell yields a better signal path than a mobile that is far away from the cell. Thus the mobiles near the cell may possibly raise too much RF interference to allow for the mobiles far away to reach the serving cell with sufficient signal-to-noise ratio. This issue can be resolved by dynamically controlling the mobile transmit power such that the serving cell observes the same signal-to-noise ratio from each mobile. The 3G-1X reverse power control (RPC) algorithm consists of an open loop as well as a nested closed loop. The RPC supports the integrated fundamental and supplemental channel power control algorithms by introducing the R-PICH. The R-PICH provides a phase reference to the base station on a per-PCG basis for coherent detection of the reverse fundamental channel (R-FCH). The power allocated to the R-PICH is related to the R-FCH power by a translation value. The R-PICH Ec/Io is also closely correlated with the transmission of the estimated R-FCH Eb/Nt value and other power control commands from the base station to the mobile. In the 3G reverse open loop, the mobile estimates the required transmitted power of the reverse link channels based on the measured aggregate received power. Similar to the 2G RPC algorithm, the 3G RPC open loop function is performed in the mobile, using necessary operating parameters supplied by the base station via signaling messages in the overhead channels and the forward traffic channel. The 3G system applies several new open loop parameters, which were not included in the 2G RPC algorithm before. These include, for example, the mobile determined R-PICH mean output power (as a function of the access channel power) and a gain-adjusting cell translation parameter, RLGAIN_ADJ. This parameter is set by the base station and sent to the mobile via the Extended Channel Assigned Message (ECAM). These allow for the mobile to compute the R-FCH mean output power to be transmitted based on the R-PICH mean output power. Both the 2G and 3G reverse closed loop power control algorithms consist of nested inner and outer loops, although there is a major difference between the 2G closed loop function and the 3G. In the 2G inner loop algorithm, the mobile reduces the average power for sub-rate frames by gating off certain PCGs, thus reducing the output power ...........................................................................................................................................................................................................................................................

6 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Reverse power control

level per frame. For example, for half-rate frames, eight PCGs are gated off. For ¼-rate frames, twelve PCGs are gated off. As the 2G mobile outputs the same power level for each non-gated PCG, the base station only measures the traffic channel Eb/Nt for the non-gated PCGs. Although the average reverse link power is regulated as required, such gating reduces the closed loop operating speed. In lieu of such PCG gating, the baseline 3G RPC algorithm applies a continuous transmission scheme by reducing power level per PCG, and avoids reducing the power control speed during sub-rate frame transmission. It should be noted that the IS-2000 protocol allows for a reverse eighth-rate gating feature, also known as R-FCH gating. When transmitting the R-FCH at 1/8 rate, in order to reduce the mobile power consumption and conserve the battery, the mobile may request for such R-FCH gating via the page response message or the origination message. The base station shall then address its response to such a request via the signaling messages. If the R-FCH gating is enabled for the 1/8-rate frame transmission, the FPC inner loop at the base station only receives half the PCGs in the reverse PC sub-channel. For the gated PCGs, the base station receives a noisy signal without any knowledge of the frame rate and gating situation. This prevents the CMS-5000 ASIC from locking the finger energy for those gated PCGs, but rather maintaining the previous F-FCH gain and thus preventing the FPC inner loop function from being impaired. In this case, the effective FPC inner loop speed for the 1/8-rate frame R-FCH is reduced by half. In the RPC inner loop, up power control commands will be sent to the mobile as the base station measures noisy finger energy for the gated PCGs. The frame rate information being available, the mobile will execute only the PC commands associated with non-gated PCGs and ignore those gated while the gated PC commands are discarded. This is based on information concerning the relative delay between the R-PICH PCG number and the F-FCH PCG carrying the PC commands associated with the R-PICH measurement, as per the IS-2000 standard. Reverse power control for voice traffic

For 3G-1X voice service, the RPC algorithm consists of an open loop, and nested inner and outer closed loops. The details for the open loop and the closed loops are provided below.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 5

Power control

Reverse power control

RPC open loop for voice traffic

The primary algorithm for the open loop resides in the mobile. This serves to adjust the mobile transmit power level to compensate for larger scaled, slow varying effects such as propagation loss and shadow fading. As per IS-2000.2, for voice calls, two equations are used for computing the open loop mean R-PICH and R-FCH output power levels from mobile respectively. First, the mean R-PICH power is computed via:

PR − PICH (dBm) = −8.5dB + PACCESS −CH (dBm) + RLGAIN _ ADJ where RLGAIN_ADJ is a base station translation parameter for initial power variation, sent to the mobile via ECAM signaling message. The mean output power of the R-FCH can then be computed based on the mean R-PICH power and other parameters as per the IS-2000 standard. These parameters include the band class constant, channel power adjustment parameters, and a parameter that is set by the base station and a power offset parameter, RLGAIN_TRAFFIC_PILOT. RLGAIN_TRAFFIC_PILOT is a translation parameter set at the base station and transmitted to the mobile via signaling messages for updating the relative power between R-PICH and R-FCH power. The detailed translation information is described in CDMA Translation Applications Note #3V. RPC closed loop for voice traffic

As stated in the “Introduction” section of this chapter, the 3G-1X reverse power control closed loop consists of a nested inner /outer loop. The inner loop algorithm primarily determines and regulates the R-FCH output power level based on the detected R-PICH signal strength and the outer loop adjusted full rate Eb/Nt set point value. This new Eb/Nt set point value is determined in the outer loop based on the monitored reverse FER. The following is a functional block diagram of the RPC closed loop function for 3G-1X voice traffic.

...........................................................................................................................................................................................................................................................

6 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Reverse power control

REVERSE LINK POW ER CONTROL CLOSED LOOP FOR 3G1X R-FCH

O UTER LOO P

ATTRIBUTE_ADJUSTMENT_GAIN RLGAIN_TRAFFIC_PILOT

FROM MOBILE STATION

FROM BASE STATION

Forward Power Control Sub-Channel in F-FCH

INNER LO OP

CONVERT R-FCH E b /N t SETPO INT TO REVERSE PILOT ENERG Y THRESHOLD

NOMINAL_ATTRIBUTE_GAIN

OUTER LOOP

INNER LOO P

MEASURE RECEIVED S/N O F REVER SE PILOT CHANNEL R-FCH FER ESTIMATION

REVERSE PILO T

R-FCH

BASE STATION

Figure 6-1

MO BILE STATIO N REVER SE POW ER CONTROL

REVERSE PILO T AND R-FCH CHANN EL TRANSMITTER

MO BILE STATION

RPC closed loop for 3G-1X voice

As shown in the above diagram, the reverse outer loop computes a new R-FCH Eb/Nt set point iteratively based on the base station detected reverse frame errors at full rate. The base station then converts this Eb/Nt set point value to a R-PICH signal-to-noise ratio (Ec/Io) set point value. This updated R-PICH Ec/Io set point is mapped to an R-PICH energy threshold provided in a lookup table in the ASIC. As an embedded ASIC function, the inner loop algorithm compares the measured R-PICH pilot energy with the above threshold and determines the reverse power control bits to be sent to the mobile via the forward power control sub-channel. As a voice call is initially set up, the F-FCH is assigned prior to the R-PICH and R-FCH assignments and this F-FCH also carries the forward power control sub-channel. During this initial period, the forward power control sub-channel sent to the mobile from each leg alternating up and down commands to maintain a zero net gain in mobile transmit power in the inner loop. If the call starts in multiple legs, the first leg acquiring the R-PICH sends special preamble frames to the frame selector, which echoes the best frame to all active legs. The outer loop is initialized upon the cell receiving the first R-FCH with a good frame, and meanwhile, the inner loop stops sending the alternating PC commands to the mobile. Consequently, upon ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 7

Power control

Reverse power control

receiving the R-PICH measurement, the inner loop begins its normal routine. For Simplex calls, the inner loop starts as soon as the R-PICH is measured. The R-PICH Ec/Io set point value used in the voice R-FCH inner loop is determined as follows:

R − PICH _ Ec I o setpoint(dB) = ξ − 10 log10 (GF ) − η F , where:

 EbF  Nt

ξ = 10 log10 

   

ξ is the R-FCH set point in dB, as R-FCH outer loop output, GF = 1228800/R-FCH (full information rate), GF is the R-FCH processing Gain,  EF  η F = 10 log10  CP   EC 

ηF is the R-FCH power to R-PICH power offset at the mobile. F in the above equation denotes the R-FCH at full rate with the data rate equal to 9.6 kbps for RC3 and 14.4 kbps for RC4. The frames are each 20 msec in length, with convolution coding. RPC for packet data traffic

The reverse power control algorithm for packet data traffic is capable of performing power control functions on the R-FCH and the R-SCH separately. When the data session is an active mode, the base station regulates the mobile output power levels for the R-FCH and R-SCH when assigned. During the dormancy periods, the RPC function is disabled. Similar to that for voice services, the RPC algorithm for packet data services consists of an open loop and a nested inner/outer closed loop. The R-FCH power control open loop algorithm for packet data is analogous to that for voice service. During the R-SCH bursts, the open loop algorithm determines and regulates the R-SCH output power. The mean R-SCH transmit power is computed based on the mean R-FCH power, mean R-PICH power and parameters similar to those for determining R-FCH power. The data calls involve an offset translation parameter that defines different percentages for reverse pilot power required for voice calls and for data to achieve the desired FER.

...........................................................................................................................................................................................................................................................

6 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Reverse power control

For 3G-1X packet data services, there is only one inner loop in the reverse power control algorithm. This inner loop is controlled by the R-FCH RPC outer loop. Similar to that for the R-FCH, an R-SCH outer loop is designed to meet the R-SCH target frame error rate. The R-SCH outer loop detects the R-SCH frame errors and generates an updated RSCH Eb/Nt set point value according to the correlation between the target R-SCH FER set point and that measured. Although the output from the R-SCH outer loop does not affect the reverse inner loop and the R-FCH outer loop, the R-SCH outer loop function depends on the performance of the inner loop and the R-FCH outer loop. The detailed R-SCH outer loop algorithm is implemented via two steps. In the first step, if the based station is in soft or softer handoff, it detects the R-SCH frame quality and sends a quality indicator to the FCH frame selector in the switch via R-FCH. The frame selector determines the best R-SCH frame and sends back to the base station via F-FCH. Based on this frame quality, the R-SCH outer loop algorithm determines an updated R-SCH Eb/Nt set point value. If in simplex mode, the R-SCH outer loop directly uses the frame quality bit for deducing a new Eb/Nt set point and bypasses the frame selector process. In the second step, the R-SCH Eb/Nt and R-FCH Eb/Nt set point values are compared in a frame-by-frame basis. If the difference for a frame relative to the difference for the previous frame is greater than an offset threshold, a signaling message will be sent to the mobile to adjust for the R-SCH mean output power relative to the R-FCH mean output power. Reverse SARA for 3G-1X packet data calls

The R-SCH bursts typically transmit much higher power than that of the low-rate R-FCH for voice or for low speed packet data traffic. As the base station receives much greater RF power from such SCH bursts than that from weaker mobiles, the reverse links for the latter may possibly be impaired. The reverse supplemental air resource allocation (R- SARA) mechanism functions to assess the impact of admitting a new R-SCH burst on the current system performance and regulate any possible new R-SCH assignments. For each new R-SCH burst request, the call-processing algorithm identifies the highest rate that may possibly be assigned based on the hardware and software resources available and other service constraints. Each leg then independently executes the R-SARA algorithm for this call, and determines the maximum R-SCH rate that can be supported based on the assessed RF performance while

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 9

Power control

Reverse power control

considering the impact of adding this new R-SCH. To assess the impact of adding a new R-SCH, the RF loading must be evaluated. The contribution to the loading from active R-SCH bursts can be significant, and this is greatly dependent on how strongly it is received at the sector. Contrary to using the assumed constant receive Eb/Nt for RC3 for the active fundamental channel, an actual measurement of the received reverse link pilot signal strength is used to estimate the loading contribution from an active R-SCH burst at each of the active legs. Such estimate is based on the number of the current active Walsh codes on the sector under consideration. Also included are the reverse fundamental active channels along with any R-SCH bursts. If the difference between the strongest pilot Ec/Io among the non-active set and that of the current strongest active set is greater than a threshold, the R-SCH request will be rejected.

...........................................................................................................................................................................................................................................................

6 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Forward power control

Forward power control ............................................................................................................................................................................................................................................................

As discussed above, in the 3G-1X forward link, each serving sector transmitter must ensure that the required Eb / Nt is achieved at each mobile within that sector. The required Eb/Nt range is significantly influenced by mobile speed and multipath conditions due especially to the fact that the mobile receiver does not employ antenna diversity. The 3G-1X forward power control algorithm is designed to compensate for the fast varying Eb/Nt and other cell interference via a fast tracking closed loop in a sub-frame interval in place of the slower FER based algorithm used in the 2G FPC algorithm. Thus, tighter base station transmit gain adjustment can be achieved and this results in an increased forward link capacity. The 3G forward power control feature is highlighted below: •

Compatible with TR45 TIA/EIA/IS-2000 Standard

•

Supports voice and low speed data (9.6 kbps and sub-rates) in F-FCH and high speed data (up to 153.6 kbps) in F-SCH

•

Supports Convolution and Turbo coding for F-SCH data rates of 19.2 kHz, 38.4 kHz, 76.8 kHz and 153.6 kHz. For Release 20 and higher, 307.2 kbps will be supported as well.

•

F-FCH can be in soft or softer handoff, while F-SCH is currently designed for single leg (anchor leg) condition namely Reduced Active Set. Softer handoff for F-SCH will be available for future release.

•

Supported by first release of 3G-1X product for voice and data traffic.

The forward closed loop power control algorithm consists of an outer loop and an inner loop and the algorithm is implemented effectively at a rate of up to 800 Hz. As per IS-2000, the FPC algorithm for 3G-1X voice and packet data traffic is designed for the mobile station to support up to two inner loops. One is the “primary inner loop” that controls operation of the F-FCH for voice and the low speed packet data (at 9.6 kbps data rate); the other is the “secondary inner loop” that controls the F-SCH packet data traffic with data rates of 19.2 kbps, 38.4 kbps, 76.8 kbps and 153.6 kbps. Additionally, if a forward link dedicated control channel ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 11

Power control

Forward power control

(F-DCCH) is assigned, its output power is also controlled by the same inner loop algorithm. Any other F-SCH assigned will also be controlled by the secondary inner loop. Though the closed loop algorithm in the mobiles has not been standardized, the most common procedure for the primary inner loop is based on the power control bits (PCB), which is corresponding to the signal-to-noise ratio (S/N) measured at the mobile. The secondary inner loop function is based on the mobile measured F-SCH traffic S/N. In the forward link, the base station configures the mobile and passes the following information to the traffic channel:

Forward power control for voice traffic

•

Forward Target FER values for the F-FCH and the F-SCH

•

Initial, minimum and maximum Eb / Nt set point values

•

Ratio of PCB power in primary channel over primary channel traffic power at full rate (denoted as FPC_SUBCHAN_GAIN)

•

Primary channel (F-FCH or F-DCCH) and secondary channel (FSCH) with possible inner loop rates at (800,0), (400, 400) Hz or (200, 600) Hz.

The FPC functional diagram for voice service is illustrated in Figure 6-1. As shown in this diagram, the main functionality of both inner and outer loops resides in the mobile. The key functional blocks include the following: •

The F-FCH Eb/Nt detector

•

The primary inner loop block that generates the PC commands sent to the base station

•

The F-FER detector

•

The main outer loop block, which adjusts the Eb/Nt target value at the mobile.

More detailed description for the voice FPC inner and outer loops are provided below.

...........................................................................................................................................................................................................................................................

6 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Forward power control

FORWARD LINK POWER CONTROL CLOSED LOOP FOR 3G1X F-FCH Forward Power Control SubChannel in F-FCH

FORWARD FUNDAMENTAL CHANNEL TRANSMITTER

MEASURE RECEIVED E b/N t OF FUNDAMENTAL CHANNEL

OUTER LOOP

INNER LOOP

BASE STATION FORWARD POWER CONTROL

Reverse Power Control SubChannel in Reverse Pilot Channel

F-FCH FER ESTIMATION

PRIMARY INNER LOOP

F-FCH OUTER LOOP

F-FCH E b/N t SETPOINT

BASE STATION

MOBILE STATION FPC_FCH_IN IT_SE TPT FPC_SUBCHAN_GAIN FROM BASE STATION

FPC_MODE FPC_PRI_CHAN FPC_FCH_FER FPC_MIN_SETP T FPC_MAX_SETP T FROM BASE STATION

Figure 6-2

Forward power control function for voice service

FPC inner loop for voice

The voice FPC inner loop algorithm is an iterative procedure. As the F-FCH traffic channel power is being transmitted from the base station to the mobile, the mobile monitors the F-FCH received PCB and estimates the Eb/Nt for the full-rate traffic bit based on the value of the base station provided FPC sub-channel gain (denoted as FPC_SUBCHAN_GAIN). When compared with the current Eb/Nt target value, if the performance degrades, the inner loop commands the base station (via the RPC sub-channel) to increase the traffic channel transmit power gain. On the contrary, if the forward link quality exceeds the updated Eb/Nt target value, then it commands the based station to reduce the transmitting power. The base station ASIC detects the forward link power control commands in the reverse power control sub-channel via R-PICH at a rate of 16 per 20 msec frame, or 1.25 msec, which amounts to 800 Hz. This 1.25 msec is the time interval of each power control group (PCG). The following initial parameters are required for executing the voice FPC inner loop algorithm: •

Forward power control initial gain, FPC_INI_GAIN

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 13

Power control

Forward power control

•

Forward minimum gain

•

Forward maximum gain

•

Gain adjustment step size (up step size and down step bias).

The units of the initial, minimum, and maximum gain values are in dB, relative to the forward pilot power. The initial gain setting is not as critical because the inner loop operates at a speed of 800 Hz which is sufficient for adjusting the forward gain to meet the updated Eb/Nt set point without much delay. However, the values for the minimum and maximum gains are critical. The values required for achieving optimal capacity are dependent on the radio configuration and the number of soft handoff legs of the call. In soft handoff, the primary leg passes the above four parameters to each active leg thus allowing for different gain constrains and power control steps for different calls in the same cell/sector. For troubleshooting and/or RF optimization, one may disable the FFCH inner loop by setting the inner loop power control step sizes (both up step and down step bias) to 0 dB via translation parameter settings. With such settings, the power control commands received and processed by the cell allows the F-FCH forward gain to remain constant via the cell ASIC. By disabling the inner loop, the forward power control is effectively turned off, regardless of the on/off status of the outer loop. FPC Outer Loop for Voice

Because the primary objective for the FPC for voice traffic is to maintain an acceptable voice quality while maximizing the system capacity, and FER is a performance measure that well characterizes the voice quality, maintaining an acceptable FER is an important part of the FPC. However, given that there is no direct close mapping between FER and the measured Eb/Nt, some adjustment in the inner loop is required in order to maintain an acceptable averaged forward link FER. Specifically, the F-FCH Eb/Nt target value used in the inner loop function must be continuously adjusted based on the detected FER value. This FER detection is performed in the outer loop. In addition, the outer loop algorithm also includes estimating FER and dynamically determining the appropriate Eb/Nt target value. These outer loop functions are implemented in the mobile on a per-frame basis at a rate of 50 Hz.

...........................................................................................................................................................................................................................................................

6 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Forward power control

At the time a call is just being set up, the outer loop is configured by the cell via layer 3 signaling via the paging channel and continuously updated via the F-FCH. The following are the required parameters for configuring the outer loop: •

Forward target FER

•

Initial Eb/Nt target value

•

Minimum Eb/Nt target

•

Maximum Eb/Nt target.

These parameters are passed to the mobile during call setup via the ECAM signaling messages, the service connect messages, and the forward power control message when the mobile is assigned an F-FCH. In handoff, the outer loop parameters are updated to reflect the new number of soft and softer handoff legs that affects the minimum and maximum Eb/Nt target values. For troubleshooting and/or RF optimization, one may disable the outer loop regardless of the inner loop status. This is achieved by “freezing” the output Eb/Nt set point value, either to the current target value or a specific base station determined value that is passed to the mobile via the power control message. The outer loop function will be resumed as the cell sends to the mobile a new power control message with updated minimum and maximum Eb/Nt set point values. Forward power control for packet data traffic

In 3G-1X packet data mode, the forward traffic data is transmitted via the F-FCH and F-SCH channels, where the F-FCH transmits signaling and low rate data (at 9.6 kbps) and F-SCH transmits packet data at higher rates as discussed above. A data session consists of one or more active periods where data is transmitted over the air interface. These active periods are separated by periods of inactive mode, or dormant mode. In dormant mode, neither the F-FCH nor the F-SCH is assigned and thus any information stored in the base station associated with the previous data call is erased. When in active mode, the F-FCH is on at all times, while F-SCH may be on or off, depending on the availability of the air interface resources and the amount of data in the buffer awaiting to be sent. For trouble shooting and/or optimization, the 3G1X F-FCH and F-SCH FPC functions can be disabled separately by setting the inner loop power control step sizes to 0 dB.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 15

Power control

Forward power control

F-FCH power control for packet data

When the 3G-1X packet data calls are in active mode, the F-FCH power control algorithm follows the same closed loop process, which consists of nested inner and outer loops as that for the 3G-1X voice calls. However, some of the required translation parameter values must be set differently because of the following reasons: •

The FER target for a voice call is dictated by the voice quality requirement, while the FER target for a packet data call is established by signaling traffic requirements (delay and reliability) as well as the radio link protocol (RLP) performance.

•

The minimum and maximum gain values are dependent on the rate at which the power control operates. The F-FCH forward power control (FPC) for voice calls operates at 800 Hz, while for packet data calls only operates at 800 Hz when the F-SCH is off. The F-FCH FPC rate reduces to 400 Hz during an F-SCH burst.

As per IS-2000, the closed loop may operate in several modes. The base station selects the mode and configures the mobile via the layer 3 messages at the instance when the F-FCH is first assigned. It also updates the mobile configuration via an in-band signaling during the F-FCH operation. The packet data FPC algorithm is designed such that the base station may configure up to two reverse power control sub-channels via the R-PICH and this closes up to two independent inner loops. When there is no F-SCH assigned, mobile is configured to support only one reverse power control sub-channel, operating at 800 Hz. During an F-SCH burst, two reverse power control sub-channels are configured in a time-multiplexed fashion via the single R-PICH, such that the combined speed of these two inner loops becomes 800 Hz. Two traffic channels, defined as primary and secondary traffic channels (as per IS-2000), are mapped to the above two inner loops. The primary channel refers to the forward traffic channel that carries the FPC sub-channel used by the primary FPC inner loop. The secondary traffic channel is only meaningful when there are two co-existing inner loops. When the secondary FPC inner loop is active, the mobile performs the Eb/Nt measurements via the secondary traffic channel. At a data rate of 9.6 kbps, the packet data FPC algorithm is basically operating with F-FCH inner/outer nested power control loops, similar to that for the voice FPC. The packet data also F-FCH supports soft handoff. As an F-SCH is assigned (with a data rate higher than 9.6 kbps), for Release 20 and below, it only operates in a simplex mode so ...........................................................................................................................................................................................................................................................

6 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Forward power control

as to optimize the burst setup time. The packet data FPC algorithm consists of two inner loops and two outer loops, whether the F-FCH is in the simplex mode or in handoff. The Primary and Secondary power control loops are shown in Figure 6-3 and Figure 6-4 respectively.

FORW AR D FUN DAM ENTAL CH AN N EL TRANSMITTER

Forward P ower Control SubChannel in F-FC H

M EASU RE REC EIV ED E b /N t O F FU NDAM ENTAL C HANN EL

O U TER LO O P

INN ER LO O P

BA SE STATIO N FORW AR D POW ER CO NTRO L

Reverse P ower Control SubChannel in Reverse Pilot Channel

F-FCH FER ESTIMATIO N

PRIM AR Y INNER LO O P

F-FCH O U TER LO O P

F-FCH E b /N t S ETPO IN T

BASE STATIO N

M OBILE STATIO N F PC_F CH_INIT_SET PT F PC_SUBCHAN_G AIN F RO M BAS E ST AT IO N

F PC_MO DE F PC_PR I_CH AN F PC_F CH_FER F PC_MIN_SET PT F PC_M AX_S ET PT F RO M BAS E ST AT IO N

Figure 6-3

Forward packet data primary closed loop for FCH FPC

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 17

Power control

Forward power control

FORWARD SUPPLEMENTAL CHANNEL TRANSMITTER

Forward Power Control Sub-Channel in F-FCH

MEASURE RECEIVED Eb/Nt OF SUPPLEMENTAL CHANNEL

OUTER LOOP

INNER LOOP

BASE STATION FORWARD POWER CONTROL

Reverse Power Control Sub-Channel in Reverse Pilot Channel

F-SCH FER ESTIMATION

F-SCH OUTER LOOP

SECONDARY INNER LOOP

F-SCH Eb/Nt SETPOINT

BASE STATION

MOBILE STATION FPC_FSCH_INIT_SETPT FROM BASE STATION

FPC_MODE FPC_FSCH_FER FPC_FSCH_MIN_SETPT FPC_FSCH_MAX_SETPT FROM BASE STATION

Figure 6-4

Forward packet data secondary closed loop for SCH FPC

Forward SARA for 3G-1X packet data calls

The forward supplemental air resources allocation (F-SARA) is a mechanism residing at the base station and it determines whether the air interface resources are sufficient to be appropriately assigned to an F-SCH when the anchor cell receives a request for F-SCH assignment. Prior to invoking the F-SARA, the call-processing algorithm estimates a maximum F-SCH data rate based on CE availability, Walsh code, packet data, and other required hardware and software resources without accounting for the RF air interface resources. This maximum F-SCH data rate serves as initial input to the F-SARA algorithm for determining a more accurate F-SCH maximum data rate that can be supported by the current RF conditions in the anchor sector/carrier. This new output data rate may be less than or equal to the earlier input data rate. Also predicted by F-SARA are the initial, the minimum, and the maximum transmitted F-SCH power, and the initial, the minimum, and the maximum F-SCH Eb/Nt set point values corresponding to the output F-SCH data rate. ...........................................................................................................................................................................................................................................................

6 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Power control

Forward power control

Under certain conditions, the call-processing algorithm may update the data rate to a rate lower than that previously determined via the FSARA algorithm. The F-SARA reassesses the power commitment as it determines an updated, maximum-allowable data rate iteratively.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

6 - 19

Power control

Forward power control

...........................................................................................................................................................................................................................................................

6 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

7

Extended carrier

Overview ............................................................................................................................................................................................................................................................

Purpose

Contents

This chapter provides guidelines for RF planning for “extended” carrier deployment. Introduction

7-3

Single extended carrier

7-6

Reverse link Forward link Forward link pilot channel Forward link traffic channel Forward Data Capacity Growth strategies Multiple extended carriers with traffic growth Additional cell sites with traffic growth Applications Low traffic areas Building penetration Concentric carriers

7-6 7-8 7-9 7-10 7-14 7-15 7-15 7-15 7-18 7-18 7-18 7-19

Core carrier reverse link Core carrier forward link Traffic density Determining mobile location Growth strategies Applications Amplifier sharing - Quasi omni

7-20 7-23 7-25 7-25 7-26 7-26 7-28

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 1

Extended carrier

Growth strategies Amplifier sharing - Asymmetric cell Growth strategies Summary

7-29 7-31 7-32 7-33

...........................................................................................................................................................................................................................................................

7 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Introduction

Introduction ............................................................................................................................................................................................................................................................

This chapter provides guidelines for RF planning for “extended” carrier deployment. An extended carrier, as used here, is a CDMA carrier that is intentionally designed to carry a limited amount of traffic in order to increase the coverage area, or reap other benefits such as enhancing building penetration or better matching the offered traffic density to subscriber demands. The concept of reducing design capacity in order to achieve extended coverage is not new; indeed, this fundamental trade-off exists in 2G CDMA and has occasionally been exploited to advantage (e.g., a modest number of large, lightly loaded cells covering a low-traffic rural area). These 2G trade-offs have been naturally limited by the low reverse link interference margins7 (about 3 to 4 dB) used in 2G designs. For example, an interference margin of 3.5 dB (55% loading with respect to pole) means that at most the cell can be expanded 3 dB relative to this footprint, with associated reduction of the interference margin to 0.5 dB (10% loading with respect to pole). Further expansion of the footprint by sacrificing capacity is not possible, since the cell capacity would be driven to zero. The use of higher (i.e., typically 5.5 dB) nominal interference margins in 3G-1X opens several new possibilities for design. These include: •

Single extended carrier. This concept embodies the standard design trade-off of capacity for coverage. This trade-off can be more extensive, since there is more dB of interference margin (loading) to trade for coverage.

•

Concentric extended. This configuration uses a modest number of large, lightly loaded single carrier cells for initial deployment. The expanded footprint of the cells is achieved by trading off capacity (interference margin) for coverage. Traffic growth is accommodated by adding fully loaded carriers (of smaller footprint) to each cell as needed. The first (extended) carrier provides ubiquitous coverage, whereas the additional (smaller footprint) carriers provide localized capacity relief.

...........................................................................................................................

7 Note that reverse link will be left off the name of the interference margin throughout the rest of this chapter. The interference margin referenced here is always a reverse link term.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 3

Extended carrier

Introduction

•

Quasi-omni. This strategy services a 3-sectored configuration with a single (as opposed to three) transmitter/receiver. The coverage penalties inherent in sharing the single transceiver via a splitter and combiner are offset by reducing the loading (the interference margin). This lightly loaded configuration can later be upgraded to full capacity at the same footprint by adding two transceivers (the additional equipment offsets the coverage penalties introduced by the additional loading).

•

Asymmetric cell (“split sector”). This strategy is similar to the previous one but services a 3-sectored configuration with two transmitters/receivers instead on just one. The first (dedicated) transceiver services the cell busy sector, which offers full capacity. The second transceiver services the remaining two sectors. The coverage of these two sectors is identical to that of the busy sector, since the penalties inherent in sharing (splitting) the transceiver are offset by reducing the sector loading. This configuration provides full (nominal) coverage for cells with asymmetric traffic distributions, at reduced cost.

In the following sections, we consider each of these methods in turn. In the "Single extended carrier" section on Page 7-6, the mechanics of basic capacity-coverage trade-offs are reviewed, with particular attention paid to required forward link adjustments in an expanded cell. The concentric configuration is discussed in "Concentric carriers" section on Page 7-19. The quasi-omni and asymmetric cell configurations are discussed in "Amplifier sharing - Quasi omni" section on Page 7-28 and "Amplifier sharing - Asymmetric cell" section on Page 7-31. Note that all strategies discussed provide a potential means for reducing the cost of initial deployment either through reducing cell or equipment (transceiver) count. The optimal strategy for a given deployment depends largely upon traffic needs and projected traffic growth. For example, a single extended carrier may not be feasible for an area with aggressive traffic growth, since the small design capacity per (large) cell would necessitate rapid addition of carriers. In such an area, a configuration that begins with quasi-omni and is later upgraded to normal (3 transceiver) configuration may be a more suitable way to contain initial deployment costs and smoothly migrate as needed to higher capacity cells. Alternatively, if the projected traffic within the area is likely to be asymmetric (one busy sector), then the best solution

...........................................................................................................................................................................................................................................................

7 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Introduction

may be deployment of the split-sector configuration. Finally, if the traffic growth is likely to be highly localized close to the cells, then concentric carrier may offer the best answer.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 5

Extended carrier

Single extended carrier

Single extended carrier ............................................................................................................................................................................................................................................................

This section describes the deployment of a carrier with extended coverage. If capacity demands exceed the capacity of the extended carrier, then either more extended carriers or more cells are required. Reverse link

The design trade-off for capacity and coverage in the reverse link is embedded in the interference margin term. The interference margin is defined as: 1 Rim = 1− µ (see Lucent document 401-703-201, PCS CDMA RF Engineering Guidelines, equation 7.6) where: µ is the ratio of the planned number of RF channels to the “pole capacity”. The reduced interference margin directly translates to an increased maximum allowable path loss. The cell radius is proportional to the maximum allowable path loss raised to the path loss slope. Therefore, changes in maximum allowable path loss can be translated to changes in cell radius as follows:   R1 = 10 R2

P1 −P2   S 

where the Rs are the respective cell radii, Ps are the respective maximum allowable path losses (in dBs), and S is the path loss slope (in dB/decade). For a typical 3G-1X system with 3-sector cells and Radio Configuration 3 (RC3), the pole capacity is 48.5 channels. The following table summarizes the capacity coverage trade-off.

...........................................................................................................................................................................................................................................................

7 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Single extended carrier Table 7-1

Capacity versus coverage

Channels

% load relative to pole

Erlangs

Interference Margin (dB)

Area rel to 72% loading

1

2%

0.02

0.1

191%

2

4%

0.223

0.2

189%

3

6%

0.602

0.3

187%

4

8%

1.09

0.4

185%

5

10%

1.66

0.5

183%

6

12%

2.28

0.6

181%

7

14%

2.94

0.7

179%

8

16%

3.63

0.8

176%

9

19%

4.34

0.9

173%

10

21%

5.08

1.0

171%

11

23%

5.84

1.1

169%

12

25%

6.61

1.2

166%

13

27%

7.4

1.4

164%

14

29%

8.2

1.5

162%

15

31%

9.01

1.6

159%

16

33%

9.83

1.7

157%

17

35%

10.7

1.9

154%

18

37%

11.5

2.0

152%

19

39%

12.3

2.1

149%

20

41%

13.2

2.3

147%

21

43%

14

2.4

144%

22

45%

14.9

2.6

142%

23

47%

15.8

2.8

139%

24

49%

16.6

2.9

136%

25

52%

17.5

3.2

132%

26

54%

18.4

3.4

129%

27

56%

19.3

3.6

126%

28

58%

20.2

3.8

123%

29

60%

21

4.0

120%

30

62%

21.9

4.2

117%

31

64%

22.8

4.4

114%

32

66%

23.7

4.7

110%

33

68%

24.6

4.9

107%

34

70%

25.5

5.2

103%

35

72%

26.4

5.5

100%

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 7

Extended carrier

Single extended carrier

The following figure shows the trade-off of capacity versus cover graphically where the coverage is expressed as a percentage of the nominal case (72% reverse link pole loading):

30 Erlang Capcity

25 20 15 10 5 0 100%

120%

140%

160%

180%

200%

Area Relative to Nominal Case

Figure 7-1

Capacity vs. coverage

Traffic Density Rel to Nominal

Of course it is important to remember that supportable Erlang density (traffic Erlangs divided by area) falls faster than the plot above (Erlang capacity), since as the cell footprint grows the capacity decreases. Thus, the density (ratio of capacity to area) is negatively impacted twice. The following plot illustrates supportable density relative to nominal case of 72% loading, versus area gain:

Figure 7-2

Forward link

100% 80% 60% 40% 20% 0% 100%

120%

140%

160%

180%

200%

Area Relative to Nominal Case

Traffic density versus coverage

It is necessary to verify that the forward link will support the extended coverage area by examining the impact on the forward link pilot and traffic channels.

...........................................................................................................................................................................................................................................................

7 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Single extended carrier

Forward link pilot channel

As the maximum allowable path loss increases, pilot power must also be increased to maintain a constant pilot channel Ec/Io at the cell edge. Pilot Ec/Io is defined as follows:

 Ec  δ ⋅ Pi   =  I o  i FNoW + ∑ Pj all _ j

where: Ec = The time chip energy received at the mobile Io = The total noise and interference from all sectors δ = The fraction of sector power allocated to the pilot channel Pj = The power received from the jth sector i = Index of serving sector F = Mobile receiver noise figure No = Thermal noise density W = The carrier bandwidth. For insight, we can consider two simple cases analytically: The completely interference-limited case, and the completely noise-limited case. In the interference-limited case, we assume the thermal noise power to be small compared to the interference power (i.e., FNoW << ΣPj). In this case, as the maximum allowable path loss is increased, both the pilot signal and the interference are reduced by the same amount. Hence, in the interference-limited case, no increase in pilot power is required, as the cell area is increased by decreasing the loading. In the noise-limited case, we assume that the interference power is small compared to the thermal noise power (i.e., ΣPj << FNoW). In this case, the pilot power will need to be increased by the same amount as the increase in path loss to maintain the same pilot Ec/Io at the cell edge. Intermediate cases (the most likely scenario) require the pilot channel power to be increased somewhere between zero and the increase in maximum allowable path loss. The actual pilot power required, as a percentage of total amplifier power, to maintain a given Ec/Io at the cell edge, was computed (via spreadsheet) for the typical 3G-1X case for various values of interference margin (resulting in various cell footprint sizes), with the following results: ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 9

Extended carrier

Single extended carrier

Pilot % of Total Power

18.0% 17.5% 17.0% 16.5% 16.0% 15.5% 15.0% 100%

120%

140%

160%

180%

200%

Area Relative to Nominal Case

Figure 7-3

Required pilot percentage of total power versus capacity

As can be seen from the graph, the pilot percentage only increases from the standard value of 15 percent at full loading of 26.4 Erlangs (72% of pole capacity) to a little less than 18 percent at no loading. The additional amount of power required for the pilot channel can be calculated on a case-by-base basis and will reduce the power available for the traffic channels. The impact on capacity is examined in the next section. Forward link traffic channel

The traffic channel is more complicated than the pilot channel since two effects of lighter loading must be accounted for: The increase in path loss, and the decreased number of users. In our analysis, we assume that all mobiles have an equal share of total base station power, which can be interpreted as all mobiles are located at the cell edge.

Pi′ n ⋅ Lp

 Eb    =  N o  FNoW + ∑ Pj + γ ⋅ Pi j ≠i

Where: i is the index of the serving sector P' is the transmitted power for all traffic channels Lp is the maximum allowable path loss n is the number of mobiles Pi = The power received from the ith sector γ is the orthogonality factor ...........................................................................................................................................................................................................................................................

7 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Single extended carrier

F = Mobile receiver noise figure No = Thermal noise density W = The carrier bandwidth. Again, for insight, we can consider two simple cases analytically: The completely interference-limited case, and the completely noise-limited case. In the interference-limited case, as the maximum allowable path loss is increased, the P' terms stays the same since the pilot channel does not require any extra power, as discussed above. The increase in path loss reduces both the serving signal and the interfering signals equally. The number of users (n) decreases to provide the reduced loading that allows for the increase in maximum allowable path loss. Therefore, the Eb/No for this case will necessarily increase. Alternatively, the total power required to maintain a given Eb/No will decrease. Accordingly, less power is required to support fewer users, even though the footprint is enlarged. In the noise-limited case, the P' term will be decreased by amount equivalent to the increase in pilot power, which, from above, would equal the increase in path loss. If we assume that the Eb/Nt achieved for the nominal case is acceptable, we can compare the achievable Eb/Nt for the expanded carrier case. Here, the achieved Eb/Nt is the Eb/Nt that is calculated for a mobile at the edge of a sector’s coverage area if that mobile receives 1/n of the available traffic power, where n is the number of channels supported by the sector. If the ratio of the extended carrier achievable Eb/Nt to the nominal achievable Eb/Nt is greater than one, we can assume that the extended carrier case has sufficient power to close the forward link. Let case 1 be the “nominal carrier case”, supporting n1 RF channels and case 2 be the “extended carrier case”, supporting n2 RF channels. P2′  Eb   N  L ⋅n  P′   L   n   o 2 = p 2 2 =  2  ⋅  p1  ⋅  1    P1′  Eb   P1′   L p 2   n2   N   L p1 ⋅ n1 o 1 If the overhead fraction is δ, then power available for all the traffic channels is: P1′ = (1 − δ 1 )⋅ Ptot

and

P2′ = (1 − δ 2 )⋅ Ptot

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 11

Extended carrier

Single extended carrier

As stated above, the overhead fraction for this noise limited case increases by an amount equal to the increase in path loss:

δ2 =

Lp 2 Lp1

⋅ δ1

Therefore, the ratio of total traffic powers can be expressed as:

P2′ (1 − δ 2 )⋅ Ptot = = P1′ (1 − δ 1 )⋅ Ptot

1−

Lp 2 Lp1

⋅ δ1

1 − δ1

Substituting back into the Eb/No ratio and simplifying gives:

Lp2    L p1   Eb  1    − ⋅ − δ1  δ 1  N  L p1  P′   L   n     L p1   n1   L p 2   n1   o 2 =  2  ⋅  p1  ⋅  1  =  ⋅ ⋅   =    ⋅  n  L  n P1′   L p 2   n2  1 − δ1 1 − δ  Eb     2 1   2   N     p2         o 1     The ratio of path losses can be related to the number of channels as follows:

Lp1 Lp 2

=

Rim2 Rim1

1 1 − µ 2 1 − µ1 = = 1 1 − µ2 1 − µ1

Substituting into the Eb/No ratio equation gives:

 1 − µ1   Eb  − δ1    N  1 − µ2  o 2  ⋅  n1  =  1 − δ 1   n2   Eb   N     o 1   For 3G-1X, the loading for the nominal case is 72%, or µ is 0.72. Also, the pilot fraction for the nominal case (δ) is equal to 15%, or 0.15.

...........................................................................................................................................................................................................................................................

7 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Single extended carrier

 Eb   1 − 0.72 − 0.15   N  1 − µ2  o 2  ⋅  0.72 ⋅ nmax  =  0.33 − 0.18 ⋅  0.72  =  1− µ   µ  n2  Eb   1 − 0.15       2  2  N     o 1  

=

0.33⋅ 0.72 0.18 ⋅ 0.72 0.11 + 0.13⋅ µ2 − = µ2 − µ22 µ2 µ2 − µ22

A plot of this function is shown in the figure below.

Eb/No relative to nominal case (dB)

5 4 3 2 1 0 -1 -2 0

0.2

0.4

0.6

0.8

Loading (mu)

Figure 7-4

Eb/Nt versus loading

From the above figure it can be observed that the achieved Eb/No for the extended carrier case will be less than for the nominal case for loadings between nominal (0.72) and about 0.16. Therefore no general conclusion that the traffic channel will achieve the required Eb/Nt for the noise limited case can be drawn. Therefore a full link budget analysis is required to examine real scenarios that fall between the noise limited and interference limited cases. The actual achieved forward link Eb/No at cell edge was computed (via spreadsheet) for a typical case (i.e. not either extreme of interference or noise limited) with the results shown in the following figure.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 13

Extended carrier

Single extended carrier

Achieved FL Eb/No (dB)

25 20 15 10 5 0 100%

120%

140%

160%

180%

200%

Area Relative to Nominal Case

Figure 7-5

Achieved forward link Eb/Nt versus capacity

As can be seen from the above figure, the achieved Eb/Nt grows with the decrease in Erlang capacity, which is to say that the required Eb/Nt is achieved, and the forward link should close. Forward Data Capacity

One would also expect that as the cell radius is increased, the data capacity of the cell would decrease. There is no simple analytical approach to deriving a data capacity versus cell radius relation; hence, simulations were run to model the behavior. The simulations focused on the impact to the forward link since data applications are expected to be asymmetric and have much lower reverse link demands relative to forward link demands. The simulations assumed a typical link budget. The simulation was a set of single rate simulations, whose outputs (per rate throughputs) were combined with a standard rate distribution. The single rate distribution was a typical forward link simulation where the number of users was increased until a certain probability of outage (defined as exceeding max amplifier power) was exceeded. The results of the simulation are shown in Figure 7-6.

...........................................................................................................................................................................................................................................................

7 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Forward Link Aggregate Sector Throughput (kbps)

Single extended carrier

112 110 108 106 104 102 100 98 96 94 92 90 100%

120%

140%

160%

180%

200%

Area Relative to Nominal Case

Figure 7-6

Growth strategies

Forward link data capacity versus cell size

Multiple extended carriers with traffic growth

The simplest way to add capacity is to add carriers. Carrier additions provide a linear growth in capacity, i.e., two carriers doubles Erlang capacity, and three carriers triples Erlang capacity, etc.8 However, adding cells has some advantage, as will be discussed next. Additional cell sites with traffic growth

Adding cells enhances capacity in two ways: 1. 2.

The greater the number of cells, the less area per cell, and hence, a higher interference margin can be tolerated. The additional cell can carry additional capacity.

The network capacity is the product of the capacity per cell and the number of cells. Adding cells increases both of these terms, and hence, provides a double benefit to network capacity. For example, doubling network capacity by adding carriers requires adding an additional carrier to all cells in the network. Doubling network capacity by adding cells does not require a doubling of cell count. For example, if the starting point was cells designed for 1.5 times the nominal cell area by reducing the capacity to 12.3 Erlangs per sector, the network capacity could be theoretically doubled by reducing cell area to 1.23 times ...........................................................................................................................

8 The capacity growth versus number of carriers is slightly greater than strictly linear due to trunking efficiency. The trunking efficiency is not the full value predicted by Erlang B, but is greater than 0.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 15

Extended carrier

Single extended carrier

nominal with a capacity of 20.2 Erlangs per sector, since 2*12.3/1.5=20.2/1.23. The increase in cell count would be 1.5 divided by 1.23, or 22%. Whether this approach is more cost effective than simply adding carriers depends on the relative costs of the hardware to support the additionally carriers versus the cost associated with the new cells (hardware, real estate, backhaul, etc.). Although the above example demonstrates the nonlinear (more than linear) gain that can be achieved by cell addition, the result is at best approximate since it assumes that the cell count is simply the total network area divided by the area per cell. However, once a network is deployed, it is unlikely that cells will be moved. So the actually increase in number of cells would probably be higher than the 22% computed. Consider the following figure that shows the typical hexagonal geometry for a 7-cell cluster. The cells are spaced at 1.5 times the nominal cell spacing. The borders shown are the nominal cell size borders. Cell/Mobile Map 3 4

3

2

1

0

5

2

1

-1

-2 6

7

-3 -3

Figure 7-7

-2

-1

0

1

2

3

Cell deployment at 1.5X typical cell radius

To fill the coverage holes would require on the order of 6 new cells, as shown in heavy red on the following figure.

...........................................................................................................................................................................................................................................................

7 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Single extended carrier

Cell/Mobile Map 3 4

3

2

1

0

1

5

2

-1

-2 6

7

-3 -3

Figure 7-8

-2

-1

0

1

2

3

Cell deployment to fill holes at full capacity

This method of cell addition can clearly be inefficient in the sense that coverage overlay inevitably occurs; however, from a traffic perspective, this method of adding cells allows selective focus on areas where traffic demand is highest. Accordingly, the net cell count for a coverage-driven area with isolated hot traffic spots is likely to be less than the number required if an initial dense array of cells were uniformly deployed. Another issue associated with adding cells is that the network may require reoptimization. However, the costs of reoptimization maybe minimized through the use of Lucent's Ocelot tool. Ocelot uses a general nonlinear optimization procedure to adjust certain parameters (e.g., antenna tilts, forward powers) of cellular networks in order to maximize a particular “objective function”. The current objective function is various combinations of coverage (the percentage of the served area where a call can be made from) and capacity (how much traffic can be carried simultaneously). When Ocelot runs an optimization, the user sees a Trade-off Curve window with different coverage/capacity points; clicking any point affords a detailed examination of the proposed design in a graphical display of the market area. It is expected that the original design and optimization will provide a baseline set of data that will allow Ocelot to generate accurate predictions of the revised optimization settings appropriate for additional cell sites.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 17

Extended carrier

Single extended carrier

Applications

Low traffic areas

The simplest application of extended carriers is to serve low traffic areas. Using larger cells will reduce cell count, reducing expenses associated with each cell such as: •

Cell site hardware

•

Real estate

•

Backhaul facilities.

If traffic grows beyond the planned capacity coverage holes will occur unless steps (e.g., added cell count) are taken. Of course this statement is true regardless of whether the carrier is “extended” or not. However, the lower capacity of an extended carrier means that the planned capacity is lower than frequently employed, and hence, extra attention must be paid to traffic growth to ensure the extended carrier does not suffer overload. Building penetration

Another use for the extended carriers is to provide building penetration margin. The extra interference margin on the reverse link and extra power on the forward link are used to provide building penetration margin rather than extended radius in this case. It would be expected that concentric carriers would be used, with the core carrier serving pedestrian traffic and vehicular traffic (and indoor traffic close to the cell site), while the extended carrier serving in-building traffic toward the cell edge. Arguably, if buildings that are important are known, it is better to place cells near or in those buildings.

...........................................................................................................................................................................................................................................................

7 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Concentric carriers

Concentric carriers ............................................................................................................................................................................................................................................................

Another application for extended carriers is a multi-carrier cell with carrier-dependent coverage. In this scenario, an extended carrier is used to extend the coverage of the cell and a “core” carrier to provide the bulk of the capacity of the cell as shown in the following figure:

Core Carrier

Extended Carrier

Figure 7-9

Core and extended carriers

In this scenario, the extended (first) carrier provides ubiquitous coverage across the region of interest with a modest number of cell sites. Each extended carrier offers low capacity only, since its capacity has been traded away for expanded coverage. As traffic increases, smaller full-capacity carriers are added as needed at selected cells. The smaller carriers address the additional capacity, which is presumed to be locally concentrated around the cell sites. Handoffs between the core and extended carriers allow mobiles to traverse between cells, while restricting the number of active mobiles on the extended carrier. This configuration alters a number of RF engineering considerations, which typically apply to carriers of identical footprint. These are discussed below, and include: •

Core carrier reverse link

•

Core carrier forward link

•

Core and extended carrier traffic densities.

Note that the RF engineering issues associated with the extended carrier reverse and forward link are identical to those in the "Single extended carrier" section of this chapter and are not re-examined here.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 19

Extended carrier

Concentric carriers

Core carrier reverse link

As explained in the Lucent documents 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines and 401-703-201, PCS CDMA RF Engineering Guidelines, the pilot Ec/Io at the edge of a cell should be equal to T_ADD9. The handoff zone is the area where one cell's pilot is above T_ADD and another cell’s pilot is above T_DROP. Therefore, it is expected that handoff zone is an area where the pilot Ec/ Io changes by the difference between T_ADD and T_DROP, which is typically 2 dB. Changes in pilot Ec/Io are not precisely equal to differences in path loss, but can be taken as an approximation. The difference between the core and extended carriers is expected to be greater than 2 dB. Therefore, little soft handoff is expected in the core carrier coverage area. The impact of no soft handoff on the link budget is to shrink the reverse link coverage by an amount equal to the soft handoff gain. The actual carrier coverages will look something like the following figure, where: •

The extended carrier coverage is increased by reducing the interference margin and maintaining full soft handoff gain

•

The nominal carrier coverage is the coverage of a carrier with nominal interference margin (i.e., corresponding to 72% loading) and full soft handoff gain

•

The core carrier coverage is the coverage of a carrier with nominal loading (i.e., corresponding to 72% loading) but with no soft handoff gain. Extended Carrier Nominal Carrier Core Carrier

Difference in Interference Margin

Loss of Soft Handoff Gain

Figure 7-10

Core, nominal, and extended carriers

...........................................................................................................................

9 For simplification the IS-95A terms are used here, but the same discussion applies to networks utilizing the IS-95B soft handoff algorithm.

...........................................................................................................................................................................................................................................................

7 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Concentric carriers

Since the coverage areas of the core carriers do not overlap or even touch the expected interference ratio is less. The reduced interference ratio will lead to an increase in reverse link capacity. This increase can be advantageous since the core carrier by design services localized areas of high traffic demand. In the following we estimate the reverse link interference ratio in order to compute the core carrier pole capacity. The loss of soft handoff gain will cause the core carrier to be 4.3 dB less in maximum allowable path loss than a nominal carrier. The difference between the nominal carrier coverage and the extended carrier coverage is the difference in interference margin and is a design parameter. If we consider 3 dB to be a typical amount for the reduction in interference margin for the extended carrier, the total difference in path loss between the core and extended carrier is 7.3 dB.

Rext

Rcore Figure 7-11

Rc-c

Definition of different distances

Therefore the ratio of the radius of the extended carrier in terms of the radius of the core carrier is: Pec−Pcore S

Rext = Rcore⋅10

where Pec is the maximum allowable path loss for the extended carrier, Pcore is the maximum allowable path loss for the core carrier, and S is the path loss slope. The difference between extended carrier and core carrier maximum allowable path losses is 7.3 dB, as stated above. Therefore, the extended carrier radius is 1.55 times the radius of the core carrier (assuming a path loss slope of 38.5 dB per decade). The distance between the centers of the cells, Rc-c, is 3.10 times the core carrier (twice the radius of the extended carrier). Therefore, the path loss from the center of one cell to the other cell in terms of the path loss to the edge of the core carrier is: ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 21

Extended carrier

Concentric carriers

R  Pc−c = Pcore + S ⋅ log c−c  = Pcore + 38.5 ⋅ log(3.1) = Pcore + 18.9  Rcore  For the nominal case (72% loading), the center-to-center distance is simply twice the core carrier distance and hence the center-to-center path loss is:

R  Pc′−c = Pcore + S ⋅ log c−c  = Pcore + 38.5 ⋅ log(2) = Pcore + 11.6  Rcore  Therefore, the extended carrier case has 7.3 dB more path loss between a cell and its first tier interferers. The reverse link interference ratio, β, is defined as the ratio of the other cell to same cell interference. As a first order approximation, we can treat the interference from other cells as coming from a point at the center of the other cells. By increasing the path loss by 7.3 dB to those other cells from the nominal case, the interference from those other cells should be reduced by 7.3 dB. So the interference ratio, β, should also be reduced by 7.3 dB. For the 3-sector case the interference ratio would then be reduced from 0.85 to 0.16. The pole capacity for this reduced interference ratio is:

nmax =

128 g +1 = + 1 = 77 RF channels 4 ) ( 10 α ⋅ d ⋅ (1 + β ) 0.58 ⋅ 10 ⋅ (1 + 0.16)

If the typical 3G-1X loading of 72% is assumed, the core carrier will support 55 RF channels. Since no soft handoff is expected on the core carrier, this number of channels needs to be increased by only a factor to account for the softer handoff links, which is 1.3. Therefore, the number of channels is 72. This value exceeds the number of Walsh codes available, which is 59. Given the 1.3 factor for softer handoff links, the Walsh code limit translates to a limit of 45 “primary” RF channels per sector. The loading cannot simply be reduced to the value associated with this number of channels since as the loading, as a percentage of pole capacity is reduced, the interference margin is decreased. However, this will change the coverage of the core carrier, and hence, our computed interference ratio. Therefore, the optimum solution can only be found through an iterative trial and error process. A solution was found for the following conditions.

...........................................................................................................................................................................................................................................................

7 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Concentric carriers

Parameter

Value

Interference Ratio

0.22

Pole capacity

73

Loading

61%

RF channel capacity

45

Erlang capacity

35.6

Interference Margin (dB)

4.1

The process outlined above can be repeated for different values of extended carrier interference margin reduction with the following results: Extended Carrier

Core Carrier

Interference Margin Reduction (dB)

Forward Link Interference Ratio (linear)

Pole Capacity

Loading (% of pole point)

RF Channel Capacity

0.5

0.329

67.06

67.1%

45

35.6

4.8

1

0.304

68.32

65.9%

45

35.6

4.7

1.5

0.281

69.64

64.6%

45

35.6

4.5

2.0

0.257

70.73

63.6%

45

35.6

4.4

2.5

0.237

72.14

62.4%

45

35.6

4.2

3.0

0.217

73.25

61.4%

45

35.6

4.1

3.5

0.197

74.22

60.6%

45

35.6

4.0

4.0

0.180

75.46

59.6%

45

35.6

3.9

4.5

0.164

76.54

58.8%

45

35.6

3.8

5.0

0.148

77.40

58.1%

45

35.6

3.8

Core carrier forward link

Erlang Capacity

Interference Margin (dB)

The core carrier forward link must be assessed on a case-by-base basis to ensure link balance. The issues affecting the ability of the forward link to support the reverse link are discussed below. The forward link traffic channel coverage of the core carrier will also suffer due to the loss of soft handoff gain. Soft handoff gain is not explicitly listed in the forward link Eb/Nt analysis, but instead is

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 23

Extended carrier

Concentric carriers

embedded in a reduced fade margin for the forward link. The fade margin listed in the forward link is actually reduced fade margin. The reduction is due to both soft handoff gain and other effects. The reduction for 95% area coverage is 6.0 dB. While the Eb/Nt analysis does not state what proportion of this is due to soft handoff and what is due to other effects (independence of fading within the cell, limited dynamic range of forward link transmit power), it is expected that the soft handoff gain would be no greater than the value for the reverse link soft handoff gain, which is 4.0 dB. However, the forward link benefits from the lack of soft handoff in that no power must be allocated for the soft handoff legs. The impact of the lack of soft handoff is manifested in the forward link Eb/Nt analysis by setting the soft handoff overhead factor to 1.3 (value for softer handoff) instead of 1.75 typically used for 3G-1X. This difference in soft handoff overhead factor leads to a corresponding increase in traffic channel power of 1.3 dB. The net impact of no soft handoff on the received traffic channel signal in the Eb/Nt analysis is a loss of 2.7 dB (4.0 -1.3). The forward link of the core carrier also benefits in terms of forward link interference ratio. The lack of soft handoff increases the interference ratio since the power from all the sectors involved in the soft handoff are excluded from the interference term. However, the fact that the border of the core carrier is within the cell border reduces the interference ratio more than the lack of soft handoff increases it. The reduction in interference ratio for the case considered here (no soft handoff on inner border and inner border 5 dB inside outer border) is believed to be up to 6 dB. The reduction in the interference ratio reduces the other cell interference term. The overall effect on the core carrier forward link depends on to the ratio of the other cell interference to the thermal noise. In a noise-limited system, the reduction in other cell interference will provide little benefit and the forward link will fall short of power. In an interference-limited system, the reduction in other cell interference will more than make up for the reduced received traffic channel signal. The typical case considered here was analyzed, and the forward link did balance.

...........................................................................................................................................................................................................................................................

7 - 24

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Concentric carriers

The forward link pilot channel does not have a soft handoff gain. So the loss of soft handoff gain does not penalize the core carrier pilot coverage. The reduced interference ratio will benefit the core carrier pilot channel, and hence the pilot channel coverage in the core carrier area is not an issue. Traffic density

By design, the traffic density in the extended carrier coverage area must be less than the traffic density of the core carrier area. The difference depends upon the extent to which the extended carrier capacity has been lowered in design in order to expand coverage. The plot below shows the design traffic density, relative to the design density in the core carrier coverage area, versus the design area of the extended carrier relative to the nominal carrier design area. As coverage of the extended carrier grows, the traffic density between the core and extended carriers becomes more imbalanced.

Extended Carrier Erlang Density Relative to Core Carrier

Extended Carrier Erlang Density 1.00 0.80 0.60 0.40 0.20 0.00 1.00

1.20

1.40

1.60

1.80

2.00

Extended Carrier Area Relative to Nominal Carrier

Figure 7-12

Determining mobile location

Extended carrier traffic density versus coverage

To make the concentric carrier approach work, it is necessary to avoid violating the design capacities of the core and extended carrier. To keep the extended carrier lightly loaded, all mobiles in the coverage area of the core carrier need to be served by the core carrier. Also, mobiles outside the core carrier coverage area need to be served by the extended carrier or they will suffer degradation (e.g., high FER, call drop, etc.).

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 25

Extended carrier

Concentric carriers

Through call processing, it is possible to have the mobile report (PPSMM message) pilot strength as measured in terms of Ec/Io and overall interference in the band, Io. Multiplying these two terms together will provide the mobile's received energy per pilot chip, Ec. The base station knows the transmitted energy per chip. The difference between transmitted and received energy per chip is the path loss. The base station can then use this path loss value to estimate whether the mobile is in the core carrier's coverage area or the extended carrier's coverage area10. The same measurement of path loss would also be used for triggering inter-frequency handoffs at the boundaries between the core and extended carriers. Currently, the capability to make this estimate of path loss does not exist in the Lucent products. A new feature is required to support this capability. The deployment of concentric carriers will cause an increase in interfrequency hard handoff. While Lucent has implemented several features that increase the robustness of inter-frequency handoffs, interfrequency handoffs are still hard handoffs are hence inherently less reliable than soft handoffs. Therefore, it is possible that some increase in call drop rate could result from the deployment of concentric carriers. This increase could be minimized by careful optimization, particularly in an area where the mobile locations are concentrated (e.g., along rural highways) and the locations of hard handoffs are well known. Growth strategies

As traffic demand grows in the core carrier region, clearly the growth path is to add carriers. As traffic demand grows in the extended carrier region the same alternatives (adding carriers or adding cells) and tradeoffs apply as in the simple extended carrier case, as discussed in "Growth strategies" section on Page 7-15. Note that since the core carriers are placed at traffic hot spots, the pattern of growth could well dictate that multiple additional core carriers are added well before a second extended carrier is required.

Applications

The concentric carrier approach makes sense for regions of low traffic density punctuated by localized hot spots, such as scattered small towns or villages surrounded by a rural area. The town would have to be small enough to fit within the footprint of the core carrier. The traffic demand ...........................................................................................................................

10 One caveat to this approach is that IS-98 does not specify how accurately the mobile must measure Io.

...........................................................................................................................................................................................................................................................

7 - 26

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Concentric carriers

from within the town would have to be small enough to be served by the number of carriers available. The areas around the town are expected to generate light traffic demand, and hence, be ideal for the extended carrier.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 27

Extended carrier

Amplifier sharing - Quasi omni

Amplifier sharing - Quasi omni ............................................................................................................................................................................................................................................................

One novel application for extended carriers is amplifier sharing. Schematically, amplifier sharing can be illustrated by the following figure:. Beta

Alpha

LA

Rx

Tx

Rx

Rx

Tx

Rx

Rx

Combiner

Radio

Rx

Splitter

Tx

Gamma

Figure 7-13

Quasi-omni illustration

In this configuration, a single amplifier and single receiver service 3 sectors (“quasi-omni”). The splitter splits the power from the linear amplifier three ways, reducing the power per antenna by 1/3 or -4.8 dB.11 The combiner combines the signals from the three faces (alpha, beta, and gamma), and hence, increases the reverse link noise figure by a factor of 3 or 4.8 dB. The coverage advantage gained by reducing capacity can be used to overcome the combiner and splitter disadvantages instead of extending the cell radius. Each sector is lightly loaded but the footprint of the cell remains the same as that of a fully loaded, conventional 3-sector cell. In the reverse link budget, the increased noise figure directly translates to a decrease in maximum allowable path loss. In the forward link, as shown previously, the decrease in capacity will be sufficient to offset the loss in power (i.e., the link will balance), typically with some margin. ...........................................................................................................................

11 Note that no insertion loss is considered here since values of insertion loss may vary widely. Once hardware is chosen and the insertion loss is known, it should be considered as well.

...........................................................................................................................................................................................................................................................

7 - 28

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Amplifier sharing - Quasi omni

To fully overcome the combiner/splitter disadvantage of 4.8 dB would require reducing the capacity to 2.9 Erlangs per cell. As traffic demand increase past this capacity, more radios and amplifiers are added and the splitter/combiners removed. The penalty for the splitters and combiners is removed from the link budget so there is no longer any need to reduce capacity. The cell can then run at full capacity of 26.4 Erlangs per sector, or 79.2 Erlangs per cell, in the same footprint. This approach has the advantage of lowering initial cost in a deployment (one as opposed to 3 transmitters/receivers per 3-sectored cell), and selectively paying over time as needed for the additional equipment required to address traffic growth. In the case analyzed here, the forward achieved Eb/No is 7.1 dB higher than the nominal case (note that the pilot was increased to 16.8% of total power); clearly, the forward link has more than enough power. This asymmetry can be reduced by the use of Tower Top Low Noise Amplifiers (TTLNA). As discussed in Chapter 8 of the PCS CDMA RF Engineering Guidelines, TTLNAs reduce the reverse link noise figure. The reduction depends on the value for cell site cable loss. Taking 2 dB as a typical value for cell site cable loss, the typical reduction in reverse link noise figure is 1.9 dB. Thus, the net increase between the signal combiner and TTLNA is 2.9 dB (4.8 - 1.9). To achieve this reduction in interference margin requires that the cell capacity be reduced to 14 Erlangs per face, or 42 Erlangs per cell. The forward link shows that there is sufficient power to achieve the same pilot channel Ec/Io and traffic channel Eb/No as the nominal case. Again, as traffic demand increases past the capacity of the cell, the combiner/splitters can be removed, as well as the TTLNA. The degree to which this approach is advantageous depends on the relative cost of a single TTLNA versus the cost of 2 amplifiers and radios. Growth strategies

The benefit of this approach is that it delays the cost of the second and third amplifiers and radios until they are needed, while maintaining the same cell footprint. Thus, a network provider can “pay as they grow”, by simply adding hardware to existing sites. As traffic increases on the cell the network operator can either grow to two amplifiers (see next section) if the traffic demand is asymmetric among the three sectors or to three amplifiers if the traffic demand is roughly equal among the sectors. This decision requires some knowledge of the traffic distribution amongst the sectors. Since all three sectors are served by the same radio, they have the same PN code, and hence, traditional service measurements will not capture per-sector traffic information. However, a network operator can use the Lucent

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 29

Extended carrier

Amplifier sharing - Quasi omni

On-Demand PSMM Collection (ODPC) feature in conjunction with post processing to determine traffic demand patterns. The ODPC feature allows the network operator to collect periodic (period from 1 to 10 minutes) pilot strength information from each mobile on several (up to 20) cells for a specified time period (up to 2 hours). The pilot strength information is stored in a file at the OMP. Post-processing of the pilot strength measurement data, for example using the Lucent EFLT (Enhanced Forward Link Triangulation) algorithm, can determine the mobile location to accuracy sufficient to determine persector traffic demand. Another issue associated with adding amplifiers is that the network may require reoptimization. This process could be required since the addition of amplifiers will clearly impact the internal interference distribution throughout the network, thus necessitating changes in such parameters as antenna downtilts, neighbor lists, and pilot power. However, the costs of reoptimization can be minimized through the use of Lucent's Ocelot tool. Ocelot uses a general nonlinear optimization procedure to adjust certain parameters of cellular networks in order to maximize a particular “objective function”. The current objective function is various combinations of coverage (the percentage of the served area where a call can be made from) and capacity (how much traffic can be carried simultaneously). When Ocelot runs an optimization, the user sees a Trade-off Curve window, with different coverage/capacity points; clicking any point affords a detailed examination of the proposed design in a graphical display of the market area. It is expected that the original design and optimization will provide a baseline set of data that will allow Ocelot to generate accurate predictions of the parameter changes required when additional equipment is added.

...........................................................................................................................................................................................................................................................

7 - 30

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Amplifier sharing - Asymmetric cell

Amplifier sharing - Asymmetric cell ............................................................................................................................................................................................................................................................

An approach similar to the quasi-omni approach of the last section is the asymmetric cell configuration. Instead of sharing a single amplifier among all three sectors, two total amplifiers are employed. One amplifier is shared among two of the three of sectors, while the remaining amplifier is devoted to the third sector. This approach is appropriate for a cell that has high traffic demand on one sector, and low traffic demand on the other two sectors. For a case of high demand on the alpha sector and low demand on beta and gamma sectors, the scheme would schematically look like the following figure.

Tx

Rx

Tx

Rx

Rx

Tx

Rx

Rx

LA

LA

Combiner

Radio

Splitter

Radio

Rx

Figure 7-14

Asymmetric cell illustration

The splitter splits the power from the linear amplifier two ways, reducing the power per antenna by 1/2 or -3.0 dB in the lightly loaded beta and gamma sectors. The remaining amplifier services the fully loaded alpha sector. The combiner combines the signals from the two lightly loaded faces (beta and gamma) and hence increases the reverse link noise figure by a factor of 2 or 3.0 dB. In this configuration, the coverage footprint of all three sectors is the same. In beta and gamma, the coverage advantage gained by reducing capacity can be used to overcome the combiner and splitter disadvantages instead of extending the cell radius.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 31

Extended carrier

Amplifier sharing - Asymmetric cell

In the beta/gamma reverse link budget, the increased noise figure directly translates to a decrease in maximum allowable path loss. In the forward link, as was shown previously, the decrease in capacity helps offset the loss in power. The standard link budget cannot be used since the standard link budget uses an interference ratio that assumes that all sectors are at equal power. The problem is under study and we’re not currently prepared to deliver a split-sector budget, even though we’re introducing the concept here. To fully overcome the combiner/splitter disadvantage of 3.0 dB would require reducing the capacity to 14 Erlangs for the two lightly loaded sectors. The third sector that is equipped with its own amplifier would support the full capacity of 26.4 Erlangs. Thus, the cell's total capacity is 40.4 Erlangs. Growth strategies

If traffic demand grows in the same pattern, i.e., the busy sector remains significantly higher loading than the other two sectors, then the logical growth path is to add carriers in the same arrangement of amplifier sharing. If traffic grows and the sectors are more uniformly loaded, then a third amplifier should be deployed with each sector being supported by its own amplifier. The penalty for the splitters and combiners is removed from the link budget, so there is no longer any need to reduce capacity. The cell can then run at full capacity of 26.4 Erlangs per sector, or 79.2 Erlangs per cell, in the same footprint. If amplifiers are added, the network may require reoptimization. However, the costs of reoptimization can be minimized through the use of Lucent's Ocelot tool, as described before.

...........................................................................................................................................................................................................................................................

7 - 32

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Extended carrier

Summary

Summary ............................................................................................................................................................................................................................................................

In CDMA, cell design capacity can be lowered in order to expand cell coverage. A carrier in which this design trade-off occurs is termed an “extended carrier”. Although this design trade-off exists in 2G systems, it is of greater interest in 3G systems since the allowed higher loading of 3G yields more dynamic range in which to trade off capacity for coverage. The concept of extended carrier may be used in several ways, lowering deployment costs by better tailoring the design to the specific needs of the network. These include: Single extended carrier. This concept embodies the standard design concept of lowering capacity to extend coverage. Such expanded, lowcapacity cells may reduce deployment costs in lightly loaded (e.g., rural areas) where traffic demand is slight. Additional extended carriers are added when needed to address growth. Concentric extended carrier. This concept uses a base-extended carrier to achieve ubiquitous, low capacity coverage over a large area. Traffic growth is addressed by adding reduced coverage, high capacity (core) carrier at the cell sites, which are centered in the traffic hot spots. Coverage is thus carrier-dependent. Mobiles crossing the boundary between core and extended carriers will hard handoff between the two carriers. This configuration is useful for large low traffic areas punctuated by traffic hot spots. To fully realize the benefits of this configuration, feature development is required to determine mobile location to trigger handoffs between core and extended carriers. Quasi-omni. This configuration services a 3-sector arrangement with a single transmitter/receiver by lowering the design capacity and using the benefit to overcome splitter/combiner losses rather than expand the coverage. The quasi-omni footprint is thus identical to that of a standard 3-sector serviced by 3 transmitters/receivers. Traffic growth can be accommodated within the footprint by adding additional transmitters/receivers as needed, thus “paying as you grow”. Asymmetric cell (split-sector). This configuration services a 3-sector arrangement with two transmitters/receivers. One transmitter/receiver services 2 sectors with low capacity, exploiting the benefit of lower traffic to overcome splitter/combiner losses rather than expanding the footprint. The split-sector footprint is thus identical to that of a standard ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

7 - 33

Extended carrier

Summary

3-sector serviced by 3 transmitters/receivers. This configuration minimizes deployment costs for cells where the traffic tends to be concentrated on a single sector. In each of the above scenarios, case-by-base analysis of the forward link is required in order to ensure link balance. Additionally, some of the growth scenarios may require re-optimization; however, use of the Lucent’s Ocelot tool to specify recommended parameter settings can minimize any associated costs.

...........................................................................................................................................................................................................................................................

7 - 34

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

8

Fixed wireless voice networks

Purpose

Contents

This chapter provides detailed analysis of system performance of 2G and 3G-1X CDMA fixed wireless voice networks. Introduction

8-2

Parameters for fixed wireless analysis

8-3

Reverse link interference ratio (βr) Required reverse link Eb/Nt for 3G Walsh code overhead Recommended loading factor Channel activity factor Reverse link coverage

8-3 8-4 8-6 8-8 8-8 8-9

System capacity calculation

8-10

Capacity calculation methodology Reverse link based capacity calculations Indoor Outdoor Power requirements of forward link

8-10 8-10 8-11 8-14 8-17

3G-1X RC3 3G-1X RC4 3G-1X with SMV Conclusions

8-17 8-21 8-21 8-23

References

8-24

............................................................................................................................................................................................................................................................ 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 1

Fixed wireless voice networks

Introduction

Introduction ............................................................................................................................................................................................................................................................

This chapter provides detailed analysis of system performance of 2G and 3G-1X CDMA fixed wireless voice networks. The method of calculating voice Erlang capacity for 2G and 3G CDMA systems (i.e., IS-95, and CDMA2000 or 3G-1X) is well understood and well documented (see Lucent document 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines, and 401-703-201, PCS CDMA RF Engineering Guidelines). The same methodology can be used to estimate the capacity of a CDMA system when the constraint is added that the subscriber units are fixed. The capacity of a fixed system is expected to be greater than for a full mobility system since the fixed condition of the subscriber unit leads to a relaxed requirement for Eb/ Nt (signal power to impairment power) of both the cell site and subscriber receiver. Fixed wireless networks are categorized into two types of applications based on whether the subscriber unit is located within a building or outside a building. For the indoor application, the subscriber unit with conventional omni-directional antenna is placed in fixed position within a building. Here, the building penetration loss has to be taken into account in network design due to signal attenuation through the wall of the building. For outdoor application, the subscriber unit with narrow beam directional antenna is likely to be on mounted at a elevated location on a building wall or roof-top and connected to telephone terminal within the building through a wired connection. The narrow beam antenna at the subscriber unit reduces the interference from a given subscriber to cells other than the serving cell. The narrow beam antenna also reduces the average number of handoff legs subscribers will use, which will benefit forward link capacity. The outdoor application has coverage and capacity advantages over the indoor application due to both the directional antenna at the subscriber location and the lack of building penetration loss.

...........................................................................................................................................................................................................................................................

8 - 2

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Parameters for fixed wireless analysis

Parameters for fixed wireless analysis ............................................................................................................................................................................................................................................................

Reverse link interference ratio (βr)

For indoor applications, the interference ratio is the same as the mobile case because omnidirectional antenna is used at the subscriber unit. It is well known that the reverse link interference ratio of a mobile system is 0.6, 0.85, and 1.2 for omni cell, 3-sector, and 6-sector, respectively.

+ +

+ + +

+ +

Figure 8-1

Directional antenna points to desired base station

For outdoor application, as discussed earlier, the main effect of the directional antenna is to reduce interference to cells other than the serving cell. In order to determine how βr depends on the antenna beamwidth and cell site sectorization, simulations were done that modeled a network like that pictured in Figure 8-1with the following assumptions: •

A total of 19 cell sites consisting of two tiers

•

Subscriber units are randomly placed over the entire service area

•

The subscriber unit antenna is correctly oriented toward the serving antenna

•

Hata propagation mode is employed and the correlated log-normal shadowing effects is added in calculating path loss

•

Perfect power control is assumed

•

The horizontal and vertical antenna patterns as well as antenna downtilt in base station and uptilt in subscriber unit are included.

Figure 8-2 shows the simulated interference ratio as function of antenna beamwidth from 300 to 600 and cell site sectorization. As we can see, βr increases with increasing antenna beamwidth. At 600 of antenna beamwidth, the interference ratio of omni cell (β=0.2) is lower ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 3

Fixed wireless voice networks

Parameters for fixed wireless analysis

than that of 3-sector (βr=0.3) and 6-sector (βr=0.27), at 450, βr of 6-sector is the same as omni cell (βr=0.15) which is lower than that of 3-sector (βr=0.22), and at 300, βr of 6-sector is slightly less than 0.1, which is interference ratio of omni cell. A explanation why 6-sector configuration has low interference ratio is that for given antenna beamwidth of subscriber unit the 6-sector configuration can suppress the interference from all other sectors due to using narrow beam antenna at base station. Compared to the indoor or mobile scenario, the interference ratio of directional antenna is significant lower in the fixed wireless system. For estimation and comparison purposes, the capacities in this paper for outdoor fixed wireless applications will assume a 500 beamwidth for the subscriber unit which leads to the following values of reverse link interference ratio (omni-directional or indoor applications are included to make the table complete): Table 8-1

Reverse link interference ratios

Subscriber Antenna Cell Sectorization Omni

2-sector (linear highway) 3 -Sector 6-Sector

Omni

0.6

0.3

0.85

1.2

Directional

0.15

0.09

0.25

0.2

Figure 8-2

Required reverse link Eb/Nt for 3G

Interference ratio as function of antenna beamwidth of subscriber unit and cell site sectorization

Reverse link required Eb/Nt is used both in capacity calculations (pole capacity equation) and in coverage calculations (link budgets). Required Eb/Nt is a function of channel condition. One of the main

...........................................................................................................................................................................................................................................................

8 - 4

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Parameters for fixed wireless analysis

Fixed wireless voice networks

characteristics of the channel condition is subscriber speed. Subscribers at zero velocity are typically thought of being in an Additive White Gaussian Noise (AWGN) channel. However, field experience shows that the AWGN value of required Eb/Nt may not properly reflect the subscriber conditions due to the fact that the subscriber receiver sees some apparent motion due to movement of its surrounding environment. Therefore we derive a required Eb/Nt for the fixed case from an interpolation of the link level simulation results for the AWGN and slow speed channel models. Link level simulation results for the 3G ASIC (CSM5000) at 1% FER show that the worst-case total reverse link traffic Eb/Nt for 2 paths at 9.6 kbps is 5.4 dB. This value corresponds to 5.4 –3 =2.4 dB traffic Eb/Nt per diversity branch, and shall be used in further calculations for the full mobility case. Note that the bit energy Eb in this value corresponds only to the traffic energy and does not include the energy embedded in the reverse link pilot signal. The total per-branch Eb/Nt that must be applied in capacity or coverage applications must include the pilot. The pilot is 3.75 dB below the traffic channel, or 42% of the traffic channel. The total per-branch Eb/Nt can be obtained from the traffic per-branch Eb/No by scaling the numerator to contain both traffic and pilot energy:

 Eb  E  TrafficPower + PilotPower  Eb  (1 + 0.42) = 100.393   =  b  =   N N TrafficPow er N  t  total  t  traffic  t  traffic We therefore take the full mobility total per-branch Eb/Nt as 4 dB. Similar calculations establish that the fixed Additive White Gaussian Noise (AWGN) total per-branch Eb/Nt is 2.15, or approximately 2.2 dB. As stated previously, the AWGN value may not properly reflect the subscriber conditions since the receiver sees some apparent motion due to movement of its surrounding environment. For example, Qualcomm 2G ASIC simulations indicated that the per-branch Eb/No for 0 velocity AWGN was 3 dB. Later field measurements indicated that the value for fixed subscribers was higher: 4.6 dB. This difference suggests that the AWGN model underestimates the fixed receiver requirements. This information can be used to estimate a reasonable 3G fixed Eb/Nt from available information. The 2G ASIC values for required Eb/Nt for the cases of AWGN and full mobility are 3.0 dB and 7.0 dB, respectively. The observed fixed receiver Eb/Nt of 4.6 dB can be ........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 5

Fixed wireless voice networks

Parameters for fixed wireless analysis

viewed as the Eb/Nt corresponding to a partial or limited mobility that corresponds to the situation of a fixed receiver within a surrounding, moving environment. Each of these values can be associated with a relative mobility index varying between 0 and 1, where AWGN corresponds to index 0 and full mobility corresponds to index 1. The mobility index for a fixed receiver in a moving environment (i.e., the mobility index corresponding to the observed Eb/Nt of 4.6 dB) can be estimated from a line fit to the AWGN and full mobility values. Table 8-2

Eb/No values and mobility index for 2G/ASIC 1.0

Condition AWGN Full Fixed, moving environment

Per-branch Eb/Nt (dB) 3.0 7.0 4.6

Mobility Index 0 1 x?

100.46 = (100.7 − 100.3 )⋅ x + 100.3 The equation is a linear fit to the values in the table. The equation can be solved for the value of mobility index, x (x=0.3) that corresponds to the 4.6 dB associated with the fixed receiver in a moving environment. A line can also be fit to the data from the CSM5000 in the same manner, since the endpoints for 0 relative mobility (AWGN) and full relative mobility (1) are known. Since x=0.3 or 30% relative mobility appears from the above to be the proper choice for a fixed receiver in a moving environment, the appropriate Eb/Nt requirement for 3G fixed wireless can be estimated by substituting x=0.3 into this equation. Table 8-3

Eb/No values and mobility index for 3G ASIC

Condition AWGN Full Fixed, moving environment

Per-branch Eb/Nt (dB) 2.2 4.0 Y?

Mobility Index 0 1 0.3 (from above)

Y = (100.4 − 100.22 ) ⋅ 0.3 + 100.22 The substitution of x=0.3 into this equation yields Y=1.92, or an estimated per-branch Eb/Nt of 10*log(1.92)=2.8 dB. This value will be used to compute uplink fixed wireless capacity for 3G. Walsh code overhead

Each soft/softer handoff leg requires a Walsh code. Based on IS-95A handoff probabilities in Reference [1] of this chapter, we can calculate the Walsh code overhead factors for 3-sector and omni configurations.

...........................................................................................................................................................................................................................................................

8 - 6

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Parameters for fixed wireless analysis

The maximum number of primary Walsh codes is the maximum number of Walsh codes available divided by this Walsh code overhead and then rounded down to the nearest integer. Note that the soft handoff Walsh code overhead differs from the forward link soft handoff power overhead (note that the latter term, power overhead, is the one that appears in forward link budgets). This difference is due to the fact that the legs consume Walsh codes in the same manner regardless of the soft handoff state, but the amount of power consumed by a leg is a function of its soft handoff state. For 3G-1X, IS-95B handoff algorithm is used. Due to the improvement in the IS-95B handoff algorithm, a 10% handoff reduction is applied to the IS-95A handoff probabilities. The Walsh code limit for traffic channels in 2G and 3G-1X is 60 (four overhead channels: pilot, sync, paging and quick paging (although quick paging is not used in 2G, the Walsh Code is reserved to avoid any possible conflicts with bordering 3G systems)). The addition of dual paging channels (FID2064) will reduce the number by 1 to 59. The calculated Walsh code overhead and the number of primary traffic channels supported with Walsh code limitation (61 for 2G and 60 for 3G) are listed in Table 8-4 and Table 8-5 below. Table 8-4

Walsh code overhead

Subscriber Application

2G (IS-95 A) Omni 3-sector

3G Omni 3-sector

mobility (for comparison)

1.4

1.76

1.36

1.68

indoor fixed

1.29

1.54

1.26

1.49

outdoor fixed

1.00

1.25

1.00

1.25

Table 8-5

Walsh Code limitation to primary traffic channels for max of 60 available

Subscriber Application

2G (IS-95 A)

3G

Omni

3-sector

Omni

3-sector

mobility (for comparison)

42

34

44

35

indoor fixed

46

38

47

40

outdoor fixed

60

48

60

48

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 7

Fixed wireless voice networks

Parameters for fixed wireless analysis

For the 3G-1X system, the Walsh code limit can be greatly increased (more than doubled) by using Radio Configuration 4 (RC4) on the forward link. Note that RC3 is still used for the reverse link, so any reverse link air interface limit still exists. The cost of the extra Walsh codes for the forward link is an approximate 1 dB penalty in required Eb/Nt. The impact of the increase in required Eb/Nt will be examined later in "Power requirements of forward link" section on Page 8-17. A single 3G-1X carrier can support both RC3 and RC4 on the forward link. Lucent has developed proprietary algorithms (FID 3747.2) that maximize forward link capacity optimizes the system capacity based on the instant value of multiple parameters such as RF Power, RC3 Walsh code usage, voice vs. data call, etc. The feature will make the RC3/RC4 assignment decision at call setup time. Recommended loading factor

Channel activity factor

In a fixed wireless system, the recommended loading factors (relative to pole capacity) are: •

72% for 3G-1X systems (the standard 3G value)

•

65% for 2G systems with pole capacities greater than or equal to 69 (the higher values of channels allows for higher loadings without risk of system instability since the larger number of subscribers tends to smooth potential instabilities)

•

55% loading otherwise (the standard 2G value).

The channel activity factor for 2G voice systems is 0.40. The value for 3G-1X must also account for the reverse link pilot and is 0.58. The Selectable mode vocoders (SMV) will result in lower channel activity factors. The following values are used by Lucent for estimating capacities of a SMV system: Table 8-6

Reverse link channel activity factors for different SMV modes

Mode

Reverse Link VAF

Mode 0

0.58 (same as EVRC)

Mode 1

0.51

Mode 2

0.47

...........................................................................................................................................................................................................................................................

8 - 8

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Reverse link coverage

Reverse link coverage ............................................................................................................................................................................................................................................................

Reverse link coverage for fixed wireless networks is estimated in the same manner as for mobile networks (see "Link budget" section on Page 2-14). However, some of the parameters for fixed wireless networks will be different, typically leading to larger predicted coverage areas. The differences that expand coverage include: •

Lower Eb/Nt requirements

•

Higher subscriber unit antenna gains for outdoor fixed wireless networks

•

Lower interference margins for cases when Walsh codes are the limiting resource

•

Lower building penetration margins for outdoor fixed wireless.

The differences that shrink coverage include: •

Higher interference margins for some 2G scenarios since the higher values of channels allows for higher loadings without risk of system instability since the larger number of subscribers tends to smooth potential instabilities (note that this item shrinks coverage as opposed to the other items that expand coverage)

•

Possibly higher building penetration margins or fade margins for indoor wireless case. Some customers may require higher building penetration losses for indoor fixed systems since all subscribers are indoors. Other customers may require that the fade margin term be increased to account for the variability of building penetration values.

For a typical indoor fixed wireless system, the coverage advantage over a mobile network will be just the difference in Eb/Nts, which is 1.2 dB for 3G-1X. For outdoor fixed wireless systems, the coverage advantage can be quite large, since it includes both the gain of the directional subscriber antenna as well as the gain due to not having a building penetration loss.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 9

Fixed wireless voice networks

System capacity calculation

System capacity calculation ............................................................................................................................................................................................................................................................

Capacity calculation methodology

The RF Engineering Guidelines (Lucent document 401-614-012, AUTOPLEX® Cellular CDMA RF Engineering Guidelines, and 401703-201, PCS CDMA RF Engineering Guidelines) explain the methodology for computing system capacity, which is a five step process as follows: 1.

Compute the “pole capacity”:

n pole =

g +1 α ⋅ d ⋅ (1 + βr )

where: g is the processing gain (bandwidth divided by channel rate) α is the channel activity factor βr is the reverse link interference ratio d is the required Eb/Nt expressed as a linear ratio (as opposed to dB) 2.

3.

4.

5.

Choose a loading factor, which is a relative amount of the pole capacity to determine maximum number of simultaneous RF channels. This loading factor is directly related to the predicted coverage through the interference margin term (sometimes called noise rise). The higher the loading, the higher the interference margin and the smaller the coverage area. This maximum number of channels must be checked against the forward link Walsh Code limit. If the Walsh Code limit is less than the computed value, the Walsh Code limit is the maximum number of channels. Choose a grade of service and use that to translate maximum number of channels to voice capacity in terms of Erlangs. Erlang B tables are typically used for this mapping. The forward link air interface capacity is then verified by checking that the forward link has sufficient power to support the number of users.

For a fixed system, two parameters (reverse link interference ratio and required Eb/Nt) of the pole capacity equation are different than the mobile case, as explained below.

...........................................................................................................................................................................................................................................................

8 - 10

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

System capacity calculation

Reverse link based capacity calculations

The capacities considering just reverse link air interface limits and Walsh code limits (to Step 4 in the 5-step process described above) are presented below. Indoor

Table 8-7

2G ASIC 1.0 reverse link capacity of indoor fixed application

ASCI 1.0 Voice

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

4.6

4

4.6

4

2.9

2.5

2.9

2.5

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.4

0.4

0.4

0.4

reverse beta

0.85

0.85

0.6

0.6

Nmax = g/(alpha*d*(1+beta))+1

61.0

46.9

70.3

54.1

% of loading

55%

55%

65%

55%

N = Nmax*% of loading

33

25

45

29

Reverse Link Channel Capacity with Walsh code limitation

33

25

45

29

Reverse Link Erlang Capacity @ 1% blocking

22.9

16.1

33.4

19.5

Reverse Link Erlang Capacity @ 2% blocking

24.6

17.5

35.6

21.0

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-8

2G ASIC 1.1 reverse link capacity of indoor fixed application

ASCI 1.1 Voice

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

4

3.4

4

3.4

2.5

2.2

2.5

2.2

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.4

0.4

0.4

0.4

reverse beta

0.85

0.85

0.6

0.6

Nmax = g/(alpha*d*(1+beta))+1

69.9

53.7

80.6

61.9

% of loading

65%

55%

65%

55%

N = Nmax*% of loading

45

29

52

34

Reverse Link Channel Capacity with Walsh code limitation

38

29

46

34

19.5

34.3

23.8

21.0

36.5

25.5

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Reverse Link Erlang Capacity @ 1% blocking Reverse Link Erlang Capacity @ 2% blocking

27.3 29.2

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 11

Fixed wireless voice networks

System capacity calculation Table 8-9

RC3 reverse link capacity of indoor fixed application

3G-1X Voice RC3

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

2.8

2.6

2.8

2.6

1.9

1.8

1.9

1.8

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.58

0.52

0.58

0.52

reverse beta

0.85

0.85

0.6

0.6

Nmax = g/(alpha*d*(1+beta))+1

63.6

49.7

73.4

57.4

% of loading

72%

72%

72%

72%

N = Nmax*% of loading

45

35

52

41

Reverse Link Channel Capacity with Walsh code limitation

40

35

47

41

Reverse Link Erlang Capacity @ 1% blocking

29.0

24.6

35.2

29.9

Reverse Link Erlang Capacity @ 2% blocking

31.0

26.4

37.5

31.9

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-10

3G-1X RC4 Reverse Link capacity of indoor fixed application

3G-1X Voice RC4

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.58

0.58

reverse beta

0.85

0.6

Nmax = g/(alpha*d*(1+beta))+1

63.6

73.4

% of loading

72%

72%

N = Nmax*% of loading

45

52

Reverse Link Channel Capacity with Walsh code limitation

45

52

Reverse Link Erlang Capacity @ 1% blocking

33.4

39.7

Reverse Link Erlang Capacity @ 2% blocking

35.6

42.1

Eb/Nt in dB d = Eb/Nt|rqd in ratio

...........................................................................................................................................................................................................................................................

8 - 12

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

System capacity calculation Table 8-11

3G-1X RC4 with SMV Mode 1 reverse link capacity of indoor fixed application

3G-1X Voice RC4 & SMV Mode 1

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.51

0.51

reverse beta

0.85

0.6

Nmax = g/(alpha*d*(1+beta))+1

72.4

83.6

% of loading

72%

72%

N = Nmax*% of loading

52

60

Reverse Link Channel Capacity with Walsh code limitation

52

60

Reverse Link Erlang Capacity @ 1% blocking

39.7

46.9

Reverse Link Erlang Capacity @ 2% blocking

42.1

49.6

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-12

3G-1X RC4 with SMV Mode 2 reverse link capacity of indoor fixed application

3G-1X Voice RC4 & SMV Mode 2

Indoor Fixed

Configuration

3-sector BS, omni terminal omni BS, omni terminal

Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.47

0.47

reverse beta

0.85

0.6

Nmax = g/(alpha*d*(1+beta))+1

78.5

90.6

% of loading

72%

72%

N = Nmax*% of loading

56

65

Reverse Link Channel Capacity with Walsh code limitation

56

65

Reverse Link Erlang Capacity @ 1% blocking

43.3

51.5

Reverse Link Erlang Capacity @ 2% blocking

45.9

54.4

Eb/Nt in dB d = Eb/Nt|rqd in ratio

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 13

Fixed wireless voice networks

System capacity calculation

Outdoor

The capacity improvement of a sector is an important benefit obtained from the reduced in the outer sector interference when the directional antenna is used. Table 8-13 2G ASIC 1.0 reverse link capacity of outdoor fixed application ASCI 1.0 Voice

Outdoor Fixed

Configuration

3-sector BS, directional terminal

omni BS, directional terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

4.6

4

4.6

4

2.9

2.5

2.9

2.5

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.4

0.4

0.4

0.4

reverse beta

0.25

0.25

0.15

0.15

Nmax = g/(alpha*d*(1+beta))+1

89.8

68.9

97.5

74.9

% of loading

65%

65%

65%

65%

N = Nmax*% of loading

58

44

63

48

Reverse Link Channel Capacity with Walsh code limitation

48

44

60

48

Reverse Link Erlang Capacity @ 1% blocking

36.1

32.5

46.9

36.1

Reverse Link Erlang Capacity @ 2% blocking

38.4

34.7

49.6

38.4

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-14 2G ASIC 1.1 reverse link capacity of outdoor fixed application ASCI 1.1 Voice

Outdoor Fixed

Configuration

3-sector BS, directional terminal

omni BS, directional terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

4

3.4

4

3.4

2.5

2.2

2.5

2.2

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.4

0.4

0.4

0.4

reverse beta

0.25

0.25

0.15

0.15

Nmax = g/(alpha*d*(1+beta))+1

102.9

79.0

111.8

85.8

% of loading

65%

65%

65%

65%

N = Nmax*% of loading

66

51

72

55

Reverse Link Channel Capacity with Walsh code limitation

48

48

60

55

Reverse Link Erlang Capacity @ 1% blocking

36.1

36.1

46.9

42.4

Reverse Link Erlang Capacity @ 2% blocking

38.4

38.4

49.6

44.9

Eb/Nt in dB d = Eb/Nt|rqd in ratio

...........................................................................................................................................................................................................................................................

8 - 14

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

System capacity calculation Table 8-15

3G-1X reverse link capacity of outdoor fixed application

3G-1X Voice RC3/RC2

Outdoor Fixed

Configuration

3-sector BS, directional terminal

omni BS, directional terminal

Vocoder

8K

13K

8K

13K

Data rate

9600

14400

9600

14400

2.8

2.6

2.8

2.6

1.9

1.8

1.9

1.8

g, processing gain

128

85.3

128

85.3

α, voice activity factor

0.58

0.52

0.58

0.52

reverse beta

0.25

0.25

0.15

0.15

Nmax = g/(alpha*d*(1+beta))+1

93.7

73.1

101.7

79.4

% of loading

72%

72%

72%

72%

N = Nmax*% of loading

67

52

73

57

Reverse Link Channel Capacity with Walsh code limitation

48

48

60

57

Reverse Link Erlang Capacity @ 1% blocking

36.1

36.1

46.9

44.2

Reverse Link Erlang Capacity @ 2% blocking

38.4

38.4

49.6

46.8

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-16

3G-1X reverse link capacity of outdoor fixed application

3G-1X Voice RC4 (FL)

Outdoor Fixed 3-sector BS, directional omni BS, directional terminal terminal

Configuration Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.58

0.58

reverse beta

0.25

0.15

Nmax = g/(alpha*d*(1+beta))+1

93.7

101.7

% of loading

72%

72%

N = Nmax*% of loading

67

73

Reverse Link Channel Capacity with Walsh code limitation

67

73

Reverse Link Erlang Capacity @ 1% blocking

53.4

58.9

Reverse Link Erlang Capacity @ 2% blocking

56.3

62.0

Eb/Nt in dB d = Eb/Nt|rqd in ratio

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 15

Fixed wireless voice networks

System capacity calculation Table 8-17

3G-1X RC4 with SMV Mode 1 reverse link capacity of indoor fixed application 3G-1X Voice RC4 & SMV Mode 1

Indoor Fixed

Configuration

3-sector BS, omni terminal

omni BS, omni terminal

Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.51

0.51

reverse beta

0.25

0.15

Nmax = g/(alpha*d*(1+beta))+1

106.7

115.9

% of loading

72%

72%

N = Nmax*% of loading

76

83

Reverse Link Channel Capacity with Walsh code limitation

76

83

Reverse Link Erlang Capacity @ 1% blocking

61.7

68.2

Reverse Link Erlang Capacity @ 2% blocking

64.9

71.6

Eb/Nt in dB d = Eb/Nt|rqd in ratio

Table 8-18

3G-1X RC4 with SMV Mode 2 reverse link capacity of indoor fixed application 3G-1X Voice RC4 & SMV Mode 2

Indoor Fixed

Configuration

3-sector BS, omni terminal

omni BS, omni terminal

Vocoder

8K

8K

Data rate

9600

9600

2.8

2.8

1.9

1.9

g, processing gain

128

128

α, voice activity factor

0.47

0.47

reverse beta

0.25

0.15

Nmax = g/(alpha*d*(1+beta))+1

115.7

125.6

% of loading

72%

72%

N = Nmax*% of loading

83

90

Reverse Link Channel Capacity with Walsh code limitation

83

90

Reverse Link Erlang Capacity @ 1% blocking

68.2

74.7

Reverse Link Erlang Capacity @ 2% blocking

71.6

78.3

Eb/Nt in dB d = Eb/Nt|rqd in ratio

As we see, the capacity of outdoor is significant higher than that of indoor with or without Walsh code limitation.

...........................................................................................................................................................................................................................................................

8 - 16

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Power requirements of forward link

Power requirements of forward link ............................................................................................................................................................................................................................................................

3G-1X RC3

To assess the forward link, we begin with the fundamental forward link equation that conserves power at the J4 (antenna connector) port at the base station (see "Forward link" section on Page 2-20):

∑α

j

x j Qmax + µQmax ≤ Qmax

all links

where α and x are the forward link voice activity and forward link power allocation, respectively, for the jth link (user). The allocation is the fraction of total transmit power allocated to the link, and is frequently referenced as Ec/Ior. Qmax is the maximum power (e.g., 16 watts for PCS Modcell) broadcast at the J4 port. The fraction µ is the (fixed) percentage of maximum power provided for overhead functions (e.g., pilot, page). At full power, the expression above reduces to:

all

∑α

j

x j ≤ 1− µ

links

The capacity is determined by the allowed number of links in the sum; i.e., for a given distribution of the random variables α and x, there is a maximum number N of links that can be supported in order to satisfy the equation above with a high degree of probability. We will use the equation above to estimate the difference between fixed and fully mobile capacity by projecting the allowed change in N when the distribution of x’s is shifted from fully mobile to fixed only. This process requires estimation of the x values for both conditions. To proceed further, we conservatively assume that all subscriber units are located at the edge of cell coverage, where “edge” in this context denotes a cell exit or entry point. This assumption can be exploited in two ways: •

At the design edge, the ratio of received pilot strength to total background interference (Ec/Io) must be optimized to be greater than the handoff add threshold T_ADD.(e.g., -12 dB). The physical boundary of the cell must correspond to this value (as opposed to T_DROP) in order to ensure that a subscriber entering the cell adds a new pilot before dropping the old one (the “make before break” rule of soft handoff).

•

Given the subscriber placement, all subscribers are in a handoff state, which (conservatively) establishes a minimum of two paths.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 17

Fixed wireless voice networks

Power requirements of forward link

The value of x (Ec/Ior) for each subscriber can therefore be obtained from curves of Ec/Ior vs. geometry for 2-path cases. These curves are available as a function of subscriber speed and for the AWGN cases. The curves are generated from link level simulations. In the reverse link analysis (see above), the AWGN values were scaled to obtain Eb/Nt requirements for a fixed receiver in a moving environment. This scaling was done by comparing measurements of 2G fixed wireless requirements to AWGN values. Since there are less empirical results on 2G forward link fixed wireless Eb/Nt requirements, the forward link 3G AWGN values shall be used without adjustment as fixed wireless requirements. The link level simulation curves12 plot Ec/Ior as a function of the geometry. Geometry is defined as Ior/(FNoW+Ioc), where Ior = the sum of received power density from the cell(s) in the active set, FNoW is the mobile noise floor, and Ioc = the received power density from all surrounding cells not in the active set. The Ec/Ior value from the curves applies to the link from the host cell only. The Ioc term does not contain the receiver noise, as the underlying simulations were interferencelimited. For the handoff case, this value becomes the ratio of the total received power density from the two handoff cells (i.e., Ior = Ior (1) + Ior (2)) to the total impairment density Ioc. The curves were produced with 20% of the host power allocated to pilot. To compute the appropriate value of Ec/Ior for the subscriber placement presumed, the value of geometry for each subscriber must be computed. The value of geometry can be computed by noting the following: •

The pilot power is a constant fraction η of the total maximum cell power (Ior1W)

•

At the cell edge, each subscriber’s ratio of pilot power to total background power is THRES

•

At the cell edge, Ior1=Ior2 due to the placement of subscribers and the assumption that all cells broadcast at full, equal power.

Accordingly:

...........................................................................................................................

12 These curves are found in the document “Simulation Study of the OTD Mode for the Voice Service Case in IS-2000”, by Qi Bi, Yung-Fang Chen, and Raafat Kamel; March 2000.

...........................................................................................................................................................................................................................................................

8 - 18

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Power requirements of forward link

pilot = ηI or1W pilot = 10THRES / 10 FNtW + I or1W + I or 2W + I ocW geometry G =

I or1W + I or 2W FNtW + I ocW

These three equations can be combined to yield:

G=

1 η  10−THRES/ 10   − 1  2

Evaluation of this expression for THRES=-12 dB and a 20% ratio of pilot to host power yields a geometry of 1.7, or 2.3 dB. Note that this analysis implies that the geometry is constant for all subscribers on the edge, regardless of effects such as lognormal fading. The constant value follows from the assumption that optimization for all possible boundary (i.e., exit/entry) positions has established a value equal to the THRES value in order to ensure handoff performance at the cell edge. The value of Ec/Ior for AWGN and 3 velocities for a geometry of 2.3 dB are tabulated below for 2 GHz. These values can therefore be used to analyze the relative increase between the 3G-1X full mobility and 3G-1X fixed wireless capacities for the PCS case. These results will serve as a conservative estimate of the relative increase at lower frequencies (e.g., 450 MHz, 850 MHz) since at longer wavelengths typical subscriber velocities will yield lower fast fading rates. These rates will bias the spread of Ec/Ior for nonzero velocities towards the upper rows of the table,13 resulting in a slightly higher relative increase when the receiver is fixed.

...........................................................................................................................

13 For example, a speed of 100 km/hr. at 2 GHz is roughly equivalent to a speed of 400 km/hr. at 450 MHz: each achieves the same fast fading rate, since the wavelength at 450 MHz is approximately 4 times greater than that at 2 GHz. The value of -15.9 dB would therefore be excluded from the 450 MHz case.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 19

Fixed wireless voice networks

Power requirements of forward link Table 8-19

Ec/Ior for 2-path, d=0 (equal strength handoff legs), geometry is 2.3 dB, frequency=2GHz Velocity

x = Ec/Ior

“0” km/hr. (AWGN)

-16.7 dB (2.1%)

3 km/hr.

-14.3 dB (3.7%)

30 km/hr.

-15.3 dB (3.0%)

100 km/hr.

-15.9 dB (2.6%)

For the full mobility case, presuming that each velocity is equally probable, the mean and standard deviation of the random variable x is 0.028 and 0.0058, respectively. In contrast, for the fixed wireless case the single constant value of x is 0.021 (equivalently, a mean of 0.021 and a standard deviation of 0). This information can now be used to evaluate the fundamental forward link equation for the fixed and fully mobile cases. The left hand side is a random variable that can be approximated as Gaussian since the sum is over a large number of independent variables (note that the voice activity and allocation x are independent). To satisfy the equation with high (e.g., 98%) probability, we require that the 98th percentile (the value corresponding to the mean plus two standard deviations) of the Gaussian distribution be less than or equal to the right hand side. In summary:

y = ∑α j x j ≤ 1 − µ N

j =1

η + 2σ y ≤ 1 − µ y

η y = E[y ]= Nηαηx σ y = N σ x2σ α2 + ηα2σ x2 + η x2σ α2 Accordingly, Nηαη x + 2 N σ x2σ α2 + ηα2σ x2 + η x2σ α2 − (1 − µ ) ≤ 0 The last equation is quadratic in √N, and can be solved for N. In solution, a value of µ=0.29 is employed since this is the fraction of total power consumed by a pilot at 20% total (the value employed in the Ec/Ior simulations) and the additional overhead channels of page and

...........................................................................................................................................................................................................................................................

8 - 20

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Power requirements of forward link

sync. Additionally, the mean and standard deviation of voice activity are estimated simply by presuming full rate (1) with a probability of 0.4 and 1/8 rate with a probability of 0.6. With these assignments, the quadratic equation in √N is evaluated for 1) the N corresponding to full mobility; and 2) the N corresponding to AWGN. The ratio of the latter to the former is 1.4, or a 40% increase in channels. Since the full-power (blocking) condition for full mobility has been estimated in simulations to be 35 channels (26.4 Erlangs), the corresponding number of primary channels at the AWGN full-power (blocking) condition should be 35(1.4)=49. This corresponds to 37 Erlangs at the 1% blocking condition and 39.3 Erlangs for the 2% blocking condition. This estimate must be treated as an upper bound on the capacity for a fixed receiver in a moving environment, since in arriving at this value no adjustment was made to the Ec/Ior value obtained from the AWGN condition. In contrast, the AWGN value in uplink analysis was scaled in order to account for motion in the surrounding environment. Nevertheless, the forward link result of 37 Erlangs is significant in that it suggests that the forward link should be able to support the 34 Erlangs estimated for the uplink. This conclusion is applicable to lower frequencies as well, since the relative increase between fixed and mobile capacity at PCS frequencies should be less than or equal to the relative increase at longer wavelengths (see above). 3G-1X RC4

In several cases, Walsh codes are the capacity limiting resource. Forward link RC4 allows for twice the number of Walsh codes relative to RC3 (128 vs. 64). However, the cost of the extra Walsh codes is higher, required power requirements to support RC4 subscriber units. The power fraction (Ec/Ior) versus Geometry curves show about 1 dB power penalty for RC4. This 1 dB can be directly applied as an Erlang capacity reduction. Thus, the 37 Erlangs for RC3 is reduced by 1 dB to 29.4 Erlangs for 1% blocking and from 39.3 to 31.2 Erlangs for 2% blocking. In this case the limiting resource is forward link air interface capacity.

3G-1X with SMV

The 1 dB Eb/Nt penalty from utilizing RC4 typically leads to the forward link air interface being the limiting resource. The SMV (Selectable Mode Vocoder) provides capacity gains for the forward link. The capacity gains for SMV in the forward link are expected to be as follows.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 21

Fixed wireless voice networks

Power requirements of forward link

Mode

Forward link gain

Mode 0

0%

Mode 1

34%

Mode 2

64%

Mode 3

80%

½ Max mode 1

70%

½ Mac mode 2

93%

Thus the increase in forward link capacity relative to the RC4 case for Mode 1 is to 39.3 Erlangs for 1% blocking, or 41.8 Erlangs for 2% blocking. For Mode 2, the capacity increases to 48.2 Erlangs for 1% blocking, and 51.2 Erlangs for 2% blocking.

...........................................................................................................................................................................................................................................................

8 - 22

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003

Fixed wireless voice networks

Conclusions

Conclusions ............................................................................................................................................................................................................................................................

The following table summarizes the capacities for various configurations. Fixed Wireless Application

Indoor (Omni Subscriber Antenna)

Base Station Antenna

3-sector

Vocoder

8K

13K

omni 8K

Outdoor (Directional Subscriber Antenna) 3-sector

omni

13K

8K

13K

8K

13K

1% blocking ASCI 1.0 Voice

22.9

16.1

33.4

19.5

36.1

32.5

46.9

36.1

ASCI 1.1 Voice

27.3

19.5

34.3

23.8

36.1

36.1

46.9

42.4

3G-1X Voice RC3/RC2 (RL & FL)

29.0

24.6

35.2

29.9

36.1

36.1

46.9

44.2

3G-1X Voice RC4 (FL)

29.4

N/A

TBD

N/A

TBD

N/A

TBD

N/A

3G-1X Voice RC4 (FL) SMV Mode 1 39.3

TBD

TBD

TBD

3G-1X Voice RC4 (FL) SMV Mode 2 43.3

TBD

TBD

TBD

2% blocking ASCI 1.0 Voice

24.6

17.5

35.6

21.0

38.4

34.7

49.6

38.4

ASCI 1.1 Voice

29.2

21.0

36.5

25.5

38.4

38.4

49.6

44.9

3G-1X Voice RC3/RC2 (RL & FL)

31.0

26.4

37.5

31.9

38.4

38.4

49.6

46.8

3G-1X Voice RC4 (FL)

31.2

N/A

TBD

N/A

TBD

N/A

TBD

N/A

3G-1X Voice RC4 (FL) SMV Mode 1 41.8

TBD

TBD

TBD

3G-1X Voice RC4 (FL) SMV Mode 2 45.9

TBD

TBD

TBD

Where the highlighting indicates the limiting resource as follows: •

No highlighting indicates that the reverse link air interface is the limiting resource

•

Pink highlighting indicates that Walsh codes are the limiting resource

•

Yellow highlighting indicates that the forward link air interface is the limiting resource.

•

Green highlighting indicates that further analysis is required to verify the forward link power requirements are satisfied.

Capacities for fixed wireless 3G-1X data networks are currently being studied.

........................................................................................................................................................................................................................................................... 401-614-040 Issue 2, February 2003

Lucent Technologies - Proprietary See Notice on first page

8 - 23

Fixed wireless voice networks

References

References ............................................................................................................................................................................................................................................................

[1]. “Range vs. number of subscribers for the forward and reverse links,” Qualcomm, July 18, 1995.

...........................................................................................................................................................................................................................................................

8 - 24

Lucent Technologies - Proprietary See Notice on first page

401-614-040 Issue 2, February 2003