Hsdpa
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View Hsdpa as PDF for free.
More details
- Words: 3,185
- Pages: 46
National Taiwan University Department of Computer Science and Information Engineering
Buffer Overflow Control for UMTS High-Speed Downlink Packet Access
Phone Lin (Ph.D.) Dept. of Comp. Sci. & Info. Engr., National Taiwan University, Taipei, R.O.C. FAX: +886-2-2362-8167 Email: [email protected]
1
National Taiwan University Department of Computer Science and Information Engineering
Outlines a Introduction to UMTS a Packet Access through WCDMA a High Speed Downlink Packet Access in UMTS a The Network Architecture of HSDPA a Overflow Control Issues in UMTS HSDPA a Four Schemes for Buffer Overflow Control a Performance Evaluation for Four Schemes a Summery 2
National Taiwan University Department of Computer Science and Information Engineering
Universal Mobile Telecommunication System (UMTS) (2/2)
3
National Taiwan University Department of Computer Science and Information Engineering
Universal Mobile Telecommunication System (UMTS) (1/2) aUMTS supports high data transmission rate (up to 2 mbps) for mobile users, which consists of Terrestrail Radio Access Network (UTRAN) Core network
aThe UTRAN consists of radio base stations (Node Bs) and Radio Network Controllers (RNCs), which Connected through an ATM network
4
National Taiwan University Department of Computer Science and Information Engineering
Radio Access in UMTS a A User Equipment (UE) communicates with UTRAN through the air interface Uu. WCDMA Radio Access Technology
a In a UE communication session, several cells (Node Bs) are defined as an active set. a If the quality of the wireless link between the UE and a cell is above some threshold, then this cell is included in the active set. a When the quality of the wireless link of a cell in the active set is below some threshold, then the cell is removed from the active set. 5
National Taiwan University Department of Computer Science and Information Engineering
Packet Data Traffic a A packet service session contains one or several packet calls depending on the application (e.g., Interactive & Background Services). During a packet call, the packet call constitutes a bursty sequence of packets.
a The parameters that define the characteristics of the packet data traffic in web-browsing Session arrival process (usually modeled as Poisson process) Number of packet calls per session Reading time between packet calls Time interval between two packets inside a packet call Packet size 6
National Taiwan University Department of Computer Science and Information Engineering
Characteristic of a Packet Service Session Packet Service Session Packet Call
Time
Reading Time
Packet Size
Packet Arrival Interval
7
National Taiwan University Department of Computer Science and Information Engineering
The Properties that are for Non-RealTime Packet Services a Packet data is bursty. The required bit rate can change rapidly from zero to hundreds of kilobits/second.
a Packet data tolerates longer delay than real-time services. Packet data is controllable traffic from the radio access network point of view.
a Packets can be retransmitted by the radio link control (RLC) layer. This allows the use of worse radio link quality and much higher frame error ratio than in case of real-time services.
8
National Taiwan University Department of Computer Science and Information Engineering
Overview of WCDMA Packet Access a Packet allocations in WCDMA are controlled by the packet scheduler (PS). (PS) Its functions are to Divide the available air interface capacity between the packet data users. Decide the transport channel to be used for each user’s packet data transmission. Monitor the packet allocations and the system load. The PS allocates a bit rate for a bearer and possibly changes this bit rate during an active connection. connection
a WCDMA packet access allows non-real-time bearers to use Common Channels Dedicated Channels Shared Channels 9
National Taiwan University Department of Computer Science and Information Engineering
The Location of Packet Scheduler a The PS is typically located in RNC where the scheduling can be done efficiently for multiple cells also considering the soft handover connections. connections
a The Node B provides the measurements of the air interface load for the PS. If the load > the target, target the PS can decrease the load by decreasing the bit rates of packet. If the load < target, target it can increase the load by allocating more data. The PS is also a part of the network load control because it can or reduce the network load.
10
National Taiwan University Department of Computer Science and Information Engineering
Packet Access in WCDMA Network
P acket allocation s B ase Station L oad m easurem en ts
RNC (P S )
P ack et allocations
B ase S tation 11
National Taiwan University Department of Computer Science and Information Engineering
Why Standard UTRAN Cannot work for High Speed Packet Access
aIn standard UTRAN, multiple paths exist between the UE and all Node Bs in the active set. aThis multi-path approach does not support high speed data transmission because multiple links for a UE may increase the overall interference within an UTRAN, and thus the data transmission rate decreases.
12
National Taiwan University Department of Computer Science and Information Engineering
The network architecture of HSDPA Frame n Frame 2 Frame 1
Frame n
Cell1
Frame n
UE
Frame 2 Frame 1
RNC Frame 2 Frame 1
Cell2 Frame n Frame 2 Frame 1
Cell3
UTRAN
RNC: Radio Network Controller UE: User Equipment UTRAN: UMTS Terrestrial RAdio Access Network
13
National Taiwan University Department of Computer Science and Information Engineering
HSDPA in 3GPP TR 25.950 (1/2) a The UE only selects one cell (referred to as the serving cell) in the active set for high-speed downlink transmission. a This “serving cell” is selected by the Fast Cell Selection mechanism based on the Common Pilot Channel Received Signal Code Power measurements of the cells in the active set.
14
National Taiwan University Department of Computer Science and Information Engineering
HSDPA in 3GPP TR 25.950 (2/2) a Two physical channels: High Speed Physical Downlink Shared Channel (HSPDSCH) for downlink packet frame transmission Dedicated Physical Control Channel (DPCCH) are used for uplink/downlink signaling, respectively.
a The RNC sends the packet frames to the cells in the active set. For the serving cell, the packet frames are forwarded to the UE. On the other hand, every non-serving cell in the active set queues the packet frames in a buffer. If the link quality between the serving cell and the UE degrades below some threshold, the UE selects the best cell in the active set as the new serving cell.
15
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Issues a Since the non-serving cells do not send packet frames to the UE, their buffers may overflow. a The buffer overflow issue is not addressed in 3GPP TR 25.950. a We propose five overflow control schemes for this issue. a Analytic and simulation models are conducted to evaluate the performance of the proposed schemes. a Our schemes guarantees that (i) when the buffer of a non-serving cell is full, the previously received packet frames in the buffer can be safely dropped, and (ii) after the UE has switched wireless link to the new serving cell, no packet frames are lost. 16
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Schemes (1/2) aBasic Over-flow Control (BOFC): The information needed for frame synchronization is carried by the uplink DPCCH. When the size of frame synchronization information exceeds the capacity of an uplink DPCCH, this information must be carried through multiple uplink DPCCH transmissions.
aNetwork-based Overflow Control (NOFC): NOFC to guarantee one uplink DPCCH transmission for frame synchronization through message exchange between the old and new serving cells. 17
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Schemes (2/2) aCombined BOFC and NOFC (COFC) COFC to take advantage of both BOFC and NOFC
aThe COFC with Counter Reset (COFCR) To reduce the possibility of these extra message exchanges, we can substract the counters CRUE, CSi, CRi and CSRNC by a number θ if CRUE >= N^{\star} when frame synchronization occurs.
18
National Taiwan University Department of Computer Science and Information Engineering
The BOFC scheme a Algorithm OFC1 (exercised by RNC) The RNC sends a packet frame to every cell , and increments CSrnc by one.
a Algorithm OFC2 (exercised by the serving cell) This algorithm performs flow control between the serving cell and the RNC. A window based flow control algorithm.
a Algorithm OFC3 (exercised between the UE and the serving cell). Algorithm BOFC3 is exercised between the UE and the serving cell. Since the wireless link is not reliable, “Stop-And-Wait Hybrid ARQ (SAW-Hybrid ARQ)” [5] is used in BOFC3 for flow control.
a Algorithm OFC4 (exercised by a non-serving cell) When a non-serving cell receives a packet frame from the RNC, it checks if the buffer overflows. If so, it drops a packet in front of the buffer.
a Algorithm BFS (UE switches the wireless link from the old serving cell to the new serving cell) 19
National Taiwan University Department of Computer Science and Information Engineering
The Relationship among OCF1-OFC4 algorithms
20
National Taiwan University Department of Computer Science and Information Engineering
The Timing Diagram Cells
w
acks
ackUE nackUE
w
ackUE
ackUE acks
w 21
N imax , max
National Taiwan University Department of Computer Science and Information Engineering
The Counters used in BOFC a Ni,max: the size of the buffer maintained in Celli a Ki: the number of packet frames currently stored in the buffer of Celli in the active set a CRUE: a counter maintained by UE to indicate the number of packet frames received by UE a CRi (CSi): a counter counts the number of packet frames received from the RNC (processed by Celli) a CSRNC: a counter maintained by RNC to record the number of packet frames that have been received by the serving cell a w: the window size used for downlink frame transmission from the RNC to the serving cell a Fs: an overflow flag maintained by the serving cell a WIs: indicates the number of packet frames received by Cells 22
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC1 (exercised by the RNC) aThe RNC sends a packet frame to Celli, and increments CSRNC by one. Case OFC1.1. If CSRNC mod w <> 0, the RNC continues to send the next packet frame to Celli. Case OFC1.2. If CSRNC mod w = 0, the RNC suspends the packet frame transmission until an ACK message is received from the serving cell.
23
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC2 (exercised by the serving cell for the flow control between RNC and serving cell)
a Step OFC2.1. When a packet frame arrives, Cells updates counters: CRs <- CRs + 1; Ks <- CRs – CSs; WIs <- CRs mod w
a Step OFC2.2. Cells checks WIs and Ks values to determine if it will receive the next frames from RNC. Case OFC2.2.1. If WIs <> 0, it continues to receive the next packet frame. Case OFC2.2.2. If WIs == 0 and Ks <= Ns,max – w, Cells replies the RNC an ACK, and the RNC will be triggered to send the packet frames of the next window. Case OFC2.2.3. If WIs == 0 and Ks > Ns,max – w, Cells sets Fs to 1, and no ACK message is sent to the RNC. 24
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC3 (between Cells and UE; SAW Hybrid-ARQ)
a When the UE receives a packet frame, it replies the status of the transmission to serving cell. a Case OFC3.1. If an incorrect packet frame is received, it replies a NACK message to Cells. a Case OFC3.2. If a correct frame is received, UE updates CRUE Ns,max – w, Cells takes no action. Case OFC3.2.3. If Fs = 0, No action is taken by Cells. 25
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC4 (exercised by a nonserving cell)
aStep OFC4.1. When Celli receives a packet frame from the RNC, it sets Ki <- CRi – CSi. aStep OFC4.2. If Ki = Ni,max (i.e., the buffer is full), Celli deletes a packet frame at the head of the buffer, and sets CSi <- CSi + 1. aStep OFC4.3. Celli adds the received packet frame at the tail of the buffer, and sets CRi
National Taiwan University Department of Computer Science and Information Engineering
The Basic Frame Synchronization (BFS) Algorithm aThe frame synchronization information (i.e., CSUE) is carried through the uplink DPCCHs in the air interface.
27
National Taiwan University Department of Computer Science and Information Engineering
Message flow for Algorithm BFS
28
National Taiwan University Department of Computer Science and Information Engineering
The NOFC Scheme aThe OCF1, OFC2, OFC3, and OFC4 algorithms are adopted in this scheme. aThe Network Frame Synchronization (NFS) algorithm is performed for frame synchronization information delivery. The frame synchronization information (i.e., CSo ) is delivered through the ATM network.
29
National Taiwan University Department of Computer Science and Information Engineering
Message flow for Algorithm NFS
30
National Taiwan University Department of Computer Science and Information Engineering
The COFC Scheme a OFC1-OFC4 algorithms are adopted in this scheme. a If the CRUE value can be fitted in one DPCCH, then the number of messages exchanged in BFS is less than that in NFS. a On the other hand, if multiple DPCCHs are required to deliver CRUE in BFS, then the cost of BFS is higher than that of NFS (because the DPCCH delay is anticipated longer and less reliable than that of message delivery in Steps NFS2-5 of NFS). a In the frame synchronization algorithm CFS of the COFC scheme, Cello determines whether BFS or NFS should be exercised when it receives the Change Serving Cell Request message. 31
National Taiwan University Department of Computer Science and Information Engineering
The Message Flow for CFS
32
National Taiwan University Department of Computer Science and Information Engineering
The COFCR Scheme aOFC1-OFC4 are adopted in this scheme. aThe CFSCR algorithm is performed for frame synchronization information delivery. aTo reduce the possibility of these extra message exchanges, we can substract the counters CRUE, CSi, CRi and CSRNC by a number θ if CRUE >= N^{\star} when frame synchronization occurs. 33
National Taiwan University Department of Computer Science and Information Engineering
The message flow for Algorithm CFS with Counter Reset
a The counters are reset at the “.” points in the time lines.
34
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (1/5)
a We investigate how the following two factors affects the affect the performance of BFS, NFS, CFS, and CFSCR. (i) traffic patterns (the session time and the packet frame arrival rate) and (ii) the serving cell residence time (the period in which the UE resides in a serving cell and receives downlink packet frames from that cell) a®ect the performance of BFS, NFS,
35
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (2/5) aWhen the UE switches from the jth cell to the j + 1st cell during a communication session, we compute (i) the number nr;j of control messages delivered in the air interface (between the UE and the cells) and (ii) the number nn,j of control messages delivered in the ATM network (among the new serving cell, the old serving cell, and the RNC).
36
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (3/5)
aLet dr and dn be the expected transmission delays in the air interface and the ATM network, respectively. athe net cost ds,j for frame synchronization and cell switching to the j + 1st cell can be expressed as ds,j = nr,jdr + nn,jdn
37
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (4/5)
aLet packet frame arrivals to the UE in a downlink transmission session be a Poisson stream with rate λ. aLet the serving cell residence time t of a UE be a random variable with the density function f(t), Laplace transform f¤(s) and the expected value 1/μ.
38
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (5/5) a Let Pr[K = k; J = j] be the probability that the transmission session is not complete at the jth serving cell, and during tj , there are k packet frames received by the UE. a We have
a Where
39
National Taiwan University Department of Computer Science and Information Engineering
Analysis for BOFC a Let Pr[M = m; J = j] be the probability that there are m uplink DPCCH transmissions when the UE switches to the j + 1st cell. a We have
40
National Taiwan University Department of Computer Science and Information Engineering
Analysis for NOFC a When the UE switches from the jth to the j + 1st cell, we have nr,j = 4 and nn,j = 4 as ds,j = 4dr + 4dn
41
National Taiwan University Department of Computer Science and Information Engineering
Analysis for COFC a If CRUE < N^{\ast}, then BFS is executed, and four messages are exchanged among the UE, Cello and Celln through the air interface. a Otherwise (i.e., CRUE >= N^{\ast}), NFS is exercised. a We have
42
National Taiwan University Department of Computer Science and Information Engineering
Effects of λ/μ
43
National Taiwan University Department of Computer Science and Information Engineering
Effects of j
44
National Taiwan University Department of Computer Science and Information Engineering
Summary aIn this work, we proposed four overflow control schemes for HSDPA in UMTS. aOur schemes guarantee that when the buffer of a non-serving cell is full, the previously received packet frames in the buffer can be safely dropped, and after the UE has switched wireless link to the new serving cell, no packet frames are lost.
aComplete analysis for performances of the four schemes are given. 45
National Taiwan University Department of Computer Science and Information Engineering
References a [1] Lin, Phone, Lin, Yi-Bing, and Chlamtac, Imrich. Overflow Control for UMTS High-Speed Downlink Packet Access. Accepted and to Appear in IEEE Transactions on Wireless Communications 2003. a [2] Lin, Phone, Lin, Yi-Bing, and Chlamtac, Imrich. Modeling Frame Synchronization for UMTS High-Speed Downlink Packet Access. IEEE Transactions on Vehicular Technology, 50(1): 132-141 (January 2003).
46
Buffer Overflow Control for UMTS High-Speed Downlink Packet Access
Phone Lin (Ph.D.) Dept. of Comp. Sci. & Info. Engr., National Taiwan University, Taipei, R.O.C. FAX: +886-2-2362-8167 Email: [email protected]
1
National Taiwan University Department of Computer Science and Information Engineering
Outlines a Introduction to UMTS a Packet Access through WCDMA a High Speed Downlink Packet Access in UMTS a The Network Architecture of HSDPA a Overflow Control Issues in UMTS HSDPA a Four Schemes for Buffer Overflow Control a Performance Evaluation for Four Schemes a Summery 2
National Taiwan University Department of Computer Science and Information Engineering
Universal Mobile Telecommunication System (UMTS) (2/2)
3
National Taiwan University Department of Computer Science and Information Engineering
Universal Mobile Telecommunication System (UMTS) (1/2) aUMTS supports high data transmission rate (up to 2 mbps) for mobile users, which consists of Terrestrail Radio Access Network (UTRAN) Core network
aThe UTRAN consists of radio base stations (Node Bs) and Radio Network Controllers (RNCs), which Connected through an ATM network
4
National Taiwan University Department of Computer Science and Information Engineering
Radio Access in UMTS a A User Equipment (UE) communicates with UTRAN through the air interface Uu. WCDMA Radio Access Technology
a In a UE communication session, several cells (Node Bs) are defined as an active set. a If the quality of the wireless link between the UE and a cell is above some threshold, then this cell is included in the active set. a When the quality of the wireless link of a cell in the active set is below some threshold, then the cell is removed from the active set. 5
National Taiwan University Department of Computer Science and Information Engineering
Packet Data Traffic a A packet service session contains one or several packet calls depending on the application (e.g., Interactive & Background Services). During a packet call, the packet call constitutes a bursty sequence of packets.
a The parameters that define the characteristics of the packet data traffic in web-browsing Session arrival process (usually modeled as Poisson process) Number of packet calls per session Reading time between packet calls Time interval between two packets inside a packet call Packet size 6
National Taiwan University Department of Computer Science and Information Engineering
Characteristic of a Packet Service Session Packet Service Session Packet Call
Time
Reading Time
Packet Size
Packet Arrival Interval
7
National Taiwan University Department of Computer Science and Information Engineering
The Properties that are for Non-RealTime Packet Services a Packet data is bursty. The required bit rate can change rapidly from zero to hundreds of kilobits/second.
a Packet data tolerates longer delay than real-time services. Packet data is controllable traffic from the radio access network point of view.
a Packets can be retransmitted by the radio link control (RLC) layer. This allows the use of worse radio link quality and much higher frame error ratio than in case of real-time services.
8
National Taiwan University Department of Computer Science and Information Engineering
Overview of WCDMA Packet Access a Packet allocations in WCDMA are controlled by the packet scheduler (PS). (PS) Its functions are to Divide the available air interface capacity between the packet data users. Decide the transport channel to be used for each user’s packet data transmission. Monitor the packet allocations and the system load. The PS allocates a bit rate for a bearer and possibly changes this bit rate during an active connection. connection
a WCDMA packet access allows non-real-time bearers to use Common Channels Dedicated Channels Shared Channels 9
National Taiwan University Department of Computer Science and Information Engineering
The Location of Packet Scheduler a The PS is typically located in RNC where the scheduling can be done efficiently for multiple cells also considering the soft handover connections. connections
a The Node B provides the measurements of the air interface load for the PS. If the load > the target, target the PS can decrease the load by decreasing the bit rates of packet. If the load < target, target it can increase the load by allocating more data. The PS is also a part of the network load control because it can or reduce the network load.
10
National Taiwan University Department of Computer Science and Information Engineering
Packet Access in WCDMA Network
P acket allocation s B ase Station L oad m easurem en ts
RNC (P S )
P ack et allocations
B ase S tation 11
National Taiwan University Department of Computer Science and Information Engineering
Why Standard UTRAN Cannot work for High Speed Packet Access
aIn standard UTRAN, multiple paths exist between the UE and all Node Bs in the active set. aThis multi-path approach does not support high speed data transmission because multiple links for a UE may increase the overall interference within an UTRAN, and thus the data transmission rate decreases.
12
National Taiwan University Department of Computer Science and Information Engineering
The network architecture of HSDPA Frame n Frame 2 Frame 1
Frame n
Cell1
Frame n
UE
Frame 2 Frame 1
RNC Frame 2 Frame 1
Cell2 Frame n Frame 2 Frame 1
Cell3
UTRAN
RNC: Radio Network Controller UE: User Equipment UTRAN: UMTS Terrestrial RAdio Access Network
13
National Taiwan University Department of Computer Science and Information Engineering
HSDPA in 3GPP TR 25.950 (1/2) a The UE only selects one cell (referred to as the serving cell) in the active set for high-speed downlink transmission. a This “serving cell” is selected by the Fast Cell Selection mechanism based on the Common Pilot Channel Received Signal Code Power measurements of the cells in the active set.
14
National Taiwan University Department of Computer Science and Information Engineering
HSDPA in 3GPP TR 25.950 (2/2) a Two physical channels: High Speed Physical Downlink Shared Channel (HSPDSCH) for downlink packet frame transmission Dedicated Physical Control Channel (DPCCH) are used for uplink/downlink signaling, respectively.
a The RNC sends the packet frames to the cells in the active set. For the serving cell, the packet frames are forwarded to the UE. On the other hand, every non-serving cell in the active set queues the packet frames in a buffer. If the link quality between the serving cell and the UE degrades below some threshold, the UE selects the best cell in the active set as the new serving cell.
15
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Issues a Since the non-serving cells do not send packet frames to the UE, their buffers may overflow. a The buffer overflow issue is not addressed in 3GPP TR 25.950. a We propose five overflow control schemes for this issue. a Analytic and simulation models are conducted to evaluate the performance of the proposed schemes. a Our schemes guarantees that (i) when the buffer of a non-serving cell is full, the previously received packet frames in the buffer can be safely dropped, and (ii) after the UE has switched wireless link to the new serving cell, no packet frames are lost. 16
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Schemes (1/2) aBasic Over-flow Control (BOFC): The information needed for frame synchronization is carried by the uplink DPCCH. When the size of frame synchronization information exceeds the capacity of an uplink DPCCH, this information must be carried through multiple uplink DPCCH transmissions.
aNetwork-based Overflow Control (NOFC): NOFC to guarantee one uplink DPCCH transmission for frame synchronization through message exchange between the old and new serving cells. 17
National Taiwan University Department of Computer Science and Information Engineering
Overflow Control Schemes (2/2) aCombined BOFC and NOFC (COFC) COFC to take advantage of both BOFC and NOFC
aThe COFC with Counter Reset (COFCR) To reduce the possibility of these extra message exchanges, we can substract the counters CRUE, CSi, CRi and CSRNC by a number θ if CRUE >= N^{\star} when frame synchronization occurs.
18
National Taiwan University Department of Computer Science and Information Engineering
The BOFC scheme a Algorithm OFC1 (exercised by RNC) The RNC sends a packet frame to every cell , and increments CSrnc by one.
a Algorithm OFC2 (exercised by the serving cell) This algorithm performs flow control between the serving cell and the RNC. A window based flow control algorithm.
a Algorithm OFC3 (exercised between the UE and the serving cell). Algorithm BOFC3 is exercised between the UE and the serving cell. Since the wireless link is not reliable, “Stop-And-Wait Hybrid ARQ (SAW-Hybrid ARQ)” [5] is used in BOFC3 for flow control.
a Algorithm OFC4 (exercised by a non-serving cell) When a non-serving cell receives a packet frame from the RNC, it checks if the buffer overflows. If so, it drops a packet in front of the buffer.
a Algorithm BFS (UE switches the wireless link from the old serving cell to the new serving cell) 19
National Taiwan University Department of Computer Science and Information Engineering
The Relationship among OCF1-OFC4 algorithms
20
National Taiwan University Department of Computer Science and Information Engineering
The Timing Diagram Cells
w
acks
ackUE nackUE
w
ackUE
ackUE acks
w 21
N imax , max
National Taiwan University Department of Computer Science and Information Engineering
The Counters used in BOFC a Ni,max: the size of the buffer maintained in Celli a Ki: the number of packet frames currently stored in the buffer of Celli in the active set a CRUE: a counter maintained by UE to indicate the number of packet frames received by UE a CRi (CSi): a counter counts the number of packet frames received from the RNC (processed by Celli) a CSRNC: a counter maintained by RNC to record the number of packet frames that have been received by the serving cell a w: the window size used for downlink frame transmission from the RNC to the serving cell a Fs: an overflow flag maintained by the serving cell a WIs: indicates the number of packet frames received by Cells 22
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC1 (exercised by the RNC) aThe RNC sends a packet frame to Celli, and increments CSRNC by one. Case OFC1.1. If CSRNC mod w <> 0, the RNC continues to send the next packet frame to Celli. Case OFC1.2. If CSRNC mod w = 0, the RNC suspends the packet frame transmission until an ACK message is received from the serving cell.
23
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC2 (exercised by the serving cell for the flow control between RNC and serving cell)
a Step OFC2.1. When a packet frame arrives, Cells updates counters: CRs <- CRs + 1; Ks <- CRs – CSs; WIs <- CRs mod w
a Step OFC2.2. Cells checks WIs and Ks values to determine if it will receive the next frames from RNC. Case OFC2.2.1. If WIs <> 0, it continues to receive the next packet frame. Case OFC2.2.2. If WIs == 0 and Ks <= Ns,max – w, Cells replies the RNC an ACK, and the RNC will be triggered to send the packet frames of the next window. Case OFC2.2.3. If WIs == 0 and Ks > Ns,max – w, Cells sets Fs to 1, and no ACK message is sent to the RNC. 24
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC3 (between Cells and UE; SAW Hybrid-ARQ)
a When the UE receives a packet frame, it replies the status of the transmission to serving cell. a Case OFC3.1. If an incorrect packet frame is received, it replies a NACK message to Cells. a Case OFC3.2. If a correct frame is received, UE updates CRUE
National Taiwan University Department of Computer Science and Information Engineering
Algorithm OFC4 (exercised by a nonserving cell)
aStep OFC4.1. When Celli receives a packet frame from the RNC, it sets Ki <- CRi – CSi. aStep OFC4.2. If Ki = Ni,max (i.e., the buffer is full), Celli deletes a packet frame at the head of the buffer, and sets CSi <- CSi + 1. aStep OFC4.3. Celli adds the received packet frame at the tail of the buffer, and sets CRi
National Taiwan University Department of Computer Science and Information Engineering
The Basic Frame Synchronization (BFS) Algorithm aThe frame synchronization information (i.e., CSUE) is carried through the uplink DPCCHs in the air interface.
27
National Taiwan University Department of Computer Science and Information Engineering
Message flow for Algorithm BFS
28
National Taiwan University Department of Computer Science and Information Engineering
The NOFC Scheme aThe OCF1, OFC2, OFC3, and OFC4 algorithms are adopted in this scheme. aThe Network Frame Synchronization (NFS) algorithm is performed for frame synchronization information delivery. The frame synchronization information (i.e., CSo ) is delivered through the ATM network.
29
National Taiwan University Department of Computer Science and Information Engineering
Message flow for Algorithm NFS
30
National Taiwan University Department of Computer Science and Information Engineering
The COFC Scheme a OFC1-OFC4 algorithms are adopted in this scheme. a If the CRUE value can be fitted in one DPCCH, then the number of messages exchanged in BFS is less than that in NFS. a On the other hand, if multiple DPCCHs are required to deliver CRUE in BFS, then the cost of BFS is higher than that of NFS (because the DPCCH delay is anticipated longer and less reliable than that of message delivery in Steps NFS2-5 of NFS). a In the frame synchronization algorithm CFS of the COFC scheme, Cello determines whether BFS or NFS should be exercised when it receives the Change Serving Cell Request message. 31
National Taiwan University Department of Computer Science and Information Engineering
The Message Flow for CFS
32
National Taiwan University Department of Computer Science and Information Engineering
The COFCR Scheme aOFC1-OFC4 are adopted in this scheme. aThe CFSCR algorithm is performed for frame synchronization information delivery. aTo reduce the possibility of these extra message exchanges, we can substract the counters CRUE, CSi, CRi and CSRNC by a number θ if CRUE >= N^{\star} when frame synchronization occurs. 33
National Taiwan University Department of Computer Science and Information Engineering
The message flow for Algorithm CFS with Counter Reset
a The counters are reset at the “.” points in the time lines.
34
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (1/5)
a We investigate how the following two factors affects the affect the performance of BFS, NFS, CFS, and CFSCR. (i) traffic patterns (the session time and the packet frame arrival rate) and (ii) the serving cell residence time (the period in which the UE resides in a serving cell and receives downlink packet frames from that cell) a®ect the performance of BFS, NFS,
35
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (2/5) aWhen the UE switches from the jth cell to the j + 1st cell during a communication session, we compute (i) the number nr;j of control messages delivered in the air interface (between the UE and the cells) and (ii) the number nn,j of control messages delivered in the ATM network (among the new serving cell, the old serving cell, and the RNC).
36
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (3/5)
aLet dr and dn be the expected transmission delays in the air interface and the ATM network, respectively. athe net cost ds,j for frame synchronization and cell switching to the j + 1st cell can be expressed as ds,j = nr,jdr + nn,jdn
37
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (4/5)
aLet packet frame arrivals to the UE in a downlink transmission session be a Poisson stream with rate λ. aLet the serving cell residence time t of a UE be a random variable with the density function f(t), Laplace transform f¤(s) and the expected value 1/μ.
38
National Taiwan University Department of Computer Science and Information Engineering
Analytic Analysis for the Four Schemes (5/5) a Let Pr[K = k; J = j] be the probability that the transmission session is not complete at the jth serving cell, and during tj , there are k packet frames received by the UE. a We have
a Where
39
National Taiwan University Department of Computer Science and Information Engineering
Analysis for BOFC a Let Pr[M = m; J = j] be the probability that there are m uplink DPCCH transmissions when the UE switches to the j + 1st cell. a We have
40
National Taiwan University Department of Computer Science and Information Engineering
Analysis for NOFC a When the UE switches from the jth to the j + 1st cell, we have nr,j = 4 and nn,j = 4 as ds,j = 4dr + 4dn
41
National Taiwan University Department of Computer Science and Information Engineering
Analysis for COFC a If CRUE < N^{\ast}, then BFS is executed, and four messages are exchanged among the UE, Cello and Celln through the air interface. a Otherwise (i.e., CRUE >= N^{\ast}), NFS is exercised. a We have
42
National Taiwan University Department of Computer Science and Information Engineering
Effects of λ/μ
43
National Taiwan University Department of Computer Science and Information Engineering
Effects of j
44
National Taiwan University Department of Computer Science and Information Engineering
Summary aIn this work, we proposed four overflow control schemes for HSDPA in UMTS. aOur schemes guarantee that when the buffer of a non-serving cell is full, the previously received packet frames in the buffer can be safely dropped, and after the UE has switched wireless link to the new serving cell, no packet frames are lost.
aComplete analysis for performances of the four schemes are given. 45
National Taiwan University Department of Computer Science and Information Engineering
References a [1] Lin, Phone, Lin, Yi-Bing, and Chlamtac, Imrich. Overflow Control for UMTS High-Speed Downlink Packet Access. Accepted and to Appear in IEEE Transactions on Wireless Communications 2003. a [2] Lin, Phone, Lin, Yi-Bing, and Chlamtac, Imrich. Modeling Frame Synchronization for UMTS High-Speed Downlink Packet Access. IEEE Transactions on Vehicular Technology, 50(1): 132-141 (January 2003).
46
Related Documents
Hsdpa
November 2019 0
Hsdpa
May 2020 0
6 Hsdpa Technology 51
May 2020 0
1 Hsdpa Concept
June 2020 0
Bandluxe Cdma Hsdpa User Manual
June 2020 0