[bmsb2009] A Transmission Architecture With Scramble Selection For Improving Cell-edge Performance And Reducing Papr In Multi-cell Mbms

1

mm09-95

A Transmission Architecture with Scramble Selection for Improving Cell-Edge Performance and Reducing PAPR in Multi-Cell MBMS Hsien-Wen Chang, Chorng-Ren Sheu, Ming-Chien Tseng, and Ching-Yung Chen

Abstract—To solve the cell-edge problem (deep and flat fading for long time) of multi-cell MBMS in SFN and reduce PAPR in OFDM systems simultaneously, a novel transmission architecture is proposed. Grid scrambling diversity scheme is adopted in the architecture to increase time and frequency diversity. Several grid scramblers are used and the time domain signal with smallest PAPR is selected for transmission. Some simulation results are provided to demonstrate the efficacy of the proposed architecture. Index Terms— multi-cell MBMS, SFN, COFDM, PAPR, time-frequency diversity, cell-edge problem

I. INTRODUCTION

M

ULTIMEDEA-broadcast multicast-service (MBMS) will become a promising service of existing and emerging cellular wireless systems. Uniform coverage-reception and high data streaming at cell-edge for multi-cell MBMS are necessary. One kind of the multi-cell networking deployments is single frequency network (SFN), which possesses several benefits, including wider cell coverage, good spectrum efficiency, and no need to switch frequency band for mobile users when entering an adjacent cell area. Orthogonal frequency division multiplexing (OFDM) transmission technique has become increasingly popular for high data rate wireless communications because it decomposes a high data rate signal into numerous low data rate signals modulating on orthogonal subcarriers. Besides, with channel coding and interleaving, OFDM systems treat channel frequency selectivity as resource of diversity since signals at subcarriers deeply faded have chance to be corrected by signals at subcarriers with good channel condition. Therefore, the Coded OFDM (COFDM)-based SFN scheme has become a mainstream scheme in the digital terrestrial broadcasting applications, such as DVB-T/H and DAB. However, at the cell edge between transmitters in an SFN, it may happen that a receiver receives the same signal from multiple transmitters almost simultaneously. If the signals have

Manuscript received April 9, 2009. Hsien-Wen Chang, Chorng-Ren Sheu and Ming-Chien Tseng are with the Industry Technology Research Institute, Hsinchu, 31040 Taiwan (phone: 886-3-5915729; fax: 886-3-5820279; e-mail: [email protected]; [email protected]; [email protected]). Ching-Yung Chen is with the Department of Computer and Communication Engineering, National Kaohsiung First University of Science and Technology, Kaohsiung, Taiwan (e-mail: [email protected]).

phase reversed to one another, their destructive combination hence results in a totally faded flat channel. It is even worse that, for a static/quasi-static receiver, this terrible situation may continue for a long time relative to the time interleaving length. Therefore, it is important to ‘create’ diversity for solving the problem without affecting the receiver design (i.e., backwards compatible). Several scrambling techniques, such as frequency group-wise scrambling (FGS) [1] and grid scrambling diversity (GSD) scheme [2], have been proposed to improve the cell-edge performance. Besides, due to the large number of subcarriers, the peak-to-average power ratio (PAPR) of conventional OFDM transmitted signal is large. The high PAPR will demand high dynamic range in the power amplifier and result in inefficient power consumption due to large output back-off (OBO). Some solutions are based on scrambling each OFDM symbol with different scrambling patterns and selecting the transmitted sequence giving the smallest PAPR, such as partial transmit sequences (PTS) and selected mapping (SLM) [3], [4]. In this paper, we propose one transmission architecture with scramble selection (SS) to improve cell-edge and PAPR performance simultaneously. This paper is organized as follows. In Section II, the multi-cell SFN system and channel model is introduced, and cell-edge as well as PAPR problem is explained. The proposed transmission architecture is described in Section III, including the introduction of GSD scheme, followed by the description of the proposed SS architecture. The performance of the proposed architecture in PAPR and error rate reduction accompanied by some discussions is treated in Section IV. Finally, some conclusions are given in Section V.

II. SYSTEM AND CHANNEL MODEL A. Multi-cell SFN A typical SFN configuration is shown in Fig. 1. There is one transmitter at the center of each cell. Each receiver may receive line-of-sight (LOS) signals as well as echoes from the transmitters. The system model is illustrated in Fig. 2. There are M BSs transmitting the same OFDM signals, and then they are received and combined at the receiver through different channels. Assuming that length of cyclic prefix (CP) of OFDM symbols is larger than the maximum path delay, the system model for n-th OFDM symbol can be described as follows.

Authorized licensed use limited to: Industrial Technology Research Institute. Downloaded on July 28, 2009 at 03:34 from IEEE Xplore. Restrictions apply.

2

mm09-95

Fig. 1. SFN system configuration. Fig. 2. Multi-cell SFN system model.. Symbol index n is dropped for simplicity.



 ,
n         

(1*



N, n$ is the received where
n  Y1, n Y2, n … YN signal after N-point point FFT operation with N subcarriers,  n  diagH 1, n H 2, n … H N, n> is the channel frequency response from the m-th th BS to the receiver,

   1,  2,  …  , ! is the encoded modulation signal to be transmitted, and    "1,  "2,  … " , ! denotes an additive white Gaussian noise. The notation #$ means the transposition of #. Moreover, % k, n is the channel frequency response of k-th subcarrier for n-th OFDM symbol from m-th BS and is related to multipath delay as: % k, n 

+

(A*

/0(1- *?@ B ()* 2  h n e-. , ) 

(2*

where L is the number of channel paths, h n and τ n is the time-varying complex gain and delay samples of l-th path, (* respectively. It’s assumed τ n  0 for simplicity simplicity. Eq. (1) can also be described as followss.  ,
n  56 7       (3* where 56 7   denotes the composite channel response of the M individual channels as: ()*



56 7        . 

()*

(4*

B. Cell-edge problem Let’s consider the situation when a receiver is located at the edge between two cells (i.e. M=2). For or ease of illustration, assume there ere is only single path (i.e. L= L=1) from both cells. From (1) and (2), the received signal is now expressed as: (* (*
n  :h n  h/ n;; · =n =  En. (5* The composite channel has become a totally flat channel due to tiny delay spread. Moreover, iff the statistically independent (* (* path gain h n and h/ n happens to be phase-reversed to each other, it becomes destructive combination. A worse case is that, for a static/quasi-static receiver, this terrible situation may continue for a long time.

C. PAPR problem The he PAPR for any BS in a multi-cell multi SFN system is defined as follows. max|xxi, n|/ PAPR  , (6* i, n|/> E|xi where 2

/0(1-*S 1 xi, n   Xk, ne n . 2 . N 1 

(7*

The PAPR of the transmitted signal is large since each sample xi, n is composed of contribution from a large number of subcarriers. The high PAPR through the power amplifier will result in nonlinear distortion ortion due to high dynamic range or insufficient power consumption due to large output back-off. back III. PROPOSED TRANSMISSION ARCHITECTURE A. Grid Scrambling Diversity Scheme To avoid flat and/or long-term term deep fading channel response in multi-cell cell SFN systems, artificial arti selectivity is created using GSD scheme as follows. In an OFDM modulator, GSD is applied after pilot symbol insertion and before IFFT. Several everal subcarriers and several symbols are grouped together to form a grid structure. Then the phase of each grid is rotated by an amount according to the scrambling code assigned to the grid. grid To assure diversity, scrambling crambling codes are uncorrelated for different grids and different cells. Fig. 3 is a simple illustration of GSD. There are two cells (i.e. M=2). In each cell, the transmitted symbols form a grid every six subcarriers and every three symbols. The received OFDM signal when GSD is applied can be represented as follows. 


OPQ n          ,  

(8*

where     1,   2,  …   , V is the scrambled signal to be transmitted from m-th cell, and

Authorized licensed use limited to: Industrial Technology Research Institute. Downloaded on July 28, 2009 at 03:34 from IEEE Xplore. Restrictions apply.

3

mm09-95 C1,t A

_ ^ ∑ W, [ u  · \ ,] · \ ,] if 1 a a 3 and 1 a W a 6, Y∑ u W,  · \ ^ · \ _ if 1 a a 3 and 7 a W a 12, ,] ,b

C1,t B

c _ ^ Z ∑ u W,  · \ ,b · \ ,] if 4 a a 6 and 1 a W a 6, Y _ ^ X∑ u W,  · \ ,b · \ ,b if 4 a a 6 and 7 a W a 12, (12* It can be observed that, by applying GSD, plenty of diversity is introduced in time-frequency dimension so that flat fading and long-term shadowing is mitigated.

h1( ) l

τ 1(l )

C1,f A h1( ) l

C1,fB

×g B , A

×g A , A

C 2,t A

C 2,t B ×g A ,B ×g B ,B

C 2,f A h 2( ) l

C 2,f B

h 2( ) l

τ 2(l ) Fig. 3. Illustration of Grid Scrambling Diversity scheme.

B. Proposed Architecture The proposed scramble selection architecture is shown in Fig. 4. The pilot insertion module performed before the scrambler module may result in channel diversity at the receiver as used in a multi-cell COFDM-based SFN. The composite channel estimated by the pilot carriers will combine the initial channel effect with the scramble pattern effect. The scramble patterns should be mutually uncorrelated for different patterns and different cells i. The adoption of proper scramble patterns and the selection of numerous scrambling patterns for the proposed architecture can help to improve the cell-edge performance and (9* reduce the PAPR in a multi-cell COFDM-based SFN.

 W,  _ ^ [ W,  · \ ,] · \ ,] if 1 a a 3 and 1 a W a 6, YW,  · \ ^ · \ _ if 1 a a 3 and 7 a W a 12, ,] ,b c  _ ^ Z W,  · \ ,b · \ ,] if 4 a a 6 and 1 a W a 6, Y _ ^ XW,  · \ ,b · \ ,b if 4 a a 6 and 7 a W a 12, _ _ ^ ^ The scrambling codes \ ,] , \ ,] , \ ,b , and \ ,b are normalized complex numbers lying on unit circle with respective values e f/gh , i j (0,2k. It can also be described as follows.
OPQ n  56 7,lmn       , (10* where 56 7,lmn   denotes the composite channel response of the M individual channels combining the effect of scrambling codes as: 

56 7,lmn      ,lmn   

 diaghopq,OPQ 1, n where hopq,OPQ k, n 

r

hopq,OPQ K, n>

(11*

IV. SIMULATIONS AND DISCUSSIONS A two-cell COFDM-based SFN system with 16-QAM modulation is considered and perfect channel estimation is assumed in the simulation. The turbo code and time interleaver (block interleaver) are also used to evaluate the turbo coded block error rate (BLER). The system parameters for each cell in the SFN can be shown in TABLE I. The number of GS patterns at each cell (S), the number of sub-carrier groups in an OFDM symbol (Gw ), and the number of groups in a sub-frame (G$ ) for the proposed transmission architecture in our simulation are shown in TABLE II. The conventional OFDM (i.e. S=1, G$  1, and Gw  1) and the FGS [1] (i.e. S=1, G$  1, and Gw  6) are used for the purpose of performance comparison. In each cell, two kinds of channels, including the single path

Fig. 4. The proposed transmission architecture for each cell.

Authorized licensed use limited to: Industrial Technology Research Institute. Downloaded on July 28, 2009 at 03:34 from IEEE Xplore. Restrictions apply.

4

mm09-95 TABLE I PARAMETERS FOR EACH CELL IN THE SFN

10

Conventional OFDM FGS (GF =6)

Parameter

Value

Number of total subcarriers Number of active subcarriers Subcarrier spacing Useful symbol duration (Ty ) Guard interval (TO )

Iteration for Turbo decoder

1,024 600 15kHz 66.67µsec 1 T 4 y 83.33µsec 16QAM 1 3 8

Block interleaver size

12 OFDM symbols

Proposed SS (GF =6, GT=12, S=10)

10

-1

Pr(PAPR > PAPR0 )

Total symbol duration (TPz{ ) Modulation Code rate for Turbo code

0

10

10

-2

-3

6

6.5

7

7.5

8

8.5

9

9.5

10

10.5

11

11.5

PAPR0 (dB)

Fig. 5. The complementary CDF for given PAPR (PAPR0) among the conventional OFDM, the FGS, and the proposed SS architecture.

Parameter

Value

Number of Grid Scrambler at each cell (S)

10

Number of subcarrier groups in an OFDM symbol (Gw )

6

Number of groups in a sub-frame (G$ )

12

TABLE III POWER DELAY PROFILE OF THE TU6 CHANNEL Tap delay (µs)

0

0.2

0.5

1.6

2.3

5

Fading gains (dB)

-3

0

-2

-6

-8

-10

channel and the 6-tap typical urban channel (TU6) are simulated. The vehicle velocity is 30 km/hr and the carrier frequency is 2.5 GHz for the both channels in our simulation. Each tap of the TU6 channel is generated by Jake’s model, and the power delay profile of the TU6 channel is shown in TABLE III. Totally 60,000 OFDM symbols are simulated in our simulation. The complementary cumulative distribution function (CDF) for given PAPRs and the turbo coded BLER for required SNRs are used to evaluated the performance of the PAPR reduction and the multi-cell diversity, respectively. Generally, we observe the given PAPR0 (dB) at the complementary CDF equals 10-3 and the required SNR (dB) at the turbo coded BLER equals 10-2. The complementary CDF among the conventional OFDM, the FGS, and the proposed SS architecture can be shown in Fig. 5. From Fig. 5, we observe that the given PAPR0 (at the complementary CDF equals 10-3) of the proposed architecture is about 2.5dB better than the given PAPR0 of both the conventional OFDM and the FGS. The performance comparisons of turbo coded BLER among the conventional OFDM, the FGS, and the proposed SS architecture in a time-varying single path channel at each cell can be shown in Fig. 6. From Fig. 6, we observe that the required SNR (at the turbo coded BLER equals 10-2) of the proposed architecture is about 5 dB and 7 dB better than the

required SNR of the FGS and the conventional OFDM, respectively. Fig. 7 shows the performance comparisons of turbo coded BLER among the conventional OFDM, the FGS, and the proposed SS architecture in a time-varying TU6 channel at each cell. From Fig. 7, we observe that the required SNR (at the turbo coded BLER equals 10-2) of the proposed architecture is about 1.5 dB and 2 dB better than the required SNR of the FGS and the conventional OFDM, respectively. The less obvious SNR improvement in time varying TU6 channel than in single-path channel is due to the fact the time varying TU6 channel owns more frequency selectivity than single-path channel.

V. CONCLUSION This paper proposes a novel transmission architecture to improve time and frequency diversity, especially at cell-edge of SFN deployment, and reduce PAPR simultaneously. Simulation results show that the proposed approach can both reduce the PAPR and make a significant improvement on the

10

Turbo Coded Block Error Rate

TABLE II PARAMETERS FOR THE PROPOSED TRANSMISSION ARCHITECTURE

0

Conventional OFDM FGS (G F=6) Proposed SS (G F=6, G T=12, S=10) 10

10

10

10

-1

-2

-3

-4

5

10

15

20

25

30

SNR (dB)

Fig. 6. Performance comparisons of turbo coded BLER among the conventional OFDM, the FGS, and the proposed SS architecture in a time-varying single path channel at each cell.

Authorized licensed use limited to: Industrial Technology Research Institute. Downloaded on July 28, 2009 at 03:34 from IEEE Xplore. Restrictions apply.

5

mm09-95

Turbo Coded Block Error Rate

10

0

Conventional OFDM FGS (G F=6) Proposed SS (G F=6, G T=12, S=10)

10

10

10

-1

-2

-3

5

6

7

8

9

10

11

12

13

14

15

SNR (dB)

Fig. 7. Performance comparisons of turbo coded BLER among the conventional OFDM, the FGS, and the proposed SS architecture in a time-varying TU6 channel at each cell.

turbo coded BLER performance in a flat-faded single path channel at each cell. It also works with realistic multipath channel, just with slighter error rare improvement, which demonstrates the robustness of the scheme. The proposed architecture is backwards compatible and can be flexibly applied to almost any existed OFDM transmission system since it doesn’t require any design change for the user equipment. REFERENCES [1]

[2]

[3]

[4]

K. Akita, R. Sakata, and N. Deguchi, “Group-Wise Scrambling Diversity for Broadcast and Multicast Services in OFDM Cellular System,” Vehicular Technology Conference, VTC 2007-Fall. IEEE 66th, pp. 174-178. Chorng-Ren Sheu, Ming-Chien Tseng, Ching-Yung Chen, and Chih-Yang Kao, “Time-frequency scrambling scheme for improving cell-edge performance in multi-Cell multimedia broadcast multicast service,” Broadband Multimedia Systems and Broadcasting, 2008 IEEE International Symposium on, 2008, pp. 1-6. S. H. Muller, R. W. Bauml, R. F. H. Fishcher and J. B. Huber, "OFDM with Reduced Peak-to-Average Power Ratio by Multiple Signal Representation", Annals of Telecommunications, Vol. 52, No. 1-2, pp. 58-67, Feb. 1997. A. Jayalath and C. Athaudage, “On the PAR reduction of OFDM signals using multiple signal representation,” Personal, Indoor and Mobile Radio Communications, 2003. PIMRC 2003. 14th IEEE Proceedings on, 2003, pp. 799-803 Vol.1.

Authorized licensed use limited to: Industrial Technology Research Institute. Downloaded on July 28, 2009 at 03:34 from IEEE Xplore. Restrictions apply.