A Multi-User Cooperative Diversity for Wireless Local Area Networks

Transcription

1 I. J. Communications, Network and System Sciences, 008, 3, Publised Online August 008 in SciRes (ttp:// A Multi-User Cooperatie Diersity for Wireless Local Area Networks Jun CHEN, Karim DJOUANI LISSI Lab., Uniersity Paris and CEDRIC-CNAM Paris, France Member, IEEE, F SATIE/TUT Pretoria, Sout Africa and LISSI Lab., Uniersity Paris, France cen_ju@auditeur.cnam.fr, djouani@uni-paris.fr Receied on Noember, 007; reised and accepted on May 3, 008 Abstract In tis paper, an idea of using space-time block coding (STBC) in multi-user cooperatie diersity as been exploited to improe te performance of te transmission in wireless local area networks. Te teoretical and simulation results sow tat, using STBC approaces can always aciee te better performance tan existing tecniques witout introducing te space-time coding. By analyzing te trougput and frame error ratio (FER) of te two different STBC cooperatie scemes, we find te trade-off between trougput and reliability. Te location of te relay is crucial to te performance, wic supposes a rule for future crosslayer design. Keywos: Multiple-input-multiple-output (MIMO), Cooperation, Space-time Block Coding (STBC).. Introduction Diersity is a powerful tecnique to mitigate fading and improe robustness to interference [], wic refers to te metod by coneying te signal to te receier oer multiple independently signal fading cannels. Te conentional iew of transmit diersity is tat a single wireless terminal transmits using an array of multipleantennas so tat te pats from eac antenna to te destination wit independently fading. Te recent researc work in tis area is te space-time coding (STC) tecniques tat ae been deeloped for multi-antenna arrays. STC is an effectie coding tecnique tat uses transmit diersity to combat te detrimental effects in wireless fading cannels [7]. Unfortunately, transmit diersity metods based on multiple-input-multipleoutput (MIMO) approac are not applicable to many wireless systems because of te size, complexity, power or oter constraints, as for instance, ad-oc networks and sensor networks. On account of tese reasons, cooperation between wireless terminals as been recently proposed as a means to proide transmit diersity as wic sown in Figure, were S, R and D represent source, relay and destination terminal, respectiely. A new metod introduced in [] and [3] to realize space diersity gain as been studied under te name of cooperatie diersity. Traditional cooperatie diersity transmits te same signals troug two different cannels as Figure. In te first time slot, te source communicates to te relay and to te destination at te same time; in te second time slot, just te relay retransmits te signal receied at te first time-slot to te destination. Te relay may simply forwa te signal receied from te source terminal or retransmit te estimates of te receied symbols, obtained by detection. We call it as repeat cooperation. In tis paper, we present a paradigm for cooperatie diersity, wic we term space-time block coding (STBC) cooperation [], integrating user cooperation wit STBC. We summarize ere te releant contributions in te area of te cooperatie diersity. Relay cannels and spacetime code form te basis for our study. Te classical treeterminal communication cannels originally examined by an der Meulen [5]. For te cannels wit multiple information sources, Kramer and an Wijngaaen [6] consider a multiple access cannel in wic te sources communicate to one destination and sare one relay. Laneman et al. examines te mode of user cooperation diersity [,3] and analyzes space time coding cooperatie diersity in nonergodic settings using outage probability as a performance measure [4]. Tey Copyrigt 008 SciRes. I. J. Communications, Network and System Sciences, 008, 3, 07-83

2 A MULTI-USER COOPERATIVE DIVERSITY FOR WIRELESS LOCAL AREA NETWORKS 67 Figure. Single-relay cooperatie diersity model. Figure. Time sequence of time slots repeat cooperatie diersity. Figure 3. Time sequence of time slots multi-user cooperatie diersity. demonstrated te extent to wic space-time coding cooperatie diersity aciees iger diersity oer tan repetition-based scemes for larger spectral efficiencies in teorem. Te model tey analyzed is a selectie ortogonal amplify and forwa (OAF) protocol, were source transmits te ector of encoded data in te first time slot and relay retransmits te receied ector by adjusting te power. Te non-ortogonal amplify-andforwa (NAF) sceme was proposed by Nabar et al. [8,9] for te single-relay cannel, were source transmits all te time but te relay only transmits on een time slots. Tey consider tree different time-diision multiple-access-based cooperatie protocols tat ary te degree of broadcasting and receie collision in eiter te amplify-and-forwa (AF) or decode-and-forwa (DF) modes. And te results indicate tat optimal space-time code design in te single relay case consists of satisfying te classical rank and determinant criteria for co-located antennas. Tese academic works sustain te possibility, existence and benefits for deploying space-time coding cooperatie diersity protocols in practice. Tis paper examines a new full-rate space-time code (Golden-Code) [] in te single-relay cooperatie NAF model. For source transmits in bot two time slots, tis protocol can aciee a iger trougput tan tat of te OAF protocol. And we ere consider tese two types of cooperatie protocols and compare te performance between Golden-Code and te classical Alamouti code [0]. Besides te distinct benefits of te space-time code, we can see te trade-off between trougput and reliability during te transmission by analyzing te results of trougput and frame error rate. At te last part, we gie a basic idea about te selection of relay. Organization of te paper. Tis paper continues as follows: Section outlines te multi-user cooperatie diersity model. Section 3 explains STBC cooperatie diersity. Section 4 sows te performance analysis by te simulation results. Section 5 summarizes our conclusions.. Multi-User Cooperatie Diersity Model We consider wireless network in wic two terminals are communicating wit a base station. Te cannel between eac terminal and te base station are independent of eac oter, and independent of te cannel between te terminals. All cannels are subject to flat (frequency nonselectie) fading in oer to isolate te benefits of spatial diersity. Considering te multi-user cooperatie diersity model, signal is to be transmitted from te source terminal S to te destination terminal D wit te assistance of te relay terminal R. All te terminals are equipped wit single antenna. Trougout te paper we assume tat a terminal cannot transmit and receie simultaneously. te cannels S D, S R and R D are known to te destination terminal. Te signal transmits procession is like following: During te first time slot, te source communicates wit te relay and destination. In te second time slot, bot te relay and source communicate wit te destination. Figure 3 sows te detail of te time sequence. In te AF relaying metod [], te relay simply amplifies and retransmits te signal receied from te source (te signal receied at te relay is distorted by fading and additie noise). No demodulation or decoding of te receied signal is performed at relay in tis case. Te signals receied by te destination and relay in te first time slot can be defined as and y = w x + n () y = w x + n () respectiely, were w and w are te aerage signal energies receied by destination oer cannel S D and S R, respectiely [9]. and are te random, complex-alued and unit-power cannel gains between S D and S R. n CN(0, N, n CN(0,N) is te Copyrigt 008 SciRes. I. J. Communications, Network and System Sciences, 008, 3, 07-83

3 68 J. CHEN ET AL. additie noises, and in general w w. Te energy of receied signal (3) is gien by ( y ) = E( w x ) + E( n ) = w N E + (3) In oer to retransmit te signal wit te same power as te sender did, te gain β for te amplification is β (4) w + N = Ten, te destination receies a superposition of relay and source during te second time slot: y = w x + βy + n (5) were is te aerage signal energy receied at te destination troug cannel R D, te definition of and n are te similar to and n. So te equation (5) can be rewritten as: y = w x + w β x n ~ (6) + were n ~ CN ( 0, N 0) wit N = N N β + 0 As te summary, te transmission function of tis cooperatie diersity is y = Hx + n (7) were y x y =, x = y x is te receied signal ector and transmitted signal ector, respectiely; n n = (8) β n + n is te noise ector; and H is te cannel matrix gien by w 0 H = (9) w β w Assuming tat te cannel coefficient matrix H is known or can be estimated, Maximum Likeliood (ML) decoding can be used at receier to fully explore te diersity adantage of te sceme. In equation (9), te noise of first time slot and second time slot do not ae te same powers, te ML estimation can not be used directly. One solution is normalizing te receied noise by a parameter ρ as follows: y ρy were w = ρw β = β N ρw 0 x n + x n ~ ρ (0) N ρ () + N Ten, equation (0) can be noted as y ~ = H ~ x+ n ~. Assuming tat te cannel coefficient matrix H ~ is known or can be estimated, te ML estimate of te transmitted packets is presented as follows: ~ xˆ = arg min n ~ Hx () x were F represents te Frobenius- norm, and x takes all possible finite alues depending on te signal constellation. 3. STBC Cooperation Model STC is a metod employed to improe te reliability of data transmission in wireless systems by using multiple transmit antennas. It relies on redundant copies of a signal to te receier in te ope tat at least some of tem may surie te pysical pat between transmission and reception. Space time codes may be split into two main types: Space-time trellis coding (STTC) [6] and STBC [7]. We are only concerned ere wit STBC wic acts on a block of data at once (similarly to block coding) and proide only diersity gain, but are muc less complex in implementation terms tan STTC. Alamouti coding [0] and Golden-Code [] are typical examples of STBC. 3.. Repeat Cooperation Firstly, we present te model sown in Figure, repeat cooperation transmits te same signals troug two different cannels. In te first time slot, te source communicates to te relay and to te destination at te same time; in te second time slot, just te relay retransmits te signal receied at te first time slot to te destination. Ten, te transmission function can be noted as follows: y = w y = x + n β w x + β F n + n (3) Te cooperatie transmission function can be written as y = x + n (4) were y w y =, = y β w n n = β n + n 3.. Alamouti Coding Cooperation Alamouti proposed a simple MIMO sceme tat aciees a full diersity gain [7] wit a simple ML Copyrigt 008 SciRes. I. J. Communications, Network and System Sciences, 008, 3, 07-83

4 A MULTI-USER COOPERATIVE DIVERSITY FOR WIRELESS LOCAL AREA NETWORKS 69 decoding algoritm. Te transmit signals are modulated using an M-ary modulation sceme, ten te encoder takes a block of two modulated signals s and s in eac encoding operation and sends it to te transmit antennas accoing to te code matrix: * x x s s C = = * (5) x3 x4 s s were * denotes complex conjugate. In tis code matrix, te first column represents te first time slot (transmission period) in a MIMO system [] and te second column represents te second time slot. Te first row corresponds to te signals transmitted from te first antenna and te second row corresponds to te signals transmitted from te second one. Tis implies tat te signals are transmitting bot in space (across two antennas) and time (two transmission interals), tat is to say, space-time coding. Te traditional Alamouti coding is designed for a two-transmit antenna system. Assuming te cooperatie metod using one-relay AF cannel, we define d = (x, x) and d = (x3, x4). Tus, in te first time slot, te source sends d, te relay and destination receie te signal; in te second time slot, te source and relay send d and xr to destination respectiely. Ten te Alamouti coding cooperatie transmission function can be written as were x X = x 3 Y = HX +N (6) x y, Y = x 4 y 3 y y 4 are te transmitted and receied signal matrix, respectiely; cannel matrix H and noise N are gien by N w H = β w n = β n + n + n 3 β n + n + n Golden-Code Cooperation Te Golden-Code is a STBC for MIMO system as Figure 5, te coding matrix for te model is: w 0 n C Figure 5. Golden-code in MIMO model. x x ( s + s ) ( s + s ) α θ α θ 3 4 = x3 x = 4 5 iα ( s3 + θ s4 ) α ( s + θ s ) (7) were s, s, s3, s4 Z[i] are te information signals, θ =, θ =, α = i iθ, α = i iθ and te factor is necessary for energy normalizing purposes 5 []. Te Golden-Code aciees te diersity multiplexing frontier [3], and in [] te Golden-Code was proposed as a full rate and full diersity code for MIMO systems. To te cooperatie metod using one-relay AF cannel, we define d = {(x,x)} and d = {(x3, x4)} wic are transmitted in first time slot and second time slot, respectiely. Te transmission function is similar to equation (6). 4. Numeral Results In tis section, some simulations are presented to sow te performances of te presented approaces. In te following simulations, Rayleig model is used for te fading cannel [0], eac cannel multi-pat is a zero mean complex Gaussian random ariable, and te distance between all te terminals is assumed to be same. Transmission energies follow te ypotesis as Table. Table. Transmission energies in simulations. Protocol st time slot nd time slot Cooperation MIMO w =.0 w =0.5 =0.5 w = w =0.5 = =0.5 Te trougput was defined as te aerage number of aailable frames tat were transmitted in a specific time slot. We performed a random experiment consisting of 0,000 repeated independent trials. Te lengt of eac frame was fixed to N = 600 bits. Considering te multipack reception, te trougput can more tan. 4.. Te Trougput Comparison between te Repeat Cooperation and STBC Cooperation Figure 4. Alamouti coding in MIMO model. We conducted comparisons between te STBC Copyrigt 008 SciRes. I. J. Communications, Network and System Sciences, 008, 3, 07-83

5 70 J. CHEN ET AL. cooperation and repeat cooperation sceme. Figure 6 and Figure 7 sow te results of trougput ersus SNR, for 6Mbps and Mbps transmit rates, respectiely. We obsere tat all te tree scemes can aciee te maximum trougput wit a ig SNR (> 5dB). Wit a special coding metod, Golden-Code cooperation sceme aciees a muc iger trougput tan te oter two. Considering te coding matrix of Golden-Code, eac row contains all te 4 original signals, wic means te full-rate of te transmission. Te cooperatie metod transmits 4 aailable signals ({s; s; s3; s4}) during time slots, wic means te maximum alue of trougput is. As to amamouti coding sceme, eac row of te coding matrix contains original signals (s and s). In eery time slot, te system transmits one signal and te conjugated signal of te oter one, were s* and s* are surly te redundancy copies of te original signals. Tat is wy only aailable signals (s and s) can be obtained at te destination in tis sceme wile 4 aailable signals ({; s; s3; s4}) can be obtained by using Golden-Code sceme. Tus, by using two pair of conjugate signals, Alamouti coding sceme transmits aailable signals during time slots of te cooperatie period, wic means te maximum alue of trougput can no more tan wit te increasing of SNR. Furtermore, as sown in Figure, repeat cooperation transmits one signal during te first time slot and retransmits te same one in te second. Clearly, repeat cooperation can just transmit signal during te two time slots. Tus, its trougput is less tan 0.5. From te simulations, we see tat wit te elp of STBC gains, te STBC cooperation is outperform repeat cooperation. And as a reasonable result of analysis and simulation, te Golden-Code cooperation can clearly aciee te best trougput among all te tree scemes. Tis also proes tat te design of te space-time code could impact te performance of te transmission. Figure 6. Trougput of STBC cooperation and repeat cooperation scemes (6Mbps). Copyrigt 008 SciRes. Figure 7. Trougput of STBC cooperation and repeat cooperation scemes (Mbps). 4.. Te FER Comparison between Nonooperation, Repeat Cooperation and STBC Cooperation Te simulation results of FER ersus SNR between Noncooperation and cooperation scemes demonstrate again tat te use of relay-assisted communication is not always beneficial wen compared to direct transmission (Non-cooperation sceme) [8]. Figure 8 and Figure 9 reeal tat te frame error rate of Non-cooperation communication is better tan tat of te simple repeat cooperation for a ig SNR (>35dB). Furter, as expected, cooperation wit STBC is always preferred oer Non-cooperation sceme. Tus from our simulations, we see tat, performance using STBC cooperation improes significantly oer Noncooperation demonstrating te adantage of using STBC cooperation. Between te two STBC cooperation scemes (Alamouti coding and Golden-Code), Alamouti coding metod sows a better performance. As we discussed, Alamouti coding transmits te redundance Figure 8. FER ersus SNR for Non-cooperation and cooperation scemes (6Mbps). I. J. Communications, Network and System Sciences, 008, 3, 07-83

6 A MULTI-USER COOPERATIVE DIVERSITY FOR WIRELESS LOCAL AREA NETWORKS 7 Golden-Code metod. Te simulation result demonstrates again tat tere is a trade-off between te trougput and te reliability Effect ia te Moements of te Relay Figure 9. FER ersus SNR for Non-cooperation and cooperation scemes (Mbps). Te main building blocks of a wireless network design are rate control, power control, medium access (sceduling) and routing. Tese building blocks are diided in layers. Typically, routing is considered in a routing layer and medium access in a MAC-layer, wereas power control and rate control are sometimes considered in a PHY-layer and sometimes in a MAC-layer. So far, te tree stations (S, R, D) were positioned equidistantly and terefore all te tree cannels ad te fixed distance. Let us denote te distance between source and destination as d; distance between source and relay as d and distance between relay and destination as d. Denote SNR, SNR, SNR as SNR between te source and destination during te time slots. We ae SNR d Figure 0. FER ersus SNR for STBC in cooperation and MIMO scemes (6Mbps). signals, a original and a conjugate. Tis is te reason tat it as a lower error rate in te destination wile GoldenCode just intersperses original signal among all parts of te transmit signals. Comparing wit te simulation results about te trougput of tese two STBC cooperation scemes, we see tat, Alamouti coding ae a lower trougput but a iger reliability tan tat of Golden-Code. As a summary, tere is always a trade-off between te trougput and te reliability. SNR d SNR d (8) were is te pat loss exponent. In te following analysis, we assume tat = 4 for urban enironment [8]. In tis section, te relay is moed, so te distance between te relay and source, te relay and destination will cange at te same time. Te effects on te signal quality wen moing te relay between te source and destination using Golden-Code cooperation wit 6Mbps and Mbps transmission rate are sown in Figure and Figure 3, respectiely. In te simulations, te distance between te sender and te destination is set to one, and terefore te SNRs sown in te X-axis is only alid for te direct link S D Te FER Comparison between STB Cooperation and MIMO Scemes Figure 0 and Figure sow us te FER of cooperation and MIMO system referring to te different SNRs. Accoing to te simulation results, te MIMO systems aciee lower FER tan te corresponding cooperatie scemes. Tis supports tat MIMO cannels allowing multiplexing gain [4,5] wic is absent in cooperatie relaying cannel since time is expended in te latter. Tus, using MIMO system always obtains te gain of spatial diersity. And as expected, te Alamouti coding metod as a better performance tan te corresponding Copyrigt 008 SciRes. Figure. FER ersus SNR for STBC in cooperation and MIMO scemes (Mbps). I. J. Communications, Network and System Sciences, 008, 3, 07-83

7 7 J. CHEN ET AL. Figure. Benefit results wen te relay is located between te source and te destination (6Mbps). Figure 3. Benefit results wen te relay is located between te source and te destination (Mbps). Te best performance is acieed wen te relay is situated in te middle of te source and destination, wic means te better cannel quality at S R and R D. And tis can be a rule for a relay-selection metod at MAC-layer using te information of PHY-layer. 5. Conclusions Tis paper describes STBC cooperation in wireless communication, a tecnique tat allows single-antenna mobiles to sare teir antennas for obtaining some benefits of multiple-antenna systems. Te diersity is realized by using a ti station as a relay and te STBC metods for information coding. We analyze te performance of two different types of STBC cooperatie metods (Alamouti coding and Golden- Code) troug te teoretical study and simulations, tere is te tradeoff between trougput and reliability during te transmission. Te results sow tat using te STBC cooperatie diersity can always increase te performance. Troug te analysis of te two metods wit te corresponding MIMO systems, we know tat te Copyrigt 008 SciRes. performance of MIMOs is always better tan tat of cooperation wit allowing multiplexing gain. Te location of te relay is crucial to te performance. Te best performance was acieed wen te relay is in te middle of source and destination. And in general te relay sould not be to far from te line between te two terminals. We beliee seeral areas of future researc on cooperatie communication will be fruitful. Firstly, te generalization of te one op space-time coded cooperation to multi-op case. Most of te researc work about cooperatie communication concerns te singleop (single-relay or multi-relay) transmission. Nowadays, multi-op ad-oc network can be found in eerywere, and te protocol adapted to multi-op enironment always deries from tat of te single-op. Secondly, te integration and interaction wit iger layer network protocols can be explored. Recently, te need for protocol adaptation and code cooperation of wireless communication system suggested a new concept of protocol arcitecture, named cross-layering arcitecture. Different protocols implemented at different protocol layers may be designed to ae mutually cooperatie reactions, based on saring te information between te different layers. Obiously, a cross-layer approac tat is based on metrics computed at pysical layer as SNR and minimal distance in te decoding process is under inestigation. Suc approac will be of a certain interest for MAC and Network leels, taking adantage of te information measured or estimated at te pysical layer. Our contribution will concern, mainly, link adaptation and frames sceduling at MAC leel. Lastly, generalization of te STBC approac to mesed network wile considering multi-cannel cooperation, radio resources management and link adaptation will be our crucial objectie in perspectie. 6. Acknowledgement Tis work comes witin te framework of a project supported by te Agence Nationale de la Recerce/ R eseau National de Recerce en T el ecommunications under name RNRT/RADIC-SF/COMSIS and reference ANR-05-RNRT References [] A. Noatinia, T. Hunter, and A. Hedayat, Cooperatie communication in wireless networks, IEEE Communications Magazine, Vol. 4, No. 0, pp , October 004. [] J. Laneman, G. Wornell, and D. Tse, An efficient protocol for realizing cooperatie diersity in wireless networks, in Proceedings IEEE ISIT, Wasington, DC, pp. 94, June 00. I. J. Communications, Network and System Sciences, 008, 3, 07-83

8 A MULTI-USER COOPERATIVE DIVERSITY FOR WIRELESS LOCAL AREA NETWORKS 73 [3] J. Laneman, D. Tse, and G. Wornell, Cooperatie diersity in wireless networks: efficient protocols and outage beaior, IEEE Transactions on Information Teory, Vol. 50, No., pp , December 004. [4] J. Laneman and G. Wornell, Distributed space-time coded protocols for exploiting cooperatie diersity in wireless networks, IEEE Transactions on Information Teory, Vol. 49, No. 0, pp , October 003. [5] E. an der Meulen, Tree-terminal communication cannels, Adanced Applications Probability, Vol. 3, pp. 0 54, 97. [6] G. Kramer and A. an Wijngaaen, On te wite Gaussian multipleaccess relay cannel, in Proceedings IEEE International Symposium Information Teory (ISIT), Sorrento, Italy, p. 40, June 000. [7] J. Q. Li, K. Letaief, and Z. G. Cao, Co-Cannel interference cancellation for space-time coded OFDM systems, IEEE Transactions on Wireless Communications, Vol., No., pp. 4 49, January 003. [8] R. Nabar and H. Bolcskei, Space-time signal design for fading relay cannels, Proceedings IEEE Globecom, San Francisco, CA, Vol. 4, pp , December 003. [9] R. Nabar, H. Bolcskei, and F. Kneubuler, Fading relay cannels: performance limits and space-time signal design, IEEE Journal on Selected Areas in Communications, Vol., No. 6, pp , August 004. [0] S. Alamouti, A simple transmit diersity tecnique for wireless communications, IEEE Journal on Selected Areas in Communications, Vol. 6, No. 8, pp , October 998. [] J. Proakis, Digital communications, Fourt edition. McGraw-Hill, 00. [] J. Belfiore, G. Rekaya, and E. Viterbo, Te Golden Code: A x full-rate space-time code wit nonanising determinants, IEEE Transactions on Information Teory, Vol. 5, No. 4, pp , April 005. [3] H. Yao and G. Wornell, Acieing te full MIMO diersity-multiplexing frontier wit rotation-based spacetime codes, Proceedings of Allerton Conference on Communication, Control and Computing, October 003. [4] L. Zeng and D. Tse, Diersity and multiplexing: A fundamental tradeoff in multiple-antenna cannels, IEEE Transactions on Information Teory, Vol. 49, No. 5, pp , May 003. [5] D. Tse, P. Viswanat, and L. Zeng, Diersity- Multiplexing tradeoff in multiple access cannels, IEEE Transactions on Information Teory, Vol. 50, No. 9, pp , September 004. [6] Vaid Tarok, Nambi Sesadri, and A. Calderbank, Space-time codes for ig data rate wireless communication: Performance analysis and code construction, IEEE Transactions on Information Teory, Vol. 44, No., pp , Marc 998. [7] V. Tarok, H. Jafarkani, and A. Calderbank, Spacetime block codes from ortogonal designs, IEEE Transactions on Information Teory, Vol. 45, No. 5, pp , July 999. [8] T. Rappaport, Wireless communications: priciples and practice, New Jersey: Prentice Hall, 996. [9] R. Lin and A. Petropulu, A new wireless medium access protocol based on cooperation, IEEE Transactions on Signal Processing, December 005. [0] Y. Zeng and C. Xiao, Improed models for te generation of multiple uncorrelated Rayleig fading waeforms, IEEE Communications Letters, Vol. 6, No. 6, June 00. [] J. Cen and K. Djouani. Space time coding in amplifyand-forwa cooperatie cannel, In Proceedings IEEE International Conference on Wireless Communications, Networking and Mobile Computing, Sangai, CN, pp , September 007. Copyrigt 008 SciRes. I. J. Communications, Network and System Sciences, 008, 3, 07-83