Inter national J our nal of Electrical and Computer Engineering (IJECE) V ol. 8, No. 3, June 2018, pp. 1684 1691 ISSN: 2088-8708 1684       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     Computationally Efficient Multi-Antenna T echniques f or Multi-User T w o-W ay W ir eless Relay Netw orks Samer Alabed Department of Electrical Engineering, American Uni v ersity of the Middle East, K uw ait Article Inf o Article history: Recei v ed Sep 24, 2017 Re vised Feb 8, 2018 Accepted Mar 16, 2018 K eyw ord: MIMO systems Multi-user tw o-w ay wireless relay netw orks Multi-antenna techniques Netw ork coding Cooperati v e di v ersity Minimum mean squared error Maximum lik elihood (ML) ABSTRA CT In this w ork, we are interested in implementing, de v eloping and e v aluating multi-antenna techniques used for multi-user tw o-w ay wireless relay netw orks that pro vide a good trade- of f between the comput ational comple xity and performance in terms of symbol error rate and achie v able data rate. In particular , a v ariety of ne wly multi-antenna techniques is proposed and studied. Some techniques based on orthogonal projection enjo y lo w com- putational comple xity . Ho we v er , the performance penalty associated with the m is high. Other techniques based on maximum lik elihood strate gy enjo y high performance, ho w- e v er , the y suf fer from v ery high computational comple xity . The Other techniques based on randomization strate gy pro vide a good trade-of f between the computational comple xity and performance where the y enjo y lo w computational comple xity with almost the same performance as compared to the techniques based on maximum lik elihood strate gy . Copyright c 2018 Institute of Advanced Engineering and Science . All rights r eserved. Corresponding A uthor: Name: Samer Alabed Af filiation: Assistant professor Address: Department of Electrical Engineering, American Uni v ersity of the Middle East, Block 3, Building 1, Eg aila, K uw ait. Phone: +965 2225 1400 Ext.: 1790 Email: Samer .Al-Abed@aum.edu.kw 1. INTR ODUCTION In a wireless netw ork, relay station adv antages can be e v aluated through tw o parameters: performance and cost. From the performance side, relay stations can be utilized to e xtend the achi v able data rate within the same cell or alternati v ely , the y can be used to e xtend the co v erage area [1]. When a relay station i s installed to e xtend the co v erage area, both, relay station and base station, use the same frequenc y at the same time which increases the spectrum reuse. Thus, the use of relay station impro v es the o v erall system throughput. By installing more base stations instead of relay stations, the same or in f act better performance can be achie v ed. Ho we v er , installing base stations is much more e xpensi v e than installing relay stations. Relay technology can be used in rural scenarios to e xtend the co v erage [2]. It can be used in the case of earthquak e or disasters where deplo ying a fix ed line backhaul link for a base station is dif ficult. In the last decade, cooperati v e di v ersity strate gies using randomly distrib uted relay nodes between the communicating terminals ha v e been e xtensi v ely studied as their impro v ements in performance do not require additional po wer or frequenc y spec- trum [3–14]. The main objecti v e of thi s w ork is to propose ef ficient relaying techniques to increase the sum rate and reduce the symbol error rate (SER) with lo w computational comple xity . 2. SYSTEM MODEL Let us consider a half duple x system which consists of M single-antenna mobile stations (MSs) commu- nicating with another M single-antenna mobile stations via a relay station (RS) ha ving N ( N 2 M ) antennas as sho wn in Fig. 1. There is no direct link between mobile stations and their communication partners. Relay station uses either the decode-and-forw ard (DF) or the amplify-and-forw ard (AF) protocol depending on the used technique. The noise at the relay stat ion and at MS nodes is assumed to be modeled as independent, zero-mean, J ournal Homepage: http://iaescor e .com/journals/inde x.php/IJECE       I ns t it u t e  o f  A d v a nce d  Eng ine e r i ng  a nd  S cie nce   w     w     w       i                       l       c       m     DOI:  10.11591/ijece.v8i3.pp1684-1691 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1685 comple x Gaussian random v ariable with v ariance 2 R S and 2 , respe cti v ely . Let us assume that the uplink and do wnlink channels are reciprocal and frequenc y flat f ading. Further , the channels are assumed to remain constant during each transmission c ycle. The maximum transmission po wer at the relay station and at t he i th MS, i.e., MS i , is gi v en by P R S and P i , respecti v ely . Furthermore, it is assumed that the channel state information (CSI) is a v ailable at the relay and mobile stations. h 1 h 2 h M h 1 h 2 h M M S 1 M S 2 M S M M S 1 M S 2 M S M R S N a n t e n n a s Figure 1. System model. The combined multiple access channel H 2 C N 2 M from all MSs to RS is gi v en by H = h 1 h 2 ::: h M h 0 1 h 0 2 ::: h 0 M (1) where h i and h 0 i , i = 1 ; :::; M are column v ectors representing channel from MS i to RS and from its corresponding partner , i.e., MS 0 i , to RS, respecti v ely . Similarly , broadcast channel from relay station to all MS nodes is gi v en by H H 2 C 2 M N . F or the gi v en system model, in the first time slot, all users transmit their data to the relay station. The signal recei v ed at relay station y R 2 C N 1 is gi v en by y R = M X i =1 h i s i + M X i =1 h 0 i s 0 i + n R (2) where s i and s 0 i are the signals transmitted from the MS i and MS 0 i nodes, respecti v ely , to the relay station and n R is the noise v ector at the relay stat ion in the first time slot. This signal needs to be processed in order to mitig ate interference and noise. Let G 2 C N N be the processing matrix at the relay station. In case of using the AF protocol, relay processing matrix is represented by a single matrix, i.e., G [15]. Whereas, in case of using the DF protocol, relay processing matrix is represented by a multiplication of three matrices G m , G b , and W , such that G = G m WG b (3) where G m is used to remo v e the ef fect of the interference occurring in the multiple access phase, i.e. during the first time slot, permutation matrix W is then used to rearrange the resulting v ector in a proper order before sending it, and G b is used to remo v e the ef fect of the interference occurring in the broadcast phase, i.e., during the second time slot. After processing the recei v ed signal y R defined in (2) at the relay by using relay processing matrix G , the signal v ector x R is obtained which is then transmitted to all mobile stations in the second time slot. 3. TECHNIQ UES T O MITIGA TE INTERFERENCE This section presents proposed relaying techniques to mitig ate interference. The orthogonal projection technique enjo ys lo w computational comple xity , ho we v er it suf fers from lo w performance in terms of SER [15]. The other technique is based on maximum lik elihood (ML) strate gy to detect the symbol v ector and then uses minimum mean square error (MMSE) s trate gy to broadcast the resulting v ector . This technique enjo ys optimal Computationally Ef ficient Multi-Antenna T ec hniques for Multi-User T wo-W ay ... (Samer Alabed) Evaluation Warning : The document was created with Spire.PDF for Python.
1686 ISSN: 2088-8708 performance in terms of SER, ho we v er it suf fers from high decoding comple xity due to the use of ML detector at the relay . More techniques are also proposed in this section in order to reduce the o v erall computational comple xity and impro v e the o v erall system performance using netw ork coding and randomization strate gy as e xplained in the ne xt subsections. 3.1. Multi-antenna technique based on ML and MMSE strategy In the technique proposed in [15], a zero-forcing strate gy is used to reduce the ef fect of the interference at the cost of noise enhancement [16]. In order to impro v e the pre vious technique, other strate gies can be used to reduce the ef fects of interference. Note that both, the interference in the multiple access phase and in the broadcast phase, need to be mitig ated. In this technique, ML detector e xplained in [17] is used to detect the recei v ed signals at the relay during the first time slot. T o mitig ate the interference occurring in the broadcast phase, MMSE strate gy is applied. 3.1.1. ML detector As e xplained in Sec. 2., we are considering a multi-user system where all users are transmitting their signals at the same time to the relay station and the recei v ed signal v ector at the relay station y R is gi v en by (2). The k e y idea of ML detector is to find the joint error for each possible combination of the transmit symbols, such as = N X n =1 j y R ( n ) M X i =1 ( h i ( n ) s i + h 0 i ( n ) s 0 i ) j 2 (4) where y R ( n ) is the n th element of the v ector y R . After calculating the v alue of for each possible c o m bination of the transmit symbols from all mobile stations, the detected transmit symbol v ector ^ s 2 C 2 M 1 which gi v es minimum v alue of is obtained. ML pro vides an optimal solution to the detect ion problem, ho we v er it suf fers from e xtremely high decoding comple xity due to the e xhausti v e search o v e r all possible combinations of symbols where its decoding comple xity increases e xponentially with the increase of the constellation si ze and the number of transmitted symbols. 3.1.2. P ermutation matrix After detecting the symbols optimally using ML detector , the symbols need to be arranged in a v ector in a proper sequence in order to recei v e them correctly at the destination node. The permutation matrix W 2 C 2 M 2 M used in (3) is gi v en by W = 0 M M I M I M 0 M M (5) where 0 M M and I M M denote an M M matrix which cont ains zeros in all its entries and an M M identity matrix, respecti v ely . The detected symbol v ector ^ s is multiplied by the permutation matrix W , such that t = W ^ s : (6) 3.1.3. MMSE strategy T o mitig ate the interference in the broadcast phase, MMSE filter is used. This filter is represented by the matrix G b used in (3). After normalization, it is gi v en by G b = ~ G b (7) where ~ G b is the MMSE filter , gi v en by ~ G b = H ( H H H + 1 = I N ) 1 ; (8) and is the f actor to fulfill po wer constraint at the relay station, gi v en by = p P R S q k ~ G b k 2 : (9) IJECE V ol. 8, No. 3, June 2018: 1684 1691 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1687 The transmitted signal v ector from relay station can then be obtained as x R = G b t : (10) The signal gi v en by (10) is then recei v ed by all mobile stations where the signal recei v ed at the i th mobile station MS i is gi v en by y i = h H i x R + n i : (11) In this technique, MS i needs only its CSI, i.e., h i . The achie v able sum rate of this system can be calculated by using R sum = 1 2 2 M X i =1 log 2 (1 + i ) (12) and the recei v e signal to interference and noise ratio (SINR) at the MS i in this case is gi v en by i = j h H i G bi j 2 2 + P j 6 = i j h H i G bj j 2 (13) where G bi represents the i th column of the relay transmit filter G b . 3.2. Multi-antenna technique based on MMSE strategy In this technique, MMSE filter is applied at the relay twice. First, as a recei v e filter at the relay during the first time slot. Second, as a transmit filter before sending the decoded data from the relay station in the second time slot. During the first time slot, the MMSE detector at the relay is gi v en by G m = ( H H H + 1 = I N ) 1 H H : (14) Using this detector , the estimated signal v ector ^ s is obtained as ^ s = G m y R : (15) This estimated v ector is then hard decoded and used for further processing. Note that the estimated v ector without hard decoding could also be used. The decoded symbols need to be rearranged to mak e sure that e v ery node recei v es symbol from its corresponding partner . This can be achie v ed by using permutation matrix gi v en by (5). The rearranged v ector t is gi v en by (6). Before transmitti ng this signal v ector , transmit MMSE filter gi v en by (7) is performed. Sec. 3.1.3. also e xplains the calculation of the po wer normalization f actor e xpressed in (9) for this filter . The final transmit v ector x R is gi v en by x R = G y R (16) where G is defined in (3). The achie v able sum rate using this technique can be calculated using (12). 3.3. Multi-antenna technique based on netw ork coding 3.3.1. Concept Netw ork coding can be useful in combining signals to transmit the m using less number of time slots [3, 7–12]. In multi-antenna scenarios, netw ork coding combines signals to reduce the number of transmitted symbols which can then be transmitted o v er less number of antennas. Otherwise, if the same number of antennas is used to transmit these combined symbols, a better performance can be achie v ed. In this technique, M-PSK modulation scheme is used. Making use of the f act that when tw o M-PSK symbols lying on the unit circle are multiplied, the resultant symbol lies on the same circle. Because of this property , E ss H is preserv ed e v en after multiplication is performed. 3.3.2. Implementation As e xplained earlier , the basic idea behind this technique is to use netw ork coding at t he relay to impro v e the performance in terms of sum rate as well as SER. The recei v ed signal v ector at the relay station is gi v en by (2). This recei v ed signal is the sum of all the signals coming from all the mobile stations. A recei v e filter is needed at the relay station to separate the signals and thus detect the correct signals from all the nodes. In this technique, MMSE filter gi v en by (14) is applied. The estimated signal v ector ^ s is obtained using (15). The symbol v ector ^ s Computationally Ef ficient Multi-Antenna T ec hniques for Multi-User T wo-W ay ... (Samer Alabed) Evaluation Warning : The document was created with Spire.PDF for Python.
1688 ISSN: 2088-8708 is then hard decoded based on t h e decision boundary of the used modulation scheme. Note that the signal v ector ^ s can also be sent without hard decoding. In the ne xt step, symbols belonging to the same communication pair are multiplied to obtain a ne w combined symbol. The ne w symbol t k , generated from the symbols of the k th pair , is gi v en by t k = ^ s k ^ s 0 k k = 1 ; ::::; M : (17) T o k eep the symbol t k generated for the k th pair separate from the symbols of the other pairs, the orthogonal projection strate gy is applied. The k e y idea of the orthogonal projection strate gy is to find a precoding matrix at the relay that groups the signals from the same pair together and eliminates the inter -pair interference [15]. The constraint for finding such a precoding matrix is N 2 M 1 which is fulfilled by our system model. In this strate gy , MS i needs to kno w the relay precoding matrix in order to remo v e the ef fect of self-interference which means that the CSI is required at the mobile stations. The precoding matrix for the k th communicating pair is gi v en by P k = h 1 ::: h k 1 h k +1 ::: h M h 0 1 ::: h 0 k 1 h 0 k +1 ::: h 0 M (18) where P k 2 C N (2 M 2) is a submatrix of channel matrix H , gi v en by (1). P k is obtained by remo ving the k th and ( k + M ) th columns from H . The orthogonal projection matrix Q k 2 C N N is obtained from interference channel P k and is gi v en by Q k = ( I N P k ( P H k P k ) 1 P H k ) k = 1 ; ::::; M : (19) Q k is then multiplied by the recei v ed signal y R gi v en by (2) to reco v er the signals of the k th pair , such as Q k y R = Q k h k s k + Q k h 0 k s 0 k + Q k n R : (20) Note that Q k h i = 0 and Q k h 0 i = 0( k 6 = i ) . Let us define a matrix 1 2 C N 1 which contains ones in all its entries. The transmitted signal from the relay is gi v en by x R = M X k =1 Q k 1 ^ t k (21) where , used to normalize the transmit signal po wer at the relay station in order to fulfill the relay station po wer constraint, is gi v en by = p P R S q P M k =1 k Q k k 2 : (22) The recei v ed signal at the i th node MS i during the second time slot is gi v en by y i = h H i x R + n i : In this technique, MS i needs only its CSI, i.e., h i . The achie v able sum rate of the system can be calculated using (12) where SINR at the node MS k is gi v en by k = j P N j =1 h H k Q k j j 2 2 (23) and Q k j is the j th column of the relay precoding matrix Q k for the k th pair nodes, i.e., MS k and MS’ k . 3.4. Randomization techniques 3.4.1. Concept In the techniques e xplained in Sec. 3.2. and Sec. 3.3., the MMSE strate gy is performed at the relay to mitig ate the interference occurring in the first time slot using (14). The techniques based on MMSE strate gy with or without netw ork coding are simple and enjo y a lo w decoding comple xity , ho we v er , the y suf fer from lo w performance in terms of SER as compared to the techniques based on ML strate gy . T o impro v e the performance of the techniques based on MMSE strate gy during the first time slot, let us search randomly for a symbol v ector ha ving same mean and v ariance as our estimated MMSE symbol v ector defined in (15). In other w ords, the randomization strate gy is used during the first time slot to find a better estimated symbol v ector at the relay and thus to reduce the o v erall error rate. IJECE V ol. 8, No. 3, June 2018: 1684 1691 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1689 The signal recei v ed at the relay y R is gi v en by (2) and the MMSE filter G m used for obtaining the estimated symbol v ector ^ s is gi v en by (14). This estimated v ector ^ s is obtained by (15). Aft er recei ving y R and obtaining ^ s using MMSE filter , a ne w random symbol v ector ^ s new is generated with mean and v ariance equal to the original estimated symbol v ector ^ s . Afterw ards, tw o error v alues for the tw o estimated symbol v ectors, ^ s and ^ s new , are generated using e = k y R H ^ s k 2 ; (24) e new = k y R H ^ s new k 2 : (25) If e new < e is true, this means that we ha v e found a better estimated symbol v ector . Therefore, the original estimated symbol v ector ^ s can then be discarded and the ne w one, i.e., ^ s new , can be stored instead. If the original estimated symbol v ector ^ s is better , then the ne wly obtai ned one, i.e., ^ s new is discarded. W e ag ain try to find an estimated symbol v ector better than the obtained one in the last step. F or that, we ag ain follo w the same procedure of finding a ne w random symbol v ector , comparing it with the e xisting one, and then storing the best one. This process needs to be repe ated for a pre-defined number of iterations to obtain a better es timated symbol v ector . The randomization technique based on MMSE without using netw ork coding is e xplained abo v e and named as the r andomization tec hnique based on MMSE str ate gy . This strate gy , i.e., randomization strate gy , can be combined also with the techniques based on netw ork coding e xplained in Sec. 3.3. in order to impro v e the estimated symbol v ectors of each pair before combining them using (17). The latter technique is named as the r andomization tec hnique based on network coding . 4. RESUL TS AND DISCUSSION This section presents the simulation settings and the results obtained for the te chniques described in Sec. 3.. The performance of the proposed techniques and the one proposed in [15] are compare d using tw o parameters: a v erage achie v able sum rate and symbol error rate. A Rayleigh flat f ading reciprocal channel is assumed for uplink and do wnlink communication. Channel v ectors are assumed to be i ndependent and identical distrib uted (i.i.d) and remain constant during the whole transmission c ycle. The whole CSI is assumed to be a v ailable at the relay station while mobile stations require their o wn CSI as e xplained in Sec. 3.. F or all simulations, the number of mobile station pairs intending to communicate with each other is set to M = 2 and number of antennas at the relay station are set to N = 4 . T ransmit signal to noise ratio (SNR) is v aried from 0 dB to 30 dB. The po wer at each mobile station is P i = 1 ; 8 i . The po wer at the relay is assumed to be proportional to the number of antennas at relay , thus, P R = N = 4 . The noise po wer is assumed to be changing in accordance with t he transmit SNR requirement. 4.1. Symbol Err or Rate T o find the SER at each SNR v alue, 500000 symbols are transmitted from each MS. In the techniques based on randomization strate gy , 30 iterations of randomization are used. Fig. 2 sho ws the performance of each technique discussed in Sec. 3. and the one proposed in [15] in terms of SER. In Fig. 2, the le gend OP , MLtxMMSE, MMSEtxMMSE, MMSESigMulOP , MMSErandSigMulOP , and MMSErandtxMMSE denote the technique pro- posed in [15], the technique based on ML strate gy e xplained in Sec. 3.1., the technique based on MMSE strate gy e xplained in Sec. 3.2., the technique based on netw ork coding e xplained in Sec. 3.3., the randomization technique based on netw ork coding e xplained in Sec. 3.4., and the randomization technique based on MMSE strate gy e x- plained in Sec. 3.4., respecti v ely . It is clearly visible that the proposed technique based on the optimal ML detector , denoted by MLtxMMSE, enjo ys the best performance and outperforms the other techniques, ho we v er , as e xplained in Sec. 3.1., it suf fers from e xtremely high decoding comple xity . On the other hand, the proposed randomization technique based on MMSE, denoted by MMSErandtxMMSE, enjo ys lo w decoding comple xity with almost the same performance as compared to the one based on the optimal ML detector . Moreo v er , the randomization strate gy can surely impro v e the performance drastically in a comparati v ely less comple x w ay . 4.2. A v erage achie v able sum rate Fig. 3 s h o ws the performance of the proposed techniques and the technique proposed in [15] in terms of achie v able sum rate where the le gend OP , MMSE, and SigMulOP denote the technique proposed in [15], the technique based on MMSE strate gy which is e xplained in Sec. 3.1. and Sec. 3.2., and the technique based on netw ork coding which is e xplained in Sec. 3.3., respecti v ely . As e xplained in Sec. 3.3., the technique based on netw ork coding reduces the number of transmitted symbols by combining them, therefore, Fig. 3 sho ws a significant g ain in the sum rate achie v ed in the case of the technique based on netw ork coding as compared to the other techniques. Computationally Ef ficient Multi-Antenna T ec hniques for Multi-User T wo-W ay ... (Samer Alabed) Evaluation Warning : The document was created with Spire.PDF for Python.
1690 ISSN: 2088-8708 0 5 10 15 20 25 30 10 −4 10 −3 10 −2 10 −1 10 0     O P [ 1 2 ] M M S E S i g M u l O P M M S E r a n d S i g M u l O P M M S E t x M M S E M M S E r a n d t x M M S E M L t x M M S E S E R S N R [ d B ] Figure 2. Symbol error rate vs SNR (dB). 0 5 10 15 20 25 30 0 5 10 15 20 25     S i g M u l O P M M S E O P [ 1 2 ] A v e r a g e s u m r a t e [ b i t / s / H z ] S N R [ d B ] Figure 3. A v erage achie v able sum rate vs SNR (dB). IJECE V ol. 8, No. 3, June 2018: 1684 1691 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 1691 5. CONCLUSION In thi s w ork, the performance in terms of SER and achie v able data rate of a v ariety of ne wly multi-antenna techniques used for multi-user tw o-w ay wireless relay netw orks are proposed and studied. The multi-antenna tech- niques based on orthogonal projection enjo ys lo w computational comple xity . Ho we v er , the performance penalty associated with them is high. Moreo v er , the techniques based on ML strate gy enjo ys high performance in terms of SER, ho we v er , the y suf fer from v ery high decoding comple xity . On the other hand, the proposed randomiza- tion technique based on MMSE strate gy enjo ys lo w decoding comple xity with almost the same performance as compared to the technique based on ML strate gy . REFERENCES [1] R. P abst, B. H.W alk e, D. Schultz, P . Herhold, H. Y anik omeroglu, S. Mukherjee, H. V isw anathan, M. Lott, W . Zirw as, M. Dohler , H. Aghv ami, D. F alconer , and G. Fettweis, ”Relay-based deplo yment concepts for wireless and mobile broadband radio, IEEE Communications Ma gazine , v ol. 42, no. 9, pp. 80-89, 2004. [2] M. Iw amura, H. T akahashi, and S . Nag ata, ”Relay technology in L TE-Adv anced, NTT DoCoMo T ec hnical J ournal , v ol. 12, no. 2: pp. 29-36, 2010. [3] S. Alabed and M. Pesa v ento, ”Distrib uted dif ferential space-time coding techniques for tw o-w ay wireless relay netw orks, the 16th International IEEE/ITG W orkshop on Smart Antennas (WSA 2012) , Dresden, German y , No v ember 2011. [4] Nasaruddin, Y unida, Khairul Munadi, ”Impro v ed model of the selection with soft and hard combining decoding strate gies for multi-user multi-relay cooperati v e netw orks, International J ournal of Electrical and Computer Engineering (IJECE) , v ol. 6, no. 4, pp. 1766 1778, August 2016. [5] C. Preetham, M. Prasad, D. Saran ya, C. Somepalli, D. Krishna, and V . Rohit, ”Performance Analysis of Cooperati v e Hybrid Cogniti v e Radio Netw ork with V arious Di v ersity T echniques, International J ournal of Electrical and Computer Engineering (IJECE) , v ol. 6, no. 5, pp. 2125 2133, October 2016. [6] Nasaruddin, et al., ”Optimized po wer allocation for cooperati v e amplify-and-forw ard with con v olutional codes, TELK OMNIKA Indonesian J ournal of Electrical Engineering , v ol. 12, no. 8, pp. 6243-6253, 2014. [7] S. Alabed, M. Pesa v ento, and A. Klein, ”Relay selection based space-time coding for tw o-w ay wireless relay netw orks using digital netw ork coding, In Pr oceedings of the T enth International Symposium on W ir eless Communication Systems , Ilmenau, TU Ilmenau, German y , Aug. 27-30, 2013. [8] S. Alabed, ”Performance Analysis of T w o-W ay DF Relay Selection T echniques, Else vier ICT Expr ess , DOI: 10.1016/j.icte.2016.08.008, September 2016. [9] S. Alabed, J. P aredes, and A. B. Gershman, ”A simple distrib uted space-time coded strate gy for tw o-w ay relay channels, IEEE T r ansactions on W ir eless Communications , pp. 1260-1265, v ol. 11, no. 4, April, 2012. [10] S. Alabed, M. Pesa v ento, and A. Gershman, ”Distrib uted dif ferential space-time coding techniques for tw o- w ay wireless rel ay netw orks, In Pr oceedings o f the F ourth IEEE Internati onal W orkshop on Computational Advances in Multi-Sensor Adaptive Pr ocessing (CAMSAP 11) , pp. 221-224, San Juan, Puerto Rico, December 2011. [11] S. Alabed, M. Pesa v ento, and A. Klein, ”Distrib uted dif ferential space-time coding for tw o-w ay relay net- w orks using analog netw ork coding, In Pr oceedings of the 1st Eur opean Signal Pr ocessing Confer ence (EU- SIPCO’13) , Marrak ech, Morocco, Sep. 9-13, 2013. [12] S. Alabed, M. Pesa v ento, and A. Klein, ”Non-coherent distrib uted space-time coding techniques for tw o-w ay wireless relay netw orks, EURASIP Special Issue on Sensor Arr ay Pr ocessing , Feb . 2013. [13] S. Alabed, J. P aredes, and A. Gershman, A lo w comple xity decoder for quasi-orthogonal space-time block codes, IEEE T r ansactions on W ir eless Communications , v ol. 10, no. 3, March 2011. [14] A. Schad, S. Alabed, H. De genhardt, and M. Pesa v ento, ”Bi-directional dif ferential beamformi ng for multi- antenna relaying, 40th IEEE International Confer ence on Acoustics, Speec h and Signal Pr ocessing , 2015. [15] Z. Zhao, Z. Ding, M. Peng, W . W ang, and K. Leung, ”A Special Case of Multi-Way Relay Channel: When Beamforming is not Applicable, IEEE T r ansactions on W ir eless Communications , v ol. 10, no. 7, pp. 2046- 2051, 2011. [16] H. Jaf arkhani, ”Space-T ime Coding: Theory and Practice, Cambridge, U.K. Cambridge Uni v . Press, 2005. [17] B. Bjeck e and J. Proakis, ”Multiple transmit and recei v e antenna di v ersity techniques for wireless communi- cations, Adaptive Systems for Signal Pr ocessing , Communications, and Contr ol Symposium 2000 , pp. 70-75, 2000. Computationally Ef ficient Multi-Antenna T ec hniques for Multi-User T wo-W ay ... (Samer Alabed) Evaluation Warning : The document was created with Spire.PDF for Python.