Performance Analysis of MIMO Satellite Communications Via Multiple Terrestrial Non-Regenerative Relay Nodes

Transcription

1 35 Pefomance Analysis of MIMO Saellie Communicaions Via Muliple Teesial Non-Regeneaive Relay Nodes Syliani Fassoi, Emmanouel T. Michailidis, and Ahanasios G. Kanaas Depamen of Digial Sysems School of Infomaion and Communicaion Technologies Univesiy of Piaeus 80 Kaaoli & Dimiiou S., 8534, Piaeus, Geece {sfassoi, emichail, Absac Muliple-inpu muliple-oupu (MIMO) saellie communicaion sysems have eceived he aenion of he eseach communiy ove he las yeas. This pape poposes a downlink MIMO saellie-o-eesial (ST) sysem aided by muliple amplify-and-fowad (AF) eesial elay nodes. This sysem inends o povide obus, eliable, and efficien communicaion links and impove he specal efficiency and he oal capaciy of he newok. In paicula, his pape mainly concenaes on invesigaing he pefomance of he poposed sysem and evaluaing he bi-eo-ae () and he channel capaciy. To model he saellie and eesial channel, he Loo and Rician saisical disibuions ae uilized, especively. One majo implemenaion difficuly of he MIMO echnology is he signal sepaaion (deecion) poblem a he eceiving side of he communicaion link due o inefeence fom muliseam ansmission. In his pape, he linea zeo-focing () and minimum mean squae eo () muli-anenna signal deecion echniques ae employed. To impove he pefomance wihou significanly inceasing he complexiy, odeed successive inefeence cancellaion (SIC) echniques ae also exploied. Keywods-Amplify and fowad (AF) elaying; muliple-inpu muliple-oupu (MIMO) sysems; saellie communicaions; signal deecion echiques I. INTRODUCTION As new equiemens fo access o compehensive boadband and boadcas/mulicas high-speed wieless communicaion sevices ae emeged, saellie communicaions can play an impoan ole in he evoluion of cuen and fuue communicaion sysems by poviding global coveage and ubiquious access [], []. Saellie newoks inend o subsanially suppo eesial backhaul newoks and povide unineuped adio coveage o fixed, poable, and mobile eesial eceives. The developmen of nex-geneaion communicaion sysems envisages he synegeic and seamless inegaion of heeogeneous eesial and saellie newoks wih diffeen capabiliies, poviding voice, ex and mulimedia sevices. ybid saellie-eesial newoks ae a ypical example of coopeaion beween diffeen achiecues. Fo he eesial infasucue, he muliple-inpu muliple-oupu (MIMO) achiecue has fulfilled he gowing demands fo high daa houghpus and enhanced link eliabiliy [3]. In ecen yeas, heoeical and expeimenal effos have been also devoed by academia and space agencies o he invesigaion of he applicabiliy of muliple-anenna echniques o saellie sysems and he poenial enhancemens ha can be achieved hough spaial and/o polaizaion divesiy [4], [5]. The advanages of MIMO echnology can be combined wih he feaues of coopeaive divesiy echniques via inemediae elays [], [6]-[9], in ode o impove he qualiy of sevice (QoS), exend he newok ange, and peseve he end-o-end communicaion beween a souce and a desinaion. The mos usual and well-defined ypes of elaying ae he non-egeneaive elaying, e.g., amplify-andfowad (AF) elaying, and he egeneaive elaying, e.g., decode-and-fowad (DF) elaying. In he fis ype, he elay is a convenional epeae, which jus amplifies he eceived signal and fowads i o he desinaion. In he second ype, he elay has an acive ole being able o decode he eceived signal, pefom baseband signal pocessing, and eansmi he signal o he final desinaion. In [0], [], he use of elaying in a single-anenna hybid saellie-o-eesial (ST) communicaion sysem was poposed, wheeas he pefomance of a single-anenna hybid ST muli-elay coopeaive sysem was analyzed in []. Besides, a MIMO ST communicaion sysem wih a single eesial elay was poposed in [3]. The benefis egading he ouage pobabiliy, he symbol-eo-ae (SER) and he egodic capaciy of a ST communicaion sysem wih a single muli-anenna elay compaed o a convenional single-anenna elay sysem was undelined in [4]. Indeed, eseach on muli-elay newoks wih MIMOenabled nodes emains limied. A coopeaive muli-elay MIMO sysem, whee evey eminal in he newok is employed wih muliple anennas was pesened in [5]. oweve, his sysem does no conside he special chaaceisics of he saellie adio channel. This pape invesigaes he pefomance of a downlink MIMO ST communicaion sysem wih muliple eesial elay nodes in ems of he bi-eo-ae () and he available channel capaciy. To model he saellie channel, he Loo saisical disibuion is used [6]. Besides, he eesial channel is modeled using he Rician disibuion. Since he eceive ofen obseves a linea supeposiion of

2 36 sepaaely ansmied infomaion ha canno be easily sepaaed, his pape uilizes he linea zeo-focing () [7] and he minimum mean squae eo () [8] signal deecion echniques, which ae chaaceized by compuaional simpliciy compaed o non-linea echniques. Moeove, he odeed successive inefeence cancellaion (SIC) echniques ae employed o enhance he pefomance wihou significanly affec he complexiy a he eceive [9]. The es of he pape is oganized as follows. Secion II pesens a MIMO muli-elay ST sysem. In Secion III, he saellie and eesial adio channels ae saisically modeled using widely acceped saisical disibuions. Secion IV focuses on signal deecion echniques. Resuls ae povided in Secion V. Finally, conclusions and fuue eseach pespecives ae dawn in Secion VI. II. SYSTEM MODEL OF TE MULTIPLE-INPUT MULTIPLE- OUTPUT SATELLITE-TO-TERRESTRIAL MULTI-RELAY COMMUNICATION SYSTEM In his secion, a downlink MIMO ST communicaion sysem is consideed, whee R full-duplex (FD) AF eesial elays (R) ae assigned o assis he souce (S), i.e., he saellie, in fowading is infomaion o he desinaion (D), i.e., eesial saion. Alhough half-duplex (D) elaying offes inefeence-fee ansmission a he cos of inefficien esouce uilizaion, FD elaying has eceived significan aenion and many sudies sugges ha by allowing a ceain amoun of loop-inefeence (LI), impoved pefomance can be havesed compaed o D elaying [0]. I is assumed he anennas a he elays ae isolaed and ha pefec LI cancellaion is feasible. I is also assumed ha he diec link beween souce and desinaion is obsuced due o high aenuaion. Τhe communicaion sysem compises R inemediae elay nodes equipped wih M ansmi and M ansmi anennas, whee M = M. Besides, N and N anennas ae used a he souce and he desinaion, especively. Fig. depics he communicaion scenaio, wheeas Fig. demonsaes he sysem model. Souce N SR M M s Relay R h Relay M M RD N Desinaion Figue. The sysem model of a MIMO muli-elay ST sysem. Noe ha he genealizaion o he case, whee each elay has a disinc numbe of ansmi and eceive anennas can also be similaly incopoaed in he following analysis bu a he expense of a moe complicaed noaion. The waves emied fom he souce anennas avel ove pahs wih diffeen lenghs and impinge he elays anennas. Then, he elay nodes amplify and fowad he eceived signal o he desinaion. The ansmied daa consiss of N independen daa seams, which ae allocaed o he coespondingly numbeed anennas a he souce and elay nodes. The link beween he souce and he elays epesens he saellie link, while he link beween he elays and he desinaion can be modeled as a eesial link. Daa ae ansmied in N-symbol packes. All wieless adio channels ae assumed uncoelaed, unless ohewise specified, wih fequency-fla block fading, whee he coheence ime is equal o he duaion of he N-symbol packe. Noe ha he enie sysem can be sepaaed ino R MIMO subsysems elaed wih he communicaion link beween he souce and each elay, as well as each elay and he desinaion. I is consideed ha each elay pocesses he eceived signals independenly. Fis, he MIMO subsysem fo he communicaion link beween he souce and he h elay is consideed. Fo his subsysem, he M eceived signal a he h elay fo he ih symbol is given by [ i] = [ i] [ i] + [ i], y x n () R SR R uses Fowad Link via Relays elays Fowad-Diec link Figue. Simple epesenaion of a muli-elay ST sysem. whee he maix SR is he h M N MIMO channel maix (analyically pesened in Secion III), x is he N inpu daa veco saisfying R x = E xx, whee Ε [ ] is he saisical expecaion opeao, and ( ) denoes he complex conjugae (emiian) anspose opeao, and n R is he M noise veco wih addiive whie Gaussian noise (AWGN) a he h elay s banches, whose vaiance is σ SR, he auocoelaion maix is σ SR I SR, and he covaiance maix is R nr = E. nr n R The

3 37 signal eceived by all he elays can be expessed using an T T T MR-elemen veco y [ ] y [ ] y [ ] y [ ] as follows [] R i i i i = R R RR [ i] = [ i] [ i] + [ i], y x n () R SR R T = SR SR SRR T T T whee [ i] [ i] [ i] [ i] SR is he MR N channel maix beween he souce and he elays, and n is an MR AWGN veco a he elays wih R R nr = E nrnr. Fo he MIMO subsysem of he communicaion link beween he h elay and he desinaion, he N eceived signal a he desinaion is he summaion of he R elayed signals [5] and is given by R y [ i] = a [ i] y [ i] + n [ i], (3) D RD R D = whee a is he amplificaion faco, which is assumed idenical fo each elay banch, RD is he h N M MIMO channel maix (analyically pesened in Secion III), y SR i is defined in (), and n D is he N noise veco wih AWGN a he desinaion s banches, whose vaiance is σ RD, and he auocoelaion maix is σ SRI SR, and he covaiance maix is R nd = E D D. n n The summaion of (3) can be expessed in a moe compac fom as follows [ i] a [ i] [ i] [ i], y = y + n (4) D RD R D whee [ i] [ i] [ i] [ i] RD = RD RD RRD is he N MR compound channel maix. The end-o-end signal-o-noise aio (SNR) of each elay banch can be consuced fom he compounded channels of he poposed sysem, as shown in [5, eq. (53)]. An impoan pospecive feaue of muli-elay MIMO communicaion newoks is an incease in he channel capaciy. The egodic channel capaciy (in bis/sec/z) of a MIMO AF muli-elay sysem is defined as he expecaion of he insananeous muual infomaion (MI) beween he souce and desinaion. Fundamenally, he MI is given by he diffeence beween he diffeenial enopy and he condiional diffeenial enopy of he eceived signal a he desinaion via he elays when he ansmi daa ae known. This can be expessed as [5] T ( ) = ( ) ( ) I x; y y y x. (5) d D D D Afe exensive manipulaions pesened in [] and [3], i is obained ha ( N RD ) ( x; ) log de I a [ ] [ ] y [ ] Id yd = + E RD i yr i R i (6) III. STATISTICAL MODELING OF TE SATELLITE AND TERRESTRIAL CANNEL The modeling of he saellie channel can be pefomed via a deeminisic o saisical appoach. Alhough he deeminisic channel models ae accuae, hei compuaional complexiy is lage. In paicula, he applicaion of he deeminisic channel models o saellie sysems is no pacically aacive, since a single saellie beam coves a wide popagaion aea and he deeminaion of all he elevan pahs beween he saellie and he eesial saion is difficul. On he conay, he saisical channel models expess he disibuion of he eceived signal by means of he fis-ode saisics, such as he pobabiliy densiy funcion (PDF) o he cumulaive disibuion funcion (CDF), and he second-ode saisics, such he level cossing ae (LCR) and he aveage fade duaion (AFD). Since mulipah and shadowing effecs ae impoan in he signal popagaion, he saisical models usually assume ha he eceived signal consiss of wo componens, he line-of-sigh (LoS) componen and he nonline-of-sigh (NLoS) componen. Then, he elaive powe of he diec, i.e., LoS, and mulipah, i.e., NLoS, componens of he eceived signal is conolled by he Rician faco and he disibuions of hese wo componens ae usually sudied sepaaely. The saisical models fo ST channels can be chaaceized ino wo caegoies; single sae and muli-sae models [4]. The single sae channel models ae descibed by single saisical disibuions and can be used fixed saellie scenaios, whee he channel saisics emain consan ove he aeas of inees. Besides, he muli-sae channel models ae used fo non-saionay ime-vaying popagaion condiions. In his secion, a single-sae saisical modeling appoach fo he saellie and he eesial channel is descibed. Specifically, he saellie channel is modeled using he Loo disibuion [6], whee he long-em shadowing due o oadside ees affecs only he LoS componen and is descibed hough a log-nomal disibuion, wheeas he NLoS componen is descibed by a Rayleigh PDF. ence, he esuling complex signal envelope is he sum of coelaed lognomal and Rayleigh pocesses. The Loo disibuion assumes ha he foliage no only aenuaes bu also scaes he adio waves. In addiion, he Rician disibuion is uilized, in ode o model he eesial channel. Then, a song LoS signal also aives a he eceive banches and he fading envelope follows a Rice disibuion.

4 38 A. Modeling of he saellie adio channel As peviously menioned, fo he communicaion link beween he saellie and he eesial elays, he Loo disibuion is used, which was veified expeimenally by conducing measuemens in ual aeas wih elevaion angles up o 30 o [5]. Using he Loo disibuion, he channel maix of he saellie link fo he envelope h ij is given by whee and i, j SR = h SR SR, ij = h ij + h ij = + (7) h = h exp jφ ij ij i j (, ) ( φ ) (, exp φ, ) = h exp j + h j (8) ij i j ij i j 0, π. The φ ae unifomly disibued ove [ ] φ, i, j fis faco epesens he log-nomal fading, while he second one descibes he Rayleigh fading. Theefoe, he Loo disibuion exaced fom (8) is he supeposiion of he log-nomal disibuion o model he lage-scale fading and Rayleigh disibuion fo he modeling of small-scale fading. Specifically, he Loo pobabiliy densiy funcion is given by h ij ( ij ) = p h b 0 πσ ( ) ln µ + hij hij h ij hij h ij exp I 0 dhij h 0 ij σ b 0 b0 (9) whee b 0 is he aveage scaeed powe esuling fom he mulipah componens, σ and μ ae he sandad deviaion and mean, especively, and I0 ( ) is he zeo ode modified Bessel funcion of he fis kind. B. Modeling of he eesial adio channel The eesial wieless adio channel is mosly chaaceized by he suounding local scaees in he viciniy of he eesial nodes, which poduce mulipah componens. Since a song LoS componen is also pesen, he popagaion envionmen can be chaaceized using he Rician disibuion as follows [6] K RD= RD + RD, (0) K + K + whee K is he Rician faco, which expesses he elaive powe of he diec and scaeed componens of he eceived signal fo he link beween he h elay and he desinaion and povides an indicaion of he link qualiy, RD is a deeminisic uni ank maix, which epesens he diec componen, and RD is he channel maix of he mulipah componens. When K = 0 he channel is descibed by a Rayleigh disibuion, wheeas a vey lage value of K, i.e., K, implies he pesence of a Gaussian channel. Recen sudies have shown ha he pefomance of MIMO sysems songly depends on he Rician faco [7]. In paicula, as he Rician faco inceases, he coelaion beween MIMO subchannels also inceases [8]. ence, efficien and accuae mehods fo esimaing he Rician faco ae of consideable inees [9]. Seveal values of he Rician faco have been epoed in he lieaue fom measuemen campaigns and sudies pefomed in he L- and S- fequency bands fo saellie communicaions sysems [30]. Accoding o hese measuemens, he value of he Rician faco depends on he elevaion angle of he saellie and he opeaing fequency. Neveheless, he value of he Rician faco also depends on he popagaion aea, and he degee of ubanizaion. Thus, he Rician faco is expeced o be lowe in highly ubanized aeas, whee he scaees ae usually dense. IV. LINEAR SIGNAL DETECTION SCEMES In MIMO sysems, spaial muliplexing is exploied, whee muliple seams of independen daa ae ansmied fom he ansmiing anennas. These seams should be hen sepaaed a he eceive by means of appopiae pocessing echniques. ence, signal deecion is equied fo he signals. In his pape, sandad linea signal deecion mehods fo MIMO spaial muliplexing sysems ae used due o hei simpliciy, vesailiy, well-undesood chaaceisics, and ease of exacing pefomance meics. In linea signal deecos, a linea ansfom is applied o he oupus of convenional mached files o poduce a new se of oupus, which may geneae bee esuls. These deecos ea all ansmied signals as inefeences excep fo he desied seam fom he age anenna a he ansmie. Theefoe, inefeence signals fom ohe anennas ae minimized o nullified in he couse of deecing he desied signal fom he age anenna. To faciliae he deecion of signals fom each anenna, he esimaed symbols ae inveed by a weigh maix W as follows [] T x = xx x N = Wy, () whee y = x + n is he eceive veco, is he channel maix, x is he ansmi veco, and n is he noise veco fo a geneic MIMO communicaion sysem. ence, a linea combinaion of he eceived signals in he desinaion node is consideed. Noe ha hee is one deecion fo each

5 39 symbol, which depends on he numbe of he ansmi anennas. The sandad linea deecion mehods include he well-defined and widely used and linea signal deecion echniques. The simples MIMO deeco is he deeco, which simply inves he channel maix and aemps o compleely emove (foced o zeo) he inefeence caused by he channel. Fo he case when he invese of he channel does no exis, he pseudoinvese of he channel maix is used. The deecion echnique assumes ha he base saion has pefec knowledge of he channel sae infomaion (CSI) of all uses equipmen pesen a he eceive. The weigh maix of he echnique is given by [] whee ( ) obain x ( ), W = () is he emiian anspose opeaion. Thus, we = W = y ( ) ( x + n) ( ) ( ) = Η x+ Η n ( ). = x+ n = x+ n (3), whee n = ( ) n. Noe ha he deeco pefoms pooly when he channel maix is close o being singula, since i amplifies he noise. On he ohe hand, when he channel maix is ohogonal, his subopimal linea deeco does no amplify he noise, and is equivalen o a decision feedback o non-linea maximum likelihood (ML) deeco [3]. The lae is consideed as an opimal complex echnique in he sense of minimum eo pobabiliy, when all daa vecos ae equally likely, and i fully explois he available divesiy. The noise enhancemen effec plaguing he deecion echnique can be educed by using he deecion echnique, which is also consideed subopimal. To maximize he pos-deecion signal o inefeence plus noise aio (SINR), he weigh maix is given by [] ( σ ) W = + I. (4) The eceive uses he saisical infomaion of noise. σ Thus, using he weigh in (), we obain x = W y ( σ I) ( x n) ( σ ) ( σ ) = + + = + I x+ + I n ( σ ) = x+ + I n. = x+ n (5), whee ( ) n = + σ I n. Alhough non-linea ecepion offes pefomance advanages in MIMO sysems by assising in he miigaion of he muli-anenna inefeence, he linea deecion mehods ae chaaceized by low complexiy in ems of hadwae implemenaion. To impove hei pefomance wihou significanly inceasing hei complexiy associaed wih ohe non-linea mehods, odeed SIC echniques can be exploied. These echniques conside a bank of linea eceives, each of which deecs one of he paallel daa seams, such ha he deeced signal componens successively canceled fom he eceived signal a each sage. The signal is fis obained in he deecion sep of each popagaion pah signal. Then, he signals ae combined o deec each subseam. Moe specifically, he deeced signal in each sage is subaced fom he eceived signal so ha he emaining signal wih he educed inefeence can be used in he subsequen sage []. Fig. 3 illusaes he odeed SIC signal deecion pocess fo fou spaial seams, i.e., N = 4. Le us denoe x i he symbol o be deeced in he ih ode, which may be diffeen fom he ansmi signal a he ih anenna, since x ( i) depends on he ode of deecion. Le x ( i) denoe a sliced value of x ( i ). In odeed SIC echniques, symbol esimaion can be obained using a linea deeco, such as o. The fis seam is esimaed wih he fis ow veco of he and weigh maix in (3) and (5), especively. y y s seam y hxˆ = () () () y () nd seam y = y hxˆ () () () () y () 3 d seam y = y hxˆ (3) () (3) (3) y (3) 4 h seam Figue 3. Illusaion of he odeed SIC signal deecion fo fou spaial seams. ˆx () ˆx () ˆx (3) ˆx (4)

6 40 Poviding ha x ( ) = xˆ ( ), he inefeence is successfully canceled in he couse of esimaing x ( ). oweve, if x( ) xˆ ( ), eo popagaion is incued, since he weigh, which was designed unde he pecondiion of he equaliy x( ) = xˆ ( ), is used fo he esimaion of x ( ). Due o he eo popagaion caused by eoneous decision in he pevious sages, he ode of deecion has significan influence on he pefomance of odeed SIC deecion. Fo he SIC- echnique, we obain x W y, (6) SIC = i whee y i = yd hi x fo he ih seam esimaion. Similaly, fo he SIC- echnique, we also obain whee yi = yd hi x x W y, (7) SIC = i fo he ih seam esimaion. This pape consides an AF DF muli-elay MIMO sysem. oweve, in a moe ealisic scenaio, he capaciy of a MIMO channel using a linea deeco is given by C LD k log ( SINRk ), (8) = + i= whee he SINR k fo each eceive is diffeen. The SINR fo he eceive fo he kh spaial seam can be expessed as [3] SINR k =, ( * ( ) ) (9) IN + SNR Rn whee I N is a N N ideniy maix and is he emiian anspose of. The SINR fo he eceive denoed by SINRk can be expessed as follows by condiioning on [7] SINR k = SNR ( ( ) ) Rn V. RESULTS kk. kk (0) This secion demonsaes he pefomance of he poposed communicaion sysem wih efeence o he and he available channel capaciy. To invesigae he pefomance of he MIMO muli-elay ST sysem, wo scenaios ae iniially examined (see Fig. 4). In he fis scenaio, a single-elay sysem is consideed wih wo anennas a he souce, elay, and desinaion. The second scenaio includes wo synchonized elay nodes each equipped wih single anennas. Fo he fis scenaio, he Rician faco is se o 0 db, wheeas fo he second scenaio, he Rician faco is se o 8 db fo he communicaion link beween he souce and he fis elay and 0 db fo he communicaion link beween he souce and he second elay, especively. Fig. 5 demonsaes he end-o-end pefomance fo he afoemenioned wo communicaion scenaios. QPSK modulaion is used, since saellie communicaions ae sensiive o daa loss due o he limied esouces. Accoding o he esuls, he bes pefomance is achieved wih SIC-, while he wos wih fo boh scenaios. In addiion, he MIMO wo-elay ST sysem oupefoms he MIMO single-elay ST sysem. In Fig. 6, he advanage of he echnique ove he echnique is depiced in ems of he channel capaciy. oweve, his advanage is nullified as he SNR inceases. SR RD Souce Relay Desinaion Souce SR SR (a) s Relay nd Relay (b) RD RD Desinaion Figue 4. (a) A MIMO single-elay ST communicaion sysem (b) A MIMO wo-elay ST communicaion sysem SIC- (M=, R=) (M=, R=) (M=, R=) SIC- (M=, R=) (M=, R=) (M=, R=) SIC- (M=, R=) SIC- (M=, R=) Eb/No (db) Figue 5. End-o-end pefomance of a MIMO ST communicaion sysem, whee a single elay equipped wih wo anennas o wo elays equipped wih single anennas ae used.

7 (M=, R=) (M=, R=) Ideal (M=, R=) Ideal (M=, R=) (M=, R=) (M=, R=) Loo-Rician Rician-Rician Rician-Rayleigh Loo-Rayleigh bis/s/z SNR[dB] Figue 6. Channel capaciy of a MIMO ST communicaion sysem, whee a single elay equipped wih wo anennas o wo elays equipped wih single anennas ae used. In Fig. 7, diffeen popagaion scenaios ae examined egading he fo a MIMO single-elay ST sysem. Specifically, he Loo-Rician, Loo-Rayleigh, Rician-Rician, and Rician-Rayleigh disibuions ae compaed. signal deecion is exploied and i is consideed ha he souce, he desinaion, and he elay ae equipped wih wo anennas. One obseves ha he pefomance is bee, as soon as he Loo-Rayleigh fading disibuion is consideed, i.e., he Loo disibuion is used fo he link beween souce and elay, wheeas he Rayleigh disibuion is used fo he link beween elay and desinaion. The effec of he Rician faco, which conols he sengh of he LoS componen is demonsaed in Fig. 8, whee idenical values of he Rician faco ae used fo he diffeen links. In paicula, he pefomance degades as he Rician faco inceases. Oveall, he esuls in Figs. 7 and 8 confim ha he MIMO advanages can be successfully exploied in popagaion envionmens, which ae chaaceized by a sufficienly lage numbe of noncoheen diffuse componens. In Fig. 9, he effec of he value of he amplificaion faco on he end-o-end pefomance of a MIMO ST communicaion sysem is illusaed, whee a single elay equipped wih wo anennas and SIC- echniques ae used. One obseves ha inceasing he amplificaion faco impoves he pefomance. Fig. 0 shows he end-o-end pefomance of a MIMO single-elay ST communicaion sysem fo diffeen digial modulaion schemes, i.e., BPSK, QPSK, 8- PSK, and 6-PSK. I is clea ha BPSK is he pefeed modulaion scheme fo he poposed sysem. Fig. demonsaes he channel capaciy as a funcion of he numbe of elays and he numbe of anennas a he elays. The capaciy inceases as he numbe of singleanenna elays inceases. oweve, when he elays ae equipped wih a lage numbe of anennas, inceasing he numbe of elays has an insignifican effec on he capaciy Eb/No [db] Figue 7. End-o-end pefomance of a MIMO single-elay ST communicaion sysem fo diffeen saisical modeling of he saellie and eesial channel K=0 db K=0 db K=0 db K=5 db K=5 db K=-0 db Eb/No [db] Figue 8. End-o-end pefomance in ems of he Rician faco of a MIMO ST communicaion sysem, whee a single elay equipped wih wo anennas and SIC- echniques ae used α= α=4 α= Eb/No [db] Figue 9. End-o-end pefomance in ems of he amplificaion faco of a MIMO ST communicaion sysem, whee a single elay equipped wih wo anennas and SIC- echniques ae used.

8 BPSK QPSK 8-PSK 6-PSK Eb/No [db] Figue 0. End-o-end pefomance of a MIMO ST communicaion sysem, whee a single elay equipped wih wo anennas and diffeen digial modulaion echniques ae used. bis/s/z M =M = M =M = M =M = Numbe of Relay Nodes Figue. Channel capaciy of a MIMO ST communicaion sysem fo diffeen numbe of elays and diffeen numbe of anennas a he elays. VI. CONCLUSION AND FUTURE WORK In his pape, he benefis of using muliple anenna echniques in elay-based ST sysems have been demonsaed. Specifically, he pefomance of a MIMO ST communicaions via single o muliple AF elays fo he fowad link has been invesigaed. The esuls have shown he gain in he and he achievable channel capaciy by applying,, SIC-, SIC- signal deecion schemes in diffeen popagaion condiions. These esuls have also undelined ha he mos pomising sysem model fo fuue eliable wieless newoks in difficul eains and/o high disances is he one ha uses BPSK modulaion and SIC- signal deecos. Neveheless, his wok could be fuhe impoved o exended ino diffeen aeas. Due o he lack of channelsounding measuemen campaigns, he conibuion of his wok has been limied o heoeical esuls. oweve, i is impoan o veify his esuls in eal-wold popagaion condiions. Moeove, ohe elaying echniques, such as DF elaying, and moe sophisicaed signal deecion echniques, such as he non-linea ML and Tomlinson- aashima Pecoding (TP) echniques, may be exploied, in ode o involve addiional signal pocessing and impove eo ae pefomance. The diec link fom he souce o he desinaion could be also consideed in addiion o he indiec souce o desinaion link via he elay nodes, in ode o consuc a coopeaive communicaion sysem and es is pefomance. Finally, muli-beam echniques based on he sufficien spaial sepaaion of he uses on gound and pope paiioning of he coveage aea can be also exploied, in ode o fuhe incease he specal efficiency of MIMO ST muli-elay sysems. REFERENCES [] S. Fassoi, D. Chisopoulos, S. Chazinoas, E. T. Michailidis, A. G. Kanaas, and B. Oesen, Teesial o Saellie Communicaions Using Muli-anenna Relays Nodes, in Poc. 7h Inenaional Confeence on Advances in Saellie and Space Communicaions (SPACOMM) 05, Bacelona, Spain, pp. 46-5, 9-4 Ap. 05. [] B. Evans, M. Wene, E. Luz, M. Bousque, G. E. Coazza, G. Maal, and R. Rumeau, Inegaion of Saellie and Teesial Sysems in Fuue Mulimedia Communicaions, IEEE Wieless Communicaions, vol., no. 5, pp. 7-80, Oc [3] A. J. Paulaj, D. A. Goe, R. U. Naba, and. Bolcskei, An Oveview of MIMO Communicaions A Key o Gigabi Wieless, Poceedings of he IEEE, vol. 9, no., pp. 98-8, Feb [4] P.-D. Aapoglou, K. Liolis, M. Beinelli, A. Panagopoulos, P. Cois, and R. De Gaudenzi, MIMO ove Saellie: A Review, IEEE Communicaions Suveys & Tuoials, vol. 3, no., pp. 7-5, Fis Quae 0. [5] P.-D. Aapoglou, E. T. Michailidis, A. D. Panagopoulos, A. G. Kanaas, and R. Pieo-Cedeia, The Land Mobile Eah-Space Channel: SISO o MIMO Modeling fom L- o Ka- Bands, IEEE Vehicula Technology Magazine, vol. 6, no., pp , Jun. 0. [6] A. Nosainia, T. E. une, and A. edaya, Coopeaive communicaion in wieless newoks, IEEE Commun. Mag., vol. 4, no. 0, pp , Oc [7] R. U. Naba,. Bölcskei, and F. W. Kneubuhle, Fading elay channels: Pefomance limis and space-ime signal design, IEEE J. Sel. Aeas Commun., vol., no. 6, pp , Aug [8] B. Paillassa, B. Escig, R. Dhaou, R., M.-L. Bouchee, and C. Bes, Impoving saellie sevices wih coopeaive communicaions, In. J. Saell. Commun. Newok., vol. 9, no. 6, pp , Nov./Dec. 0. [9] Y. Fan and J. Thompson, "MIMO Configuaions fo Relay Channels: Theoy and Pacice," IEEE Tans. on Wieless Communicaions, vol. 6, no. 5, pp , May 007. [0] M. K. Ai, Channel Esimaion and Deecion in ybid Saellie Teesial Communicaion Sysems, IEEE Tans. on Vehicula Technology, vol. 65, no. 7, pp , Jul. 06. [] V. K. Sakaellos, C. Kouogiogas, and A. D. Panagopoulos, Coopeaive hybid land mobile saellie eesial boadcasing sysems: Ouage pobabiliy evaluaion and accuae simulaion, Wiel. Pes. Communic., vol. 79, no., pp , Nov. 04. [] Y. Bu, M. Lin, K. An, e al., Pefomance Analysis of ybid Saellie Teesial Coopeaive Sysems wih Fixed Gain Relaying, Wiel. Pes. Communic., vol. 89, no., pp , Jul. 06. [3] Y. Dhungana, N. Rajaheva and C. Tellambua, Pefomance Analysis of Anenna Coelaion on LMS-Based Dual-op AF MIMO Sysems, IEEE Tans. on Vehicula Technology, vol. 6, no. 8, pp , Oc. 0.

9 43 [4] A. Iqbal and K. M. Ahmed, Impac of MIMO enabled elay on he pefomance of a hybid saellie-eesial sysem, Telecommun Sys., vol. 58, no., pp. 7-3, Jan. 05. [5] P. Clake and R. C. de Lamae, Tansmi Divesiy and Relay Selecion Algoihms fo Mulielay Coopeaive MIMO Sysems, IEEE Tans. on Vehicula Technology, vol. 6, no. 3, pp , Ma. 0. [6] C. Loo, A saisical model fo a land mobile saellie link, IEEE Tansacions on Vehicula Technology, vol. 34, no. 3, pp. -7, 985. [7] R. Xu, F.C.M. Lau, Pefomance analysis fo MIMO sysems using zeo focing deeco ove fading channels, Communicaions, IEE Poceedings, vol. 53, no., pp.74-80, Feb [8] D. Chisopoulos, J. Anau, S. Chazinoas, C. Mosquea, and B. Oesen, pefomance analysis of genealized mulibeam saellie channels, Communicaions Lees, IEEE, vol. 7, no. 7, pp , 03. [9] M. Mandloi and V. Bhaia, Odeed ieaive successive inefeence cancellaion algoihm fo lage MIMO deecion, in Poc. IEEE Inenaional Confeence on Signal Pocessing, Infomaics, Communicaion and Enegy Sysems (SPICES) 05, Kochi, Keala, India, pp. -5, 9 - Feb. 05. [0] I. Kikidis,. A. Suaweea, P. J. Smih and Chau Yuen, Full-duplex elay selecion fo amplify-and-fowad coopeaive newoks, IEEE Tans. Wiel. Commun., vol., no., pp , Dec. 0. [] Y. Fu, L. Yang and W. P. Zhu, "A nealy opimal amplify-and-fowad elaying scheme fo wo-hop MIMO muli-elay newoks," IEEE Communicaions Lees, vol. 4, no. 3, pp. 9-3, Ma. 00. [] Y. S. Cho, J. Kim, W. Y. Yang, and C. G. Kang, MIMO-OFDM Wieless Communicaions wih Malab, Wiley, 00. [3] I. Telaa, Capaciy of muli-anenna Gaussian channels, Eu. Tans. Telecommun., vol. 0, no. 6, pp , Nov [4] A. Abdi, W.C. Lau, M.-S. Alouini, M. Kaveh, A new simple model fo land mobile saellie channels: Fis- and second-ode saisics, IEEE Tans. Wieless Commun., vol., no. 3, pp , 003. [5] C. Loo and J. S. Buewoh, Land mobile saellie channel measuemens and modeling, in Poceedings of he IEEE, vol. 86, no. 7, pp , Jul [6] N. Lezepis and A. Gan, Capaciy of he muliple spo beam saellie channel wih Rician fading, IEEE Tans. Inf. Theoy, vol. 54, no., pp. 50-5, Nov [7] V. Taokh, N. Seshadi, and A. R. Caldebank, Space-ime codes fo high daa ae wieless communicaion: Pefomance cieion and code consucion, IEEE Tans. on Infomaion Theoy, vol. 44, no., pp , Ma [8] A. Abdi and M. Kaveh, A space-ime coelaion model fo mulielemen anenna sysems in mobile fading channels, IEEE Jounal on Selec. Aeas in Commun., vol. 0, no. 3, pp , Ap. 00. [9] A. Abdi, C. Tepedelenlioglu, M. Kaveh, and G. Giannakis, On he esimaion of he K paamee fo he Rice fading disibuion, IEEE Commun. Lees, vol. 5, no. 3, pp. 9-94, Ma. 00. [30] A. Jahn, Popagaion consideaions and fading counemeasues fo mobile mulimedia sevices, In. Jounal on Saellie Commun., vol. 9, no. 3, pp. 3-50, 00. [3] W. Peng, S. Ma, T. S. Ng, and J. Wang, A novel analyical mehod fo maximum likelihood deecion in MIMO muliplexing sysems, IEEE Tans. on Communicaions, vol. 57, no. 8, pp , Aug [3] M. Jing, Z. Ying, S. Xin, and Y. Yan, On capaciy of wieless ad hoc newoks wih MIMO eceives, IEEE Tans. on Wieless Communicaions, vol. 7, no., pp , Dec. 008.