Poceedings of he 6h WSEAS In. Conf. on Eleconics, Hadwae, Wieless and Opical Communicaions, Cofu Island, Geece, Febuay 16-19, 7 116 A Coopeaive MIMO Mobile Mulihop Relay fo Cellula Newoks Jong-Moon Chung¹, Hyung-Weon Cho², Ki-Yong Jin¹, Min-Hee Cho¹, Ho Kim² ¹ School of Elecical & Eleconic Engineeing Yonsei Univesiy Seoul 1-749, Republic of Koea hp://www.yonsei.ac.k/jmc ² Samsung Thales, Co. Ld. Yongin-Ciy, Gyeonggi-Do, 449-71, Republic of Koea Absac: In his pape, we pesen an analysis of mobile mulihop elay (MMR) echnology using coopeaive muliple-inpu-muliple-oupu (MIMO) ansmission. New wieless boadband poocols, such as WiMAX, and 3G-LTE apply MIMO echnology, and mobile communicaion poocols unde sandadizaion ae invesigaing MMR as a fuue echnology o use. These wo echnologies ae expeced o be among he mos impoan echniques in he nea fuue. Howeve convenional MIMO echnology can no be applied diecly in some applicaions. Theefoe, we invesigae a novel MMR scheme o which a coopeaive MIMO ansmission scheme can be applied. Since MMR is conduced nomally fo fa oueaching uses, one of he main facos ha influence a MMR ansmission scheme is he inefeence, which compises of inne and oue co-channel inefeence (CCI). Theefoe, his pape focuses on he pefomance analysis of he MMR scheme whee he analysis is based on he signal o inefeence powe aio (SIR) ouage pobabiliy. The esuls of his pape show ha he CCI conolled MMR scheme can povide a pefomance impovemen compaed o he convenional MMR scheme in cellula mobile communicaions. Key-Wods: Mobile Mulihop Relay, Coopeaive MIMO, Cellula Newoks 1 Inoducion In ecen yeas, wieless asymmeic daa sevices have been equesed heavily and ae expeced o be even moe popula in he fuue consideing he end of he gowing mulimedia mobile communicaion make. One of he soluions o asymmeic affic ae suppo ove wide sevices aeas is mulihop elay echniques, which ake advanage of dynamic affic dispesion [1]. The auhos of [1] show ha a numbe of ad hoc elaying saions can ovecome affic congesion by balancing he load among diffeen cells in a cos-effecive way. Fuhemoe, hee ae many invesigaions ha show he gea advanages of MMR echnology, such as, low powe consumpion, moe cos efficien sysem capaciy enhancemen, and highe sevice coveage [], [3]. Howeve, he exa elay ovehead, moe complex esouce allocaion schemes ae equied fo elaying, and incemens in he implemenaion complexiy ae dawbacks of mulihop elay echnology. Fo MIMO sysems, one of he mos impoan feaues is is abiliy o ansmi daa a highe daa aes han ha of single-inpu single oupu (SISO) sysems using he same ansmission powe. This means, in un, ha we can ansmi a fixed amoun of daa wih less powe if we use MIMO sysems. In cellula communicaion, he fac ha we ae able o ansmi daa wih less powe is impoan, because he impac of inefeence depends on he level of he ansmission powe, and also he baey lifeline can be exended. Theefoe MIMO sysems have an advanage in being used in cellula mobile communicaions due o he effec of esuling in less inefeence. In some cases, howeve, MIMO echnology can no be applied o an applicaion diecly due o sysem equiemens. Fo example, in wieless senso newoks MIMO echnology can no be used due o he node size being oo small o have muliple anennas. In addiion, he hadwae and sofwae complexiy and he sysem enegy consumpion of MIMO sysems ae highe han SISO sysems. Howeve, if we le muliple nodes opeae coopeaively we can solve his poblem. Tha is, afe a node boadcass infomaion o ohe local nodes, each local node can be pogammed o opeae like one of he anennas of a MIMO sysem,
Poceedings of he 6h WSEAS In. Conf. on Eleconics, Hadwae, Wieless and Opical Communicaions, Cofu Island, Geece, Febuay 16-19, 7 117 and can elay he signal o (o owads) he desinaion node [4], [5]. Using a coopeaive MIMO scheme, MMR can povide enhanced advanages of enegy efficiency, eliabiliy, high daa ae, and end-o-end guaaneed QoS fo shadowed aeas. Fuhemoe, aking advanage of he highe houghpu he dawback of MMR sysems equiing exa elay ovehead can be paially solved using coopeaive MIMO echnology. The emainde of he pape is oganized as follows. Secion pesens he sysem model. In Secion 3, he deivaion of he signal o inefeence powe aio is pesened. The expeimenal esuls ae pesened in Secion 4, and finally, Secion 5 concludes he pape. Sysem Model.1 Cellula Sucue We assume a mulihop cellula sucue, which consiss of a hexagonal paen, as illusaed in Fig. 1. Thee is one base saion (), seveal mulihop elay saions (s), and a numbe of mobile saions (MSs) in each cell. MSs ae assumed o be unifomly disibued. In his model, he fequency euse faco is 7 and he inefeence is compised of inne and oue inefeence. Inne inefeence is due o he mulihop elay, and he oue inefeence esuls fom co-channel inefeence, which is caused by cells using he same fequency. Fig. 1. Cellula Sucue 1s ie nd ie Refeence Cell. Cellula Sucue The adio model is based on he pah-loss powe fo adio popagaion [6] and he added ansmission pobabiliy. Accoding o his model, signal powe will depend on he local mean powe and shadowing effecs in popagaion [7] which can be descibed as, P μ + ς (1) μ P + G + G + log ( q ) 1nlog ( ) () 1 1 1 R ς N(, σ ) (3) whee P is he mean value of he eceived signal powe a he efeence cell, μ is he local mean powe as descibed in (), and ς is he shadowing paamee as epesened in (3). P is he ansmie powe, G and G ae he ansmie and eceive anenna gains especively, q is he ansmission pobabiliy, n is he pah loss exponen, R is he disance beween Tx and Rx, and σ is he sandad deviaion..3 MIMO Sysems Accoding o he link budge elaionship [8], we can calculae he ansmi powe P as (4πd ) P EbRb M l N f (4) GGλ whee E b is he equied enegy pe bi a he eceive fo a given bi eo ae (BER) equiemen, R b is he bi ae, d is he ansmission disance, G is he ansmie anenna gain, G is he eceive anenna gain, λ is he caie wavelengh, M l is he link magin, N f is he eceive noise figue defined as N f (N /N ) wih N -171dBm/Hz which is he single-sided hemal noise powe specal densiy (PSD) and N is he PSD of he oal effecive noise a he eceive inpu [4]. Fo MIMO sysems based on he Alamoui schemes wih BPSK modulaion in Rayleigh-fading channels, we can assume he insananeous eceived SNR ( γ ) and he aveage BER ( P b ) ae given by [4], [9] H F Eb γ, M 1, (5) M N { Q( )} P b ε H γ (6) whee H is a scala fading maix.
Poceedings of he 6h WSEAS In. Conf. on Eleconics, Hadwae, Wieless and Opical Communicaions, Cofu Island, Geece, Febuay 16-19, 7 118.4 Mulihop Coopeaive MIMO The sysem achiecue of he mulihop coopeaive MIMO scheme is shown in Fig.. The infomaion ha is o be sen o he MS is boadcased o a neighboing aound he fis. Then he encodes accoding o he Alamoui code and ansmis he daa bis o he MS in he nex hop (Fig.. (a)). Fo he case of uplink (Fig.. (b)), he pocedue is he same wih ha of he downlink. We assume he Rayleigh fading channel and he pah loss is modeled as a powe falloff popoional o he disance squaed. In a cell, i is also assumed ha he ansmission powe of he is esiced unde which a given aveage BER is guaaneed. 3 SIR Pefomance Deivaions 3.1 Desied Signal Powe Calculaion Desied signal powe can be calculaed using (1) fo omni-diecional anennas and nomalized anenna gain in local mean powe a he efeence cell [6]. Thus, he desied signal powe is epesened as Ps μ + ς (7) n 1 d ς N σ, and he pah loss exponen beween he MS and is 4. Then, he pobabiliy densiy funcion of he desied signal powe is expessed as a Gaussian andom vaiable X. whee 1 log ( R ) μ, (, ) fp S 1 ( γ ) N( μ σ ) X ( γ ) (8), Fig.. Mulihop Coopeaive MIMO.5 Inefeence Model The inefeence scenaios ha will be discussed in his pape ae he inefeence caused by MSs and s in he uplink. Sysem degadaion in he downlink cycle a he efeence cell is less sensiive han ha in he uplink cycle. This sems fom he fac ha he popagaion fom o suffes less aenuaion han ha fom a MS o o o. Theefoe, sysem pefomance evaluaion in he uplink cycle is easonable. Fig. 5 shows inefeing cells ha use he same fequency and inefeing eniies obseved fom he efeence cell. Fig. 3. Inefeence model 3. Aggegae Inefeence Calculaion The aggegae inefeence P I in db consiss of all inefeence scenaios MS P I 1log1 pi 1log1 ( pi ) (9) MS whee is he ansmission powe fom he MS a p i P i he neighboing cell ci, is ansmission powe P i fom he in he neighboing cell ci, is he ansmission powe fom he in he neighboing cell ci, and is he numbe of neighboing cells. Each ansmission powe fom diffeen eniies is given in (1)-(1), and is expessed as he sum of co-channel inefeence fom wihin he same cell and neighboing cells. A neighboing cell is defined as a cell in he 1s and nd ies, and uses he same fequency as he efeence cell. Nex we obain, MS ( p ) X N( μ, σ ) 1log1 MS i MS, P MS MS (1) 1 1log1 ( p i ) X N( μ, σ ), P (11) 1 + 1 + 1 + 1 1log 1 ( pi ) X N( μ, σ P ), (1)
Poceedings of he 6h WSEAS In. Conf. on Eleconics, Hadwae, Wieless and Opical Communicaions, Cofu Island, Geece, Febuay 16-19, 7 119 Eniy whee P is he ansmission powe in he efeence cell, and heefoe, P MS P and P. The sandad deviaion of he andom vaiables σ, MS σ, and σ ae se especively o 6, 8, 6 db [9]. The pobabiliy densiy funcion of he aggegaed inefeence can be epesened as using he MIMO scheme unde he cicumsance of no MMR. Fig. 5 shows ha he pdf of he aggegaed inefeence when MMR is applied (wih γ R.4). f P I MS ( γ ) 1log1 pi 1log1 ( pi ) X MS γ + X i + ( ) ( γ ). (13) 1 3.3 SIR Ouage Pobabiliy The signal o inefeence powe aio (SIR) ouage pobabiliy used in his pape is defined as { } { } ( ) P Ouage P SIR τ f γ dγ τ (14) SIR Fig. 4. PDF of he aggegaed inefeence,. whee he ouage pobabiliy is defined as he pobabiliy when he SIR is lowe han a heshold value τ and he pobabiliy densiy funcion (pdf) of he SIR can be epesened as a convoluion of he pdf of he signal powe and he pdf of he inefeence powe. f f γ f γ (15) ( ) ( ) SIR P S PI 4 Expeimenal Resuls The mulihop cellula newok pefomance was evaluaed using he pdf of inefeence and he ouage pobabiliy of he signal o inefeence powe aio. The sysem pefomance was evaluaed in 3 diffeen envionmens: (1) wihou mulihop elaying, () mulihop elaying in a pa of he fame, and (3) mulihop elaying using coopeaive MIMO. Moeove diffeen mulihop elay aes ae consideed in he pefomance evaluaion. The paamees used in he simulaion ae defined as follows. The numbe of uplink ime slos is 33, up o 3 mulihop elays exis in each cell, and he numbe of ime slos fo mulihop elay is defined as H. The mulihop elay ae is defined by he pecenage of H ove N and is values γ R have been se as.1,.,.3,.4,.5, and.6 in he pefomance analysis. Fig. 4 shows he pdf of he aggegaed inefeence Fig. 5 PDF of he aggegaed inefeence,.4. Fig. 6. Ouage pobabiliy of he SIR. Compaing hese wo gaphs, we can obseve ha inefeence can be educed if we use he MIMO
Poceedings of he 6h WSEAS In. Conf. on Eleconics, Hadwae, Wieless and Opical Communicaions, Cofu Island, Geece, Febuay 16-19, 7 1 scheme o ansmi daa. This is due o he naue of he low powe consumpion of he MIMO scheme. Fig. 6 shows he ouage pobabiliy of he SIR adoping mulihop elaying. Inefeence fom he, I, is dominan when γ R, bu when γ R inceases, he inefeence due o he s becomes moe significan and heavily effec he sysem. In ohe wods, when adoping mulihop elay, I becomes he dominan inefeence faco and he sysem pefomance is seveely degaded by I. 4 Conclusion In his pape a coopeaive MIMO MMR scheme is pesened and is pefomance has been analyzed. The MMR scheme povides seveal echnical advanages. Howeve, as he esuls demonsae, as he mulihop elay ae inceases, he co-channel inefeence level inceases, and hee needs o be a scheme o conol (i.e., manage) he inefeence, o he advanages of using MMR echnology will disappea. To solve his poblem, his pape poposes a coopeaive MIMO scheme combined wih MMR echnology. In his combined model, he base saion boadcass daa o he elay saion fis, and hen he elay saion encodes he daa and ansmis i o he desinaion node. Though his pocedue, he elay saions ae opeaed as a single anenna of he MIMO ansmie, and he advanages of MIMO sysems can be obained. Using he mahemaical analysis povided, we can see ha he aggegaed inefeence can be educed by adoping he coopeaive MIMO scheme. This is due o he fac ha he MIMO scheme consumes less ansmission enegy when ansmiing he same amoun of daa. Howeve, as he ae of using MMR inceases, he ouage pobabiliy also inceases. Theefoe, he mulihop elay ae should be chosen caefully no o cancel ou he gain which is obained by using coopeaive MIMO. The analysis and esuls of his pape povide a mehod o accuaely esimae he pefomance influencing facos of coopeaive MIMO in mobile mulihop applicaions, such ha fuue boadband mobile communicaion echniques can be developed wih an impoved oveall sysem pefomance and wide sevice ange hough MMR echnology. 5 Acknowledgmens This eseach was sponsoed by Samsung Thales, Co. Ld. The echnologies pesened in his pape ae paen pending. Refeences: [1] H. Wu, C. Qiao, S. De, O. Tonguz, Inegaed Cellula and Ad Hoc Relaying Sysems: icar, IEEE JSAC, vol. 19, pp. 15-115, Oc. 1. [] M. DeFaia, E. S. Sousa, Effec of Inecell Inefeence on he SINR of a Mulihop Cellula Newok, Poc. IEEE 61s. VTC5-Sping, vol. 5, pp. 317-3111, May 5. [3] Z. Dawy, S. Davidovic, I. Oikonomidis, Coveage and Capaciy Enhancemen of CDMA Cellula Sysems via Mulihop Tansmission, IEEE Globecom 3, pp. 1147-1151, 3. [4] S. Cui, A. J. Goldsmih, A. Bahai, Enegy-Efficiency of MIMO and Coopeaive MIMO Techniques in Senso Newoks, IEEE JSAC, vol., no. 6, pp. 189-198, Aug. 4. [5] Y. Yuan, Z. He, M. Chen, Viual MIMO-Based Coss-Laye Design fo Wieless Senso Newoks, IEEE Tans. Vehicula Technol., vol. 55, no. 3, pp. 856-864, May. 6. [6] W. C. Jeong and M. Kavehad, Cochannel Inefeence Reducion in Dynamic-TDD Fixed Wieless Applicaions, Using Time Slo Allocaion Algoihms, IEEE Tans. Commun. vol. 5, no. 1, pp. 167-1636, Oc.. [7] R. Pasad, Univesal Wieless Pesonal Communicaions, Aech House, 1998. [8] J. G. Poakis, Digial Communicaions, 4 h ed. New Yok, NY: McGaw-Hill,. [9] V. Kühn, Wieless Communicaions ove MIMO Channels Applicaions o CDMA and Muliple Anenna Sysems. New Yok, NY: Wiley, 6. [1]B. S. Bae, D. H. Cho, An Inegaed Scheduling Algoihm in a Wieless Packe Newok Wih Dynamic TDD, Poc. IEEE 5h Vehicula Technology Confeence (VTC 1999-Fall), 1999. [11] J. Li, S. Faahvash, M. Kavehad, Dynamic Time-Division Duplex Wieless Local Loop, Poc. IEEE VTC-Fall, vol. 3, pp.178-185, Boson, MA, Sep..