Advanced Scence and Tecnology Letters Vol.3 (CST 06), pp.78-83 ttp://dx.do.org/0.457/astl.06.3.34 A SINR Improvement Algortm for DD Communcaton Underlayng Cellular Networks Ceng uan, Youua Fu,, Jn Wang 3 Key Lab of Broadband Wreless Communcaton and Sensor Network Tecnology, Mnstry of Educaton, Nanjng Unversty of Posts and Telecommuncatons, Nanjng, 0003, Cna Natonal Moble Communcatons Researc Laboratory, Souteast Unversty, Nanjng, 0096, Cna 3 Nanjng Unversty of Informaton Scence and Tecnology, Nanjng, 0044, Cna Abstract. In ts paper, we study te nterference scenaro were multple Devce-to-Devce (DD) pars and one cellular use sare te same spectrum resources, and propose a novel DD sgnal-to-nterference-plus-nose rato (SINR) mprovement algortm called DSIA from te perspectve of precodng and decodng. Numercal results sow tat n comparson wt te tradtonal spectrum ortogonal sceme, te DSIA wll enable DD to aceve sgnfcant SINR gans. Keywords: DD, cellular networks, precodng, decodng, SINR. Introducton Recently, Devce-to-Devce (DD) communcaton underlayng cellular networks as been consdered as a promsng tecnology to mprove te network spectrum utlzaton, reduce te network loadng, ncrease te cellular coverage, and decrease te battery consumpton of users []. ence, DD s becomng a researc otspot. To guarantee te performance of DD communcaton, one of mportant ssues s to control te nterference. In ts paper, we nvestgate te nterference scenaro were multple DD pars reuse te same resources allocated to te cellular user, and tus te mutual nterference between DD pars and te cellular nterference to DD are bot nvolved. Accordng to ts consderaton, we try to andle te nterference problem from te perspectve of precodng and decodng, and based on tat, a sgnalto-nterference-plus-nose rato (SINR) mprovement algortm called DSIA s proposed for DD. Based on te DD system wt te ntroducng of a green AF relay, we frst formulate te nterference control problem as an optmzaton problem wt multple varables to mze te SINR of eac DD recever. Ten, te DSIA s proposed to work out ts problem and obtan te optmzed precodng and decodng vectors wc make te SINR of eac DD recever beng mzed wc aceves te goal of controllng nterference. In order to verfy te performance of te proposed novel algortm, we execute Monte Carlo smulatons for t. Numercal results sow tat n comparson wt te ISSN: 87-33 ASTL Copyrgt 06 SERSC
Advanced Scence and Tecnology Letters Vol.3 (CST 06) tradtonal spectrum ortogonal sceme [, 3] and te case wt no nterference control, te DSIA wll enable DD nvolved to obtan sgnfcant performance gans n terms of SINR. System Model and Problem Formulaton ere we consder a sngle cell nterference scenaro, tere exst one base-staton, one cellular user ( C ), one green AF relay ( R ), and M DD pars nvolvng te transmtter ( S ) wt ts correspondng recever ( D ), were and,,m. Te optmzaton problem to mze te SINR of eac DD recever s descrbed n te rest of ts secton. Frst, te sgnals receved by D n two tme slots are expressed n vector form as S D xs S D xs CD xc n D y D j j j, j D y D SD x S S jd x S j CD x C nd RD y R j, j y () were y D denotes te receved sgnal at D troug te drect lnk n tme slot, y D denotes tat at D troug te relay lnk and drect lnk n tme slot, and yr S RxS S jrxs j CRxC nr denotes te receved sgnal at R. Besdes, j, j t AB denote te cannel gans between user A ( A S,S j,r,c ) and B ( B D,R ) n tme slot t ( t, ), wc are assumed to be known at all users and modeled as t t AB cab dab, d AB are te dstances of A-to-B lnks, t c AB are te cannel t fadng coeffcents of tese lnks, and s te pat loss exponent. x Z denote te t transmtted sgnals from user Z, n B denote te addtve noses at user B followng ndependent (0, ), and s te AF relay amplfcaton factor [9]. T Ten, defnng x S xs, x S s v, were s s te data symbol transmtted by S wt te expectaton P (te transmt power of S ), and v s te precodng S vector of S wt te power constrant v v. Besdes, defnng x C x C, x C, were x C and x C are ndependent, and ter expectatons s equal to P (te C transmt power of C ). By te above defntons, () can be rewrtten as y x x x n () D SD S S jd S j CD C D j, j T Copyrgt 06 SERSC 79
Advanced Scence and Tecnology Letters Vol.3 (CST 06) were ZD 0 n D ZD n RD ZR D. ZD nd RD n R u as te correspondng and Fnally, defnng decoded sgnal at D can be obtaned va multplyng () by decodng vector of D, and te u. Based on ts, te SINR of eac DD recever can be obtaned, and tus te optmzaton problem can be formulated as SINR v,, v, u k k s.t. v v, M k (3) were SINR v,, v, u 0 N 0. RD M S u S D vv S D u P u v v N PS j S D j js D P j C j CD CD u j, j and 3 Te SINR Improvement Algortm for DD Communcaton Apparently, t s dffcult to work out te optmzaton problem (3) drectly because multple optmzed varables exst. Terefore, we frst smplfy te object functon of te optmzaton problem accordng to te generalzed Rayleg quotent and Rayleg-Rtz teorem [4], wc s expressed as SD SD v K v k k k s.t. v v, were K PS S D v j v j S D PC CD CD N. Ten, we can fnd out te j, j j j j relatonsp between te optmzed varables n accordance wt te smplfed results, e denotes te mal.e., wen u e M and v e Q (were egenvector, M PS K S D v v S D and SD SD functon n (4) can aceve te mal egenvalue (4) Q K ), te objectve Q, by wc te orgnal optmzaton problem can be transformed to one of solvng te nonlnear equatons wc s expressed as v e Q vm e Q M (5) 80 Copyrgt 06 SERSC
Advanced Scence and Tecnology Letters Vol.3 (CST 06) Fnally, we combne te dea of Smulated Annealng (SA) metaeurstc [5] wt te relatonsp between varables to propose te DD SINR mprovement algortm (.e., DSIA) solvng (5), wc s sown n Fg.. Algortm: DSIA Intalzng te precodng vector of eac DD transmtter as arbtrary 0 v k ( k ) wt te 0 0 power constrant v k v k, and temp T 0 ( T 0 denotes te ntal temperature). wle temp Tmn ( T mn denotes te lower lmt of temperature) for l to ( denotes te amount of nner loop) Calculatng v l l k e Q k and l l k Q k Q k f k 0 l l Updatng vk v k else l l Updatng v k v k wt te probablty expk temp end f l l end for 0 Resetng vk v k Annealng as temp r temp ( r denotes te annealng control parameter) end wle opt opt Outputtng te optmzed results as,, M calculatng u opt opt k e M k v,, v M. Fg.. Te DD SINR mprovement algortm: DSIA v v and 4 Numercal Results In ts secton, we present several smulatons and numercal results to verfy te t performance of te proposed algortm. ere we assume tat (0,), M, PS P S P R PS ( P R s te transmt power of R, PS Pˆ ˆ S P S s te total transmt power of S and S n te scenaro wtout te ntroducng of te relay). For smplcty, we set PS P S and ˆ ˆ 5 P S P S. Furtermore, T 0, T mn 0, r 0.8 and 0 are set n te DSIA accordng to [6]. Oter key smulaton parameters are gven below:, ds D ds D 0m, dcd dcd 50m, d d 0m, drd drd 0m and SR SR 0 k AB v, ( k ). Copyrgt 06 SERSC 8
Advanced Scence and Tecnology Letters Vol.3 (CST 06) Fg. sows te SINR of bot DD recevers wt dfferent values of PS, were PS PC and PR PS 5. From ts fgure, we can see tat te DSIA wll enable DD to obtan obvous performance gans n terms of SINR. For example, wen PS 0 db, DD usng te DSIA can aceve SINR gans of 7.64% and 3.0% over tat usng te spectrum ortogonal sceme and te case wt no nterference control, respectvely. Fg.. SINR of DD recevers wt varous DD transmt SNR 5 Concluson In ts paper, we propose a novel DD SINR mprovement algortm (DSIA) n te nterference scenaro were multple DD pars and one cellular user coexst. Numercal results sow tat DD usng te DSIA wll make all DD recevers obtan te same SINR performance, wle compared wt tat usng te tradtonal scemes, sgnfcant SINR gans can be aceved as te DD transmt power ncrease. References. Cen, Y. C., e, S. B., ou, F., S, Z. G., Cen, X.: Optmal user-centrc relay asssted devce-to-devce communcatons: an aucton approac. IET Communcatons, 9(3), pp. 386-395 (05). Doppler, K., Rnne, M., Wjtng, C., Rbero, C., ug, K.: Devce-to-devce communcaton as an underlay to LTE-advanced networks. IEEE Communcatons Magazne, 47(), pp. 4-49 (009) 3. Elkotby,. E., Elsayed, K. M. F., Ismal, M..: Explotng nterference algnment for sum rate enancement n DD-enabled cellular networks. In: IEEE Wreless Communcatons and Networkng Conference, pp. 64-69. IEEE Press, Pars (0) 8 Copyrgt 06 SERSC
Advanced Scence and Tecnology Letters Vol.3 (CST 06) 4. Zang, X. D.: Matrx analyss and applcatons. Tsngua Unversty Press, Bejng (004) 5. Glover, F., Kocenberger, G. A.: andbook of Metaeurstcs. Kluwer, New York (003) 6. Bandyopadyay, S., Saa, S., Maulk, U., Deb, K.: A smulated annealng-based multobjectve optmzaton algortm: AMOSA. IEEE Transactons on Evolutonary Computaton, (3), pp. 69-83 (008) Copyrgt 06 SERSC 83