PERFORMANCE ANALYSIS OF THE LINK-ADAPTIVE COOPERATIVE AMPLIFY-AND-FORWARD RELAY NETWORKS WITH OPPORTUNISTIC RELAYING STRATEGY

Transcription

1 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 PERFORMACE AALYSIS OF HE LIK-ADAPIVE COOPERAIVE AMPLIFY-AD-FORWARD RELAY EWORKS WIH OPPORUISIC RELAYIG SRAEGY Bhuvan Mod, O. Olaby and A.Annamala Center of Excellence for Communcaton Systems echnology Research Department of Electrcal and Computer Engneerng, Prare Vew A & M Unversty, X Unted States of Amerca ABSRAC hs paper analyzes the performance of cooperatve amplfy-and-forward (CAF) relay networks that employ adaptve M-ary quadrature ampltude modulaton (M-QAM)/M-ary phase shft keyng (M-PSK) dgtal modulaton technques n akagam-m fadng channel. In partcular, we present and compared the analyss of CAF relay networks wth dfferent cooperatve dversty and opportunstc routng strateges such as regular Maxmal Rato Combnng (MRC), Selecton Dversty Combnng (SDC), Opportunstc Relay Selecton wth Maxmal Rato Combnng (ORS-MRC) and Opportunstc Relay Selecton wth Selecton Dversty Combnng (ORS-SDC). We advocate a smple yet unfed numercal approach based on the margnal moment generatng functon (MGF) of the total receved SR to compute the average symbol error rate (ASER), mean achevable spectral effcency, and outage probablty performance metrcs. KEYWORDS Cooperatve communcatons, adaptve M-QAM/MPSK modulaton, Opportunstc relay selecton. IRODUCIO he current and the future network desgn s hghly challenged n every front due to ncreasng connectvty and data rate requrements. he global nternet traffc has experenced exponental growth n the past 0 years and ths plummetng growth s expected to contnue n the future. Csco has predcted that annual global nternet traffc s expected to reach zettabyte threshold by 05 from current 5 bllon network connectons (ncludng machne-to-machne connectons) []. hs means, an average of more than two devces are expected to be n use per person on earth. hs surge n connectvty s attrbuted to the prolferaton of the communcaton devces such as tablets, moble phones, connected applances, and other smart machnes. Snce most of these devces are moble n nature, the ncreased connectvty requrement wll be placng a huge demand on already lmted wreless network access resources. Also, as most of the predcted nternet traffc s expected to be domnated by vdeo contents, there s a need to fnd more cost effectve ways of delverng these hgh data rate servces to the end users wthn the lmted wreless channel bandwdth. herefore, the development of very hgh-speed wreless access system s mperatve. Most of the ongong communcaton research and ndustral standard efforts are dedcated to solvng ths problem. In fact the evoluton of moble networks from /.5G (GSM, GPRS, EDGE, IS95/IS98) to 3G (WCDMA/HSPA/CDMA000) and to 4G (LE/ HSPA+/ WIMAX) has been n response to address ths ssue []. Whle the current 4G access DOI: 0.5/jcnc

2 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 network holds the promse of delverng up to GBps data rate to end users (mostly avalable at the macrocell base staton), the prolferaton of moble devces has lead to very small-sze "hot spot" and therefore requre extensve dmensonng of network resources n terms of coverage. hese hot spots could be homes, trans, arports and possbly buses wth hgh data rate wreless access requrements. he range extenson (wth hgh speed connectvty) has been a major pont of nterest n LE advance standard, and femtocells (very small base statons nstallable by the end users) have been dentfed as the most promsng potental soluton. Snce femtocells wll act as relays between the end user and the macrocells (regular base staton) wth multple femtocells expected to cooperate, the 4G network can greatly beneft from the ongong research on the new communcaton paradgm of cooperatve relay system. he deployment of a large number of femtocells can then be used to mprove coverage, capacty (area and system spectral effcency), and energy effcency of the wreless network access system. he cooperatve relay communcaton system takes advantage of the broadcast nature of the wreless channel to mprove the communcaton between the source node and a destnaton node wth the ad of one or more relay nodes. he system harnesses the new form of spatal dversty and combat multpath fadng, thereby mprovng the spectral effcency, and reducng the transmsson error, system outages, and transmsson power. he reduced transmsson power (of wreless devce, femtocell and macrocell) effectvely lowers the nter and ntra macrocell nterference, thereby mprovng both the system and the area spectral effcency. In another development, the cooperatve dversty system concept has ganed research mpetus owng to ts nherent ablty to overcome the practcal mplementaton ssue of packng a large number of antenna elements (to explot the benefts of multple-nput-multple-output (MIMO) space-tme processng technques) n small form-factor devces. In general, there are three cooperatve relayng protocols: amplfy-and-forward, decode-and-forward, and compress-and-forward [3-7]. he other varatons nclude ncremental, opportunstc, blnd and sem-blnd relays. In ths artcle, we advocate the mplementaton of the amplfy-and-forward protocol on femtocell. he advantages of ths nclude ts smplcty, lower mplementaton cost (.e., relay nodes (femtocells) do not have to decode and then re-encode the nformaton receved pror transmsson) and possbly the better prvacy. he man drawback of the regular cooperatve amplfy-and-forward (CAF) dversty system whch employs the maxmum rato combnng (MRC) or selecton dversty combnng (SDC) at the destnaton's recever s that each relay has to transmt on the orthogonal channels (DMA/CDMA). herefore, the spectral effcency s scaled by /(+), where s the number of relays, whch reduces the capacty wth ncreasng number of relays. In order to mprove ths, Bletsas et al. and Zhao et al. [8-], proposed relay selecton method otherwse known as the opportunstc relayng system (ORS). Here, the best relay s selected pror to relay-to-destnaton transmsson to lmt the number of orthogonal transmssons to two. he destnaton would then employ ether MRC (subsequently referred to as ORS-MRC) or SDC (subsequently referred to as ORS-SDC) dversty scheme on the two fnal dversty paths. In addton to reducng the number of ndependent transmssons, the ORS protocols have been shown to acheve full dversty just lke regular relay system [7]. Adaptve transmsson s yet another powerful wreless communcaton strategy for mprovng the spectral utlzaton effcency, wheren the sgnal constellaton sze, power level and/or the codng rate are matched to the prevalng channel condtons based on the acqured channel-sdenformaton (CSI) on the feedback channel. Several artcles have nvestgated the combnaton of the lnk adaptaton and the regular cooperatve dversty system, both from theoretcal lmt (ergodc capacty) and practcal mplementaton (usng dgtal modulaton schemes) perspectves.

3 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 For nstance, the ergodc capacty of cooperatve amplfy-and-forward (CAF) relay networks wth the lmted CSI were derved n [-9] for dfferent source-adaptve transmsson polces n a myrad of stochastc fadng envronments. he performance of practcal adaptve dgtal modulaton scheme wth regular relay system was consdered n [0-3]. he performance of CAF relay wth constant power M-QAM adaptve rate transmsson, when average bt error rate (ABER) n Raylegh fadng s constraned to be below a specfed target bt error rate (BER) s examned n [] and [] for fxed and optmum mode swtchng thresholds, respectvely. In [3], the performance of dscrete-rate adaptve M-QAM for a sngle ncremental relay n akagam-m envronment was examned, whle [] nvestgates the performance of a cooperatve decode-and-forward relay network wth fve-modes adaptve M-QAM transmsson n an..d akagam-m wreless fadng envronment. All these artcles and related references ndcate the advantage of the adaptve cooperatve dversty system over the non-adaptve and/or the non-cooperatve system especally at low and medum sgnal-to-nose rato (SR). However, the half-duplex nature of regular relay system makes the performance worse at hgh SRs. Wth the ntroducton of ORS n [8-], several artcles have been publshed on ts performance, but mostly focusng on non-adaptve system (.e. fxed modulaton). Average symbol or bt error rate (ASER/ABER) performance for ORS-MRC scheme over Raylegh, ndependent and dentcally dstrbuted (..d) akagam-m, and ndependent and non-dentcally dstrbuted (.n.d) akagam-m fadng channels was nvestgated n [6-7], [8], and [9] respectvely, whle [30] nvestgated ASER of both ORS-MRC and ORS-SDC scheme over..d akagam-m fadng channel. Outage capacty for ORS-MRC and ORS-SDC has been consdered n [3]. It s mportant to note that untl now; only a few artcles have consdered the lnk adaptve ORS system. For nstance, ergodc capacty wth the source adaptaton technques have been consdered n [3-35] for the ORS-MRC scheme over Raylegh fadng channel, whle the varable rate constant power adaptve M-QAM modulaton wth ORS-MRC and ORS-SDC schemes over Raylegh fadng have been consdered n [35-36] and [37] respectvely. Also, [38] analyzed the performance of lnk adaptve ncremental opportunstc relayng over.n.d Raylegh fadng channels. However [39] studed the performance analyss of cooperatve communcaton wth only ORS-SDC scheme. Careful study of the adaptve ORS schemes [3-39] ndcates that, the analyses of both the ergodc capacty and the practcally achevable spectral effcency have been obtaned usng the probablty densty functon method, whch can be very tedous and may not always yeld compact solutons. In contrast, n ths artcle we develop a new analytcal framework based on the margnal MGF method for evaluatng the ASER, mean spectral utlzaton effcency and outage probablty performance metrcs (.e., snce the MGF of total receved SR may be easer to compute or readly avalable for CAF relay networks).he developed analytcal framework s thenused to analyze and compare the performance of regular MRC, SDC, opportunstc MRC and SDC CAF relay schemes that employ constant power adaptve dscrete rate M-QAM/M-PSK transmsson. he proposed analytcal framework s general, and can be appled to any arbtrary fade dstrbuton as long as the MGF of the end-to-end SR s avalable unlke channel specfc dervatons n [3-38]. For completeness purpose, the ergodc capacty wth optmal rate adaptaton was also presented. umercal analyss ndcates that ORS-MRC leads the pack consderng the ergodc capacty and achevable spectral effcency, whle the regular MRC s the best n terms of the outage probablty and ASER. In the overall, the ORS-MRC s the best due to the addtonal power savng as t consumes only /(+), of the total power consumed by regular MRC. herefore, not only does t perform well as a communcaton paradgm, t also supports green technology due to ts low transmt power requrement. Careful lterature search ndcates that, ths s perhaps the frst tme such comprehensve analyss of regular and opportunstc CAF relay network wth the lnk adaptaton s beng reported. 3

4 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 he remander of ths artcle s organzed as follows. In Secton, the system model s dscussed. Secton 3 derves the performance metrcs for the CAF relay networks wth adaptve M-QAM/M- PSK modulaton. Selected numercal results are presented n Secton 4. Our conclusons are gven n Secton 5.. SYSEM MODEL Fgure. Lnk-adaptve cooperatve relay networks. OAL MGF OF CAF RELAY EWORKS In ths secton, we wll present the moment generatng functon (MGF) of the relayed path of dfferent cooperatve dversty and opportunstc routng protocols that wll be utlzed n evaluatng the end-to-end ASER, mean spectral effcency and the outage probablty performance metrcs of the proposed network over a myrad of fadng channels wth the adaptve M-QAM/M-PSK modulaton schemes... REGULAR COOPERAIO: CAF RELAYIG WIH MAXIMUM RAIO COMBIIG (MRC) A HE DESIAIO In ths protocol, as shown n Fg, source node S whch communcates wth a destnaton node D va a drect-lnk and through amplfy-and-forward relays, R, {,,..., }, n two transmsson phases. Durng the ntal Phase I, S transmts sgnal x to D as well as to the relays R, where the channel fadng coeffcents between S and D, S and the -th relay node R, and R and D are denoted by α s, d, αs, and α, d, respectvely. Durng the second phase of cooperaton, each of the relay nodes transmts the receved sgnal after amplfcaton va orthogonal transmssons (e.g., DMA n a round-robn fashon and/or FDMA). Hence, the channel usage per source transmsson, U ( ) = +. ow, consder that the maxmum rato combner (MRC) s employed at a D to coherently combne all the sgnals receved durng Phase I and Phase II, the total receved SR at output of the MRC detector can be shown to be (e.g., [9,0, 5]) M R C s,, d ( H M ) = s, d + =,,,, s d + s d + = s + d + = = () 4

5 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 ( HM ) where = + s the harmonc mean SR, s,, d s,, d, nstantaneous SRs of lnk a-b, nose varance. a, b α a, b Es o = corresponds to the Es denotes the average symbol energy and corresponds to the 0 he approxmaton of () s obtaned by recognzng that the nstantaneous SR of a two hops path can be accurately estmated to be the harmonc mean of the ndvdual lnk SRs, especally at moderate/hgh SR regmes [43]. Hence, t s straght-forward to show that the MGF of end-toend SRs gven by φ ( ) ( ) ( ) MRC s = φ s φ s () s, d = where φ ( s) and φ ( s) s, d are the MGFs of the SR for the S-D lnk and the relayed paths, respectvely. he MGF of SR for sngle channel recepton s readly avalable n the lterature. It has been shown n the lterature that the evaluaton of the MGF, PDF and CDF of s a dauntng task wth exstng results lmted to Raylegh fadng [4] and akagam-m [4] fadng channels wth nteger m and even n such cases the expressons are too complcated, and mostly useless for ( HM ) the system level analyss. However, t has been shown n [43] that, n () can effectvely approxmate especally at medum and hgh SRs. Also, n ths case the MGF expressons are stll dffcult to obtan wth the exstng results lmted to Ralegh fadng [44] and..d. akagamm [43] channels. Due to ths lmtaton, the bounds have been developed for and t s gven by = mn(, ). ( UB ) s,, d For nstance, the closed-form formula for the MGF of fadng statstcs s gven by [5] ( UB) n a akagam channel wth.n.d ( UB) φ mk Γ ( mk + mj ) jmk ( s k + mk ) j j k k k {( s, ),(, d)} mk Γ( mk ) Γ( mj) sj k + jmk +k mj sj k + jmk +k mj j k ( s) = F m, m ; + m ; (3) where = E[ ] corresponds to the mean SR of lnk q, Γ(.) s the Gamma functon and m q q q s the akagam-m fadng ndex... REGULAR COOPERAIO: CAF RELAYIG WIH SELECIO DIVERSIY COMBIIG (SDC) A HE DESIAIO In ths knd of protocol mplementaton, the best route s selected at the destnaton node based on the end-to-end relay SR. Here n the case of SDC, the channel usage per source transmsson s smlar to the MRC case (.e., U( ) = + ). For ths protocol, the effectve SR at the output of the SDC detector (.e., all the sgnals receved durng Phase I and Phase II) s gven by ( ) = max(,,,... ) max,mn( ),mn( ),...mn( ) (4) SDC s, d s, d s,, d s,, d s,, d SDC For nstance, the MGF of for a specal case of ndependent and dentcally dstrbuted (..d) akagam-m fadng statstcs can be obtaned from Appendx A as [45] 5

6 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 φ SDC j p m m ( ) s ( ) m m m ( ) m p s Γ λ + m+ m,,...! m λ λ+ + ( ) ( ) m ( ) m p= p p j= j Γ m+ s + + p s + + p ( s) ( + ) + ( ) F, + m+ ; m + ; (5) where F (.,.;.;.) s the Gauss hypergeometrc functon... CAF RELAYIG WIH OPPORUISIC ROUE SELECIO AD SDC (ORS-SDC) A HE SOURCE he (ORS-SDC) relayng protocol focuses on lmtng the number of cooperatng relay to one. Addtonally, the choce of the approprate route s selected by the source (assumng that the relay wth the best transmsson parameter s already determned durng the route dscovery process) or can be accomplshed n a dstrbuted fashon among the relays as proposed n [8]. Compared to the regular cooperatve dversty protocol dscussed above, the best route s selected at the source based on the end-to-end relay SR. herefore, the statstcs of the best route selecton s the same as the selecton dversty combnng at the destnaton as the best among + lnks s beng selected. However, the channel usage per source transmsson n the case s U ( ) = (.e., source wll broadcast the sgnal to all the relays and destnaton n frst tme slot; and n the second tme slot, one of the best relay amplfy and forward the sgnal to the destnaton). herefore, the spectral effcency here does not reduce wth the ncreasng number of relays as n the case of regular cooperatve dversty protocols dscussed above (.e., MRC and SDC only). Also, the amount of channel sde nformaton and the mplementaton complexty s hghly reduced... CAF RELAYIG WIH OPPORUISIC RELAY SELECIO AD MRC (ORS-MRC) A HE DESIAIO hs protocol mplementaton takes advantage of the half duplex nature of the relay transmsson to acheve better performance than the ORS-SDC protocol. Here, snce the source transmts n the frst transmsson phase, and due to the broadcast nature of the wreless channel, the destnaton can be close enough to receve ths sgnal before recevng the sgnal from the relay. hs s partcularly true n the dstrbuted ORS protocol mplementaton proposed n [8]. herefore, f the channel sde nformaton of both the lnks s avalable, the receved sgnal can be combned wth the MRC scheme at the destnaton. ote that the transmsson channel usage U ( ) = each source transmsson, but the statstcs s slghtly dfferent from the pure ORS-SDC protocol. he effectve end-to-end SR of the ORS-MRC protocol can be expressed as ( s s, ) = + max(,,... ) + max mn( ),mn( ),...mn( ) (6) ORS MRC s, d s, d,, d,, d s, d ORS MRC he MGF of for a specal case of ndependent and dentcally dstrbuted (..d) akagam-m fadng statstcs can be obtaned from Appendx B as [45] s m ( s) ( + ) + ( ) m φ ORS MRC m p m j p ( ) sγ ( λ + ) ( ),,...! m λ + p= p ( p j= j s + p ) (7) Hence, by utlzng the above mentoned closed-form MGFs of the four protocol schemes, we can easly analyse and compare the performance of CAF relay networks n terms of mean achevable spectral effcency, outage probablty and average symbol error rate. 6

7 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06.CUMULAIVE DISRIBUIO FUCIO (CDF) OF HE OAL ED-O-ED SR o evaluate the varous performance measures, the knowledge of the CDF of total effectve SR of the CAF relay networks s requred. Snce the analytcal CDF expresson for the CAF relay networks s dffcult to obtan, the alternatve s to compute the CDF from the MGF expresson n (). One of the most effcent frequency nverson method s the Abate s Fxed-albot method (.e., mult-precson Laplace transform nverson) [6], vz., r + jσ ( θ ) F ( x) ( r) e Re{ e s( ) } s( θ ) Z rx k xs ( θk ) φ + φ ( θ ) (8) X X X k k = Z Z k where r = Z / (5 x), θk = kπ / Z, s( θk ) = rθ k ( j + cot( θk )), σ ( θ k ) = θk + ( θk cot( θ k ) )cot( θk ), and postve nteger Z can be chosen to get the desred accuracy. Utlzng (8), we can easly compute the CDF expressons from ts respectve total MGF expresson for the CAF relay networks. 3. ADAPIVE MODULAIO In ths secton, we develop a new analytcal framework based on the margnal MGF for evaluatng the ASER, mean spectral utlzaton effcency and outage probablty performance metrcs.it wll shown that the mean achevable spectral effcency of ADR M-QAM/M-PSK and ASER of CAF relay networks wth adaptve source transmsson can be expressed n terms of dfference of two CDF terms.he ADR M-QAM/M-PSK schemes are frst explaned, followed by the outage probablty, the mean spectral effcency, and the average SER analyss. 3. ADR M-QAM/M-PSK SCHEMES In the adaptve modulaton technques, the destnaton node needs only to compute the total receved SR, select the approprate transmsson rate, and feedback ths nformaton to the transmtter. In context of CAF relay system, the destnaton node only needs to compute and convey the nformaton on the total (effectve) receved SR to the source node for t to select an approprate transmsson rate. hs results nto a hgher mean achevable spectral effcency wthout havng to sacrfce the error rate performance. For ths reason and more specfcally due to several other practcal advantages of the adaptve rate modulaton, we consder both the adaptve M-QAM and the M-PSK dgtal modulaton schemes n ths paper to mprove the performance of the CAF relay networks. In order to smplfy the analyss of the adaptve modulaton, there s a need to express nstantaneous error rate n desrable exponental form, smlar to the one n the exstng lteratures [-5]. Here, we employ the exponental-type representatons of the nstantaneous (SER) for the M-PSK and the M-QAM schemes and are respectvely gven by [40, able II] S sn sn b s π M b s π M P a e + c e (9) ( ( ) ) ( ) 3 b s 3b 6 9 s b s b s M M M M P ka e + kc ka e kc e k a c e (0) S 7

8 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 k = M M M s the constellaton sze and parameters a = 0.938; b =.0483; where ( ), c = are obtaned from [40, able I]. he above exponental forms partcularly facltate the averagng of the SER over the fade dstrbuton. he resultng average symbol error rate (ASER) expresson (takng advantage of Laplace transform property), whch can be evaluated as the weghted sum of the MGF of end-to-end SR of fadng channel, can be expressed as P M PSK a b M c b M = φ sn ( π ) φ sn ( π + ) () 3b 3b 6b 9b PM QAM = kaφ ( kc ( ka ) ) φ ( kc ) φ k ac φ ( M ) + M M ( M ) () where φ (.) s the MGF of SR for sngle channel. In the ADR M-QAM/M-PSK system, the range of the effectve receved SR s dvded nto + fadng regons. When the fadng causes the total receved SR to fall nto the n-th regon (n = 0, n,), the constellaton sze M n = s employed for the transmsson. Also, the SR thresholds for parttonng of the total receved SR depends on the target SER level, P s. he regon boundary n s chosen for the correspondng transmsson mode n to be the mnmum SR requred to acheve P s, whch can be easly obtaned by nvertng (9) and (0) for M-PSK and M- QAM modulaton schemes respectvely a + a + 4c P S n ln b sn ( π M ) c n (3) Mn ( Mn )( S ) a a 4c P ( M ) + + n n ln 3b c where, n =,,3..., and + = +. (4) It s worth to menton here that, the two exponental terms are more accurate than the exstng nvertble expressons n the lterature [46-48]. Representatve example has been shown n Fg., where a comparson has been made between the proposed approxmaton and the sngle and two exponental term approxmaton n [47] and [48] respectvely. hs fgure hghlghts that our proposed exponental approxmaton (-), yelds a very good estmate of the actual ASER performance over a wde range of average lnk SRs, dfferent fadng severty ndces and for dfferent constellaton szes. It s evdent from the fgure that, the proposed approxmaton performs better than [47] and [48] for the M-QAM and better than [48] for the M-PSK. herefore, for the rest of the analyss n ths artcle, the proposed two exponental term approxmaton wll be utlzed. 8

9 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May PSK, m = 0-6QAM, m = 3 Avearage Symbol Error Rate PSK, m = 3 4QAM, m = 0-5 [47, Eq.(7)] Exact Proposed Method [48, Eq.(0)] Mean SR Per Symbol (db) Fgure ASER of M-PSK and M-QAM over akagam fadng channels (m = and 3) wthout dversty 3. OUAGE PROBABILIY When the total receved SR falls below the regon boundary threshold ( can be obtaned by settng, n = n (3) or (4)), we cease the transmsson, because the prescrbed target SER cannot be satsfed even wth the smallest constellaton sze. he probablty of such an outage event s gven by P = F ( ), where the CDF term can be evaluated effcently usng (8). out 3.3 MEA SPECRAL EFFICIECY he normalzed average achevable spectral effcency for ADR M-QAM/PSK s gven by the weghted sum of the data rates n each of the parttoned regons [, 5] vz., Radr = npn B U ( ) (5) n = where p n denotes the transmsson mode selecton probablty (.e., probablty that the total receved SR falls n the n-th partton regon): n+ p = ( ) ( ) ( ) n f d = F F (6) n+ n n Hence, usng the approprate MGF expressons derved n [43-44, 49] n (8), we can readly compute the mean spectral effcency of the ADR M-QAM/PSK n a myrad of wreless fadng envronments. 9

10 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May ASER OF ADAPIVE M-QAM/PSK he ASER of ADR M-QAM/PSK can be calculated as the rato of the average number of error bts per transmsson dvded by the average number of bts per transmsson [47], vz., ASER adr = n = n SERn n =, n F ( ) F ( ) n+ n (7) where SER s the average SER n a specfc SR regon of n n [ +, n] and can be represented as n+ SER = P f ( ) d (8) n n M where P M s the CEP of the modulaton scheme n AWG channel. For the specal case of M- PSK scheme, we can derve the SER by substtutng (9) n to (8) vz., { n+ ( ( )) ( ) n ( )} SER = a exp b sn π M + c exp b sn π M f ( ) d n n φ b sn ( π ) φ b sn ( π ) n φ b sn ( π ) φ b sn ( π ) n = a ( M, ) ( M, ) c ( M, ) ( M, ) n+ + n+ β and the term φ ( β, α ) = e f ( ) d n (9) denotes the margnal MGF of the total receved SR. α ote that although ths quantty s avalable n closed-form for non-cooperatve system (e.g. [48]), smlar expressons do not appear to be readly avalable or generalzed for the CAF relay networks, partcularly n a generalzed wreless fadng envronment. Utlzng [5, Appendx C], we can compute the desred margnal MGF as a dfference between two CDF terms of an auxlary functon, vz., (9) SER = n a F ( ˆ n ) F ( ) ˆ n c F ( ˆ n ) F ( ) a + a ˆ n b + + b (0) where F ( ) ˆ x and F ( x) n (0) can be evaluated effcently va (8), but usng the MGF formulas a ˆ b of the auxlary functons (.e., ( ) ( sn φ s = φ s b ( M) ) a + π and φ ( ) ( sn ( )) s = φ s b M b + π ). Smlarly treatment can also be appled to the M-QAM modulaton scheme by substtutng (0) nto (8). Eq. (0) allows us to generalze the evaluaton of ASER over arbtrary multpath/shadowng adr fadng as long as the MGF of SR of fadng channel s avalable. hs s n sharp contrast wth channel specfc PDF methods n [3-34] whch lmted ther analyss to Raylegh fadng channel. 4. UMERICAL RESULS In ths secton, selected numercal results are provded for the normalzed mean achevable spectral effcency, outage probablty and ASER performance metrcs of CAF relay networks wth both the adaptve dscrete rate M-QAM and M-PSK dgtal modulaton schemes. In partcular, we are comparng the performance of four dstnct cooperatve dversty and 0

11 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 opportunstc routng protocols: () ORS-MRC () ORS-SDC () MRC and (v) SDC, over the..d akagam fadng envronment (ncludng specal case of Raylegh fadng). Moreover, the ergodc capacty analyss wth the optmal rate adaptaton polcy s also presented for the theoretcal performance lmt of the above protocol schemes. o generate plots, the mean lnk SRs are chosen arbtrarly as: = = s, s, s,3 = =,d =,d 3,d = s, d = E s / 0 and the fadng ndexes on the each lnks s chosen as m = 3 (unless states otherwse). For the ADR system, the target SER of 0-3 s arbtrarly chosen. Fg. 3 llustrates the ergodc capactes of the ORA polcy usng four dfferent protocols (.e., ORS-MRC, ORS-SDC, SDC and MRC). o generate the plots, we have used the followng generalzed expressons n terms of the MGF of end-to-end SR of CAF relay networks [] C y O RA = B U ( ) ln 0 y e φ ( y ) dy () Expresson () ndcates that the ORA capacty evaluaton requres only the knowledge of the MGF of SR of the fadng channel. By substtutng the total MGF of the above mentoned protocols nto (), we can easly generate the curves as shown n the fgure. From the fgure, we can observe that the performance of the opportunstc relay scheme (.e., ORS-MRC and ORS- SDC) s better than the regular cooperaton (.e., MRC and SDC) respectvely. hs mprovement n the performance of the ORS scheme s due to the utlzaton of the two orthogonal slots for the total transmssons compared to the three slots n the regular cooperaton. Moreover, t s nterestng to note that, the authors n [8], compares the ergodc capactes of the CAF relay network usng best relay selecton and the regular MRC scheme. However, ther framework does not lend tself to the analyss of the ORS-SDC or the SDC case, whereas our framework encapsulates the performance of all the four protocols. ormalzed Channel Capacty C/B (bts/sec/hz) ORS-MRC ORS-SDC SDC MRC E s / 0 (db) Fgure 3 Ergodc channel capactes of the optmal rate adaptaton (ORA) polcy wth two cooperatve relays ( =)

12 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 Fg. 4 shows the comparson n terms of spectral effcency of the dfferent cooperatve dversty and the opportunstc routng protocols, wth the adaptve M-PSK modulaton. It s worth to menton that for a sngle relay case, the ORS-MRC gves the same performance as the MRC scheme, whle the ORS-SDC scheme also gves the same performance as the SDC scheme. However, as the number of relay ncreases, ORS-MRC performs better than the regular MRC protocol at medum and hgh SRs, whle the ORS-SDC protocol performs better than the regular SDC protocol throughout the SR range. hs s because rrespectve of the number of relays n the partcpaton, the total channel usage for the ORS-MRC and the ORS-SDC s kept constant at two tme slots per source transmsson; whereas the channel usage for MRC and SDC schemes ncreases wth ncreasng number of relays. Moreover, to further mprove the spectral effcency, we ncorporate the adaptve M-PSK modulaton scheme (compared to fxed modulaton schemes n the prevous lteratures) to adapt the transmsson rate wth the varyng channel condtons. It s evdent from Fg. 4 that by ncreasng the maxmum constellaton sze (transmsson modes) n the ADR M-PSK drectly translates nto mproved spectral effcency. However, ths mprovement s acheved at the expense of the ncreased ASER (see Fg. 5). In summary, the lnk adaptve ORS-MRC scheme gves the best overall performance. Hence, t can be concluded that the ORS-MRC protocols are recommended for the cooperatng nodes at the tactcal edge or at the cell boundary, where the receved sgnal strength s weak. ormalzed Average Spectral Effcency (bts/sec/hz) ORS-MRC ORS-SDC MRC SDC = 3 = = 4 = 3 = 4 = SR E s / o (db) Fgure 4 Comparson of dfferent cooperatve dversty and opportunstc routng protocols wth adaptve M-PSK modulaton ( = 3 and 4) Fg. 5 llustrates the average symbol error rate (ASER) of a CAF relay network wth the adaptve M-PSK modulaton (usng = 3 and 4). We observe that the ASER of the MRC scheme s the lowest, whereas, the ORS-SDC scheme s the hghest. hs s due to the avalablty of the total + dversty paths n the MRC scheme. However, ths s acheved at the expense of the power effcency as the ORS-SDC and the ORS-MRC requres only /(+) and /(+) of the total power of the regular MRC scheme respectvely.

13 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May = 4 Average Symbol Error Rate = = 3 ORS MRC ORS SDC MRC = SR E s / o (db) Fgure 5 Average symbol error rate of a CAF system wth dfferent cooperatve dversty and opportunstc routng protocols usng adaptve M-PSK modulaton ( =3 and 4) ormalzed Average Spectral Effcency (bts/sec/hz) m = 4 m = ORS-MRC ORS-SDC MRC SDC SR E s / o (db) Fgure 6 Comparson of dfferent cooperatve dversty and opportunstc routng protocols wth adaptve M-MQAM modulaton ( = 5) consstng of two relays Fg. 6 shows the spectral effcency performance comparson of the four cooperatve dversty and the opportunstc routng protocols wth the ADR M-QAM modulaton scheme wth = 5. hs fgure hghlghts the nfluence of the channel fadng severty on the performance of the lnk adaptve cooperatve system. Whle the performance trend among the protocols s smlar to the one obtaned n Fg. 4, Fg. 6 n partcularly shows that, as the channel condton mproves, the 3

14 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 achevable spectra effcency mproves. o the best of our knowledge, ths effect has never been reported n the earler lterature, and t also demonstrates the versatlty of our mathematcal framework MRC ORS-MRC ORS-SDC Outage Probablty = = SR E s / o (db) Fgure 7 Probablty of outage of a CAF system wth dfferent cooperatve dversty and opportunstc routng protocols usng adaptve M-PSK modulaton ( = 3) (ote that for =, the ORS-MRC and MRC, whereas for any values of, ORS-SDC and SDC schemes are the same) Fgure 7 depcts the outage probablty as a functon of SR at the target SER of 0-3 and t hghlghts the beneft of the cooperatve dversty to maxmze the performance of the wreless communcaton system. From fgure 7, we conclude the followng mportant observatons. Frst, we notce that the case wth cooperatve dversty (.e., = 3) evdently outperforms the case wth =. Second, the outage probablty wth the MRC protocol has a better performance than all the other protocols (smlar to the case n fg. 5 for the ASER analyss wth the MRC protocol). hs performance gan s due to the addtonal dversty path offered by all the relays and drect path n the system, but stll at the expense of the power effcency. 5. COCLUSIOS hs paper analyzes the performance of cooperatve amplfy-and-forward (CAF) relay networks that employ the adaptve M-ary quadrature ampltude modulaton (M-QAM)/M-ary phase shft keyng (M-PSK) dgtal modulaton technques n the akagam-m fadng channel model. In partcular, we present and compared the analyss of the CAF relay networks wth dfferent cooperatve dversty and opportunstc routng protocols such as Maxmal Rato Combnng (MRC), Selecton Dversty Combnng (SDC), Opportunstc Relay Selecton wth Maxmal Rato Combnng (ORS-MRC) and Opportunstc Relay Selecton wth Selecton Dversty Combnng (ORS-SDC).We advocate a smple yet unfed numercal approach based on the margnal moment generatng functon (MGF) of the total receved SR to compute the average symbol error rate (ASER), mean achevable spectral effcency, and the outage probablty performance metrcs. hese analytcal frameworks and results wll facltate the choce of cooperaton protocol and confguratons that can be employed n the desgn and deployment of femtocells. 4

15 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 APPEDIX A hs secton provdes the dervaton for the MGF of end-to-end SR of CAF relay system wth SDC protocol at the destnaton. hs s also applcable to the ORS scheme wth SDC at the destnaton. he CDF of the end-to-end SR gven n (4) can be expressed as ( ) F ( ) F ( ) F ( ) F ( ) [ F ( )][ F ( )] SDC = s, d r= r s, d (A.) r= s, r r, d where F ( ), F ( ) s, d and F ( ) s, r are the CDFs of the source-to-destnaton, source-to-relay and r, d relay-to-destnaton lnks, respectvely. he effectve MGF can then be evaluate usng the dfferentaton property of the Laplace transform va a sngle ntegral expresson s φ ( s) = s e F ( ) F ( ) d SDC 0 0 s, d r= r ( ) s s e F ( ) [ F ( )][ F ( )] d s, d r= s, r r, d (A.) For specal case of ndependent and dentcally dstrbuted (..d.) akagam-m channel, the MGF can be reduced to [30] m p m λ ( ) j m (, ) ( m s G m p p ) φ SDC ( s) s e ( ) e d + 0 Γ( m) p= p,,...! p j= j (A.3) where G(.,.) s the lower ncomplete gamma functon and = p λ j= j Usng the dentty [5, Eq. (6.455.)], after few algebrac manpulatons, the closed-form MGF expresson can be obtaned as shown n (5). APPEDIX B hs secton provdes the dervaton for the MGF of end-to-end SR of ORS CAF relay system wth MRC protocol at the destnaton. he effectve MGF of (6) can be evaluated usng the addton and dfferentaton propertes of the Laplace transform va a sngle ntegral expresson gven by s φ ( s) = sφ ( s) e F ( ) d ORS MRC s, d 0 r= r ( ) s sφ ( s) e [ F ( )][ F ( )] d s, d 0 r= s, r r, d (B.) 5

16 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 For specal case of ndependent and dentcally dstrbuted (..d.) akagam-m channel, the MGF can be reduced to p m j m p ( ) m ( m s λ s p ) φ ORS MRC ( s) ( + m ) + ( ) s e e d 0 p= p,,...! p j= j (B.) Usng the dentty [5, able 7.3], after few algebrac manpulatons, the closed-form MGF expresson can be obtaned as expressed n (7). It s worth to menton that the expresson n (7) s much more compact and smpler than the equvalent expresson n [30, Eq. (7)]. ACKOWLEDGME hs work s supported n part by fundng from the atonal Scence Foundaton SF/HRD and the US Ar Force Research Laboratory (Contract #FA ). REFERECES [] Csco Vsual etworkng Index: Forecast and Methodology, c-48360_ns87_etworkng_solutons_whte_paper.html. Accessed February 0, 0 [] IU global standard for nternatonal moble telecommuncatons IM-Advanced. advanced & lang=en. Accessed February 0, 0 [3]. Laneman, D. se, and G. Wornell, (004) Cooperatve dversty n wreless networks: effcent protocols and outage behavor, IEEE rans. Info. heory, vol. 50, no., pp [4]. Lanemanand G. Wornell, (003) Dstrbuted space-tme coded protocols for explotng cooperatve dversty n wreless networks. IEEE rans. Info. heory, vol. 49, no.0, pp [5] A. Sendonars, E. Erkp and B. Aazhang, (003) User cooperaton dversty, part I: system descrpton IEEE rans. Communcatons, vol. 5, no., pp [6] A. Sendonars, E. Erkp and B. Aazhang,(003) User cooperaton dversty, part II: mplementaton aspects and performance analyss,ieee rans. Communcatons, vol. 5, no., pp [7] M. Khojastepour, A. Sabharwal and B. Aazhang, (004) Lower bounds on the capacty of Gaussan relay channel,conf. Informaton Scences and Systems (CISS), Prnceton, J, pp [8] A. Bletsas, A. Khst,D. P. Reedand A. Lppman, (006) A smple cooperatve dversty method based on network path selecton, IEEE Journal on Selected Areas on Communcatons, vol. 4, no.3, pp [9] Y. Zhao, R. Adve, and. J. Lm, (006) Improvng amplfy-and-forward relay networks: optmal power allocaton versus selecton,ieee Internatonal Symposum on Informaton heory, pp [0] Y. Zhao, R. Adve and. J. Lm, (006) Symbol error rate of selecton amplfy-and-forward relay systems,ieee Communcatons Letters, vol. 0, no., pp [] A. Bletsas, H. Shn, and M. Z. Wn, (007) Cooperatve communcatons wth outage-optmal opportunstc relayng, IEEE ransactons on Wreless Communcatons, vol. 6, no. 9, pp. -. [] B. Mod, A. Annamala, O. Olaby and R. Palat, (03) Ergodc capacty analyss of cooperatve amplfy-and-forward relay networks over generalzed fadng channels,wley Journal of Wreless and Moble Computng, vol. 5, no. 8, pp [3]. echporenko, K. Phan, C. ellambura and H. guyen, (009) Capacty of Raylegh fadng cooperatve systems under adaptve transmsson, IEEE rans. Wreless Comm., vol. 8, no.4, pp

17 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 [4] A. Annamala, R. Palat and J. Matyjas, (00) Estmatng ergodc capacty of cooperatve analog relayng under dfferent adaptve source transmsson technques, IEEE Sarnoff Symposum, pp. -4. [5] A. Annamala, B. Mod, R. Palat and J. Matyjas, (00) ght-bounds on the ergodc capacty of cooperatve analog relayng wth adaptve source transmsson technques,ieee Internatonal Symposum on Personal, Indoor, and Moble Rado Comm., pp [6] R. Palat, A. Annamala and J. Reed, (008) An effcent method for evaluatng nformaton outage probablty and ergodc capacty of OSBC systems, IEEE Comm. Letters, vol., no.3, pp [7] M. D. Renzo, F. Grazos, and F. Santucc, (00) Channel capacty over generalzed fadng channels: A novel MGF-based approach for performance analyss and desgn of wreless communcaton systems, IEEE rans. Vehcular echnology, vol. 59, no., pp [8] B. Mod, O. Olaby, A. Annamala and D. Vaman, (0) On ergodc capacty of cooperatve nonregeneratve relay networks n Rce fadng envronments, IEEE GLOBECOMWorkshop, pp [9] B. Mod, O. Olaby, A. Annamala and D. Vaman, (0) Improvng the spectral effcency of adaptve modulaton n amplfy-and-forward cooperatve relay networks wth an adaptve ARQ protocol, IEEE Global elecommuncatons Conference, pp. -5 [0] M. Hasna, (005) On the capacty of cooperatve dversty systems wth adaptve modulaton, Internatonal Conference on Wreless and Optcal Communcaton etworks, pp []. echporenko, K. Phan, C. ellambura and H. guyen, (008) Performance analyss of adaptve M-QAM for Raylegh fadng cooperatve systems, IEEE Internatonal Conference on Communcatons, pp []. echporenko, P. Kalansurya and C. ellambura, (009) Performance of optmum swtchng adaptve M-QAM for amplfy-and-forward relays, IEEE rans. Vehc. ech., vol. 58, no. 5, pp [3] K. Hwang, Y. Ko and M. Aloun, (008) Performance analyss of opportunstc ncremental relayng wth adaptve modulaton over cooperatve networks, IEEE Internatonal Symp. Wreless Pervasve Comput.,pp [4] P. Kalansurya and C. ellambura, (009) Performance analyss of decode-and-forward relay network under adaptve M-QAM, IEEE Internatonal Conference on Communcatons, pp [5] B. Mod and A. Annamala, (0) Improvng the spectral effcency of amplfy-and-forward cooperatve relay network wth adaptve M-QAM modulaton, 0th IEEE Internatonal Conference on Computer Communcatons and etworks, pp. -6. [6] B. Barua,. Quoc and H. Shn, (008) On the SEP of cooperatve dversty wth opportunstc relayng, IEEE communcaton letters, vol., no. 0, pp [7] S. Ikk, and M. H. Ahmed, (008) Performance of multple-relay cooperatve dversty system wth best relay selecton over Raylegh fadng channels,eurasip Journal on Advances on Sgnal Processng, artcle ID [8] S. I. Hussan, M.S. Aloun and M. O. Hasna, (00) Performance analyss of best relay selecton scheme for amplfy-and-forward cooperatve networks n dentcal akagam-m channels,ieee Sgnal Processng Advances n Wreless Communcatons Conference, pp. -5. [9] S. I. Hussan, M.O Hasna and M.S. Aloun, (00) Performance analyss of best relay selecton scheme for fxed gan cooperatve networks n non-dentcal akagam-m channels, IEEESgnal Processng Advances n Wreless Communcatons Conference, pp [30] O. Waqar, D. C. McLernon and M. Ghogho (009) Performance analyss of non-regeneratve opportunstc relayng n nakagam-m fadng,ieee Internatonal Symposum on Personal, Indoor, and Moble Rado Comm., pp [3] S. Valentn, S. I. Hermann, H. Karl, L. Loyola and J. Wdmer, (008) Opportunstc relayng vs. selectve cooperaton: analyzng the occurance condtoned outage capacty,th Internatonal Symposum on Modelng Analyss and Smulaton of Wreless and Moble Systems, pp [3] S. Ikk, and M. H. Ahmed, (00) Performance analyss of adaptve decode-and-forward cooperatve dversty networks wth best relay selecton IEEE ransactons on Communcatons, vol. 58, no., pp [33] S.Ikk, and M. H. Ahmed, (00) On the capacty of relay selecton cooperatve dversty networks under adaptve transmsson,ieee Vehcular echnology Conf., pp. 5. 7

18 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 [34] M. orab, andd. Heccoun, (00) Capacty analyss of opportunstc relayng n cooperatve systems wth outdated channel nformaton,ieee Communcaton Letters, vol. 4,no., pp. -3. [35] M. orab, J. F. Frgon and D. Haccoun, (0) On the performance analyss of AF opportunstc relayng wth adaptve transmsson over Raylegh fadng channels, IEEE pacfc Rm conference on communcatons, computers and sgnal processng, pp [36] E. S. Altubash, and X. S. Shen, (0) Spectrally effcent varable-rate best-relay selecton scheme for adaptve cooperatve system, IEEE GLOBECOM Conference, pp. -5. [37] M. orab, J.F. Frgon, and D. Haccoun, (0) Performance analyss of varable rate adaptve modulaton for AF opportunstc relayng under outdated CSI IEEEInternatonal Symposum on Personal, Indoor and Moble Rado Communcaton, pp [38] K.S. Hwang, Y.C. Ko, and M.S. Aloun, (009) Performance analyss of ncremental opportunstc relayng over dentcally and non-dentcally dstrbuted cooperatve paths, IEEE ransactons on Wreless Communcatons, vol. 8, no. 4, pp [39] H.Y. Lateef, D. C. McLermon and M. Ghogho, (00) Performance analyss of cooperatve communcaton wth opportunstc relayng, IEEE thinternatonal Workshop on Sgnal Processng Advances n Wreless Communcatons, pp. -5 [40] O. Olaby and A. Annamala, (0) ASER analyss of cooperatve non-regeneratve relay systems over generalzed fadng channels, IEEE Internatonal Conference on Computer Communcaton and etworks, pp. -6. [4] R. H. Y. Loue, Y. L, and B. Vucetc, (008) Performance analyss of beamformng n two hop amplfy and forward relay networks,ieee Internatonal Conference on Communcatons, pp [4] D. Senarante and C. ellambura, (00) Unfed exact performance analyss of two-hop amplfy-andforward relayng n akagam fadng, IEEE rans. Veh. ech., vol. 59, no.3, pp [43] M. Hasna and M.S. Aloun (004) Harmonc mean and end-to-end performance of transmsson system wth relays, IEEE rans. Communcatons, vol. 5, no, pp [44] Wefeng Su, K. S. Ahmed andk. J. Ray Lu, (008) Cooperatve communcaton protocols n wreless networks: performance analyss and optmum power allocaton, Sprnger lnk Wreless Personal Communcaton, vol. 44, no., pp.8-7. [45] O. Olaby, A. Annamala, O. Odejdeand E. Adebola, (0) Integrated desgn of APP/E/PHY/MAC layers for cooperatve relay networks. Under revew for Publcaton, Internatonal Journal of Wreless and Moble Computng [46] O. Olaby and A. Annamala, (0) Effcent symbol error rate analyss of cooperatve nonregeneratve relay systems over generalzed fadng channels,internatonal Journal of Wreless and Moble etworks, vol. 4, no., pp. -0 [47] A. Goldsmth and S. Chua, (997) Varable-rate varable-power M-QAM for fadng channels, IEEE rans. Commun., vol. 45, no. 0, pp [48] M. Chan, D. Dardar and M. K. Smon, (003) ew exponental bounds and approxmatons for the computaton of error probablty n fadng channels, IEEE rans. On Wreless Commun., vol., no. 4, pp [49] M. D. Renzo, F. Grazos and F. Santucc, (009) A unfed framework for performance analyss of CSI-asssted cooperatve communcatons over fadng channels,ieee rans. Communcatons, vol. 57, pp [50] A. Annamala, G.Deora and C. ellambura, (005) heoretcal dversty mprovement n GSC (, L) recever wth non dentcal fadng statstcs, IEEE rans. Commun., vol. 53, no. 6, pp [5] B. Mod, A. Annamala, O. Olaby and R. Chembl Palat, (0) Ergodc capacty analyses of cooperatve amplfy and forward relay networks over Rce and akagam fadng channels, Internatonal Journal of Wreless and Moble etworks, vol. 4, no., pp [5] I. S. Gradshteyn and I. M. Ryzhk, able of Integrals, Seres and Products, 5th ed., San Dego, CA: Academc,

19 Internatonal Journal of Computer etworks & Communcatons (IJCC) Vol.8, o.3, May 06 AUHORS Dr. Bhuvan Mod receved PhD. degree from Prare Vew A & M Unversty, exas A & M Unversty System, n 0. He earned hs M.S. degree n Electrcal Engneerng from Lamar Unversty, Unted States of Amerca, M.S. degree n Electroncs and Communcaton Engneerng from Dharmsnh Desa Unversty, Inda, and the B.S. degree n Electroncs and Communcaton Engneerng from orth Gujarat Unversty, Inda, n 009, 00 and 00, respectvely. He s currently workng as a Senor Member of echncal Staff at A& Moblty Lab, Seattle, WA. He receved Student ravel Grant Award to present hs work at the IEEE MILCOM. Over last few years, Mr. Mod has publshed over a dozen peer revewed conference and journal artcles. Hs current research nterests nclude cross-layer desgn/optmzaton for adaptve-lnk cooperatve relay networks, Software Defne etworkng and 4G VoLE. Dr. Oluwatob O. Olaby receved the B.Sc. degree n Electronc and Electrcal Engneerng from Obafem Awolowo Unversty, Ile-Ife and M.S. and PhD degree n Electrcal Engneerng from Prare Vew A&M Unversty, exas. Over the last few years, he has co-authored approxmately two-dozen peer-revewed conference and journal artcles. He was the recpent of the Roy G. Perry College of Engneerng Outstandng Masters Student of the Year Award (0) and the atonal Socety of Black Engneer s Golden orch Award for Graduate Student of Year (0). Hs research nterests nclude dynamc spectrum access, MIMO, cooperatve communcatons, statstcal sgnal processng, compressve sensng, machne-learnng and optmzaton technques. Dr. Annamala s presently the Drector of Center of Excellence for Communcaton Systems echnology Research, a exas A&M Board of Regents approved Unversty Research Center at the Prare Vew A&M Unversty, and a tenured faculty member n the Department of Electrcal and Computer Engneerng. He has over 0 years of research/teachng experence n wreless communcatons at Motorola, Unversty of Vctora, Ar Force Research Laboratory, Vrgna ech and PVAMU wth approxmately 00 peer-revewed publcatons and 5 book chapters. Dr. Annamala has been honored by hs colleagues on numerous occasons for hs excellence n research ncludng wnnng the 0 Roy G. Perry College of Engneerng Outstandng Faculty (Research) Award, IEEE Leon Krchmayer Prze Paper award, ASEE/AFOSR Summer Faculty Fellowshps, SERC Doctoral Prze, CAGS/UMI Dstngushed Doctoral Dssertaton Award, IEEE VS/Motorola Danel E. oble Fellowshp, among others. He had served on the Edtoral Boards of four IEEE journals/transactons n the last 5 years, and has helped to organze a few major IEEE conferences on wreless communcatons ncludng servng n the capacty of echncal Program Char of the 00 IEEE Vehcular echnology Conference n Vancouver, Canada. Hs current research nterests nclude cooperatve spectrum sensng, compressve sensng, cross-layer desgn for scalable multmeda transmsson and cooperatve wreless communcatons. 9