Applictio of Superpositio Codig for Subcrrier d Bit Alloctio i Dowli OFDM Systems r Nthwi, Member, IACSIT, Adity Trivedi, d lp Goyl, Member, IACSIT Abstrct Ohogol frequecy divisio multiplexig (OFDM) is well ow for multicrrier modultio techique i which high dt rte strem is divided ito differet prllel lower dt strem. I this pper, we study the mrgi dptive - superpositio codig (MA-SC) lgorithm d rte dptive - superpositio codig (A-SC) lgorithm, where t most two user c shre ech subcrrier s compred to MA d A lgorithm, where ech subcrrier is shred by oly sigle user without SC scheme. We pply SC scheme over MA d A to chieve mximum system throughput with seprte power costrit for rel d o rel time users, esurig tht QoS requiremet for rel time d propoiol firess mog o rel time users is stisfied. The overll computtiol complexity of proposed modified A-SC lgorithm is sme s A lgorithm. I the modified proposed MA-SC, complexity icreses i some of the steps due to dditio of SC scheme over MA lgorithm, but overll complexity remis the sme. Mote-Crlo simultios re beig crried out to show tht the proposed lgorithm improves the performce i terms of power required d throughput. Idex Terms MA-SC, OFDM, A-SC. I. INTODUCTION Ohogol frequecy divisio multiplexig (OFDM) is efficiet techique for providig high dt rte system d its bility to icrese the symbol time lrge eough so tht the iduced delys produced by chel re oly isigifict poio of the symbol durtio, which mes OFDM system cpble of reducig or elimitig itersymbol iterferece (ISI) [1]. esource lloctio re geerlly of two types-first fixed or sttic resource lloctio, where we do ot utilize the owledge of chel stte iformtio (CSI) d users hve predetermied owledge of subcrriers o which they c trsmit. Aother oe is dymic resource lloctio, where ech user te dvtge of CSI i selectig the best subcrrier for them. Agi bsed o differet optimiztio problem, resource lloctio c be divided i two ps-first is mrgi dptive (MA) optimiztio for rel time users discussed i [2]-[6], which miimizes the overll trsmit power ssurig tht bit error rte (BE) d bit rte requiremets is stisfied, secod is rte dptive (A) Muscript received Jury 28, 2012; revised Mrch 5, 2012. This wor ws suppoed by ABV-IIITM Gwlior. r Nthwi is with the Idi Istitue of Techology, pur. (e-mil: r046@gmil.com). Adity Trivedi is with the Depmet of ICT, ABV-Idi Istitute of Iformtio Techology d Mgemet, Gwlior, Idi (e-mil: trivedi@iiitm.c.i). lp Goyl is with the Depmet of Electroics, Mul Azd Ntiol Istitue of Techology, Bhopl, Idi (e-mil: lp.goyl19@gmil.com). optimiztio lgorithm for o-reltime users discussed i [8]-[11], with priciple of mximizig the system throughput hvig costrits of overll trsmit power, propoiol firess d BE requiremet. Both MA for rel time user d A for o rel time user is cosidered i pper [7], which utilizes heterogeeous services to mximize the overll system throughput uder the totl trsmit power costrit mig sure tht the propoiol firess mog o-reltime users d QoS requiremet of rel time users is stisfied. I [8], i order to chieve the effective cpcity for multicst trsmissio SC techique is used without tig multiuser ccess techique ito ccout. Simple wtermrig system is proposed i [9], which llows two clsses to be trsmitted o the sme chel with the sme frequecy pl. A suboptiml lgorithm with objective to chieve mximum system throughput is discussed i [10] with trffic requiremet of users. SC is used i [11] to perform subcrrier, bit d power lloctio by llowig t most two users to shre sme subcrrier. But they hve pplied this scheme for oly rel time users without tig o rel time users ito ccout. I [11] d [12], uthor first llocte subcrrier to weest li or fr user d llow these subcrrier to be shred by etil user (er to the BS) usig SC scheme. I [6], resource lloctio lgorithm without iteger bit costrit for heterogeeous services i prcticl dptive OFDMA systems is cosidered, which is fuher ivestigted i [7] for dptive resource lloctio problem i OFDMA systems with heterogeeous services. I this pper, we exted the lloctio strtegy proposed i [7] by itroducig the SC scheme [11] o both MA d A lgorithm, whe cosiderig the rel time d o rel time users respectively d we ssume perfect owledge of CSI t BS. The proposed MA-SC d A-SC lgorithm chieves mximum throughput with power costrit of rel d o rel time users, stisfyig QoS requiremet d propoiol firess of rel d o rel time users respectively. We lso dd some ew costrits to [7], explied i sect. II to stisfy the requiremet of SC theorem. (MA-SC) lgorithm bsed o mrgil utility of subcrrier, determies power required by rel time users P by llowig first rded rel time user (fr from BS) to loo for the best subcrrier for him o the bsis of chel gi the etil rel time user (er to BS) is llowed to shre the sme subcrrier by trsmittig smll mout of power without deteriortig the performce of rded user. (A-SC) lgorithm gives the throughput of o rel time user by utilizig the remiig power P d SC scheme s followed i MA-SC. I MA-SC, iclusio of etil user lgorithm i fouh step icreses the complexity to O(W N) from O(N) of MA lgorithm [7], where W is totl umber of rel time users d N is totl 278
umber of subcrriers d N>W. However, the overll complexity of MA-SC remi sme s MA lgorithm [7]. I the proposed A-SC, the complexity remis the sme s tht of A [7]. The lyout of pper s follows: Sect. II describes optimiztio problem of bit d subcrrier lloctio. Sect. III explis SC theorem. Sect. IV explis (MA-SC) d (A-SC) lgorithms. Sect. V presets simultio results. Filly, coclusios re drw i Sect. VI. II. OPTIMIZATION POBLEM The optimiztio problem of the subcrrier d bit lloctio lgorithm with iclusio of SC for both rel time (rded d etil) user d o rel time (rded d etil) is to mximize system throughput, esurig the bit rte requiremet of reltime users d the propoiol firess mog o-reltime users subject to power costrit for rel d o rel time users. Let us hve W users tht iclude W reltime (rded d etil) users d W o-reltime (rded d etil) users with requiremet of BE & bit rte d without requiremet of bit rte respectively. The complete set of reltime users d o-reltime users which coti both rded d etil users re deoted by U d U respectively. The dowli chel gi of the th (1<< M) user o the th (1<<N) subcrrier is give by h. The optimiztio scheme [1] is modified with icresed costrits described below: N T = mx [0,1] b ε b εl εu =1 The mximiztio is subjected to the costrits give below: C1: = 1 2, C2: = req, ε U C3: i / Ω i = / Ω, i, j ε U C4: S ε C5: p 0, C6: C7: = 1 β, C8: j j p P T, pwr 0 β 1,, = 1 εs β, =0,, = 1 εs pwr β, 1, p P (1) P Here, L deotes the set of turl umbers; C1 gurtees tht ech subcrrier c be shred by t most two user. The bit rte requiremet req of reltime users is idicted by C2. Oe c chieve the propoiol firess for o-reltime users usig C3 with rte weight of user deoted by Ω stisfies E Ω =1. I C4, the sum of power o ech rel d o rel time subcrriers should be less th equl to P T d P respectively. C5 idictes power o ech rel d o rel time subcrriers should be positive. C6 deotes subcrrier lloctio idictor for the etil d rded user. C7 explis tht oe user cot be cosidered s etil d rded o the sme subcrrier. C8 is mximum power costrit for o rel time users. III. SUPEPOSITION CODING SCHEME Superpositio codig [11] llows t most two users to shre ech subcrrier for their trsmissio. Idepedet resultt iformtio S correspodig to the rded d etil user S 1 d S 2 respectively is trsmitted to two or more receivers t the sme time. S= S 1 +S 2 /A (2) where A equl to 8 is rtio of djusted power betwee the rded d etil user sigl. Sigle tx d rx te is used. The sigl received t receiver Y is decoded ito S^1 d S^2 (expressed below), which deotes the received sigl correspodig to the rded d etil user side respectively. S^1=h S+Z (3) S^2=h S+Z (4) where h d h re chel gi, Z d Z re (AWGN) oise for rded d etil user respectively. The mechism to decode the sigl of rded d etil user t receiver is similr to [11]. U 's sigl is cosidered s oise by U d the U decodes its ow sigl from S^1. Now the tur for the etil user to decode its ow sigl. Successive iterferece cceltio (SIC) is performed by U who hs the best chel gi. This mes tht t etil user side, whe S 1 is decoded, U moves wy from S^2 to decode S 2. Some prcticl simplifictio [11] is beig doe to reduce the complexity which re s follows: 1) Potetil user c oly trsmit 2 bits usig QPS modultio, otherwise etil user S 2 is ot llowed to trsmit y iformtio. 2) I order to ow tht piculr user is etil or ot o specific subcrrier, predetermied threshold give i Tble I is used for compriso with user's SN. Fig. 1. Trsmissio d eceptio Scheme for SC [9]. IV. THE SUBCAIE AND BIT ALLOCATION TECHNIQUE I this sectio, we develop (MA-SC) d (A-SC) lgorithm i order to determie the power required by the rel time user d throughput of o-rel time user respectively, where rel d o rel time user cosist of etil d rded users. The totl power vilble t the bse sttio 279
P T is utilized i such wy tht power required by the rel time (rded d etil) user usig MA-SC comes out to be P. The remiig power P is obtied by subtrctig P from totl power P T d remiig subcrrier is obtied s S=(1,2,...N)-S. This remiig power d subcrrier is used i A-SC for clcultig the throughput of o rel time (etil d rded) users. A. Mrgi Adptive Algorithm with Superpositio Codig scheme The objective of (MA-SC) lgorithm is to chieve miimiztio of totl required power stisfyig the requiremet of bit rte for reltime (Potetil d rded) users d costrit E p P T. S is totl subcrriers vilble to the rel time users. I this lgorithm, joit subcrrier d bit lloctio b is cosidered. MA-SC is bsed o the otio of mrgil utility of subcrrier [7], which gives the mximum reductio i power, whe subcrrier is dditiolly llocted to user, which is deoted s pwm. MA lgorithm [7] is modified by itroducig SC scheme [11]. The proposed modified lgorithm is described below: Iput S, P T d Output P. 1. Iitiliztio ) set = 0, b =0, β, =0 d β, =0 for E U d E S. Set S=S, S =Ø d S= Ø,. b) Power iitiliztio: the mout of power o ech rel time subcrrier S is p = P T / S, where P T is the totl power vilble t the bse sttio. 2. for ech E U ) So h i scedig order b) Fid =rg mx E S h c) set =1,β, =1 d b = req, S=S-{}, S =S +{}. d) looig for etil user i. From b, determie the umber of bits trsmitted mi by user o subcrrier, the obti SN from Tble I ii. Adjust trsmit power p to get SN,=(p (h ) 2 )/(σ 2,) equl to SN mi from Tble I. iii. for ech user recompute the SN iv. cll etil user fidig lgorithm described i sectio IV(c). 3. for ech E U ) Fid =rg mx E S h. b) Clculte mrgil utility Δpwm, usig the pproch explied i [7]. 4. While S Ø ) Fid (,)=rg mx, E S Δpwm, b) set, =1,,β, =1, S=S-{}, S = S +{} c) edistribute the required bits of user to subcrrier set S. d) looig for etil user s doe i 2(d) of sme lgorithm with is replced by. e) for ll user whose Δ pwm usig the pproch explied i [7]. 5. d b is fil subcrrier d bit llocrio idictors. The overll trsmit power is give s P = εu S ε 2 Γσ (2 h b 1) Expltio: The subcrrier d bit lloctio idictors is iitilized i first stge log with the seprte lloctio coefficiets for both rded β, d etil user β,. S hve the owledge of the ullocted subcrrier, S d S will depict the subcrrier set ssiged to rded users d etil users respectively. Also i first stge, equl mout of power P T is llocted o differet subcrriers S, where. deotes the crdility of set. I the secod stge, subcrrier which is ot beig utilized d lso hs the mximum chel gi is ssiged to user t ech itertio. These users re rded rel time users, who c trsmit the totl umber of bits vilble to them by bit rte requiremet o subcrrier recetly foud t ech itertio. The we loo for the etil user who c trsmit 2 bits o sme subcrrier usig the SC. To loo for etil user, the o. of bits trsmitted by rded user d 2 bits by etil user is compred i the Tble I d correspodig SN mi is foud. The trsmit power p is compested iorder to djust SN, =p (h, ) 2 /(σ 2,) equl to SN mi so tht costrit is stisfied d the we recompute the SN o the subcrrier for ech user. After tht, we mi compre the SN of ll the etil users with the SN to decide tht if shre lloctio o the subcrrier is possible or ot d if possible i.e etil user is foud, the we dd to S. It is lso im to ote tht oe user cot be cosidered s etil d rded o the sme subcrrier. I the third stge, we clculte the mrgil utility [7] of the ussiged subcrrier tht re hvig the highest chel gi for ech user. I followig subcrrier selectio stge, we would select oly those subcrrier tht re hvig lrgest mrgil utility. I the fouh stge, we joitly select ullocted subcrrier d piculr user tht re hvig the lrgest mrgil utility, d we llocte this subcrrier to the selected user. The we redistribute p of the required bits of the user to the recetly dded subcrrier. After this, we loo for the etil user s doe i step 2(d). There my be possibility tht selected subcrrier is the best subcrrier for other users, the i tht cse the mrgil utility of other ullocted subcrrier for those users re ecessry to be clculted. Filly, we obti subcrrier d bit lloctio results d the overll power required to llocte ll the subcrriers. I MA-SC, iclusio of etil user lgorithm i fouh step icreses the complexity to O(WN) from O(N) of MA lgorithm [7]. However, the complexity of remiig step would remi sme s MA [7], with o chge i overll complexity. B. te Adptive Algorithm with Superpositio Codig Scheme A-SC lgorithm mximizes the totl throughput of o-reltime (Potetil d rded) users by llowig both etil d rded user to shre the sme subcrrier, subject to o rel time power costrit C8 of optimiztio 280
problem log with costrit E P d user rte propoiolity. I this lgorithm, we seprtely perform subcrrier d bit lloctio procedure. A lgorithm [7] is modified by itroducig SC scheme [11]. Modified subcrrier lloctio step icludig SC is give below: Iput: S, P Output: T. 1. Iitilistio ) set = 0, b =0, β, =0 d β, =0, b =, S =Ø S= Ø d = Ø, for E U d E S. Set S=S,. Clculte b s = h P b log2 (1+ ) 2 Γσ S b) power iitiliztio: the mout of power o ech rel time subcrrier S is pwr=p / S, where P is remiig power vilble t the bse sttio 2. While S Ø ) Fid =rg mi ( /Ω ) b) Fid =rg mx b c) set = 1, β, =1 d S=S-{}, S = S +{}, = +b. d) looig for etil user s doe i step 2(d) of MA-SC lgorithm, with d p is replced by d pwr respectively. 3. is the totl subcrrier lloctio idictor, S d S is the sucrrier set ssiged to rded d etil user respectively. Bit lloctio step usig iformtio from bove s follows. Iputs, S, S d P. Output T. 1. Iitilistio ) Set b =0, =0,for E U d E U d E S, Iitlise P=0. 2 Γσ b Δ P = (2 ) h b) Trsmit 2 bits to ech user ccordig to subcrrier lloctio set S obtied from subcrrier lloctio step. These users re etil users. 2. While (1) ) Fid = rg mi (( /Ω ) b) Fid =rg mi E S Δ P. c) if P+ Δ P P updte b = b +1, P=P+ Δ P, = +1, 2 Γσ b Δ P = (2 ) h else exit while loop; ed if-else. 3. The fil bit lloctio of the th user o the th subcrrier is give s b. The overll throughput of o rel time user is give by = b. T S ε U ε Expltio: The subcrrier lloctio step is repeted up to the legth of S, where. deotes the crdility of set d we iitilize the subcrrier lloctio idictor log with seprte lloctio coefficiets for both rded d etil user. S d S re subcrrier set llocted to rded d etil users respectively. S eeps the owledge of the ullocted subcrriers.the remiig power P is eqully divided mog the remiig subcrrier S. User rte propoiolity is gurteed by llowig tht user to choose subcrrier which hs miimum ormlized trsmit rte /Ω. Now, selected user will choose tht subcrrier which hs highest trsmit rte d these users re med s rded users. The we will loo for etil user, who c trsmit 2 bits o sme subcrrier usig SC. To loo for etil user, we eed to follow the sme step s followed i 2(d) d 4(d) of (MA-SC) lgorithm. Oe user cot be cosidered s etil d rded o the sme subcrrier. After ll the subcrriers i S re ssiged, we obti the totl subcrrier lloctio idictor d the fil ssiged subcrrier set S d S of rded d etil users respectively. These three, S d S re the iput to bit lloctio step. The bit d correspodig power is llocted by bit lloctio step o ech subcrrier. Bit lloctio step is similr to the subcrrier lloctio step which ssures to stisfy the user rte propoiolity by lloctig oe bit to the user,who hs lest ormlized rte /Ω Ad we lso llocte 2 bits to etil user ccordig to subcrrier lloctio set S. The ext step llocte bits to rded users. I ech itertio, the user will llocte dditiol bit to tht subcrrier, which requires the lest power. The bit lloctio step ceses, whe power llocted reches the o rel time power costrit P. With the dditio of power iitiliztio i 1st step, etil user lgorithm i 2d step of A-SC (subcrrier lloctio lgorithm) d trsmissio of 2 bits to etil user of A-SC (bit lloctio lgorithm) requires the complexity of O(N), O(W N) d O(W N) respectively. The overll complexity of A-SC lgorithm O(N 2 ) would remi sme s A lgorithm [7]. C. Potetil User Fid Algorithm The lgorithm for serchig etil user [11] is described below: Set m=1, Potetil user=not foud, while (m ) d (Potetil User=Not foud) do if SNm, SNmi the Potetil User=Foud d m =m, bm=, Set m =1 β m=1 S ( m ) = S ( m ) { } else m=m+1 ed TABLE I: THESHOLD VALUE OF SN MIN USING SC WITH SE=10-3 mi SN mi SN 2 2 11.0 db 28.4 db 4 2 18.3 db 35.4 db 6 2 24.6 db 41.6 db 8 2 30.7 db 47.1 db 10 2 36.9 db 53.7 db 12 2 42.9 db 59.8 db 14 2 48.9 db 65.8 db 16 2 54.9 db 71.8 db 281
The Tble I of [11] is expded for =16 bits, where d re bits trsmitted by rded d etil user respectively. We ssume tht sme service is brodcsted to ech user, which stisfy the followig coditio: SE rd (S 1 ) 10-3, mesured t rded user side. SE et (S 2 ) 10-3, mesured t etil user side. V. SIMULATION EVALUATION A. Simultio Setup MATLAB tool hs bee utilized to evlute the performce of the proposed MA-SC d A-SC lgorithm. Three-pth ryleigh fdig chel hs bee cosidered for simultio with expoetil power dely profile, whose vrices re computed by d -α pth-loss model for ryleigh fdig chel, with α=3. The totl power P T t BS is 6.4 W. Bdwidth ssiged for OFDM system is 5MHz, over this bdwidth BS will llocte 64 subcrriers. The verge chel gi for both reltime (etil d rded) users or o reltime (etil d rded) users is chose rdomly withi (0.1,1). The bit error rte cosidered for ech user is equl to 10-3 the SN gp of OFDM system is bout 5.48 db. AWGN's vrice t ech receiver is ssumed to be equl, i.e σ 2 =1mW. The umber of rel time users d o rel time users re req equl to 8, d ech user hs the bit rte requiremet equl to 16 bits/ofdm symbol, d the rte weight of ech o rel time user stisfies: Ω =1/8, forll ε U. Avergig is doe over 1000 etwor topologies over 100 OFDM symbols for ech results. B. Simultio esults I Fig. 2, W =8 d req =16 bits/ofdm symbol forll ε U.. I proposed (MA-SC), power required to trsmit required umber of bits for ech rel time user is reduced by (57.14 %), whe 19 subcrriers re used d reduced by (64.28 %), whe 21 subcrriers re used d reduced by (71 %), whe 24 subcrriers re used d reduced by (30%), whe 32 subcrriers re used i compriso to MA [7], but ow it requires to fid the etil user t ech itertio to trsmit 2 bits to the user o the sme subcrrier which hd lredy bee used by the rded user. Whe we llocte 56 subcrriers, power required by rel time user usig MA d MA-SC lgorithm comes out to be sme, i.e., power become costt for both MA-SC d MA fter 56 subcrriers hve bee llocted to rel time users. I Fig. 3, W =8 d Ω =1/8, forll ε U. We otice tht there is sigifict icremet i the system cpcity by usig A-SC. With A [7], throughput t 16 subcrriers is zero but with A-SC, it is 235 bits/ofdm symbol d whe 24 subcrriers re used, throughput icreses from 185 bits/ofdm symbol to 310 bits/ofdm symbol d whe 32 subcrriers re used, throughput icreses from 193 bits/ofdm symbol to 258 bits/ofdm symbols d throughput icreses fuher i compriso to A lgorithm [7] for the remiig subcrriers. Fig. 2. equired power by rel time user versus umber of subcrriers for reltime user Fig. 3. Totl throughput for o-reltime user v umber of subcrriers used by rel time user Fig. 4. Averge required power per subcrrier vs required bit rte I Fig. 4, verge required power per subcrrier give s P /(N σ 2 ) is reduced s the power required by rel time users is reduced. W vries from 4 to 8. Whe 16 bits re used the verge power per subcrrier is reduced by 26 % d whe 20 bits re te the verge power per subcrrier is reduced by 48.7 % d whe 24 bits re te verge power per subcrrier is reduced by 48.18 % d it decreses fuher, whe we go upto 28 bits. The bits required for Fig. 4 re te i rge from 16 to 28 bits/ofdm symbol for ech user d elsewhere bits required hve bee te equl to 16 bits/ofdm symbol. 282
VI. CONCLUSION I this pper, we first study MA-SC lgorithm, which joitly cosiders subcrrier d bit lloctio for clcultig the power required by rel time (etil d rded) users d the A-SC lgorithm, which seprtely llocte bit d subcrrier for mximizig the throughput of the o rel time (etil d rded) users utilizig the P & S. I both lgorithms, ech subcrrier is shred by t most two users s compred to MA d A lgorithm without SC scheme. System throughput is oly domited by throughput of o rel time users, s throughput of the rel time user is fixed d provided. The results show tht our proposed lgorithm performs better th MA d A lgorithm without SC [7]. I MA-SC, iclusio of etil user lgorithm i fouh step icreses the complexity to O(W N) from O(N) of MA lgorithm [7]. The complexity of remiig step is sme. The overll computtiol complexity of A-SC lgorithm is sme s A lgorithm [7]. SC c be esily pplied to y lloctio lgorithm for OFDM system s: wheever subcrrier is llocted to user, the we clculte SN, d compre this SN, to SN mi from Tble I so tht we c hve c ide tht how my bits c be trsmitted d we tret this user s rded. After this, we serch for the etil user u, who c use this remiig power for trsmissio of 2 more bits o the sme subcrrier. The we c declre user s etil if its SN, is greter th SN mi determied i Tble I. [9] S. Njeh, H. Besbes, d A. Boullegue, ``A simple wtermrig system,'' i Proc. ISCCSP 2004, Hmmmet Tuisi, Mrs, 2004, pp. 531-534. [10] S. Njeh, H. Besbes, d A. Boullegue, ``Greedy lgorithm for dymic resource lloctio i dowli of OFDMA system,'', i Proc. 2d ISWCS 2005, Sie Itly September 5-7, 2005. [11] H. Besbes, d S. Njeh, ``A Simple Superpositio Codig Scheme for Optimizig esource Alloctio i Dowli OFDMA Systems,'' Wireless Pers Commuictios, pp. 1-24, Spriger 2010. [12] H. Besbes, S. Njeh, A. Jdidi, d S. Smiri, ``esource lloctio lgorithm i dowli of OFDMA systems bsed o users clssifictio d superpositio codig,'' i Proc. SCS 2009, Djerb Tuisi, 2009, Nov. [13] W. hee, d J. M. Cioffi, ``Icresig i cpcity of multi user OFDM system usig dymic sub chel lloctio''. I Proc. IEEE It. Vehiculr Tech. Cof., vol. 2, Sprig, 2000, pp. 1085-1089. [14] Z. She, J. G. Adrews, d B. L. Evs, ``Optiml power lloctio i multi user OFDM systems,'' i Proc. IEEE GLOBECOM '03, vol. 1, S Frcisco, CA, Dec. 1-5, 2003, pp. 337-341. [15] I. C. Wog, Z. She, J. G. Adrews d B. L. Evs, ``A low complexity lgorithm for propoiol resource lloctio i OFDMA systems,'' i Proc. IEEE SIPS 2004, Austi, Texs, Oct. 13-15, 2004, pp. 1-6. [16] G. Yu, Z. Zhg, Y. Che, P. Cheg d P. Qiu, ``Sub crrier d bit lloctio for OFDMA systems with propoiol firess,'' i IEEE WCNC 2006, vol. 3, Ls Vegs, USA, Apr. 3-6, 2006, pp. 1717-1722. r Nthwi received the B.Tech ree i Electroics & Commuictio from UPTU Uiversity i 2009 d M.Tech. ree i Digitl Commuictio from ABV-IIITM Gwlior i 2011. He is ow Ph. D. cdidte i Electroics depmet i IIT pur. His reserch iterests iclude speech processig, wireless commuictio etc. ACNOWLEDGMENT The uthors re thful to ABV-Idi Istitute of Iformtio Techology d Mgemet Gwlior, Idi for extedig the ficil suppo for this pper. EFEENCES [1] J. G. Adrews, A. Ghosh d. Muhmed, Fudmetls of WiMAX PreticeHll,USA,2007. [2] G. Yu, Z. Zhg, Y. Che, J. Shi d P. Qiu, ``A ovel resource lloctio lgorithm for rel time services i multiuser OFDM Systems,'' i IEEE 63 vehiculr techology coferece, vol. 3, Melboure, Austrli, My 7-10, 2006, pp. 1156-1160. [3] C. Y. Wog, C. Y. Tsui,. S. Cheg d. B. Letief, ``A rel time sub-crrier lloctio scheme for multiple ccess dowli OFDM trsmissio,'' i Proc. IEEE VTC 1999-Fll, vol. 2, Sept. 1999, pp. 1124-1128. [4] G. Zhg, ``Sub crrier d bit lloctio for rel-time services i multi user OFDM systems,'' i Proc. IEEE ICC' 04, Vol. 5, Pris, Frce, Jue 20-24, 2004, pp. 2985-2989. [5] D. ivc, G. Li, d H. Liu, ``Computtiolly efficiet bdwidth lloctio d power cotrol for OFDMA,'' IEEE Trsctios o Wireless Commuictios, vol. 2, o. 6, pp. 1150-1158, November 2003.. [6] W. Wg,. C. Hwg,. B. Lee, d S. Bh, ``esource lloctio for heterogeeous services i multiuser OFDM systems,'' i Proc. IEEE Globecom' 2004, vol. 6, Dlls, Texs, 2004, pp. 3478-3481. [7] G. Yu, Z. Zhg, d P. Qiu, ``Adptive subcrrier d bit lloctio i OFDMA systems suppoig hetrogeeous services,'' Wireless Pers Commuictios, vol. 43, o. 4, pp. 1057-1070, Spriger 2007. [8] Q. Du, d X. Zhg, ``Effective cpcity of superpositio codig bsed mobile multicst i wireless etwors,'' i Proc. of IEEE ICC 2009, Dresde, Germy, Jue 14-18, 2009 Dr. Adity Trivedi is Professor i the ICT Depmet of ABV-Idi Istitute of Iformtio Techology d Mgemet, Gwlior, Idi. He received his bchelor ree (with distictio) i Electroics Egieerig from the Jiwji Uiversity d M.Tech. ree (Commuictio Systems) from Idi Istitute of Techology (IIT), pur. He obtied his doctorte (PhD) from IIT ooree i the re of Wireless Commuictio Egieerig. His techig d reserch iterest iclude Digitl commuictio, CDMA systems, Sigl processig, d Networig. He is fellow of the Istitutio of Electroics d Telecommuictio Egieers (IETE) d member of Istitutio of Electricl d Electroics Egieers (IEEE),USA. He hs published more th 50 ppers i vrious prestigious Itertiol/Ntiol jourls d cofereces. I 2007, he ws give the IETE s.s. rish Memoril Awrd for best system orieted pper. lp Goyl received the B.E. ree i Electroics d Commuictio Egieerig from jiv Gdhi Proudyogii Vishwvidyly, Bhopl i 2008 d M.Tech. ree i Digitl Commuictio from ABV-Idi Istitute of Iformtio Techology d Mgemet, Gwlior, Idi i 2011. She is ow Assistt Professor i Mul Azd Ntiol Istitute of Techology, Bhopl, Idi. Her reserch iterests iclude MIMO OFDM systems, cross lyer desig, wireless commuictios etc. 283