1Energy-Efcient Relay Aided Ad Hoc Networks Using Iteratively Detected IrregularConvolutional Coded, Unity-Rate Coded and Space-Time Trellis Coded TransceiversJing Zuo, Hung Viet Nguyen, Soon Xin Ng and Lajos HanzoSchool of ECS, University of Southampton, SO17 1BJ, UK.Tel: +44-23-8059 3125, Fax: +44-23-8059 4508Email: {jz08r,hvn08r,sxn,lh}@ecs.soton.ac.uk; http://www-mobile.ecs.soton.ac.ukAbstractThe nodes of ad hoc networks are typically battery-powered, hence requiring the employment of energy-efcientschemes. Near-capacitycodingschemes allowasingle linktocommunicate usingthe lowest possible transmit power, whileachieving a lowFrame Error Ratio (FER). With the samemotivation of conserving battery energy, numerous power-awarerouting algorithms have been proposed for improving the overallmultiuser networks energy-efciency instead of improving a sin-gle link. Taking both these design factors into consideration, weemploy Multi-Antenna aided Relays (MA-Rs) in the context of adhoc networks, which use a three-stage concatenated transceiverconstituted by an Irregular Convolutional Code, Unity-Rate Codeand Space-Time Trellis Code (IrCC-URC-STTC) equipped withtwo antennas. All other mobiles are single-antenna assisted nodesand refrain fromrelaying, hence they dispense with STTCschemes anduse atwo-stage concatenatedIrCC-URCcodingscheme. It is conrmed that as expected, in a high-node-densityscenariothe average energyconsumptionper informationbitand per node becomes about a factor two lower than that in theequivalent Single-Antenna Relay (SA-R) aided networks.I. INTRODUCTIONEnergy-efcient wireless networkdesignhas recentlyat-tractedwide-spreadresearchattention. Diverseerror-resilientForward Error Correction (FEC) schemes were proposed in [1]for achievingalowBit Error Ratio(BER) at near-capacitySignal-to-NoiseRatio(SNR)values. Therefore, theeffectivetransmissionrangecanbeimproved, whentherequiredre-ceived signal power is reduced. A novel Distributed Concate-natedIrregularConvolutional Coded, Unity-RateCodedandSpace-Time Trellis Coded (DC-IrCC-URC-STTC) scheme hasbeen proposed for cooperative communications in [2]. SeveralSingle-AntennaRelays (SA-Rs) wereactivatedbetweenthesource and the destination. The relays roaming closest to theiroptimal locationswereactivatedbasedonanovel techniquerelyingonExtrinsic InformationTransfer (EXIT) charts inconjunction with near-capacity code design principles.Numerous power-aware routingprotocols were proposedin [3] for improving the energy efciency from a multiuser net-workingperspective. Moreover, cross-layeroptimizedpowercontrol hasbeenwidelyexploited[48]formaintainingtherequiredtarget-integrityat alowpowerinrealisticpropaga-tion environments. A physical-layer-oriented routing protocolsupported by sophisticated power control was proposed in [4]for a Line-Of-Sight (LOS) and shadow faded scenario, wherethe estimated end-to-end BER of a multi-hop path was used astherouteselectionmetric. Furthermore, anadaptiverelayingstrategy switching between the Amplify-and-Forward (AF)The research leading to these results has received funding from the Euro-peanUnionsSeventhFrameworkProgramme(FP7/2007-2013)undergrantagreement no214625. Thenancial support of theChina-UKScholarshipCouncil, andoftheRC-UKundertheauspicesoftheIU-ATCinitiativeisalso gratefully acknowledged.and the Decode-and-Forward (DF) schemes was proposedin[5]forreducingboththeenergyconsumptionaswell asthedelayof thesystem. As afurther designdilemma, theinuence of the small number of long hops versus the manyshort hops philosophy on the energy consumption was studiedin [68]. It was indicated in [6] that the small number of longhopsroutingschemewasbetterthanthemanyshorthopsroutingschemeprovidedthat near-capacitycodingstrategiescombined with a relatively short block length were employed,becauseasubstantial SNRlosswasexhibitedbythemanyshort hopsbasedroutingscheme. Moreover, it wasdemon-strated in [7] that many short hops perform well in energy-limited scenarios relying on spatial reuse, even in the absenceofinterferencecancellation, whileusingasmall numberoflonghopsismoresuitableforbandwidth-limitedscenarios.Therefore, the routing algorithms should be carefully designed,when jointly considering both the achievable energy-efciencyand the attainable bandwidth-efciency. The tradeoffs betweenenergy-andbandwidth-efciencywerestudiedin[8], whereit was foundthat at highend-to-enddata rates the routesassociated with fewer hops minimize the energy consumption,whileat lowerend-to-enddataratestherouteshavingmorehops mitigate it.Against this backdrop, the novel contributions of this trea-tise are A near-capacity IRCC-URC-STTC transceiver is de-signedformultiple-antenna-relayingaidedadhocnet-works, which is shown to reduce the networks totalenergy-consumption. This is achieved with the aid of sophisticated cross-layer-operationaidedrouting, wheretheroutesaresetuponthe basis of using the lowest number of potentially longerhops, which become sufciently reliable as a benetof the powerful near-capacity IRCC-URC-STTCrelay-transceivers.Toelaboratealitter further, wecharacterizethebenetsof Multi-Antenna aided Relays (MA-Rs) inad hoc networksin terms of the attainable energy efciency. The MA-RsareassistedbytheQuadraturePhase-Shift Keying(QPSK)basedIrCC-URC-STTCschemeproposedin[2], whiletheSA-Rs employan8-aryPhase-Shift Keying(8PSK) basedIrCC-URC scheme. The introduction of the MA-Rs inuencesthe route selection. More specically, MA-Rs allowus toreduce the number of hops required as a benet of theirhigher effective transmission range at a given transmit powercomparedtoSA-Rssystems. Simulationsareconductedforinvestigating the inuence of both the number of MA-Rsand that of the node density on the energy consumptionof the entire network. As the ultimate benchmarkscheme,2theDiscrete-input Continuous-output MemorylessChannels(DCMC) capacity [9] is considered.The rest of the paper is organized as follows. In Section II,our IrCC-URC-STTCscheme androutingstrategyare de-tailed. In Section III, various practical scenarios are analyzed,leading to our conclusions in Section IV.II. THEORETICAL ANALYSISA. Near-Capacity Coding SchemesIrregular Convolutional Codes (IrCCs) [10] are constitutedby a family of convolutional codes, which use different code-rates for encoding the appropriately selected fractions of theinput stream. TheseIrCCs areemployedas theouter codeofourserial-concatenatedtransceiverscheme, sincetheyarecapable of approaching the channel capacity [10, 11]. Theappropriately selected fractions are determined using theprocedures proposed in [12] and detailed in [13] by exploitingthefact that alowBERmaybeachievedbytheIrCCatSNRs near the capacity, when its EXIT-chart exhibits anopen but narrow tunnel. Furthermore, as also detailed in [13],for the sake of guaranteeing a lowBERand lowFrameError Ratio(FER), aninniteimpulseresponseUnity-RateCode (URC) shouldbe employedas anintermediate codeinorder toensurethat thethree-stageconcatenatedschemeindeed becomes capable of reaching the (1, 1) point of perfectconvergence to a low BER in the EXIT chart [12]. The Space-TimeTrellisCode(STTC)employedisaMulti-InputMulti-Output(MIMO)arrangement, whichiscapableofachievingboth a coding gain and a spatial diversity gain [13]. Asaresult, theIrCC-URC-STTCcodingschemeadvocatediscapable of approaching the DCMC capacity of the SA-R aidedcooperativesystem[2]. Inthispaper, wereplacethepairofSA-Rs designed for the cooperative system of [2] with a singleMA-RanddemonstratethattheIrCC-URC-STTCschemeiscapable of approaching the DCMC links capacity derived forMA-R aided MIMO systems.We assume that each MA-R is equipped with two antennas.If more than one MA-R exist in the multi-hop ad hoc networkconsidered, then four different types of links may appear.Specically, we have the SA-R to SA-R, SA-R to MA-R, MA-R to SA-R and nally the MA-R to MA-R links. All the MA-Rsemploytheabove-mentionedQPSK-assistedIrCC-URC-STTC scheme, while all the SA-Rs employ the 8PSK-assistedIrCC-URC scheme.First, we analyze the attainable FERperformance of allthe four links, assumingthat theyare capable of reachingthe DCMC capacity of the uncorrelated Rayleigh fadingchannel [9]. TheSingle-Input Single-Output (SISO) DCMCcapacity is given by [14]:CDCMC = log2(M)1Mm=M

m=1E_log2z=M

z=1exp(m,z)|Xm_, (1)wheretheexpectationterminEq. (1) iscomputedbyem-ploying the Monte Carlo averaging method. More specically,MisthenumberofmodulationlevelsandE (A|Xm)istheexpectation ofA conditioned onXm, whilem,zis given bym,z = |h(Xm Xz) +nm|2+ |nz|2N0. (2)Hence we canreadilyndthe correspondingreceive SNRvalue SNRr|BPS for a given Bits Per Symbol (BPS) through-put from the DCMC capacity curve computedusing Eq. (1).Thecalculationof theDCMCcapacityof theSingle-InputMulti-Output (SIMO), Multi-Input Single-Output (MISO) andMulti-Input Multi-Output (MIMO) links canalsobefoundin[14]. IftheSNRr|BPSishigherthantheSNRthresholdof the corresponding capacity curve, the FERtends to 0,otherwise it tends to 1. The FER performance of all the fourlinks is portrayed in Fig. 1.DCMC22DCMC21DCMC12DCMC11IrCC22IrCC21IrCC12IrCC11SNR(dB)FER15 10 5 0100101102103104105Fig.1. FERperformanceofthefourtypesoflinksattheframelengthof1500 bits of the uncorrelated Rayleigh fading channel evaluated from Eq. (1),T R meansTtransmit antennas andR receive antennas.Theinversesecond-powerfree-spacepathlosslawandanuncorrelated Rayleigh fading channel model are considered inthispaper. Thereceivedsignal power prcanbeformulatedas [15]pr = pt_Gl4d_2E_ H 2_, (3)whereptis the transmitted power, Glis the product of thetransmit andreceiveantennagainsintheLOSdirection, His the matrix of Channel Impulse Responses (CIRs), d is theaveragedistancefromthetransmitter andisthecarrierswavelength. We assume Gl= 1, E_ H 2_ = 1and =c/f, wherecisthespeedof light invacuumandf isthecarrier frequency. Hence, we haved =_ptpr4. (4)The Signal-to-noise ratioSNR (dB) is calculated asSNR = 10 log10_prN0_[dB] , (5)where N0is the thermal noise power and N0(dB) =10 log10N0. Then received signal power pr may be expressedaspr = 10SNR +N0(dB)10. (6)Hence, giventhe above-mentionedparameters, the averagedistance from the transmitter, where the SNR requirement may3just be satised, is given byd =pt410SNR +N0(dB)20. (7)Naturally, if wexthevalueof ptandN0(dB), thenitmay be readily seen how the adequately illuminated distance,wheretherequiredtarget-FERmaybemaintained, willvaryas afunctionof theSNRvalue. As seenfromFig. 1, themaximumadequatelycoveredcommunicationdistancefromMA-R to MA-R is the highest, while that from SA-R to SA-Risthelowest. Conversely, if ptanddarexed, thentheFER is the lowest for the MA-R to MA-R link, while it is thehighest for the SA-R to SA-R link.B. Routing AlgorithmsIt was shown in [2] that the IrCC-URC-STTCschemeis capable of operating near the links capacity, hence asubstantial power saving may be attained. When this scheme isemployed by the MA-Rs of the ad hoc network considered, thedifferent error correction capability of the four different typesof links will inuence the routing strategy. Fig. 2 provides anexample on how the routing strategy is inuenced, as detailedbelow.S A B C E F G D(a)S A B C E F G D(b)S A B C E F G D(c)Fig. 2. Theinuenceof MA-Rsontheroutingstrategy: (a)h=5hopswithout MA-Rs; (b)h =4 with1 MA-R at pointF; (c)h =3 with2 MA-Rsat pointsBandF.As seen from Fig. 2, the network consists ofN= 8 nodes,whereSis the source andDis the destination. In Fig. 2(a),all nodes are equipped with a single antenna, hence all linksare SA-R to SA-R links, which yieldsh = 5 hops fromStoD. A single MA-R is employed at point Fin Fig. 2(b), wherethe packets arriving at node C are directly transmitted to nodeF. Then, nodeFwill relay its received packets further to thedestinationD. Morespecically, theC-to-FlinkisaSA-Rto MA-R link, while theF-to-Dlink is an MA-R toDlink,wherethe F Ddistanceis higher thanthat betweenthesingle-antennanodesofFig. 2(a). Consequently, thenumberof hops fromStoD is decreased toh = 4. In Fig. 2(c), twoMA-Rs, namely B and F, are employed. The number of hopsis further decreased toh = 3 as a benet of using MA-Rs fornodesBandF.This paper employs the DYnamic Manet On-demand(DYMO) [16] routingprotocol inthenetworklayer, whichcombines most of the benets of the Ad-hoc On-demandDistanceVector(AODV)[17]andDynamicSourceRouting(DSR) [18] protocols. TheDYMOroutingprotocol alwaysopts for the specic route having the lowest number of hops tothe destination. When employing the MA-R aided IrCC-URC-STTC scheme, we will demonstrate that the route selected maybeexpectedtohaveafurtherreducednumberofhops. TheDYMOroutingprotocol isconstitutedbytwomainstages,namely the route discovery and route maintenance. During theroutediscovery, theRouteREQuest (RREQ) andtheRouteREPly (RREP) packets are used for identifying a route fromthesourcetothedestination. Bycontrast, duringtheroutemaintenance phase, a Route ERRor (RERR) packet is returnedto the source, when a broken link is detected. We assume thatevery node has the same transmit power ofpt. Consequently,the sum of the energy dissipated by all nodes in the networkis given byET=

ERouteDiscovery+

_ptndatardata_+

_ptnwlanACKrwlanACK_+

ERouteMaintenance, (8)where ndatais the length of every packet arriving fromthe MediumAccess Control (MAC) layer, except for theACKnowledgment1(wlan-ACK) packet. The length of thewlan-ACKpacket is nwlanACK. Theinformationbit ratefor transmitting data and wlan-ACKpacket is denoted byrdata and rwlanACK, respectively. We assume that the wlan-ACKpacket is protectedbypowerful, low-ratecodingandmodulationschemes andseldomhas erroneous bits, hencerwlanACKis different fromrdata. Furthermore, on onehand, ERouteDiscoverydenotes the sumof energycon-sumedbytheRREQ, theRREPandthewlan-ACKpack-ets during the route discovery phase. On the other hand,

ERouteMaintenanceincludes all the energy except for theenergydissipatedbythedatapacketsandthecorrespondingwlan-ACK packets during the route maintenance phase. Onlythe transmit power is considered in Eq. (8).Furthermore, letBPSMAandBPSSAdenote the bits persymbol throughput of theMAandSAnodes, respectively.The IrCC, URC and STTC component codes have code ratesof0.5, 1.0and1.0, respectively. SinceQPSKisusedattheMAnodesand8PSKattheSAnodes,wehaveBPSMA=2BPSSA/3andthenrMA= 2rSA/3, whererMAandrSAare the bit rate at the MA-Rs and SA-Rs, respectively. Hence,Eq. (8) may be simplied as:ET=

ERouteDiscovery+ 3ptndata 12rSArma +_ptndatarSA+ptnwlanACKrwlanACK_h+

ERouteMaintenance, (9)whereagainhisthenumberofhopsoftheselectedroute.First, we assume thatpt,ndata,nwlanACKandrwlanACKare constant values. Then we introduce the short-hand of A =3ptndata 12rSAandB=_ptndatarSA+ptnwlanACKrwlanACK_. Fur-thermore, ERouteDiscoveryis replaced byf1 (, rma, v ),1The Acknowledgement packet is the one, which is returned to the transmit-ter as the acknowledgement of the correctly received data in the MAC layer,hencewerefertoit aswlan-ACKinthispaper, wherewlan-ACKmeansthe ACK packet employed in the IEEE802.11 standard. We assume that thereis no Request-To-Send (RTS)/Clear-To-Send (CTS) mechanism employed.4and ERouteMaintenanceis replaced by f2 (, rma, v ),because they are inuenced by the node density , the numberofMA-Rsrmaandthemobilespeed v . Then, Eq. (9)canbe written as:ET= f1 (, rma,v ) +Arma +Bh +f2 (, rma,v ) . (10)WefurthernormalizeETbytheoverallnumberofbitsLreceivedintheapplicationlayer of thedestinationandthenumber of nodesn of the entire network, wheren is relatedtoandLis relatedtondata. Finally, theoverall energyconsumptionENTof the entire network can be expressed as:ENT=ETnL= F(, rma,v , h, BPSSA, L). (11)III. PERFORMANCE STUDYAs seenfromEq. (11), the overall energyconsumptionoftheentirenetworkisdependent onnumerousparameters,suchasthenodedensity, thenumberofMA-Rs, themobilespeed, the number of hops of the selectedroute, the BPSthroughput and the amount of bits received in the applicationlayer of the destination. In order to reduce the dimensionalityof our investigations, wexsomeof theparameters whencharacterizing the benets of MA-Rs on the nodes achievabletransmission range and FER performance. More explicitly, wewill mainlyfocusour attentionontherelationshipbetweenthenumberofMA-Rsandtheenergyconsumption, aswellas that between the node density and the energy consumption.The simulation congurations considered are as follows: The application layer generates P= 5 packets per secondand the length of each packet isLp = 504 bits; The User Datagram Protocol (UDP) [19] is employed inthe transport layer; TheDYMO[16]routingprotocol ischoseninthenet-work layer; IEEE802.11b [20] is used in the MAC layer; The IrCC-URC-8PSK [2] and IrCC-URC-STTC-QPSK[2] schemesareemployedinthephysical layerfor SA and MA nodes, respectively; Every node transmits at 1 mW power.A. Energy Consumption versus The Number of MA-RsAssuming that the node density , the mobile speed v , theBPS throughput BPSSA and the number of bits L received inthe application layer of the destination are constant, Eq. (11)can be simplied to:ENT= F(rma, h). (12)Four scenarios are considered in order to study the relation-ship between the number of MA-Rs and the energy consump-tion, wheren = 60 stationary nodes are randomly located ina 500m500m eld, hence the node density is = 240 nodesper square kilometers. The source and the destination arelocated in the position (0, 0) and (499, 499), respectively. WeincreasethenumberofMA-Rs2fromrma = 0torma = 60insteps of 5. Thereceivers sensitivity[21] is set to 852The number of MA-Rs also includes the source and the destination. Again,we denote the multi-antenna aided nodes as MA-Rs and single-antenna nodesas SA-Rs.dBm. Anysignal receivedat apower lower thanthenoisesensitivity threshold is deemed to be undetectable. The powerrequired is quantied for the IrCC-URC-STTC scenario3andbenchmarked against the DCMC capacity at the thermal noiselevels of 92 dBm and 86 dBm. We also quantify the PacketLossRatio(PLR) andthenetworkcontrol loadinorder tocharacterize the performance of the entire network. We denethenetworkcontrol load4asthenumber of routingcontrolbitstransmittedintheentirenetworkdividedbythenumberof data bits delivered to the destination [22].coding-86dBmcapacity-86dBmcoding-92dBmcapacity-92dBmNumberofMA-RsPacketLossRatio60 50 40 30 20 10 0100101102103104Fig. 3. PLR versus the number of MA-Rs.coding-86dBmcapacity-86dBmcoding-92dBmcapacity-92dBmNumberofMA-RsReciprocalofNetworkControlLoad60 50 40 30 20 10 00.60.50.40.30.20.10Fig. 4. Reciprocal of the network control load versus the number of MA-Rs.coding-86dBmcapacity-86dBmcoding-92dBmcapacity-92dBmNumberofMA-RsEnergyConsumption(J/bit/node)60 50 40 30 20 10 02e-091.5e-091e-095e-100Fig. 5. Energy consumption versus the number of MA-Rs.As seen from Fig. 3, Fig. 4 and Fig. 5, the PLR, the networkcontrol loadandtheenergyconsumptionoftheIrCC-URC-STTC scheme and of the benchmark scheme decrease upon in-creasing the number of MA-Rs. As mentioned in Section II-A,3IntheIrCC-URC-STTCscenarioweassumethat thenetworkconsistsof the QPSK-assisted IrCC-URC-STTC aided MA-Rs and the8PSK-assistedIrCC-URC aided SA-Rs.4In order to avoid the appearance of an innite network control load, whennodataarecorrectlyreceivedinthenetworklayer, thereciprocal of thenetwork control load is used instead of the control load itself.5a lowFERand a relatively high transmission range i.e.coverage area may be ensured by using the IrCC-URC-STTCschemeadvocated. Furthermore, asjustiedinSectionII-B,the specic routes having the lowest number of hops tend to beactivated in the MA-Rs aided network considered. Therefore,as expected, thePLRdecreases, whenthenumber of MA-Rs increases. Moreover, havinga highphysical-layer FERresultsinanincreasednumberofretransmissionsandhencemay trigger route re-discovery, which results in more controlpackets being transmitted. Hence, more energy per payload bitisrequiredforsuccessfullydeliveringthesourcedatatothedestination, as demonstrated in Fig. 5.B. Energy Consumption versus Node DensityAssuming that the number of MA-Rs rma, the mobile speedv , theBPSthroughput BPSSAandthenumber of bits Lreceived in the application layer of the destination are constant,Eq. (11) can be simplied to:ENT= F(n, h). (13)Four scenarios are considered in order to study the re-lationshipbetweenthenodedensityandtheoverall energyconsumptionof theentirenetwork, whereall thenodesarestationary and randomly located in a 500m500m eld. Thesource and the destination are located in the position (0, 0) and(499, 499), respectively. Weincreasethenodedensityfrom = 80to = 400instepsof80. Thereceiverssensitivityis set to 85 dBm and the thermal noise is set to 92 dBm.Theenergyconsumption, thePLRandthenetworkcontrolload are quantied for both the IrCC-URC-STTC scheme andthe benchmarker based on the DCMC capacity when rma = 0,5, 10 and all nodes are equipped with two antennas in the adhoc network considered.rma=all,codingrma=all,capacityrma= 10,codingrma= 10,capacityrma= 5,codingrma= 5,capacityrma= 0,codingrma= 0,capacityNodeDensity(nodes/km2)PacketLossRatio400 350 300 250 200 150 100100101102103104Fig. 6. PLR versus node density.Fig. 6showsthat thePLRdecreasesuponincreasingthenode density. A high node density implies having short averagedistances between the pairs of nodes, hence a low PLR wouldbe achieved as a benet of having a high receive SNR.However, ahighnodedensityisalsoexpectedtoimposeahigh interference on each node, which is expected to reduce theSignal-to-InterferenceRatio(SIR). Toexplicitlycharacterizetheunderlyingtrends, Fig. 7showsthatthenetworkcontrolloadisreducedfor thecapacity-basedbechmarker - i.e. itsreciprocalisincreased-uponincreasingthenodedensityto=160, sincethebenetsoftheincreasedreceivedsignalpoweroutweighthedetrimentaleffectsofthedegradedSIR.rma=all,codingrma=all,capacityrma= 10,codingrma= 10,capacityrma= 5,codingrma= 5,capacityrma= 0,codingrma= 0,capacityNodeDensity(nodes/km2)ReciprocalofNetworkControlLoad400 350 300 250 200 150 1000.80.70.60.50.40.30.20.10Fig. 7. Reciprocal of the network control load versus node density.rma=all,codingrma=all,capacityrma= 10,codingrma= 10,capacityrma= 5,codingrma= 5,capacityrma= 0,codingrma= 0,capacityNodeDensity(nodes/km2)EnergyConsumption(J/bit/node)400 350 300 250 200 150 1002e-091.5e-091e-095e-100Fig. 8. Energy consumption versus node density.This trend was reversed above =160 for the DCMC-capacity-based benchmarker. By contrast, for the proposedIrCC-URC-STTC scheme, we observe that in the presence ofrma = 0 or 5 MA-Rs the network control load tends to gentlydecrease uponincreasingthe overall node density. Finally,theoverall energyconsumptionexpressedinJoule/bit/nodeis portrayed in Fig. 8. The benchmarker based on the DCMCcapacity could be deemed to represent the so-called perfectcodingscheme, whichis capable of communicatingat thecapacity limit. Therefore, it may be readily seen from Fig. 8thatnear-capacitycodingschemesbenetmoresubstantiallyin high-node-density scenarios, when aiming for a high energyefciency.IV. CONCLUSIONSA near-capacity three-stage concatenated IrCC-URC-STTCrelay-transceiver equippedwithtwoantennas was proposedfor theadhocnetworkconsidered, sinceit wascapableofachieving a low FER at a low transmit power. The high effec-tive transmission range of the IrCC-URC-STTC aided MA-Rsallows us to use cross-layer optimization for activating routeshaving the lowest number of longer hops. The simulationresultsconrmthat theenergyconsumptionperbit andpernode in the MA-R aided network becomes lower than that intheequivalentSA-Raidednetwork, especiallyinhigh-node-density scenarios.REFERENCES[1] M. C. VuranandI. F. Akyildiz, Errorcontrolinwirelesssensornet-works: A cross layer analysis, IEEE/ACM Transactions on Networking,vol. 17, no. 4, pp. 1186 1199, Aug. 2009.6[2] H. V. Nguyen, S. X. Ng, and L. Hanzo, Distributed three-stageconcatenatedirregular convolutional, unity-rateandspace-timetrelliscoding for single-antenna aided cooperative communications, in Proc.72ndIEEEonVehicularTechnologyConferenceVTC2010-Fall,Sept.2010.[3] J. G. Li, D. Cordes, andJ. Y. Zhang, Power-awareroutingprotocolsin ad hoc wireless networks, IEEE Wireless Communications, vol. 12,no. 6, pp. 69 81, Dec. 2005.[4] G. Ferrari, S. A. Malvassori, andO. K. Tonguz, Onphysical layer-oriented routing with power control in ad hoc wireless networks, IETCommunications, vol. 2, no. 2, pp. 306 319, Feb. 2008.[5] J. C. Fricke, M. M. Butt, andP. A. Hoeher, Quality-orientedadap-tiveforwardingforwirelessrelaying,IEEECommunicationsLetters,vol. 12, no. 3, pp. 200 202, March 2008.[6] M. Haenggi and D. Puccinelli, Routing in ad hoc networks: a case forlong hops, IEEE Communications Magazine, vol. 43, no. 10, pp. 93 101, Oct. 2005.[7] M. Sikora, J. N. Laneman, M. Haenggi, D. J. Costello, and T. E. Fuja,Bandwidth- andpower-efcient routinginlinear wirelessnetworks,IEEETransactionsonInformationTheory, vol. 52, no. 6, pp. 26242633, June 2006.[8] C. Bae and W. E. Stark, End-to-end energy and bandwidth tradeoff inmultihop wireless networks, IEEE Transactions on Information Theory,vol. 55, no. 9, pp. 4051 4066, Sept. 2009.[9] L. Hanzo, S. X. Ng, T. Keller, andW. Webb, QuadratureAmplitudeModulation: FromBasicstoAdaptiveTrellis-Coded, Turbo-Equalisedand Space-Time Coded OFDM, CDMA and MC-CDMA Systems. JohnWiley and Sons, Ltd, 2004.[10] M. Tucher and J. Hagenauer, Exit charts of irregular codes, in Proc.36thAnnual ConferenceonInformationSciences andSystems 2002,March 2002.[11] M. Tuchler, Design of serially concatenated systems depending on theblocklength,IEEETransactionsonCommunications, vol. 52, no. 2,pp. 209 218, Feb. 2004.[12] S. X. Ng, J. Wang, M. X. Tao, L. L. Yang, and L. Hanzo, Iter-atively decoded variable length space-time coded modulation: Codeconstruction and convergence analysis, IEEE Transactions on WirelessCommunications, vol. 6, no. 5, pp. 1953 1963, May. 2007.[13] L.Hanzo, O.Alamri, M.E.Hajjar,andN. Wu,Near-CapacityMulti-Functional MIMOSystems. NewYork, USA: JohnWileyandSons,2009.[14] S. X. NgandL. Hanzo, OntheMIMOchannel capacityof multi-dimensional signal sets,IEEETransactionsonVehicularTechnology,vol. 55, no. 2, pp. 528 536, March 2006.[15] A. Goldsmith, Wireless Communications. Cambridge University Press,2005.[16] Dynamic MANET On-demand (DYMO) Routing draft-ietf-manet-dymo-21, http://tools.ietf.org/html/draft-ietf-manet-dymo.[17] Ad hoc On-Demand Distance Vector (AODV) Routing,http://tools.ietf.org/html/rfc3561.[18] The Dynamic Source Routing Protocol (DSR) for Mobile Ad HocNetworks for IPv4, http://tools.ietf.org/html/rfc4728.[19] User Datagram Protocol, http://tools.ietf.org/html/rfc768.[20] Corrigenda to 802.11b-1999 Higher Speed Phys-ical Layer Extension in the 2.4 GHz band,http://standards.ieee.org/getieee802/download/802.11b1999Cor1-2001.pdf.[21] J. Zuo, S. X. Ng, andL. Hanzo, Fuzzylogicaideddynamicsourcerouting in cross-layer operation assisted ad hoc networks, in Proc. 72ndIEEE on Vehicular Technology Conference VTC 2010-Fall, Sept. 2010.[22] C. E. Perkins, E. M. Royer, S. R. Das, and M. K. Marina, Performancecomparison of two on-demand routing protocols for ad hoc networks,IEEE Personal Communications, vol. 8, no. 1, pp. 16 28, Feb. 2001.