Cooperative Comm unication and Cognitive Radio （1）

Luo Tao,Hao Jianjun，Yue Guangxin

（Beijing University of Posts and Telecommunications, Beijing 100876, P. R. China）

1 Cooperative Communication

I nmulti-user communication environment,cooperative communication technique enab les the neighboring mobile usersw ith sing le antenna to share theirantennas in some way for cooperative transm ission,which is sim ilar to a d istributed virtual multi-antenna transm ission environment and combines the advantages ofboth diversity technology and relay transm ission technology.As a result,the spatialdiversity gains can be achieved and the system's transm ission performance can be imp roved in a cooperative communication system w ithoutadd ing any antennas.Itcan be app licab le to such networks as cellular mobile communication systems,wireless Ad hoc networks,Wireless LocalArea Networks（WLANs）and w ireless sensor networks.Therefore,it is ofsignificant value for research and w illbe one ofhot top ics thatmay have greatim pacton the developmentof future wireless communications,follow ing multi-carriermodulation technology and multi-antenna technology.

Moreover,being highly flexib le,cooperative communication can be integrated w ith other technologies w ithoutsacrificing their respective advantages.Forexam p le,itcan be integrated w ith Orthogonal Frequency Division Multip lexing（OFDM）technology,making fulluse ofcapability ofcombating frequency selective fading.Itcan be combined

1.1 History and Current Status of Research

▲Figure 1. Relay channel model.

The origin of cooperative communication can be traced back to the work ofCover and ElGamalon the relay channelin 1979.Their relay channelmodelinc ludes a source node,a relay node and a destination node,as shown in Figure 1.Thismodelcan be decom posed into a b roadcastchannel（where the source node A transm its the signals,the relay node B and the destination node C receive the signals）and amultip le access channel（where the source node A transm its the signals,the relay node B retransm its the signals received from node A,and the destination node C receives signals from node A and node B）.

Coverand ElGamal'works demonstrates that the capacity ofa discretememoryless Add itiveWhite Gaussian Noise（AWGN）relay channelis better than thatof the source-destination channel.They develop lowerbounds on channelcapacity via three structurally d ifferent random cod ing schemes:

·Facilitation,in which the relay does notactively help the source,but rather facilitates the source transm ission by inducing as little interference as possib le;

·Cooperation,in which the relay fully decodes the sourcemessage and retransm its,jointly w ith the source,a bin w ith cod ing orspace-time coding scheme to obtain cod ing gain.Integrated w ith cognitive rad io technology,itcan im p rove the spec trum detec tion p robability orgetmore spec trum access chances.This partdiscusses the cooperative communication techniques in terms ofhistory,current research and cooperative schemes.index（in the sense of Slep ian-Wolf cod ing）of the p revious sourcemessage;

·Observation,in which the relay encodes a quantized version of its received signal,using ideas from source cod ing w ith side information.

These random coding schemes are the basicmethods for relay nodes of various cooperative communication systems to p rocess information.

Cooperative communication originates from the relay channelmodel butd iffers significantly from themodelin many aspects.First,cooperative communication technology is app lied in the fad ing channel for the purpose of reducing multipath fad ing,while Cover and ElGamal'relay channelmodel focuses on analyzing the capacity of AWGN channel.Second,the relay node in the relaymodelis only used to help the source node send information,while in cooperative communication,the entire system resources are fixed,each user can actas notonly a relay node to help the source node butalso a source node to transm it its own information.Therefore,they focus on different research fields.Combining the advantages ofd iversity technology and relay transm ission technology,cooperative transm ission enab les a distributed virtual Multip le-InputMultip le-Output（MIMO）system（in cooperative communication,virtualMIMO refers to a transm ission system where severalrelay nodes naturally form a virtualantenna array,and they simulate conventionalMIMO app lication environmentby coordinating and communicating w ith each other,thus achieving jointspace-time cod ing）,overcom ing such lim itations as coherence d istance.Consequently,it can ob tain a transm ission gain app roximating to thatofmulti-antenna and multi-hop transm ission w ithout add ing antennas.Meanwhile,the destination node receives the signals not only d irectly from the source node,but also forwarded by the relay nodes.With sufficientvalid information p rovided,e.g.the rad io link status and the signal quality,the destination node can choose a p ropermethod to combine these signals,thus achieving d iversity gains and greatly im p roving data transm ission rate.

Since Sendonaris etal.p roposed the concep tof cooperative communication in 1998,related research has been sparked and is flourishing worldwide.

Internationally,related sub jects have been in p rogress.Forexam p le,Wireless World Research Forum（WWRF）has estab lished a vision comm ittee to specially study cooperative networks and has pub lished severalwhite papers.On January 1,2004,European Union started a p rojec tnamed WirelessWorld Initiative New Rad io（WINNER）under its Sixth Framework Prog ramme（FP6）,aim ing to develop a ubiquitous w ireless system which excels existing system s in term s of performance,efficiency,coverage and flexibility.In this p roject,relay-based concep ts are involved.In add ition,many well-known internationalperiod icals and conferences have also opened special issues,technicalsymposia orworkshop for reporting cooperative communication technology,inc luding IEEE Communication Magazine,IEEE InternationalConference on Communications（ICC）,IEEEWireless Communications and Networking Conference（WCNC）,and IEEEGlobal Telecommunications Conference（G lob leCom）.In 2006,the academ ic journalSp ringer pub lished a book on cooperative communicationwhich was co-authored bymany researchers.

The labs ofmany universities in the world have started the research,too.For exam p le,the Wireless Communication Group ofCommunication Technology Lab ofSw iss RoyalAcadem y ofScience has started the＂Cooperative MIMO Wireless Network＂p roject,and European Communication Councilhas launched the Information Society Technologies-Resource Management and Advanced Transceiver Algorithm s forMultihop Networks（IST-ROMANTIK）p rojec t.Many p rofessors and researchers also havemade d istinguished contributions to cooperative communication research,inc luding Dr.J.N.Laneman of Massachusetts Institute of Technology,associate p rofessor E.Erkip ofUS Polytechnic University,associate p rofessorM.DohlerofKing's College of the University of London,Dr.T.E.Hunter from the Med ia Communication Lag of University of Texas and Dr.M.O.Hasna ofUniversity ofMinnesota.

In China,cooperative communication technology has attracted w idesp read attentions recently.The NationalNatural Science Foundation ofChina（NSFC）and the Ministry ofEducation ofChina have sponsored many p rojects.Many Chinese universities also have started their research,inc luding Beijing University of Posts and Telecommunications,Zhejiang University,and XiDian University.

1.2 Cooperative Scheme

By cooperative ob ject,cooperative communication is c lassified into two kinds:cooperation among heterogeneous networks and cooperative communication w ithin homogeneous network.

Forhistoricalreasons,many access networks are now coexisting,such as WLAN,Worldw ide Interoperability for Microwave Access（WiMAX）,cellular network,and satellite communication network.Although these networks have theirown advantages and meet the requirements of the users in one or severalaspects regarding data transm ission rate,coverage area and support for term inalmobility,so far,there is notyeta network thatcanmeetall these requirements.In order to satisfy the demands ofvarious users ford ifferent app lications,future communication networksmusthave the capability of integ rating existing networks onto one united information p latform.The ITU,3GPPand 3GPP2 from the perspective of telecom network,and IETF from the perspective of IP packetnetwork describe the future communication networks.Desp ite their differences,these descrip tions have one common point that their solutions are IP-based and adop t cooperation ford ifferentaccess networks so as to p rovide users w ith d iversified circuitand packetswitching services.Therefore,in the future telecom network development,one inevitab le trend is to imp lementcooperation among d ifferent access networks,i.e.cooperation among heterogeneous networks.Obviously,the focus ofcooperation ismobility managementamong heterogeneous networks,mainly inc lud ing inter-network handoverand roam ing.

Cooperative communication w ithin homogeneous networkmeans allnodes work in the same kind ofnetwork.Itcan be realized w ith two cooperation schemes:fixed relaying and user cooperation.The fixed relaying scheme is very sim ilar to the relay channelmodel shown in Figure 1.In this scheme,a fixed relay node is p laced between the source node and the destination node in advance and itis w irelessly connected w ith the source node and the destination node.This relay node does nothave any information to transm it,but forwards the information it receives.Unlike fixed relaying,user cooperation ismore flexib le.The source nodes can actas relay nodes to forward the information of cooperative partners in addition to sending theirown information.As a result,these term inals should have such func tions as signal forward ing and simp le routing.By the ways the relay node p rocesses the source node's information,user cooperation can be furtherd ivided into the follow ing schemes:Am p lify-and-Forward（AF）,Decode-and-Forward（DF）,Coded Cooperation（CC）,Space-Time Coded Cooperation（STCC）and Network Coded Cooperation（NCC）.In the follow ing sec tions,we w illdiscuss the basic p rincip les of these schemes and com pare them roughly.To sim p lify our descrip tion,we only d iscuss the case w ith sing le relay node here.The cases w ithmultip le relay nodes are sim ilar.

▲Figure 2. AF scheme.

1.2.1 AF

AFscheme was firstp roposed by Laneman etal,as shown in Figure 2.

As shown in Figure 2,signal p rocessing in AF scheme can be sim p lified into three phases:In Phase 1,the source node transm its the signals by way ofb roadcasting,while the destination node and the relay node receive the signals.In Phase 2,the relay node am p lifies the powers of the signals received from the source node and forwards them to the destination node.In Phase 3,the destination node combines and decodes the signals received from the source node in Phase 1 and the relay node in Phase 2 so as to restore the original information.AF is also called nonregenerative relaying scheme and it is basically a p rocessing method foranalog signals.Compared w ith other schemes,AF is the sim p lest.Besides,as the destination node can receive independent fad ing signals from the source and relay nodes,fulld iversity gain and good performance can be achieved w ith this scheme.However,AF scheme is p rone to noise p ropagation effectbecause the relay node amp lifies the noise on the source-relay channel when the retransm itted signals are am p lified.

1.2.2 DF

DF scheme was firstp resented by Sendonaris etal.Sim ilarly to AF,signal p rocessing in DF scheme can also be simp lified into three phrases,as shown in Figure 3.

In Phase 1 and Phase 3,DFscheme p rocesses the signals the same way as AF.In Phase 2,the relay node decodes and detects the signals received from the source node before it forwards the signals to the destination node.Hence,DF is also called regenerative relaying scheme.Obviously,DF is essentially a d igitalsignalp rocessing scheme.Although noise p ropagation p rob lem w ill not take p lace,the signalp rocessing in DF largely depends on transm ission performance ofsource-relay channel.If Cyc lic Redundancy Check（CRC）is not im p lemented in cod ing,fulldiversity orders can notbe ob tained.Moreover,the errors broughtby the relay node during signaldemodulation and decoding w illaccumulatew ith the increase ofhops,thus affec ting d iversity advantage and relay performance.All these demonstrate that the transm ission characteristics ofsource-relay channel have g reatim pacton the performance of DF communication system s.

AFand DFaforementioned are often called fixed cooperationmodes because the relay node always participates in cooperative communication nomatter what the channel transm ission characteristics are.As amatterof fact,cooperation does notalways b ring benefits.Forexam p le,in a halfdup lex mode,the data transm ission rate and the utilization of the deg rees of freedom w ill decrease.This ind icates when to cooperate is a criticalissue.To add ress the p rob lem,selectionmodes and incrementalmodes are put forward based on AFand DF,which are listed in Tab le 1.

Figure 3.▶DF scheme.

Selec tionmodes compare transm ission charac teristics of the source-relay channelagainsta p redefined threshold.Only when the characteristic value is g reater than the threshold,cooperative communication is im p lemented;otherw ise,the source node direct transm ission again.Hence,the key in selectionmodes is the cond itions of source-relay channel.In inc remental modes,the feedback of the destination node is used to determ ine whether the direct transm ission is successful.If the data are correctly detected,source node w illsend new data;otherw ise,the relay node w illparticipate in the cooperative communication p rocess.This p rocess is equivalent to adding redundancy mechanism orautomatic detection and retransm issionmechanism in the relay transm ission.Obviously,the key issue involved in inc rementalmodes is the conditions ofsource-destination channel.In fixed orselectionmodes,the relay node has to repeated ly send the information received from the source node,whichmay lead to decreased usage of the deg rees of freedom;while in incrementalmodes,cooperative communication is used only when itis needed,avoid ing repeated transm ission,buta feedback channelis required.

▼Table 1. Three cooperation modes

From the perspec tives of reliability and effec tiveness,IncrementalAF（IAF）perform s best.In terms of the com p lexity ofalgorithm,AF is sim p lestand can achieve fulldiversity gain;DFperforms poorand cannotob tain fulldiversity gain;Selection DF（SDF）can achieve full diversity gain butit ismore comp licated than AF.Analyses show thatboth Selection AF（SAF）and Inc rementalDF（IDF）can notachieve good performance:Selectivemode paysmuch attention to the transm ission characteristics of the source-relay channel,but in AF,the source-relay channeland the relay-destination channelare of the same importance because the relay node only am p lifies,notdecodes,the information received from the source node;the incrementalmode focuses on the source-destination channel,but in DFscheme,errors w illaccumulate and b roadcastw ith information forward ing if serious fad ing takes p lace on the source-relay channeland lots oferrors are resulted from decod ing.Therefore,selectionmode ismore suitab le forDF scheme,while incrementalmode ismore suitab le for AF scheme.

1.2.3 CC

In AFand DF,the relay node always repeated ly forwards the information received from the source node,which often leads to decreased usage of the deg rees of freedom.To solve the p rob lem,Hunteretal.integ rated channel cod ing into cooperative communication and p roposed CC scheme.Signal processing in CC scheme is shown in Figure 4（a）.

In CC scheme,d ifferentsegments of each user's codewords can be sentvia two d ifferent fad ing paths.Each user correc tly decodes the information received from cooperative partners and then re-encodes them before forwarding them.With redundantinformation bits being repeated ly transm itted through d ifferentspaces,the system performance is im p roved.In CC scheme,eachmobile term inalachieves diversity and cod ing gains by re-encod ing and transm itting d ifferent redundantbits,thus the system performance is g reatly enhanced.Moreover,this scheme does not require information feedback between cooperative nodes.When a relay node cannotcorrectly decode the information bits,itautomatically reverts back to non-cooperativemode,ensuring the system efficiency.

Figure 4（b）gives an exam p le ofCC scheme.Themobile term inal first encodes the information bits to be sent by b locks and then adds the CRC codes.During cooperative transm ission,the encoded information is divided into two segments,containing the information bit N1and the add itionalpunctured bit N2（the leng th oforiginalcode word is N1+N2bits）respectively.Apparently,two time slots（i.e.frames）are needed to send N1and N2,respectively.In the first frame,eachmobile term inalsends its own N1information,and at the same time it tries to decode the information bits transm itted in the first frame of its partner.If the partner is correctly decoded,which is verified w ith CRC check,the term inal then sends N2bits of its partner in the second frame.If the term inalcannot correctly decode the partner's information,it transm its its own N2information.In this way,eachmobile term inalalways sends the information b lock of N1+N2bits in two time slots.Finally,the destination node decodes the information b locks it receives.Unlike SDF,CC schememanages to automatically sw itch between cooperative and non-cooperativemodes through code design,regard less of transm ission charac teristics of the source-relay channeld irectly.

▲Figure 4. CC scheme.

Currently,there are alreadymany channelcoding schemes thatare integrated w ith cooperative communication,for instance,convolutionalcode,Turbo code,and Low Density Parity Check（LDPC）code.In case ofslow fad ing,the CC scheme can im p rove the biterror rates of two communicationmobile term inals even if the transm ission characteristics of the channelbetween twomobile term inals are very poor.Besides,if the two cooperative partners can correctly decode each other's information,the system can achieve fulldiversity gain.Butin case of fast fad ing,the CC scheme sacrifices the performance of the term inal whose up link channelis ofgood transm ission characteristics.To solve this p rob lem,STCC is b rought forward.

1.2.4 STCC

STCC is to app ly space-time cod ing into coded cooperation scheme.Themain difference between STCC and CC is that STCC allows eachmobile term inalto simultaneously send data over multi-access channels of its own and of its cooperative partner;while in CC,each mobile term inalcan only send information over its ownmulti-access channel.Many researchers have successively suggested theirown STCC im p lementationmethods but their solutions are almost the same.Figure 5 com pares signalp rocessing ofSTCC,CC and non-cooperation schemes.

In Figure 5,axis-X and axis-Y rep resent time and frequency,respec tively,and Frequency Division Multip le Access（FDMA）method is adop ted.In STCC and CC schemes,the time originally taken in non-cooperation scheme is d ivided into two time slots,i.e.Phase 1 and Phase 2.As shown in Figure 5,STCC and CC work in the same way in Phase 1,and their difference lies in Phase 2.For the sake ofdescrip tion,let's take the exam p le for User2 to exp lain the work p rocess of STCC.In phase 1 ofSTCC scheme,User1 and User2 use Channel1 and Channel2 respectively to send theirsource node information to their cooperative partners and the destination node in a b roadcasting way.In phase 2,ifUser2 decodes the information sentby User1 correctly,itsends its partner's（i.e.User1）information and its own information over Channel1 and Channel 2 respectively to the destination node;otherw ise,ituses both channels to send its own information.Comparatively,if the information of the cooperative partner is successfully decoded in Phase 2,each useronly sends the decoded information over its own channelin CC scheme,but in STCC scheme,each usersends its own information as wellas the information of its partner.Stud ies show thatSTCC scheme can achieve fulldiversity gain even in a fast fad ing environmentw ithout sacrificing the performance of themobile term inalw ith better channelquality.

▲Figure 5. Comparison of signal processing in STCC, CC and non-cooperation scheme.

Currently,many space-time cooperation schemes have been developed by integ rating different space-time cod ing w ith cooperative communication technology,inc lud ing Space Time Block Coding（STBC）and Space Time Trellis Cod ing（STTC）.Among them,STBC has attracted special attentions due to its sim p le design.

1.2.5NCC

NCC is formed by incorporating network coding into CC scheme.

Network cod ing is amulti-cast technology.The core idea ofnetwork cod ing is thatan intermed iate node no longerperforms simp le store-and-forward function butencodes and forwards the received information,thus im p roving the capacity and robustness of thewhole network.The network cod ing concep twas originally used forw ired networks,so ismostof current related work.But the b roadcast characteristics of rad io channels are suitab le forapp lication ofnetwork coding in wireless networks,and the information interac tion between w ireless nodes can be fully achieved via network cod ing.As a result,the combination ofnetwork cod ing and cooperative communication can effec tively imp rove the performance ofw ireless communication system s.

CurrentNCC research focuses on the relay node's network coding schemes and basic communicationmethods.By the network cod ing schemes used by the relay node,NCC can be d ivided into two categories:linearand non-linear;by basic communicationmethods adop ted,NCC can be d ivided into fixed relaying,opportunistic relaying,recip rocal relaying,and bi-directionalrelaying.The signalp rocessing in NCC are illustrated in Figure 6.

Figure 6（a）illustrates the fixed relaying scheme,where R acts as a fixed relay node ofUser A and User B.The relay node itselfhas no data to transm it butperforms network cod ing on the data it receives from User A and User B.Figure 6（b）shows recip rocal relaying scheme,which is also called user cooperationmode.In this scheme,User A and User B actas cooperative partners ofeach other.Hence,in addition to transm itting theirown data,they perform network cod ing on theirown data and their partner's data.Figure 6（c）illustrates opportunistic relaying,where s1sends data to d1,and s2sends data to d2.The relay node R1does notparticipate in relay transm ission only when an error occurs in the d irec ts1-d1or s2-d2transm ission link.During its participation,R1perform s network cod ing.Figure 6（d）is bi-d irectionalrelaying,where User A and User B communicate w ith each other and there aremany relay nodes between them.The relay nodes perform network coding on the data exchanged between User A and User B to imp rove the system throughput.

▲Figure 6. NCC scheme.

◀Figure 7.BER performances of different cooperation schemes.

Figure 7 com pares the Block Error Rate（BER）ofseveralcooperative schemes.For the sake ofcom parison,the case ofno cooperative is also illustrated.Itcan be easily seen from Figure 7 that the BER performances of cooperative communication system s are better than thatofnon-cooperative system.Among the above-d iscussed cooperative schemes,performance of CC,NCC and STCC are often better than thatof AFand DF,but they involve comp licated algorithms as wellas d iverse cod ing technologies.As a result,the signalp rocessing time and delay at the relay node is increasing,which is not good for the developmentofmodern w ireless communication systems.Therefore,w ith all factors being considered,AFand DFare regarded more p racticalthan the other schemes.

（To be continued）