LDPC Based Cooperative Codes Design In Decode-and-Fonvard Half-duplex Relay Channels

In cooperative communication systems, the idle antennas in the network can beused as relays to assist the communication between the source nodes and the destination,which overcomes the fading effects with spatial diversity. Compared with directlyforwarding the information of the source nodes, cooperative coding has higher spectralefficiency. At present, the study on coding, decoding, and the parity check matrixconstruction for point-to-point channels is quiet mature. As the characteristics of relaychannels are different to that of point-to-point channels, LDPC codes designed bytraditional methods for the point-to-point channels is not suitable for the relay channels.Considering the practical problems confronted in cooperative coding technology, thisdissertation studied the optimizing methods of design LDPC codes for cooperativecommunication systems, which is based on puncturing and Tanner graph extensionunder DF and SDF protocols. Our researching work is mainly focused on theoptimization of the parity check matrix structure, the convergence condition and degreedistribution of the cooperative LDPC codes. The arrangements of this dissertation are asfollows:1. After an explicit analysis on the features of punctured LDPC codes andcooperative coding method, we point out the problems of puncuring. To improve thecooperative coding gain and reduce decoding complexity, the cooperative LDPC codeson puncturing is optimized by two steps. In the first step, the parity check matrix of themother code is optimized to improve the coding gain of the cooperative coding systems.To improve the asymptotic performance of the mother codes, the degree distribution ofthe mother codes is optimized under the rate-compatible constrains. Based on therecovery-tree of punctured nodes, a new structure of the parity-check matrix to themother codes is devised to facility the puncturing, which will improve the errorperformance of the punctured codes from the source nodes. In the second step, amodified BP algorithm is proposed to accelerate the decoding convergence of the ratecompatible puncturing LDPC codes by a sequential scheduling of the puncturedvariable nodes. The variable nodes with high reliability are priority processed whiledecoding, which will accelerate the speed of convergence. When decoding with lowiterations, this method outperforms both the BP decoding algorithm and randomshuffled BP decoding algorithm, especially when the relay is close to the destinationnode, the performance improvement is more obvious.2. To avoid the performance losses by puncturing, an optimised extension of parity-check matrix of the LDPC(OEPCMC-LDPC)codes is presented for cooperativecoding. Considering the relationship between the source LDPC codes and theOEPCMC-LDPC codes, the structure of parity-check matrix to the OEPCMC-LDPCcodes and its corresponding tanner graph are derived. Combining two different SNR ofthe source-destination channels and relay-destination channel, we presented a mehod toanalysis the decoding convergence conditions of OEPCMC-LDPC codes. Finally,considering the relays transmitting the information bits of themselves or not, two kindsof cooperative LDPC codes are optimised based on dynamic and fixed time slotallocation, which are called DTSA_LDPC and FTSA_LDPC. The simulation resultsverify the validity of both methods.3. In order to approach the theoretical limit of decode-and-forward strategy forhalf-duplex multiple access relay channel, a parity jointly coded multi-edge type LDPC(PJCMET-LDPC) coding structure is proposed. Firstly, we presented a tanner graph ofLDPC codes with multi-edge structure, which makes the bits from multi-sources and therelay as parts of the PJCMET-LDPC codes. The overall coding gain of PJCMET-LDPCcodes is improved when the joint structure is used for decoding in the destination.Considering the channel states that different parts of the PJCMET-LDPC codes frommulti-source nodes and the relay node experienced, we derive the decoding convergenceconditions of the PJCMET-LDPC codes. The degree distribution of the PJCMET-LDPCcodes is optimised using differential evolution algrithm, which improves the asymptoticperformance of PJCMET-LDPC codes. The simulation results show thatPJCMET-LDPC codes always outperform the independently cooperative codes.4. To improve cooperative coding gain and the reliability of soft messages forwardby the relay node, a new type of cooperative LDPC codes under softdecode-and-forward (SDF) protocol is proposed by modeling the soft messages as theoutput of the fast fading channel. Firstly, considering the feature of rate-compatiblebeween the source codes and cooperative LDPC code, we construct the parity-checkmatrix with low triangle structure. Then, reliability of the soft information is improvedby BP iterative decoding algorithm, soft parity bits is derived by simple substitutealgorithm. Finally, in order to improve reliability of the destination LLR (log-likelihoodratio) which receive signal from source and the relay in MAC (multiple-access) mode,the soft information is modeled as the output signal of the fast fading channel using theparity bits as input. The transfer probability of this channel is obtained by usinghistogram statistics method. An accurate calculation of the destination LLR message isobtained under this model. Further more, we conclude that the influence of errormessage forward by the relay node on LLR of the destination message under MAC mode can be controlled by power controlled on the relay nodes. The simulation resultsverify the validity of the method.