Research On Asymmetric Distributed Source Coding

Distributed source coding compresses each independent source at the encoderand decodes jointly by the correlation between the sources at the decoder.Distributed source coding has many applications, which range from the wirelesssensor network technology to video image processing technology. This dissertationbases on the theory of distributed source coding and focuses to various statisticalcorrelation characteristics and the actual noisy channel conditions. Asymmetricdistributed source coding will be further researched and analyzed.This dissertation first presents the rate-adaptive Low-Density Parity-Check(LDPC) codes to solve the various statistical correlation characteristics between thecorrelation sources in asymmetric distributed source coding system. Thisdissertation presents and constructs the rate-adaptive LDPC code sequence, enableto change the coding rate flexibility for the various statistical correlationcharacteristics between sources. The simulation indicates that the rate-adaptiveLDPC codes can approach to the Slepian-Wolf bound within the entire compressionregion. Besides, the well-designed rate-adaptive LDPC codes are superior to therate-adaptive Turbo codes which have been widely used in asymmetric distributedsource coding system.Secondly, a novel rate-adaptive asymmetric distributed source coding schemebased on LDPC codes is preserved for high compression rate. In low compressionrates region, we continue with the traditional rate-adaptive LDPC codes. However,in high compression rates region, the original rate-adaptive LDPC codes arereconstructed by selectively reducing the proportion of variable nodes withmaximum degree and by redesigning the degree distribution using density evolutionalgorithm. Simulation result shows that the proposed asymmetric distributed sourcecoding scheme is superior at high compression rate.Finally, asymmetric distributed source coding under actual noisy channelconditions is studied and analyzed. While the actual noisy transmission channel isconsidered, asymmetric distributed source coding consists of two channels, i.e. theactual noisy channel and the virtual correlation channel, which is called distributedjoint source-channel coding (DJSCC). This dissertation based on distributed sourcecoding using syndrome (DISCUS) Slepian-Wolf coding scheme to obtain effectivedistributed source coding compression and error protection. At the encoder,rate-adaptive LDPC codes are still used. At the decoder, a novel check matrix basedon the decoding matrix is redesigned to adapt to the various channel conditions and the statistical correlation characteristics. Simulation result shows that the proposeddecoding method is well-designed in the distributed joint source-channel codingscenario.