Research On Distributed Joint Source-Channel Coding Using Protograph LDPC Codes

With the driving of the Internet of Things and Mobile Internet,distributed joint source-channel coding schemes have broad application prospects.This dissertation focuses on the problem of distributed joint source-channel coding for memoryless correlated sources and correlated Markov sources.For the memoryless correlated sources,this dissertation firstly describes the distributed joint source-channel coding scheme and the corresponding encoding and decoding algorithms.Then,by analyzing the probability density distribution of the log-likelihood information of the variable nodes in Tanner Graph,the finite-length EXIT algorithm is used to analyze the convergence behavior of the code pattern for this distributed joint source-channel coding scheme.Finally,based on the EXIT analysis and simulation results,good codes can be obtained.The simulation results show that for the AWGN channel communication,channel codes with excellent performance in channel coding have better performance than other code types in the DJSC coding scheme of this paper.For the memory-correlated source,this dissertation mainly studies the distributed joint source-channel coding of two correlated Markov sources.In order to exploit the time correlation of Markov sources,a new joint source-channel decoding algorithm where a sum-product decoder is used to cascade a BCJR decoder is proposed.Moreover,to further improve the BER performance,an improved scheme is presented in considerable detail.For the proposed improved scheme,considering that the time correlation is disturbed by noise for the second distributed Markov source,a new correlated source model is designed to preserve its Markov attributes as much as possible.Thus,the decoding performance is improved for both the correlated sources due to more accurate information transferred at decoding.In addition,in order to further tap the gain brought about by the space correlation between the sources,global iterations are introduced to dynamically update the space correlation between sources.Finally,the simulation results show that compared with the reference group,the proposed improved scheme has more 1.7dB gain at BER 10-6 orders of magnitude for two sources.