Research On Decoding Convergence Rate Of Ldpc Codes

LDPC (Low-Density Parity-Check) codes are a kind of channel codes limiting to Shannon limits, which has a very low decoding complexity. With the growing maturity of theoretical research, the encoding complexity of LDPC codes has been improved a lot. This makes it used more widely in the wireless communications, deep space communications, optical fiber communications, media storage and other domains.Base on BP (Belief Propagation) decoding algorithm, several improved decoding algorithms are given in this thesis. Main functions of this thesis include:(1) The basic principle of LDPC codes is illuminated. The principles of coding and decoding for low-density parity-check codes are briefly summarized.(2) Parallel BP decoding algorithms in probability domain and LLR domains are given. By serializing the information process, serial BP decoding algorithm is obtained. Simulation results show that the serial BP decoding algorithm has been improved in terms of convergence rate and BER performance.(3) Instead of processing the information statically, a method of dynamically choosing the next node whose information is going to be propagated is given, which is called RBP (Residual Belief Propagation), results show a faster convergence rate is obtained in case of increasing some complexity.