Research On Decoding Algorithm Of LDPC Codes Based On Difference-Map And Belief Biasing

As a powerful class of error-correcting codes based on graphical models anditerative decoding algorithms, low-density parity-check (LDPC) codes have recentlyreceived much attention due to their capacity approaching performance and relativelylow decoding complexity. With the development of the theory research and theexperiment, LDPC codes have been used in deep space communication, optical networkand digital storage.In this paper, based on the research of decoding algorithm, an advanced algorithmhas been implemented in this thesis as well a combination of the difference-mapsmethod, post-processing method and the belief propagation algorithm to improve theperformance of decoding. Some results are obtained and summarized as follow:1. The basic principles of LDPC codes are introduced, and the Tanner graphicalmodel for LDPC codes is provided. The belief propagation decoding algorithm isanalyzed in detail. Error floor is studied based on some properties of LDPC codes, suchas cycles and connectivity, and the harmfulness of cycles is explained.2. Principle of DC algorithm and the feasibility of its applicaition on LDPC codesare thoroughly analyzed. This paper combines the difference map of DC algorithm withLDPC codes, Difference Map belief propagation algorithm (DMBP) is given, and theemulation results show the effectiveness of this method.3. A method to eliminate (8,8) absorbing sets in LDPC codes is proposed in thispaper. This algorithm can destroy the combined structure of absorbing sets in LDPCcode by biasing some selective nodes. This method can effectively eliminates theinfluence of (8,8) absorbing set in LDPC code and reduces the error floor of LDPCcode. The simulation results show that only need a small amount of the cost caneffectively improve the decoding performance.