Simulation Research Of Encoding And Decoding Algorithms Of Quasi-Cyclic LDPC Codes Based On DMB-TH System

Low-density-parity-check codes (LDPC codes) which was put forward by Gallager in the early 1962, silent for many years, inspired by Turbo codes ,it became a hot communication technology again.LDPC codes is a type of linear block codes with very sparse check matrix, and it has wonderful features: capability of approaching the Shannon limit, convenient analyse and research in theory, simple decoding, ability of parallel performance and suitable for implementation with hardware. Quasi-Cyclic LDPC (QC-LDPC) codes is an important type of LDPC codes. In practical application, QC-LDPC codes is a superior channel coding because of its lower complexity and lower error-floor.DMB-TH system with independent IP (Information Property) is independently presented by Tsinghua University. Channel coding occupies very special position in many key technologies of DMB-TH system. In DMB-TH system, the LDPC-BCH concatenated codes were used in FEC module, LDPC as inner codes, BCH as outer codes.The basic theories of transmission module and FEC module of DMB-TH systenu the features of QC-LDPC codes,the principles of encoding and decoding algorithms of QC-LDPC codes on graphs were summarized in this thesis. Simulations of LLR BP,Min-Sum and modified Min-Sum algorithms over AWGN channel were presented to compare the performance and complexity of decoding algorithms of three rates of QC-LDPC codes for DMB-TH system. In order to improve the performance of QC-LDPC codes while reducing complexity further, layered decoding algorithm was used in DMB-TH system, then layed decoding algorithm was combied with modified Min-Sum algorithm to simulate its performance. The result showed that layered modified Min-Sum decoding algorithm did not impove noise threshold but it could reduce the decoding iteration by half.