Research On Iterative LDPC Decoding Algorithms

Low-density parity-check(LDPC) codes is a class of linear block code which could provide near Shannon-Limit performance as well as low error-floor. For some long LDPC codes, their performances are even better then the Turbo code. Iterative decoding techniques have been receiving more and more attentions with the invention of turbo codes and the rediscovery of LDPC codes. Although the optimum decoding algorithms could provide very good performance for LDPC codes, complicated operations are involved in the optimum decoding, and prohibit the wide applications of LDPC codes in the next generation digital communication and storage system.In this paper, we reviewed the fundamental knowledge and rising history of the channel coding. We descript the characterization of LDPC code, showed how to express it as a Tanner Graph and discussed how to make a good LDPC code from many aspects. Then, we expounded two families of iterative decoding algorithm for LDPC codes, one is bit-flipping algorithm which enables very low complexity in decoding, the other is belief propagation algorithm which could reach very good performance. We concretely introduced BF, WBF, IWBF, SPA, Log-SPA and analyzed them by simulating their performance. At last, we introduced several lately and better way of LDPC decoding algorithms including MIWBF, RRWBF, MMS, FMS, OMS, and proposed several new algorithms like PIWBF, FBF and LMMS, which could enable less iteration or reach higher performance while still keep low complexity.