Research On Performance And Construction Of Low-Density Parity-Check Codes

Low-density parity-check (LDPC) code is a type of error-correcting code which is now widely used in various kinds of communication systems, such as the European DVB-S2 system and China's Digital TV terrestrial broadcasting system (CDTV-T). Construction of LDPC codes with high performance becomes a research hotspot. Based on density evolution theory which determines the performance of a code under certain degree distribution, the optimal degree distribution pairs can be found which makes it possible to construct good codes. When constructing a parity-check matrix, PEG can maximize the girth length, thus lowering error-floor, while quasi-cyclic structure bears other advantages. With the above analyses, LDPC codes with good performance and low implementation complexity can be constructed.This dissertation first gives a deep study on LDPC decoding algorithms including BP, MSA, and then a comparison between these different decoding algorithms is shown. On the basis of density evolution, a degree optimization algorithm under fixed code rate is presented which then proves to be able to find better degree distribution pairs than differential evolution.Afterwards a method of constructing LDPC codes with good performance and low implementation complexity is proposed. Through degree optimization and PEG, the code performance is increased and error-floor is lowered, while one the other hand, cyclic unit structure brings great advantage in implementation. A simulation which compares the codes designed this way with the codes adopted in CDTV-T then shows that the construction method in this paper is superior to CDTV-T in terms of both better performance and lower error-floor.