Research On Applications Of LDPC Codes In Storage And Wireless Communication Systems

This paper mainly studies the application of LDPC codes in storage andwireless communication systems, as two main fields where LDPC codes areused, their system requirements are being changed because of the developmentof science and technology, thus this study has some practical value.For the problem of fading in wireless communication systems, this papermainly studies a new coded modulation system using LDPC. Based on theresearch of the Tanner graph of LDPC coded modulacion systems, in this paper,a new mapping scheme is given, which change the one bit node to one singlesymbol node mapping scheme into a one bit node to multi symbol node scheme,and through iterative decoding and demapping, more mutul information willexchange between bit nodes, check nodes and symbol nodes, thus greatlyimprove the error correcting performance in Rayleigh fading channels. Besidesthis, in order to break the girth four cycles, which may greatly influence theperformance, between check node and symbol node in the Tanner graph, adesigning rule for the interleaver is also given. Comparing to the LDPC basedbit interleaved coded modulation with interative decoding and demappingscheme, the new system has a performance gain of about0.75dB in Rayleighfading channels with256QAM modulation, while the complexity increase isvery little.For the application of LDPC code in storage systems, this paper mainlystudies a LDPC coded magnetic recording system. In the paper, a briefintroduction about magnetic recording channels, its equalization and channeldetecting algorithms (BCJR and SOVA) is given. Because of the partial response characteristic of magnetic recording channels and the increase of recordingdensity, the inter-symbol interference and burst error phenomenon becomesmore serious. Since nonbinary LDPC codes own the ability of correcting bursterrors, so nonbinary LDPC code is used in the magnetic recording system.Based on the drawback of nonbinary LDPC codes’ interative decoding algorithm,two improving decoding scheme is given. One is to use punching; the other ispartial concatenation of RS code. Since the improved systems partially solvedthe problem, that is, some symbol nodes of the nonbinary LDPC code arealways being in error at high SNR, thus greatly improve the system performance.Between them, the partial concatenation of RS code scheme performs better,judging from the error floor point of view, this scheme reduces it by at least anorder of magnitude.