Theory Research And Performance Analysis Of LDPC Codes

LDPC codes are progressive good codes and very close to the Shannon limit performance, its performance is better than Turbo codes especially under the conditions of long codes. It is an iterative good decoding algorithm based on belief propagation, and the decoding complexity is linear to code length. LDPC codes is widely attracted by academia and IT industry based on a number of advantages of LDPC codes, it became another research focus areas in channel coding, and the application of LDPC codes was put on the agenda.In this paper, the following tasks were done:1. The expression and classification of LDPC codes are introduced, several kinds algorithms for the construction of the check matrix of LDPC codes are concluded, and a new algorithm without short cycles for the construction of check matrix is researched .2. Decoding algorithms are analyzed from the terms of hard-decision and soft-decision. BP decoding algorithm, minimum sum algorithm and improved minimum sum algorithm are studied.3. According to performance simulation results of improved minimum sum algorithm and BP algorithm, the performance of the prior is approximate to the latter and the complexity of the prior deceased comparing to the latter. The decoding performance is better while the code rate is lessen under different length of codes based on improved minimum sum algorithm, and the frequent numbers desceased. The decoding performance under the AWGN channel is better than the Rayleigh channel. The coding performance of construction algorithm without short cycles is better than random codes.