A study of decoding algorithm of low density parity check code

摘要

The maximum a posteriori probability decoding of Turbo Code and Low Density Parity Check Code is one of the most active subjects in research of error correcting code. Sum-Product(SP) algorithm is used for those, but the performance of SP becomes bad when a small loop of length 4 is in the Bayesian Network An algorithm proposed by Matsushima on the probabilistic model equal to Junction Graph does not have such a fault. In this paper we propose a method of construction of (Normalized) Junction Graph using the algorithm for decoding of LDPC code, and compare the performance of it and SP.