Gaussian Approximation for LDPC Codes

Today, LDPC codes have become one of the hottest topics in coding theory. Unlike many other classes of cods LDPC codes are already equipped with very fast (probabilistic) encoding and decoding algorithm. The decoding of LDPC codes are implemented through message-passing algorithm. Passed messages can be modeled as Gaussian variables. In this paper, the new approximate method for calculation of statistical parameters of these Gaussian variables is proposed. Based on proposed method, we can be able to estimate the threshold level of noise of regular LDPC codes over AWGN channels.