An iterative algorithm for calculating posterior probability and model representation

Abstract

We introduce a representation method of probability models that can be applied to any code such as turbo, low density parity check (LDPC) or tail-biting code. Moreover, we propose an iterative algorithm that calculates marginal posterior probabilities on the introduced probability model class. The decoding error probability for the LDPC codes of the proposed algorithm is less than that of the sum-product algorithm.