Analysis of LDPC decoding for correlated and uncorrelated block fading channels

Abstract

This paper presents a density evolution analysis of the sum-product algorithm used for channel estimation and decoding of low density parity check (LDPC) codes on correlated and uncorrelated two-state block fading channels. The channels under consideration use binary symmetric channels and binary-input Gaussian channels as components, and the thresholds for regular LDPC codes on these channels are calculated. The analysis shows that for both correlated and uncorrelated block fading channels, the threshold increases with the memory length. If the memory length is fixed, introducing correlation between successive blocks increases the threshold; as the memory length increases, this effect diminishes