Efficient Iterative Decoding of LDPC in the Presence of Strong Phase Noise

TL;DR

This paper introduces an efficient message passing algorithm for LDPC decoding over channels with strong phase noise, utilizing an improved canonical model and clustering via directional statistics to enhance performance.

Contribution

It presents a novel approximate Bayesian inference method that incorporates phase slip handling, improving decoding accuracy and complexity over existing algorithms.

Findings

Superior performance demonstrated through simulations

Effective handling of phase slips in decoding process

Reduced computational complexity compared to state-of-the-art methods

Abstract

In this paper we propose a new efficient message passing algorithm for decoding LDPC transmitted over a channel with strong phase noise. The algorithm performs approximate bayesian inference on a factor graph representation of the channel and code joint posterior. The approximate inference is based on an improved canonical model for the messages of the Sum & Product Algorithm, and a method for clustering the messages using the directional statistics framework. The proposed canonical model includes treatment for phase slips which can limit the performance of tracking algorithms. We show simulation results and complexity analysis for the proposed algorithm demonstrating its superiority over some of the current state of the art algorithms.