Hybrid window decoding for joint source channel anytime coding system

Abstract

Joint source channel anytime coding (JSCAC) is a kind of joint source channel coding (JSCC) systems based on the causal spatially coupled coding and joint expanding window decoding (JEWD) techniques. JSCAC demonstrates greatly improved error correction performance, as well as higher decoding complexity. This work proposes a joint hybrid window decoding (JHWD) algorithm for JSCAC systems, aiming to reduce the decoding complexity while maintaining comparable error correction performance with the state of the art. Unlike the traditional JEWD technique and its variants, the proposed JHWD algorithm utilizes a hybrid window structure. It achieves this by implementing adaptive local expanding window decoding in the sliding window region, guided by syndrome-check-based detection. The hybrid window decoding characteristics of the proposed JHWD algorithm can both effectively reduce the decoding complexity caused by window expanding, and also mitigate the error propagation caused by window sliding. An improved density evolution algorithm is proposed for the asymptotic performance analysis of the proposed JHWD algorithm. Both the analytical and simulation results indicate that the proposed JHWD decoding scheme is a good low-complexity implementation option for JSCAC systems, which is also suitable for other JSCC systems with the spatially coupled coding structure.