Compression of a Markov source with side information at the decoder using turbo codes

In this paper, a turbo-based distributed joint source-channel coding scheme, for efficient compression and communication of two dependent sources with time memory over noisy channels, is presented. Specifically, we consider the asymmetric case for two binary sources, where one of the sources is available at the decoder as side information. The problem is of particular interest for low complexity compression and communication in data gathering sensor network and multimedia applications. Aiming at exploiting the spatial dependency of the sources, first, a distributed joint source-channel turbo (DJSCT) decoder is developed. Then, a novel decoding scheme is proposed, which benefits from both time memory and spatial dependency between samples of the sources. The proposed decoder is composed of two DJSCT constituent decoders and a maximum a-posteriori constituent decoder, which operate in an iterative manner. As the results validate, the proposed schemes provide effective solutions for compression and robust reconstruction of distributed dependent sources with time memory.