Joint Source-Channel Decoding for MDC-encoded Sources Transmitted Over Relay Systems

This paper presents a new transmission strategy combined with a source and channel coding and decoding method to deliver a source of information to two users requesting different quality-of-service and having different channel states. Source coding operation relies on Multiple Description Coding (MDC) and channel coding uses LDPC codes. To achieve a better quality for the high-quality-requiring user, we consider cooperative communication where another user acts as a relay. We propose a three-dimensions iterative decoding algorithm that exploits the source-relay correlation and the descriptions correlation. We demonstrate how the lossy forwarding process can help the user with bad channel conditions to recover a first quality even at very low source-destination SNRs. If the channel conditions for the second user are fair enough, it is possible to decode the second description, and achieve a very low distortion based on combining the two descriptions. It is also shown that exploiting the residual correlation between the two descriptions can lead to better results especially for medium-to-high SNRs.