On Joint Source-Channel Coding for Correlated Sources Over Multiple-Access Relay Channels

TL;DR

This paper introduces a novel joint source-channel coding scheme for transmitting correlated sources over multiple-access relay channels with side information, combining CPM and Slepian-Wolf coding to improve performance.

Contribution

It proposes the first source-channel coding scheme using CPM for encoding to both relay and destination, with new sufficient and necessary conditions for reliable transmission.

Findings

The new scheme outperforms existing methods in certain cases.

It provides the first use of CPM for encoding to two different nodes.

Derived conditions are at least as tight as previous necessary conditions.

Abstract

We study the transmission of correlated sources over discrete memoryless (DM) multiple-access-relay channels (MARCs), in which both the relay and the destination have access to side information arbitrarily correlated with the sources. As the optimal transmission scheme is an open problem, in this work we propose a new joint source-channel coding scheme based on a novel combination of the correlation preserving mapping (CPM) technique with Slepian-Wolf (SW) source coding, and obtain the corresponding sufficient conditions. The proposed coding scheme is based on the decode-and-forward strategy, and utilizes CPM for encoding information simultaneously to the relay and the destination, whereas the cooperation information from the relay is encoded via SW source coding. It is shown that there are cases in which the new scheme strictly outperforms the schemes available in the literature. This is the first instance of a source-channel code that uses CPM for encoding information to two different nodes (relay and destination). In addition to sufficient conditions, we present three different sets of single-letter necessary conditions for reliable transmission of correlated sources over DM MARCs. The newly derived conditions are shown to be at least as tight as the previously known necessary conditions.