Interference Channels With Arbitrarily Correlated Sources

TL;DR

This paper establishes a sufficient condition for lossless transmission of correlated sources over interference channels, extending known regions and demonstrating the suboptimality of separation approaches in certain cases.

Contribution

It introduces a new achievable region for correlated sources over interference channels, generalizing existing results and highlighting limitations of separation strategies.

Findings

The new region reduces to Han-Kobayashi for independent sources.

For specific correlation structures, it simplifies to known regions with common information.

Separation approach with Slepian-Wolf encoding can be suboptimal.

Abstract

Communicating arbitrarily correlated sources over interference channels is considered in this paper. A sufficient condition is found for the lossless transmission of a pair of correlated sources over a discrete memoryless interference channel. With independent sources, the sufficient condition reduces to the Han-Kobayashi achievable rate region for the interference channel. For a special correlation structure (in the sense of Slepian-Wolf, 1973), the proposed region reduces to the known achievable region for interference channels with common information. A simple example is given to show that the separation approach, with Slepian-Wolf encoding followed by optimal channel coding, is strictly suboptimal.