Interference Channel with Generalized Feedback (a.k.a. with source cooperation). Part I: Achievable Region

TL;DR

This paper introduces an achievable communication strategy for interference channels with generalized feedback, leveraging source cooperation for message relaying and interference pre-cancellation, applicable to wireless networks with multiple user pairs.

Contribution

It proposes a novel achievable region for IFC-GF that combines relay and pre-cancellation techniques, generalizing prior results and applicable to Gaussian channels.

Findings

Achievable region encompasses relay and interference pre-cancellation strategies.

Generalizes several known regions for interference channels with feedback.

Reduces to known regions in special cases.

Abstract

An Interference Channel with Generalized Feedback (IFC-GF) is a model for a wireless network where several source-destination pairs compete for the same channel resources, and where the sources have the ability to sense the current channel activity. The signal overheard from the channel provides information about the activity of the other users, and thus furnishes the basis for cooperation. In this two-part paper we study achievable strategies and outer bounds for a general IFC-GF with two source-destination pairs. We then evaluate the proposed regions for the Gaussian channel. Part I: achievable region. We propose that the generalized feedback is used to gain knowledge about the message sent by other user and then exploited in two ways: (a) to {\em relay} the messages that can be decoded at both destinations--thus realizing the gains of beam-forming of a distributed multi-antenna system--and (b) to {\em hide} the messages that can not be decoded at the non-intended destination--thus leveraging the interference "pre-cancellation" property of dirty-paper-type coding. We show that our achievable region generalizes several known achievable regions for IFC-GF and that it reduces to known achievable regions for some of the channels subsumed by the IFC-GF model.