Delay-rate tradeoff for ergodic interference alignment in the Gaussian case

TL;DR

This paper explores the tradeoff between delay and data rate in ergodic interference alignment for Gaussian channels, proposing methods to reduce delay at the expense of some rate loss, and analyzing the rate-delay relationship.

Contribution

It introduces a strategy to mitigate delay by using multiple channel realizations and demonstrates the rate-delay tradeoff through a time-sharing approach.

Findings

Delay can be reduced by using more than two channel realizations.

A rate-delay tradeoff exists and can be managed via time-sharing.

The analysis applies to Gaussian channels with potential extension to finite fields.

Abstract

In interference alignment, users sharing a wireless channel are each able to achieve data rates of up to half of the non-interfering channel capacity, no matter the number of users. In an ergodic setting, this is achieved by pairing complementary channel realizations in order to amplify signals and cancel interference. However, this scheme has the possibility for large delays in decoding message symbols. We show that delay can be mitigated by using outputs from potentially more than two channel realizations, although data rate may be reduced. We further demonstrate the tradeoff between rate and delay via a time-sharing strategy. Our analysis considers Gaussian channels; an extension to finite field channels is also possible.