Retrospective Interference Alignment

TL;DR

This paper introduces retrospective interference alignment, a novel scheme for distributed transmitters, demonstrating improved degrees of freedom in interference channels with delayed CSIT or feedback.

Contribution

The paper proposes the retrospective interference alignment scheme, enabling interference alignment with delayed CSIT in distributed transmitter settings, achieving new DoF results.

Findings

Achieves 8/7 DoF for 2-user X channel with delayed CSIT.

Achieves 9/8 DoF for 3-user interference channel with delayed CSIT.

Achieves 4/3 and 6/5 DoF with delayed output feedback.

Abstract

We explore similarities and differences in recent works on blind interference alignment under different models such as staggered block fading model and the delayed CSIT model. In particular we explore the possibility of achieving interference alignment with delayed CSIT when the transmitters are distributed. Our main contribution is an interference alignment scheme, called retrospective interference alignment in this work, that is specialized to settings with distributed transmitters. With this scheme we show that the 2 user X channel with only delayed channel state information at the transmitters can achieve 8/7 DoF, while the interference channel with 3 users is able to achieve 9/8 DoF. We also consider another setting where delayed channel output feedback is available to transmitters. In this setting the X channel and the 3 user interference channel are shown to achieve 4/3 and 6/5 DoF, respectively.