Achievable diversity gain of K-user interference channel

Abstract

Interference alignment (IA) is a powerful technique to handle interference between links that share the same wireless channel. It has been shown that IA can asymptotically help each link to achieve half of the degrees of freedom (DoF). This result solves one endpoint of the optimal diversity and multiplexing tradeoff (DMT) for interference channels, namely, the maximum multiplexing gain. In this paper, we shall focus on the other endpoint to investigate the maximum diversity gain. For an interference channel with K links where each link has a channel freedom (independent channels) of L, zero-forcing algorithm can achieve a diversity gain of L-K +1 for each link and a sum diversity gain of K(L − K + 1) for the whole network. The question we want to answer is whether interference alignment can provide a higher diversity gain, which will help determine the optimal DMT of interference channels. It will be shown that IA can increase the sum diversity gain by K − 1, where at most K − 2 of them can be utilized by one link.