On throughput gain of interference alignment in multi-hop MIMO networks

Abstract

In this paper, we investigate the potential throughput gain brought by cooperative interference alignment (IA) in multi-hop MIMO networks. Unlike the fully-connected K-user interference network where IA is usually considered, IA cooperations in multi-hop MIMO networks are limited by node density, leading to an inevitable coexistence of IA and non-IA links. Though cooperating links within the same IA group can nullify intra-group interference via IA, uncoordinated inter-group interferences as well as the interferences from non-IA links retain with noises. Hence, whether or not IA can brought throughput gain for such networks is still an open question. In allusion to this question, we firstly extend a cross-layer PHY-MAC model for MIMO networks to jointly abstract the IA and non-IA links with a set of transmission patterns. Then we formulate the problem of throughput-optimal topology control for multi-hop MIMO networks with IA and that without IA and prove their NP-hardness. We further propose integer linear mathematical models for the formulated problems and investigate the benefits of IA cooperations via simulations. Simulation results show that regardless of the number of antennas, IA cooperations do not offer noticeable throughput gain in random networks.