Optimal Use of Antennas in Interference Networks: A Tradeoff between Rate, Diversity and Interference Alignment

The tradeoff between diversity, interference alignment and rate for a K user multiple-antenna interference network is analyzed. It is assumed that the sources employ a space-time code in combination with linear preceding, while the receiving nodes use linear detectors. We show that interference alignment is needed if the system is operating at or close to the maximum achievable rate. For low rates the preferred strategy is to utilize all antennas in order to achieve high diversity gains, rather than using some of the antennas to align the interference. For the case of K = 3 users, an exact characterization of the tradeoff is provided. We also investigate the impact of channel estimation errors on the diversity of the system. It turns out that small channel estimation errors can be tolerated, while larger errors reduce the diversity gain significantly.