Antenna Selection in RF-Chain-Limited MIMO Interference Networks Under Interference Alignment

In multiple-input multiple-output (MIMO) systems, the deployment of multiple radio-frequency (RF) chains is much more expensive than that of multiple antennas. Antenna selection (AS) is a low-cost low-complexity method to exploit the diversity gain in MIMO systems with sufficient antennas but limited RF chains. Among the AS algorithms, decremental AS has been shown to be near-optimal compared with exhaustive search. In this paper, we study AS in MIMO interference networks applying the interference alignment (IA) scheme. IA aligns and zero-forces the interference via transmitter-receiver beamforming and achieves a maximum degree of freedom. The feasibility and alignment topology of IA is dependent on the active antenna configuration. For users in tightly feasible systems, IA can be resumed after AS by adapting the beamformers locally, and the exact transmission rate can be known. For users in superfeasible systems, we derive the expected rate under IA conditioned on local channel state information (CSI). Adopting the rate or expected rate as selection criterion, we propose the decremental AS algorithms achieving a twofold complexity reduction. Simulation results and complexity analysis show that the algorithms achieve near-optimal performance at low complexity.