Orthogonalization to Reduce Overhead in MIMO Interference Channels

Interference channels are useful analytical models for distributed wireless communication networks with multiple simultaneously transmitting users. The degrees-of-freedom optimal transmit strategy for interference channels is interference alignment, which requires substantial channel state knowledge throughout the network. As the network grows, the sum capacity, theoretically, increases linearly. This result, however, neglects overhead from training and feedback. This paper accounts for overhead in the MIMO constant-coefficient interference channel with linear precoding and proposes an orthogonalization approach to maximizing sum throughput when overhead is considered. The optimization’s solution, assuming each group uses interference alignment, is found to require full channel state information and a brute-force search, so a greedy partitioning method with reduced CSI requirements is proposed.