Interference alignment for multi-user MIMO interference channels via a Riemannian optimization approach

Abstract

Interference alignment (IA) is a technique shown to be able to achieve a significant overall throughput in a μ — user multiple-input multiple-output (MIMO) interference channel (IC). In this paper, we try to modify the conventional IA designs to achieve enhanced sum-rate performance. We jointly design transmit and receive IA filters (precoding and suppression filters) at a central unit (CU) to reduce the channel state information (CSI) feedback/sharing overhead. At the CU, in a one-way iterative strategy (left ‘L’ to right ‘R’, see Fig. 1), the precoding filters are optimized (on side ‘L’) in the direction of the gradient of the sum-rate, and the suppression filters are chosen (on side ‘R’) to minimize the leakage interferences. Then, in a two-way iterative strategy, the proposed scheme alternates between ‘L’ and ‘R’ sides to design the IA filters through the joint sum-rate maximization and interference leakage minimization on each side. Finally, a modified version of the two-way iterative strategy is proposed to design the IA filters on the basis of signal-to-interference-plus-noise ratio (SINR) (instead of the interference minimization) and the sum-rate maximization techniques. Comparing to other conventional IA designs, the proposed designs with provable convergence show a significant sum-rate performance improvement and often outperform other widely used algorithms, as shown in the simulation examples.