Low-Resolution Precoding for the Massive-MIMO Downlink and Frequency Selective Channels

We propose an efficient algorithm for low bit resolution multiuser downlink massive MIMO frequency selective channels. Our approach is to split the high dimensional nonlinear and non-convex problem into low dimensional sub-problems which can be solved efficiently. We show that by iterating the procedure we can achieve performance close to that obtained over frequency flat channels. Also we extend our previous framework and demonstrate that our greedy iterative precoding algorithm (MAGIQ-M) matches best known techniques in terms of achievable mutual information rates while maintaining low complexity and offering an easy to grasp solution formulation.