Near-Optimal Designs of Hybrid Precoding and Combining for Massive MIMO Systems From Lattice Decoding

Abstract

Near-optimal designs consisting of a pair of finite resolution analog precoder and combiner and a baseband encoder for hybrid massive MIMO communication systems are proposed in this paper. Firstly, for any configuration of hybrid MIMO systems, several powerful upper bounds on the maximal achievable rates are presented and are used as guidelines for the proposed designs. Armed with the insights of upper bounds, designing the coefficients for finite resolution analog precoders and combiners is then regarded as a lattice decoding problem, where low-complexity lattice decoders and convex solvers are employed to yield optimal solutions with the best structure for partial connection. Simulation results show that the proposed design achieves within a negligible gap from the rate upper bound using phase shifters with low resolution, while reducing the number of required phase shifters by roughly 20% and significantly outperforming many existing designs at the same time.