Joint Iterative Training-based Hybrid Precoding and Combining for Millimeter Wave Systems

Abstract

Hybrid precoding and combining is a key solution to reduce the hardware cost and power consumption of millimeter wave MIMO systems. However, estimating the entire channel matrix required for designing such a solution is very challenging. In this paper, we propose a joint iterative training based algorithm to design the digital precoder and combiner in the hybrid structure. We show that the proposed method can eliminate channel estimation and SVD computation needed for obtaining the precoders and combiners. Simulation results show that the proposed approach performs comparable to existing SVD-based methods in terms of spectral efficiency but with much less computational complexity.