On the spectral-efficiency of low-complexity and resolution hybrid precoding and combining transceivers for mmWave MIMO systems

Abstract

Millimeter wave (mmWave) multiple-input-multiple-output (MIMO) systems will almost certainly use hybrid precoding to realize beamforming with few numbers of RF chains to reduce energy consumption, but require low complexity technique to improve spectral efficiency. While energy-efficient hybrid analog/digital precoders and combiners designs can subdue the high pathloss inherent in mmWave channels, they assume the use of infinite- (or high-) resolution phase shifters to realize the analog precoder and combiner pair which results in high hardware cost and power consumption. One promising solution is to employ the use of low-resolution phase shifters. In this paper, we first diverse the exploration of multiple candidates of array response vectors, to propose low-complexity hybrid precoder and combiner (LcHPC) design via stage-determined matching pursuit (SdMP) namely, LcHPC-SdMP for pursuing better achievable rate for mmWave MIMO systems. We initially decouple the joint optimization over hybrid precoders and combiners into two separate sparse recovery problems. Specifically, LcHPC-SdMP algorithm revises the identification step of orthogonal matching pursuit (OMP) to the selection of multiple "correct" column indices of the matrix of array response vectors, per iteration. Then adds a pruning step-after satisfying a sparsity level condition, to iteratively refine the sparse solution which aids in further accelerating the algorithm, by requiring fewer iterations. We then propose an algorithm which iteratively designs low-resolution (two-bit) hybrid analog-digital precoder and combiner (LrHPC), for pursuing efficiency while maximizing spectral efficiency. Simulation results demonstrate that the proposed LcHPC-SdMP algorithm performs very close to its full-digital precoding and achieves better spectral efficiency over state-of-the-art algorithms with a substantially reduced number of iteration than the recently proposed schemes. In addition, simulation results also reveal that the achievable rate of the proposed LrHPC algorithm is higher than those of the existing algorithms under consideration.