Hybrid Beamforming for mmWave Massive MIMO Systems Employing DFT-Assisted User Clustering
Maliheh Soleimani, Robert C. Elliott, Witold A. Krzymień, Jordan Melzer, Pedram Mousavi
Abstract
Hybrid analog-digital precoding has been recognized as a promising potential solution to facilitate implementation of massive multiple-input multiple-output systems in the millimeter wave spectrum. Hybrid transceivers reduce the radio frequency chain count and cost by dividing the task of precoding between the digital and analog processing domains, but near-optimal design of the hybrid precoders is non-trivial and their optimal design is analytically intractable. This is due to the joint optimization of digital and analog precoding matrices as well as extra constraints for analog precoders due to their realization by analog phase shifters. Additionally, hybrid transceivers can be implemented via two architectures, either fully-connected or partially-connected. In this paper, we make use of projection algorithms to greatly simplify the design problem of the digital baseband and analog precoders into two optimization subproblems whose optimal solutions can be found. We also incorporate in the design a simplified user clustering algorithm based on the discrete Fourier transform, where users within the same and overlapping angular bins in the beam-space domain are grouped together. Simulation results demonstrate the proposed hybrid precoding and user clustering algorithm outperforms existing algorithms.