Energy-Efficient Hybrid Beamforming Design for Intelligent Reflecting Surface-Assisted mmWave Massive MU-MISO Systems

This study proposes a joint design approach for hybrid beamforming and reflecting beamforming in an intelligent reflecting surface (IRS)-assisted millimeter-wave massive multiuser multiple-input single-output system. The goal is to maximize energy efficiency using energy- and hardware-efficient hybrid beamforming architectures at the base station and low-resolution (e.g., 1–2 bits) phase shifters at the IRS. However, the problem of maximizing energy efficiency is complicated by the high coupling of design variables. To address this, we use a zero-force (ZF) beamforming technique as the digital component of hybrid beamforming and develop a probability learning algorithm based on a cross-entropy optimization (CEO) framework to determine the weights of the analog part of hybrid beamforming as well as IRS phase shifts simultaneously. Additionally, we seek to maximize spatial reuse benefits by increasing the size of the IRS while selecting only a limited number of IRS elements to improve spectral and energy efficiency while minimizing power consumption. This involves joint optimization of hybrid beamforming, IRS element selection, and phase shifts associated with the chosen IRS elements. Solving this problem is challenging, but the proposed ZF-assisted CEO algorithm can still be applied with slight modifications. Simulation results demonstrate that our algorithms achieve significantly better energy efficiency than competitors while maintaining reasonable spectral efficiency.