High-Resolution Hybrid Beamforming by Selective Offset Phase Shifters in mmWave MIMO Systems

This work introduces a novel hybrid beamforming architecture for multiuser downlink mmWave MIMO systems, leveraging dynamically switched offset phase shifters (OPSs). Unlike traditional hybrid beamforming, which relies on identical high-resolution PSs at RF chain outputs for analog beamforming, our design uses low-resolution PSs with varying offsets to achieve a broader range of phase values. A switch network then dynamically connects OPSs to antennas, allowing each antenna to link to the OPS that best generates its desired phase. This approach approximates the performance of highresolution PSs with the same number of low-resolution PSs. We propose a hybrid beamforming design where digital and analog beamformers jointly approximate the zero-forcing solution of a fully digital beamformer. The digital beamformer is set as the zero-forcing solution of the effective channel (dependent on the analog beamformer). Then, the phase selection and switch network connections of the analog beamformer are optimized to minimize error relative to the fully digital solution. The optimization is solved iteratively using Dinkelbach's transformation and the majorization-minimization algorithm, with the Hungarian algorithm determining the optimal switch configuration in each iteration. Numerical results demonstrate the effectiveness of the proposed selective OPS architecture over conventional singleresolution and dynamic twin-resolution systems.