Low-resolution phase shifter-based hybrid precoding for mmWave massive MIMO

As the demand for high-capacity and energy-efficient wireless communications grows, the exploration of cost-effective and power-optimized hybrid digital-analog beamforming techniques for millimeter-wave massive MIMO systems has gained substantial traction. Adopting low-resolution phase shifters offers a more practical solution in terms of reducing power consumption and implementation costs. We investigate hybrid beamforming strategies for mmWave massive MIMO systems that employ low-resolution phase shifters, with a specific focus on designing and optimizing the analog beamforming component. To exploit the full potential of analog phase shifter network (APSN) in mmWave massive MIMO system while maintaining low-power consumption, we propose a two-layer APSN with low-resolution PSs to replace the conventional single-layer APSN with high-resolution PSs. Under the constraints imposed by the two-layer APSN hardware architecture, we propose a two-level codebook structure and an adaptive-weighted cross-entropy optimization algorithm that offers superior performance while substantially reducing the computational burden compared to the conventional cross-entropy optimization algorithm. We further develop two alternative schemes through the maximal gain harvest method to reduce complexity. We also investigate the case that only one steering vector is shared in the inner beamformer and present two low-complexity beamforming algorithms. Through simulations, the proposed two-layer APSN structure and the proposed algorithms have been shown to outperform existing methods, with notable effectiveness in scenarios where low-resolution phase shifters are utilized.