Adaptive Hybrid Beamforming for Uplink Multi-user Millimeter Wave Systems

Abstract

Without the requirement of the fully estimated channel state information, this paper proposes a novel iterative training scheme for a uplink multi-user millimeter wave (mmWave) system with the hybrid beamforming (HBF). By introducing the multi-user reference signals, objective functions are formulated for the optimization of the phase-only analog beamforming (ABF). With limited radio frequency (RF) resources, we implement a time-division training framework for sub-arrays to adaptively exploit the optimal ABF vectors. Moreover, zero-forcing (ZF) equalization is performed to further eliminate inter-user interference in the digital beamforming (DBF). Simulation results indicate that the proposed schemes outperform the direct path steering scheme in the multi-user scattering environment.