Two‐stage beamforming training for multi‐user millimetre wave systems

Abstract

Hybrid analogue/digital precoding technique is widely considered for multi-user millimetre wave (mmWave) systems to reduce the cost and power consumption of radio frequency chains, where beamforming (BF) training is required for the configuration of analogue BF/combining vectors of multiple user equipments (UEs). However, the conventional BF training algorithms suffer from high training overhead increasing with the number of UEs. To address this problem, a two-stage BF training mechanism is proposed for multi-user mmWave systems with time division duplex mode by dividing the BF training process into downlink broadcasting and uplink training. Consequently, multiple UEs can be trained simultaneously and the feedback operation is eliminated. Simulation results demonstrate that the proposed two-stage BF training scheme approaches the optimal performance of exhaustive search with much reduced training overhead.