Hybrid Analog-Digital Beam Training for mmWave Systems with Low-Resolution RF Phase Shifters

Abstract

Millimeter wave (mmWave) wireless technologies are expected to exploit large-scale multiple-input multiple-output and adaptive antenna arrays at both the transmitter and receiver to deal with unfavorable radio propagation and realize sufficient link margin. However, the high cost and power consumption of mmWave radio components prohibit the use of fully-digital precoding/combining architectures, which incurs one dedicated RF chain per antenna element. This paper proposes a practical design of multi-beamwidth codebook exploiting hybrid analogdigital architectures with a number of RF chains much lower than the number of antenna elements and 2-bit RF phase shifters. The proposed solution relies on the orthogonal matching pursuit algorithm enhanced by a dynamic dictionary learning mechanism. Simulation results show that the designed hybrid codebooks are able to shape beam patterns very close to those attained by a fully-digital beamforming architecture. Furthermore, when leveraged in the framework of an adaptive, multi-resolution beam training protocol, our hybrid codebooks are able to estimate the most promising angle-of-departure and angle-of-arrival directions with extreme accuracy, yet requiring lower complexity hardware compared to the state of art.