AoA-Based Low Complexity Beamforming for Aerial RIS Assisted Communications at mmWaves

Abstract

Reconfigurable intelligent surface (RIS) mounted on mobile devices (aerial RIS), such as unmanned aerial vehicles, is envisioned to increase coverage for future millimeter wave (mmWave) communications. In literature, RIS beamformer is designed using channel state information (CSI) or by scanning through gNodeB and user codebooks over different training RIS weight matrices. Both of these methods incur significant training overhead. Also, the traditional subspace-based AoA estimators require several power-intensive radio frequency (RF) chains, which is not practically affordable to users. As a low-power and low-complexity alternative, we propose to design RIS beamforming weights based on angle-of-arrival (AoA) from gNodeB-to-RIS and from RIS-to-user with a single RF chain coupled to an antenna array. The proposed estimator is a combination of maximum likelihood and maximum correlation estimator and uses a single RIS training matrix. The simulation results show that, though a slightly larger number of antenna elements is required for the same spectral efficiency, the proposed approach offers a significantly energy- and computationally-efficient beamforming alternative compared to the existing subspace-based AoA estimator and CSI based techniques.