Sensing-Assisted Predictive Beamforming with NLoS Identification

Sensing-assisted predictive beamforming is a promising technique for reducing the communication overhead and latency in millimeter wave (mmWave) communication systems. In this paper, we propose a robust sensing-assisted predictive beamforming scheme that performs non-line-of-sight (NLoS) identification based on the reflected joint communication and sensing signal. Specifically, the time delay, Doppler shift and reflection coefficient are estimated from the echo signal, based on which a hypothesis test problem is formulated to determine whether the echo signal is reflected by the target vehicle or obstacles. Besides, based on the estimated channel parameters, extended Kalman filtering (EKF) is employed to estimate and predict the motion parameters of the target vehicle. If the echo signal is reflected by obstacles, it implies that the LoS link between the base station (BS) and the target vehicle is blocked. In this case, the proposed scheme can adjust the communication mode and vehicle tracking mechanism accordingly to enhance link reliability. Numerical results demonstrate that the proposed predictive beamforming scheme with NLoS identification can substantially enhance the achievable communication rate as compared to the existing techniques without taking blockage into account.