Beam Prediction Based on Multi-Modal Joint Environment Sensing

In this letter, we propose a cross-band beam prediction method that leverages multi-modal environmental sensing. The environmental point cloud surrounding the base station (BS) and the partial mmWave channel state information (CSI) at the user terminal (UT) are considered as joint environmental sensing information (JESI). An environment-sensing enhanced network is designed to fuse JESI with sub-6 GHz CSI for accurate prediction of the optimal mmWave beams at both the BS and the UT. Simulation results demonstrate the effectiveness of the proposed method, which improves the joint beam prediction accuracy by 7.25% at a cost of 0.78% of the mmWave overhead.