Image-to-Image Translation-Based Millimeter-Wave Beam Search Using Out-of-Band Spatial Information

The large-scale antenna array brings great gains to the millimeter wave (mmWave) communication system, but the consequent beam search overhead is evident. A feasible solution is to exploit the spatial congruence between signals in different frequency bands and use spatial information to help establish a mmWave link. Using out-of-band information, an image-to-image translation-based beam search method is proposed. Paired power azimuth spectrum images are used to train the model, and the mmWave angle information is predicted using spatial information of the sub-6 GHz band. Prediction results can help the mmWave communication system largely compress the beam search range. Simulation results verify the feasibility of the proposed method.