An Improved Design of Beam Training Scheme in Vehicle Communication System

This paper aims to study various technologies based on LIDAR, including distributed architecture, vision assisted wireless communication framework, multimodal machine learning framework, and self-calibrating illumination learning framework. The optimal algorithm will be selected and improved according to the specific situation of the actual dataset. The general plan is to first enhance low brightness images affected by weather and other environmental factors, and then use deep learning models to find the mapping relationship between visual image information and optimal codewords. In formal beam training, the accuracy of codeword prediction is improved to maximize the received power and improve the performance of vehicle networking communication.