Deep Learning-Based Path Loss Prediction with Satellite Maps

In recent years, with development of the six generation (6G) communication, higher requirements have been placed on high accuracy and low complexity of channel modeling. Deep learning techniques applied to channel modeling are promising solutions because of the high nonlinear fitting capability and low computational complexity. In this paper, we propose a deep learning-based model to predict path loss with satellite images. The model is analyzed and validated by using millimeter wave channel measurement data at 26 GHz in a vehicle-to-infrastructure (V2I) scenario.