AI-Driven Wireless Propagation Models and Applications

We are presenting examples of artificial intelligence enabled algorithms for radiowave propagation modeling and their application to practical problems of interests. We use computational electromagnetics techniques to generate physics-based training data for neural networks. Emphasis is given on ray-tracing and the vector parabolic equation method, for indoor and tunnel propagation problems, respectively. For the latter, the use of artificial neural networks for the efficient computation of uncertainties inherent in propagation models is discussed.