Generalizable Diffusion Models for Site-Specific Radio Propagation Modeling

Machine learning (ML) methods have been successfully employed to derive efficient radiowave propapagation models in recent years. Recent efforts have focused on artificial and convolution neural networks. In this paper, we present the first diffusion model for indoor propagation. We show that diffusion models have a unique advantage when compared to other ML models, because they can solve at once three problems in propagation modeling: received signal strength prediction, calibration with measured data, and placement of access points.