A Radio Wave Propagation Modeling Method Based on High-Precision 3-D Mapping in Urban Scenarios

Abstract

Knowledge of the radio channel in urban environments is of key importance in many critical applications, such as wireless communication and autonomous driving. This study proposes a promising radio wave propagation model for complex urban scenarios. First, electrical parameters of different object classes are estimated by utilizing deep-learning-based semantic segmentation applied to a point cloud. Second, a highly accurate 3-D geometrical and electromagnetic (EM) model of the environment is obtained by surface reconstruction. Finally, ray tracing, based on geometrical optics and the uniform theory of diffraction (UTD), is applied to the environmental model to generate path-loss predictions. A comparison with measured data shows that the ray tracing based on accurate 3-D building reconstruction is more accurate than when applied to simpler geometric models.