Path Loss Prediction in Urban Environments With Sionna-RT Based on Accurate Propagation Scene Models at 2.8 GHz
Guozhen Xia, Chen Zhou, Fubin Zhang, Zeyu Cui, Chengkai Liu, Hao Ji, Xinmiao Zhang, Zhengyu Zhao, Yin Xiao
Abstract
Ray tracing (RT), when applied with precise propagation scene models, delivers accurate path loss predictions. The increase in communication frequencies renders it necessary to take the impact of vegetation and small structures into account on radio wave propagation. This necessitates their inclusion in propagation scene models. In this study, we introduce a robust propagation scene construction method using photogrammetric point clouds. A deep learning method is adopted to semantically segment these point clouds into categories: ground, buildings, vegetation, fences, street furniture, and cars. Tailor-made surface reconstructions are adopted for different categories to balance the geometric accuracy and complexity of the reconstructed models. Path loss values are predicted and compared to actual measurements, utilizing Nvidia’s open-source RT simulation library (Sionna-RT), which is based on the established propagation scene model with millions of meshes. This method is confirmed to have a significant enhancement of RT’s path loss prediction accuracy.