OpenGERT: Open Source Automated Geometry Extraction with Geometric and Electromagnetic Sensitivity Analyses for Ray-Tracing Propagation Models

Accurate RF propagation modeling in urban environments is crucial for developing high-fidelity digital spectrum twins and optimizing wireless communication systems. This paper introduces OpenGERT, an open-source automated Geometry Extraction tool for Ray Tracing. OpenGERT automates the collection and processing of terrain and building data from multiple sources, including OpenStreetMap, Microsoft Global ML Building Footprints, and USGS terrain elevation data. Leveraging the Blender Python API, the tool creates detailed urban models necessary for high-fidelity ray-tracing simulations specifically designed for NVIDIA Sionna RT. Moreover, we conduct sensitivity analyses to assess the impact of variations in building height, position, and electromagnetic material properties on the accuracy of ray-tracing models. Specifically, we present pairwise dispersion plots of channel statistics—such as path gain, mean excess delay, delay spread, link outage, and Rician K-factor—in response to perturbations to analyze the covariance of different channel statistics. The analyses also explore how the sensitivities of these statistics change as a function of distance from the transmitters. Additionally, we provide visualizations of the variance of the channel statistics within the scene for selected transmitter locations to offer deeper insights. Our study reports results from the Munich and Etoile scenes, each featuring 10 transmitter locations. For each transmitter location, we apply perturbations across five different types, 50 perturbations for each: material, position, height, height and position combined, and all combined. The findings reveal that, assuming the initial material properties of buildings are roughly accurate, minor perturbations in permittivity and conductivity do not significantly alter the channel statistics. In contrast, variations in building height and position have a considerable impact on all the statistics, even with a noise standard deviation of 1 meter for building heights and 0.4 meters for building positions. These results highlight the importance of precise environmental modeling in achieving reliable propagation predictions, which are essential for the deployment of digital spectrum twins and advanced communication networks. Finally, we share the code for geometry extraction and sensitivity analyses at https://github.com/serhatadik/OpenGERT/ to facilitate further experimentation and development.