Predicting GNSS Availability in Urban Canyons: A Scalable Automated 3D Ray-Tracing Approach Integrating OpenStreetMap and Terrain Elevation Data

Continuous and reliable GNSS positioning is a requirement for autonomous driving and Intelligent Transportation Systems. While techniques such as RTK provide centimeter-level accuracy in open-sky conditions, their performance severely degrades in urban canyons due to Non-Line-of-Sight (NLOS) reception. Traditional autonomous routing algorithms primarily optimize travel time, disregarding GNSS signal quality, which can lead vehicles into areas with critical positioning difficulties. To address this gap, this paper presents a highly scalable, automated 3D ray-tracing framework designed to predict GNSS signal availability across urban environments. Many urban models rely on extruded building footprints assuming flat terrain; the proposed methodology integrates 3D building geometries from OpenStreetMap with a Digital Elevation Model (DEM). This approach better accounts for both artificial structures and natural topographic occlusions. By calculating the mean number of Line-of-Sight (LOS) satellites and Dilution of Precision (DOP), the framework generates a map of GNSS signal quality, providing a geometric upper bound of signal availability because diffraction is not modeled. The system is evaluated across Genoa, Italy, and validated against the KLT (Kowloon Tong) Dataset collected in Hong Kong; with the latter, the model achieves an NLOS classification precision of 95.0% and a LOS recall of 99.2%.