Machine Learning Correction for Improved PVT Accuracy

Gianluca Caparra, Paolo Zoccarato, Floor Melman

Abstract:	The number of applications relying on accurate positioning has been rapidly increasing over the recent years, demanding usage of GNSS positioning in challenging environments, like for example urbanized areas, where the performance of GNSS is typically degraded. Indeed, in such environments, GNSS receivers are prone to positioning errors mainly due to multipath and interference. When a GNSS receiver tracks a signal affected by multipath, e.g., because it is reflected by close obstacles, it erroneously estimates the distance from the transmitting satellite. This phenomena is present particularly in urban environments, where fewer Line-of-Sight (LoS) signals are typically available and several signals (especially those which are Non-Line of Sight) are potentially affected by multipath. This contribution introduces a novel method for improving Position, Velocity and Timing (PVT) accuracy of GNSS receivers exploiting a Machine Learning (ML) algorithm. The ML model exploits the post-fit residuals/innovations, which are readily available after the position computation from the PVT engine, thus adoptable by existing receivers without requiring any modification. The performance is demonstrated using data collected from mass market receivers and on the Google public dataset with contains data collected from Android smartphones.
Published in:	Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021) September 20 - 24, 2021 Union Station Hotel St. Louis, Missouri
Pages:	3392 - 3401
Cite this article:	Caparra, Gianluca, Zoccarato, Paolo, Melman, Floor, "Machine Learning Correction for Improved PVT Accuracy," Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021), St. Louis, Missouri, September 2021, pp. 3392-3401. https://doi.org/10.33012/2021.17974
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In