On the Use of Machine Learning Algorithms to Improve GNSS Products

Aug 3, 2026

Uploaded by Anda

This paper reports key findings from an investigation into machine learning techniques for processing data in Global Navigation Satellite Systems. Supported by the European Space Agency, the work targets multiple data elements across the positioning chain and evaluates different ML approaches. Results highlight improved prediction of ionospheric maps for correcting pseudorange errors and reliable forecasting of fast corrections in EGNOS messages when they are absent. Historical data and temporal correlation enable ML to outperform simple regression, enhance GNSS user-level positioning performance, and support automatic outlier detection from ionospheric scintillation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Introduction

GNSS ML Demonstrator (GMLD)

Motivation and data quality in positioning

Abstract and contributions

Ionospheric map prediction

Forecasting fast corrections

Outlier detection for ionospheric scintillation

POLITECNICO DI TORINO Repository ISTITUZIONALE On the Use of Machine Learning Algorithms to Improve GNSS Products Original On the Use of Machine Learning Algorithms to Improve GNSS Products / Nardin, Andrea; Dovis, Fabio; Valsesia, Diego; Magli, Enrico; Leuzzi, Chiara; Messineo, Rosario; Sobreira, Hugo; Swinden, Richard. -ELETTRONICO. - (2023) .(Intervento presentato al convegno 2023 IEEE/ION Position, Location and Navigation Symposium (PLANS) tenutosi a Monterey, California, USA nel 24-27 Aprile 2023) [10 . 1109/PLANS53410 .2023. 10139920] . Availability: This version is available at: 11583/2977410 since: 2023-03-23T19:02:48Z Publisher: IEEE Published DOI:10.1109/PLANS53410.2023.10139920 Terms of use: This article is made available under terms and conditions as specified in the corresponding bibliographic description in the repository Publisher copyright IEEE postprint/Author's Accepted Manuscript ©2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collecting works, for resale or lists, or reuse of any copyrighted component of this work in other works. (Article begins on next page) 18 September 2024 On the Use of Machine Learning Algorithms to Improve GNSS Products Andrea Nardin Politecnico di Torino, DET Turin, Italy andrea.nardin@polito.it Fabio Dovis Politecnico di Torino, DET Turin, Italy fabio.dovis@polito.it Diego Valsesia Politecnico di Torino, DET Turin, Italy diego.valsesia@polito.it Enrico Magli Politecnico di Torino, DET Turin, Italy enrico.magli@polito.it Chiara Leuzzi Rosario Messineo Hugo Sobreira ALTEC ALTEC European Space Agency, ESTEC Turin, Italy Turin, Italy Noordwijk, The Netherlands chiara.leuzzi@altecspace.it rosario.messineo@altecspace.it [hugo.sobreira@esa.int](hugo.sobreira@esa.int) Richard Swinden European Space Agency, ESTEC Noordwijk, The Netherlands [richard.swinden@esa.int](richard.swinden@esa.int) Abstract—This paper presents the most relevant results on the investigation of possible uses of machine learning based techniques for the processing of data in the field of Global Navigation Satellite Systems. The work was performed under funding of the European Space Agency and addressed different kind of data present in the entire chain of the positioning process, as well as different kind of machine learning approaches. This paper presents the most promising results obtained for the prediction of ionospheric maps for the correction of the related error on the pseudorange measurement and for the forecast of fast corrections normally present in the EGNOS messages, when the latter might be missing. Results show how, based on the historical data and the time correlation of the values, machine learning methods outperformed simple regression algorithms, improving the positioning performance at GNSS user level. The work results also confirmed the validity of this approach for the automatic detection of outliers due to ionospheric scintillation phenomena. Index Terms—Machine-learning, GNSS, Ionosphere, Positioning I. INTRODUCTION It is well known that machine learning (ML) methods are a powerful tool demonstrating their value when dealing with large amount of data. In particular, they show their effectiveness for the prediction of future evolution of the data series or to identify, in an automated way, “patterns” in the data or outliers. The GNSS ML Demonstrator (GMLD) aimed at investigating possible applications in the Global Navigation Satellite System (GNSS) domain that could benefit from ML capabilities, to improve some elements of the entire GNSS process to provide better positioning, navigation, and timing (PNT) services. Indeed, GNSS-based PNT has become fundamental for a wide range of applications, from critical infrastructures [1],[2] to

FAQs

What problem does this paper address in GNSS data processing?

It investigates how machine learning can process GNSS-related data to improve positioning quality, especially when needed information is missing, outdated, or degraded.

Which main GNSS improvements are demonstrated using machine learning?

The study shows promising results for predicting ionospheric maps to correct pseudorange errors and for forecasting fast corrections normally carried in EGNOS messages when they may be missing.

How do the proposed machine learning methods compare with simple regression?

Using historical data and time correlation, the results indicate that machine learning methods outperform simple regression algorithms and improve user-level GNSS positioning performance.

Aug 3, 2026

Uploaded by Anda