Enhancing GPS Accuracy with Machine Learning - A Comparative Analysis of Algorithms - Traitement du Signal Vol. 41 No. 3

Aug 4, 2026

Upload by Ezra

Global Positioning System (GPS) accuracy is undermined by signal attenuation and environmental obstacles in dense, structurally complex areas, making high-cost precision receivers necessary for reliable tracking. This study reduces positioning errors by applying a machine learning framework to vehicle-tracking datasets collected using Novatel and Ublox devices. Ten prediction algorithms are compared, including single and ensemble methods, randomness-based stability, proximity-driven prediction, and regularization to reduce overfitting. Extra Trees achieves the best results with R²=99.6% and the lowest error metrics, improving average deviation by about 6.8 meters.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Motivation

Traitement du Signal Vol. 41 , No. 3, June, 2024, pp. 1441-1450 Journal homepage: [http://iieta.org/journals/ts](http://iieta.org/journals/ts) | Enhancing GPS Accuracy with Machine Learning: A Comparative Analysis of Algorithms
Metin Zontul 1, Ziya Gokalp Ersan2, Ilkay Yelmen3* , Taner Cevik4 , Ferzat Anka5 , Kevser Gesoglu6
1 Department of Computer Engineering, Faculty of Engineering and Natural Sciences, Sivas University of Science and
| | --- | | Technology, Sivas 58000, Türkiye
2 Department of Software Engineering, Faculty of Engineering and Architecture, Istanbul Gelisim University, Istanbul 34310, | | Türkiye
3 Department of Computer Engineering, Faculty of Engineering and Natural Sciences, Istinye University, Istanbul 34396, Türkiye | | 4 Department of Computer Engineering, Faculty of Engineering, Istanbul Arel University, Istanbul 34537, Türkiye | | 5 Data Science Application and Research Center (VEBIM), Fatih Sultan Mehmet Vakif University, Istanbul 34445, Türkiye | | 6 R&D Center, Turkcell Technology, Istanbul 34854, Türkiye
Corresponding Author Email: [ilkay.yelmen@istinye.edu.tr](ilkay.yelmen@istinye.edu.tr) | | Copyright: ©2024 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license
([http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/)). | [https://doi.org/10.18280/ts.410332](https://doi.org/10.18280/ts.410332) ABSTRACT Received: 22 October 2023 Revised: 29 March 2024 Accepted: 6 May 2024 Available online: 26 June 2024 Keywords: map matching, machine learning, location estimation, Global Positioning System (GPS) In the realm of wireless communications, the Global Positioning System (GPS), integral to Global Navigation Satellite Systems (GNSS), finds extensive applications ranging from vehicle navigation to military operations, aircraft tracking, and Geographic Information Systems (GIS). The reliability of GPS is often compromised by errors particularly prevalent in dense and structurally complex environments, where signal attenuation by environmental obstacles like mountains and buildings is common. These challenges necessitate the deployment of high-cost, precision GPS receivers capable of enhanced signal tracking and acquisition. This study investigates the reduction of GPS positioning errors by implementing a machine learning framework, utilizing a dataset from vehicle tracking devices equipped with Novatel and Ublox technologies. Ten machine learning prediction algorithms were evaluated, focusing on techniques that introduce randomness for stability, employ proximity for predictions, incorporate regularization to prevent overfitting, and leverage both single and ensemble methods to refine analyses. Among the evaluated algorithms, the Extra Trees algorithm was distinguished by its superior performance, achieving a coefficient of determination (R²) of 99.6%, with the lowest error rates compared to its counterparts. The errors were quantified as Root Mean Square Error (RMSE) at 1.01E-4, Mean Absolute Error (MAE) at 4.14E-5, and Mean Square Error (MSE) at 1.03E+0 for normalized data. A comparative assessment across ten scenarios demonstrated that the machine learningenhanced approach deviated by approximately 6.8 meters on average, markedly improving accuracy over traditional GPS methods and reducing positional deviations to a scale of meters. This advance represents a significant stride towards minimizing GPS inaccuracies in complex environments, providing a robust framework for enhancing navigational precision in critical applications. 1. INTRODUCTION Recently, there has been increasing interest in the GNNS [1] framework in which positioning and geolocation processes are carried out. At least three satellites must be used to measure location in the GPS. The data acquired from multiple satellites is then processed and transformed into a single point [1] . GPS data is used in many fields, such as

FAQs

Why does GPS accuracy degrade in complex environments?

GPS reliability is reduced by signal attenuation caused by obstacles such as mountains and buildings, which leads to errors in dense and structurally complex areas.

What machine learning approach is used to improve GPS positioning?

A machine learning framework evaluates ten prediction algorithms on a vehicle-tracking dataset, comparing techniques such as randomness for stability, proximity-based predictions, regularization, and single vs. ensemble learning.

Which algorithm performs best and what metrics support this?

Extra Trees delivers the best performance, reaching R²=99.6% and the lowest error rates, with reported RMSE, MAE, and MSE values on normalized data.

Aug 4, 2026

Upload by Ezra