Drone localization based on 5G signal fingerprinting and particle filtering in GPS-denied environments

• Main Author: Zhang, Lin
• Other Authors: Chau Yuen
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2025
• Subjects: Engineering
• Online Access: https://hdl.handle.net/10356/182154
• Institution: Nanyang Technological University

With advancements in technology and social development, drones have found widespread applications across various sectors in both urban and rural areas, leading to a significant increase in operational demand. Traditional drone positioning and navigation systems primarily rely on the well-established GPS system. However, in real-world applications, many environments, such as dense urban areas or indoor spaces, may experience limited or even nonexistent GPS signals. Thus, achieving precise positioning and stable navigation for drones in GPS-denied environments has become a critical challenge. To address this issue, we propose an innovative drone localization method that integrates 5G base station signal fingerprinting with an advanced particle filtering algorithm, enabling high-precision positioning for drones in GPS-restricted environments. The system leverages measurements of Received Signal Strength (RSS) from 5G signals and fuses them with odometry data from the drone, thereby enhancing localization accuracy and robustness. Specifically, a Gaussian based fingerprint matching technique is used to compare real-time 5G RSS data against a pre-constructed signal fingerprint database, while the particle filter continuously refines position estimates based on the drone’s kinematic model and noise characteristics. In this study, we validate the proposed method using a real drone flight dataset collected from multiple flight experiments in a complex urban environment. Results indicate that even amid dynamic environmental changes, the method achieves high-precision positioning with minimal error. Compared to traditional GPS based positioning or methods that rely solely on RSS, the proposed system significantly improves the drone’s localization capability in unreliable GPS en vironments by effectively combining signal fingerprinting and particle filtering, providing a reliable alternative for drone navigation. The application potential of the proposed method is broad, encompassing scenar ios such as delivery services, remote infrastructure inspections, and search-and rescue missions in GPS-constrained areas. This research provides technological support for drone applications in urban and other GPS-denied environments. Keywords: Drone Localization, GPS-Denied Environments, 5G Signal Finger printing, Particle Filtering, Drone Navigation, Urban Environment, Positioning Accuracy