Enhanced unmanned aerial vehicle-assisted localisation for urban ground sensors

Citation

Elngar, Ahmed Mohamed Abdellatif Abdelrahman (2026) Enhanced unmanned aerial vehicle-assisted localisation for urban ground sensors. Masters thesis, Multimedia University.

Full text not available from this repository.
Official URL: http://erep.mmu.edu.my/

Abstract

Accurate localisation of ground sensors in dense urban areas is a major problem due to severe multipath propagation effects, non-line-of-sight (NLoS) propagation, and signal shadowing effects due to buildings and structures. Existing methods based on Received Signal Strength Indicator (RSSI) localisation techniques, which use log-distance path-loss models, suffer from ranging inaccuracies caused by significant variations between actual and ideal signal strength. The geometry of measurement points used in multilateration is a critical factor in accurate localisation, although this aspect is not considered in existing techniques. The main contribution of this thesis is a two-stage unmanned aerial vehicle (UAV)-assisted localisation framework that simultaneously counteracts measurement distortion and geometric instability to achieve robust sensor localisation in dense urban environments. In the first stage of the proposed scheme, a spatial refinement method is employed on the path-loss measurements along the UAV flight path. This method leverages the spatial correlation among consecutive waypoints in suppressing multipath-induced fluctuations in the path-loss measurements without affecting the inherent distance-dependent path-loss pattern. A reference sensor with known position is employed in calibrating the log-distance path-loss model for a particular environment using the least squares method, and the calibrated path-loss model is used in estimating the distances between the UAV waypoints and the unknown ground sensors. In the second stage, a geometric method of selecting a set of waypoints is incorporated, which enhances the conditioning of the multilateration problem. A set of candidate waypoint subsets is produced based on the principle of Geometric Dilution of Precision (GDOP) and filtered using two real-world constraints: non-collinearity and non-adjacency. Only geometrically stable subsets are selected for localisation, which is carried out separately using Weighted Least Squares (WLS) and Particle Swarm Optimisation (PSO) methods. The final position estimate is computed by combining the results from all the subsets for each method. The framework was assessed through a high-fidelity ray-tracing simulation conducted using Wireless InSite. The results show the state-of-the-art baseline produces a mean absolute error (MAE) of over 150 m, highlighting the inaccuracy of raw RSSI-based ranging in dense urban settings. The application of spatial refinement alone reduces the MAE to 62 m. However, the proposed two-stage framework results in an MAE of 8.5 m for both WLS and PSO optimisation methods, a reduction of more than 94% compared to raw-data results. The fact that both optimisation methods perform similarly under this proposed framework suggests that geometric filtering has a regularisation effect on the error function, which allows the computationally efficient WLS method to perform as well as the more computationally intensive PSO method. These findings demonstrate that accurate UAV-assisted localisation in dense urban environments is achievable when both propagation-related measurement distortion and geometric conditioning of waypoint configurations are addressed in a unified manner.

Item Type: Thesis (Masters)
Additional Information: Call No.: TK7872.D48 E46 2026
Uncontrolled Keywords: Wireless sensor networks
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800-8360 Electronics
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 05 Nov 2026 06:55
Last Modified: 05 Nov 2026 06:55
URII: http://shdl.mmu.edu.my/id/eprint/17068

Downloads

Downloads per month over past year

View ItemEdit (login required)