Geo-Intelligent Handover and Dynamic Selection PRx-DOP: Maximizing 5G UAV Positioning Accuracy in Complex Real Urban Areas

This study proposes an adaptive positioning algorithm for UAVs in real urban 5G networks, leveraging existing cellular infrastructure with a practical approach. Unlike previous studies that rely on idealized scenarios, this work addresses the challenges of operating in real-world environments, where networks are designed for cellular coverage rather than positioning. The algorithm combines Received Signal Power (PRx) with the Dilution of Precision (DOP) from base stations to optimize dynamic station selection, improving system geometry and reducing estimation errors. To predict the UAV’s future position and calculate DOP in real-time, Kalman filters with Interacting Multiple Models (IMM), based on Constant Acceleration (CA) and Constant Velocity (CV) models, are employed. Additionally, a least squares approach with Taylor expansion, complemented by Tikhonov regularization, mitigates errors arising from ill-conditioned matrices. The methodology is applied to a predefined clustering of base stations in the city of Valencia (Spain), where each cluster is analyzed and optimized. The impact of the handover threshold on system accuracy is studied, proposing an inter-cluster handover scheme based on centroids. Experimental results obtained in a real urban environment demonstrate that the hybrid model significantly improves accuracy in handover zones compared to approaches relying solely on PRx.