A Secure and Robust Approach for Distance-Based Mutual Positioning of Unmanned Aerial Vehicles

TL;DR

This work addresses secure, distance-based mutual positioning for multi-UAV systems in environments where GPS and terrestrial anchors may be unreliable. It combines a robust gradient-descent mutual-positioning estimator (rgd) with a recursive data anomaly detector (rdad) to counteract inaccurate and malicious distance and position measurements. The authors develop an error-conversion model that maps position errors into an effective distance-error term, derive weighted estimators, and provide a gradient-based update rule with normalization and momentum. Numerical simulations show that rgd is robust to deterioration, variance, and some bias attacks, while rdad enhances detection and resilience, especially against coordinated or manipulation attacks, offering a practical secure localization framework for autonomous aerial networks.

Abstract

Unmanned aerial vehicle (UAV) is becoming increasingly important in modern civilian and military applications. However, its novel use cases is bottlenecked by conventional satellite and terrestrial localization technologies, and calling for complementary solutions. Multi-UAV mutual positioning can be a potential answer, but its accuracy and security are challenged by inaccurate and/or malicious measurements. This paper proposes a novel, robust, and secure approach to address these issues.

Figures (6)

• Figure 1: Concept of the multi-agent mutual positioning
• Figure 2: Error model mixing position and distance errors
• Figure 3: Paraboloid fitting of the Gaussian model for $n^\Delta_{\mathrm{d},i}$
• Figure 4: Convergence of the proposed methods under attacks
• Figure 5: $\Xi$-sensitivity of rdad against coordinated bias attacks