A gaussian mixture model-based detection algorithm for an unmanned aerial vehicle control system

This paper presents a Gaussian mixture model (GMM) based detection algorithm for an unmanned aerial vehicle (UAV) control system. In the considered system, the UAV may suffer from false data injection (FDI) attacks, which would deteriorate the estimation performance of the unmanned aerial vehicles system. To handle this, a novel detection algorithm is proposed to detect whether the UAV is under attack. Firstly, a cyber-physical UAV model integrates state estimation algorithms and data attack dynamics. Error covariance recursions are rederived under adversarial conditions, with a detection statistic constructed from innovation sequences. A GMM-based detection mechanism is formulated within an extended Kalman filtering framework to identify system compromises. Numerical simulations validate the algorithm’s efficacy in UAV operational scenarios.