A framework for detection of sensor attacks on small unmanned aircraft systems

The work presented in this paper is part of an overall effort to design a secure autopilot, resilient against malicious attacks on both the cyber and physical layers, for a small unmanned aircraft system (UAS). This paper specifically deals with identification of malicious attacks on the sensors of a small UAS. A framework is presented wherein techniques from statistical analysis are used in a probabilistic setting to detect sensor attacks. The paper describes in detail the design of anomaly detectors and the Bayesian network. A case study involving detection of a spoofing attack on the GPS is used throughout the paper to illustrate the proposed approach. The anomaly detectors are designed based on a simulation dataset, and are re-tuned based on flight tests conducted on a small fixed-wing UAS platform. The performances of the detectors are studied under different external disturbances and conclusions are drawn.