Detection of Fault Data Injection Attack on UAV Using Adaptive Neural Network

A resilient and secure control system should be designed to be as safe and robust as possible in face of different types of attacks such as fault data injection (FDI) attacks; thus, nowadays, the control designers should also consider the probable attacks in their control design from the beginning. For this reason, detection of intentional faults and cyber-attacks attracts a great concern among researchers. This issue plays a great role in the safety of unmanned aerial vehicles (UAVs) due to the need of continuous supervision and control of these systems. In order to have a cyber-attack tolerant (CAT) controller, the attack and the type of attack should be detected in the first step. This paper introduces a new algorithm to detect fault data injection attack in UAV. An adaptive neural network is used to detect the injected faults in sensors of an UAV. An embedded Kalman filter (EKF) is used for online tuning of neural networks weights; these online tuning makes the attack detection faster and more accurate. The simulation results show that the proposed method can successfully detect FDI attacks applied to an UAV.