Sensor Cyber-Attack and Fault Detection and Isolation of Nonlinear Cyber-Physical Systems

In this paper, we consider a discrete-time nonlinear Cyber-Physical System (CPS) under system faults, adversarial cyber-attacks, as well as process and measurement noise. We propose an approach to estimate system states and faults for the nonlinear CPS through an augmented unscented Kalman filter (AUKF), involving the stability analysis of the estimation error dynamics. Through cyber-attack and fault detectors that are placed in the command and control (C&C) center, the cyber-attacks occurring in the measurement channels and system faults are detected and isolated. It is shown that by utilizing fault detectors, which is only sensitive to the corresponding faults, we can distinguish between the faults and cyber-attacks in the cyber-attack detector. Simulation studies do illustrate the effectiveness of the proposed methodology.