Resilient adaptive control of switched nonlinear cyber-physical systems under uncertain deception attacks

• Published in: Information sciences 2021-01, Vol.543, p.398-409
• Main Authors: Li, Zhanjie, Zhao, Jun
• Format: Article
• Language: English
• Subjects: Dynamic surface control; Resilient adaptive control; Switched nonlinear cyber-physical systems; Uncertain deception attacks
• ISSN: 0020-0255; 1872-6291
• DOI: 10.1016/j.ins.2020.07.022

•Resilient adaptive dynamic surface control for switched nonlinear cyber-physical systems under uncertain deception attacks are proposed.•Uncertain sensor and actuator deception attacks are considered.•Sensor attack compensators are constructed to mitigate the effects caused by the sensor attacks.•Unknown nonlinearities and state-dependent actuator attacks are estimated via neural networks.•Common Lyapunov function is constructed for all the subsystems. This paper addresses the problem of resilient adaptive dynamic surface control against uncertain sensor and actuator deception attacks for a class of switched nonlinear cyber-physical systems. The concerned system dynamics suffer from both unknown switching mechanisms and more general nonlinearities. Furthermore, it is our aim to deal with deception attacks as adversaries can corrupt sensor and control data, resulting the conventional error surfaces inaccessible for feedback control design. To this end, we construct sensor attack compensators to mitigate the effects caused by the sensor attacks. In addition, neural networks are utilized to approximate the nonlinear terms and compensate the state-dependent actuator attacks. Then, we construct a common Lyapunov function and propose a dynamic surface-based resilient adaptive strategy, under which the equilibrium point of the resulted closed-loop system is semi-globally uniformly ultimately bounded under arbitrary switchings. Finally, we provide a continuously stirred tank reactor system under uncertain deception attacks to validate the effectiveness of the proposed control method.