Optimal Resilient Tracking Control for Networked Systems With Multichannels Under Attacks

This article is concerned with the problem of resilient output tracking control for a class of discrete-time linear cyber-physical systems (CPSs) subject to denial-of-service attacks. The remote information exchange between the sensor and controller in CPSs employs a multichannel network because of its inherent advantages of resistance to destruction as well as to attacks. In order to design defense strategies to mitigate the negative impact of network attacks on the tracking performance of CPSs, considering the goals and energy constraints of both sides, the interaction process between them is modeled as a zero-sum stochastic game. Then, the Nash equilibrium of the problem is found with the help of the proposed Q-learning algorithm through an artfully designed state-based adaptive one-stage reward function. On the basis of the obtained optimal mixed strategy, the resilient output tracking controller is designed. Different from the work focusing on remote state estimation by means of game theory, this study makes the first effort to design a resilient output tracking control algorithm for CPSs with multiple channels using a stochastic game approach. Finally, the effectiveness of the proposed resilient controller with the optimal defense strategy is validated on a renewable energy generation system prototype by extensive case studies.