Event-Triggered Cognitive Control for Networked Control Systems Subject to DoS Attacks and Time Delay

This paper focuses on the co-design of event-triggered and cognitive control for networked control systems in the presence of denial-of-service (DoS) attacks and time delay. The designed cognitive control is established and improved by using the reinforcement learning and sliding mode control. The cognitive control has two controllers, namely the cognitive controller and sliding mode controller. The cognitive controller is constructed on the basis of reinforcement learning. The event trigger is adopted in the feedback communication channel, and it can determine whether the system state should transmit over communication networks, as well as can save communication resources. Furthermore, the characteristics of DoS attacks and event-triggered are described by Bernoulli distribution. It is shown that the cognitive controller and sliding mode controller utilize the updated state stored in the memory in case of DoS attacks or failed event trigger. This helps to improve the performance of system. Finally, simulation results illustrate that the co-designed event-triggered cognitive control can not only effectively decrease the influences of both DoS attacks and time delay, but also save the resources of communication networks.