Koopman-based Model Predictive Control for Nonlinear Systems Subjected to Joint DoS Attacks

This paper presents a Koopman-based model predictive control (MPC) approach for nonlinear systems subject to dual-channel denial-of-service (DoS) attacks. For systems with unknown dynamics, the Koopman operator theory is utilized to develop a linear control model via the extended dynamic mode decomposition (EDMD) method. The study addresses the scenario where asynchronous DoS attacks occur simultaneously on both the sensor-controller and controller-actuator communication channels. Compensation strategies are proposed to mitigate the effects of these DoS attacks. A predictive model within the controller is used to estimate the actual system’s dynamics. The optimal control problem is solved based on the updated system states received by the controller. The resulting optimal control sequence is sent periodically to the actuator, with control inputs updated as new information becomes available. Throughout the process, the control inputs of both the predictive model and the actual system remain aligned. Numerical simulations are conducted to validate the effectiveness of the proposed method.