Cloud-Edge Model Predictive Control of Cyber-Physical Systems Under Cyber Attacks

In this paper, a cloud-edge model predictive control (MPC) framework is proposed for cyber-physical systems in the presence of deception attacks and Denial-of-Service (DoS) attacks. In the proposed framework, the original MPC optimization problem is decomposed into cloud and edge layers by using an efficient parameterized control input sequence. Then, a novel controller updating mechanism is developed by discontinuously comparing the optimal value functions of the modified optimization problem and the original optimization problem, which saves the communicational and computational resources. Specifically, the control performance is optimized over all possible uncertainties and deception attack realizations using a min-max optimization technique, while the DoS attacks can be tackled with the parameterization feature of the control input sequence. Besides, the closed-loop system is guaranteed to be input-to-state practical stable (ISpS) under the proposed MPC strategy. Simulation studies and comparisons are performed to verify effectiveness of the proposed method.