Resilient networked control systems subject to cyber threats: a data encryption and cloud computing based model predictive control approach

This work presents a resilient model predictive control framework designed for constrained cloud-based networked control systems, addressing false data injection attacks that affect both the controller-to-actuator and sensor-to-controller communication channels. By taking advantage of the robustness of the receding horizon control philosophy and the confidentiality property of homomorphic cryptosystems, an innovative strategy based on their join action is conceived to secure data packets, i.e., ensuring a promptly detection in response to different malicious external actions, designing an array of attack countermeasures for the on-line operations. Finally, the recursive feasibility of the resulting scheme is formally proved in terms of sufficient conditions derived by resorting to a joint use of set-theoretic arguments and integrity properties of the shared data. A case study, taken from a part of the Barcelona drinking water network, is used to validate the whole architecture in simulation.