Attack-Defense-Model-based Secure Control Of Cyber-physical Systems

Recently,advanced computing technologies have boosted the deep coupling between the cyber network and the control system.The communication channels allow data to be shared efficiently and the controller can fuse the global information to make intelligent decisions over a large physical space.Meanwhile,the complexity of systems increases considerably,which inevitably leads to series of security challenges and a vast accumulation of loopholes in Cyber-physical systems.This leaves space for adversaries to perform malicious attacks.Security incidents are increasing significantly and secure control problem has become fierce and crucial.For CPS control system with Denial-of-Service(DoS)attacks,we analyze the security theoretically and develop corresponding control algorithms to guarantee the system stability.Firstly,we consider the security constrained optimal control for discrete-time,linear dynamic CPS under DoS attacks,where the attacker can jam the limited communication channel to disrupt the effective communication.We verify the vulnerable process corrupted by attacks with a simulation of 9-bus power grid system under attacks.Secondly,to characterize the impact and behavior of DoS attacks,we propose a novel DoS attack model in power grids based on Stackelberg game,where the controller and the attacker are two players.Moreover,optimal problems are built for both the attacker and the defender to maximize their own profits.We verify them by the proof of necessary and sufficient conditions and analytic methods in a simulation.Finally,we propose a discrete event-triggered predictive control scheme with game-theoretic control law,which ensures the system to get access to control signals when packet dropouts occur due to DoS attacks.A ball and beam system is studied numerically to demonstrate the compensation effect for packet dropouts.