Predictive Control Of Cyber-Physical Systems Under Denial-of-Service Attacks

Cyber-physical System(CPS),as a new complex control system closely combined with computing resources and physical resources,has become the preferred framework of many large-scale systems.While the openness of network structure brings convenience to the system,the security problem it faces should not be ignored.As one of the most common methods of network attacks,Denial-of-Service(DoS)attack is a kind of resource-exhausted attack,which can cause server paralysis or the occupation of network bandwidth resources,and the security and real-time performance of system data transmission is affected.Therefore,this thesis studied the predictive control strategy of CPS under DoS attack.The specific contents are as follows:Firstly,the CPS model under time-constrained DoS attack is described,and the restricted conditions of DoS attack are given from two aspects as frequency and duration.By analyzing the influence of the DoS attack,the sufficient condition for the input-to-state stability of CPS under time-constrained DoS attack is given.Secondly,for the situation that the DoS attack occurs on the control channel from the controller to the actuator,considering that the control signal is transmitted in the form of data packet,the predictive control algorithm based on data packet is designed and so as the predictive controller.At each sampling time,the controller node calculates a set of predicted control signal sequences based on the historical state signal data and transmits them to the actuator in the form of data packets.When the system is attacked,the actuator replaces the lost control signal with the predicted control signal,so that the plant could be kept under control.Finally,considering the effective utilization of communication network resources,the predictive control algorithm based on event-triggered strategy and the predictive controller is designed for the situation that DoS attack occurs on the measurement channel from sensor to controller.When the system is attacked,the controller node calculates the prediction signal data according to the existing historical state data,and generates the corresponding control signal based on the prediction controller,thus the system is controlled.For each case of DoS attacks,the corresponding simulation example is given to illustrate the effectiveness of the algorithm.