Detection And Resist Of False Data Injection Attacks In Networked Control System

Networked control system is a feedback control system in which the control loop is closed via a communication network.Because of the advantages of sharing information resources,saving system wiring costs,improving system flexibility,extending and maintaining the system easily,networked control system have been developed rapidly and widely.However,the security problems of networked control system become more and more serious.False data injection attack is a kind of deception attack.The detection and resist of false data injection attack are analyzed in this thesis,and the main work is as follows:Firstly,for the feedback channel false data injection attack,the limitations of residual test are analyzed and a feedback channel stealthy false data injection attack is introduced,which can successfully avoid residual test.In order to detect this attack,a detection method combining K-S test and residual test is proposed.This method builds a new detection index to judge whether the system is attacked.When the system detects that the new detection index exceeds the normal threshold,it is judged that there is an attack in the system.By discarding the error data which is affected by the attack,the system is not damaged by the attack.By using True Time toolbox to build a networked control system,the simulation results show that the feedback channel stealthy false data injection attack can be detected and resisted by this method.Secondly,the differences and connections between two types of two-channel stealthy false data injection attacks are analyzed.The same point of these two attacks is that the system cannot detect these two attacks when there is only an attack detector in the feedback channel.The difference between these two attacks is that the conditions for the implementation of the two attacks are different.Therefore,in order to detect this attack,it is necessary to have attack detectors in both the forward channel and the feedback channel of the system.In order to resist this attack,a predictive compensation controller based on the autoregressive moving average model is designed.Compared with the predictive compensation controller that directly uses the system state equation to predict the system state,the controller reduces the prediction error of the control input and improves the ability to resist attacks.By using True Time toolbox to build a networked control system,the simulation results show that the controller can make sure that the system under attacked will work normally.Thirdly,for the long-term two-channel stealthy false data injection attack,an improved predictive compensation controller with attack compensation is designed.When an attack is detected in the system,the controller eliminates the impact of the attack by compensating the attack and the system can still work normally.By using True Time toolbox to build a networked control system,the simulation results show that the controller can resist long-term two-channel stealthy false data injection attack and the system can still work normally without the impact of the attack.Finally,the work is summarized and the future research direction is prospected.