Linear Deception Attack Analysis And Secure Strategies For A Cyber Physical System

Cyber-physical system has broad application prospects and important theoretical research significance.Because it is often used in large-scale projects and infrastructure construction,security issues have been paid more attention.In this thesis,a linear deception attack has been analyzed and the corresponding secure defense strategies have been proposed for a cyber physical system.The main results are as follows:First,the related theories,development history and future development trends of cyber physical systems have been analyzed,and the existing research results at home and abroad have been briefly summarized.Then,cyber physical system model,Kalman filter and X2 detector,which are related content of this thesis,and the linear deception attack model used in this thesis have been given.Based on the above system introduction,in a single-sensor system,we explored the impact of linear deception attack launched by the attacker on the remote estimation error covariance under the KL divergence detector in two cases,and obtained the analytic solution for a one-dimensional optimal attack.Then in a multi-sensor system,we propose a watermark encryption method against a linear deception attack under the X2 detector.According to the attacker's understanding of the water-mark,the discussion can be divided into two situations.The first is when the attacker does not know the existence of the watermark and launches a linear deception attack.Our watermark can help the X2 detector detect the original attack that is not detected by the detector.The second type is that when the attacker knows that there is a watermark,the attack launched by the attacker will not affect the estimation and control performances of the system if we set the parameters of the watermark reasonably.In addition,based on different situations,simulations are given to further verify the effectiveness of the proposed watermark encryption.At the same time,we compare the watermark encryption method in this thesis with other methods proposed by scholars at home and abroad.The watermark method in this thesis can greatly improve the detection rate and protect the estimation and control performance of the system,which reflects the advantages of the proposed watermark in this thesis.Finally,for an event-based stealthy attack in an event-triggered system,we analyze the situation that the attacker can tamper with the sensor transmission decision but cannot tamper with the feedback information.A detection mechanism is proposed,that is,adding an improved X2 detector to the sensor side of the system to detect such an attack.Simulation results show that our method can detect such an intelligent deception attack.