Research On Linear Deception Attack And Detection Strategy For Multi-Sensor Cyber-Physical System

With the advancement of technology,the interaction and integration between the information world and the physical world has become more and more profound,gradually forming the Cyber-Physical System(CPS),which integrates computing,communication and control.CPS has a wide range of applications in industrial manufacturing,traffic management,power systems and medical services,which have greatly changed the way of production and life of human beings.However,the interconnection of everything in the world often requires the use of technologies such as network communications,which makes CPS require higher-level openness,and higher-level openness means that the system will be exposed to more potential threats of cyber-attacks.Linear deception attack as a typical form of False Data Injection(FDI)attack is able to achieve stealth attacks while causing damage to system performance.Once the system is attacked,it will cause unpredictable consequences.Therefore,it is of great theoretical value and practical significance to conduct an in-depth study of linear deception attack.From the perspectives of attack and detection,this paper focuses on the security issues related to linear deception attack under multi-sensor CPS and the main research contents are as follows:(1)From the attacker’s perspective,this paper proposes a new method for designing the parameters of the linear deception attack strategy.Firstly,the basic form of the linear deception attack based on residuals and the traditional feasibility constraint are given.In view of the volatility of the detection function values obtained by the chi-square detector,the constraint is considered to be appropriately widened to allow a certain degree of deviation in the designed residual distribution.In conjunction with Kalman Filter,the evolution of the estimated error covariance in this case is derived and used to quantify the impact of the attack on the system performance.On this basis,an objective function is constructed based on the pre-set deviations,and the problem of designing the attack parameters is transformed into an optimization problem to be solved.By comparing with existing attack strategy,it is verified that this attack scheme is able to evade the detector to achieve a stealthy attack while causing stronger damage to system performance.(2)From the defender’s perspective,for the security detection problem of multisensor CPS under linear deception attack,this paper proposes a detection scheme based on state estimation.Firstly,different cases of whether there is a secure sensor set in the system are classified.Combined with Kalman Filter,the Euclidean distance between different state estimation is used as the basis for detection,and two detection judgment rules are designed to detect the security of the system.By comparing with existing detection method,it is verified that the proposed scheme can improve the detection rate and maintain high accuracy rate.(3)For the security detection problem of multi-sensor CPS under linear deception attack,this paper proposes a detection scheme based on the statistical characteristics.Firstly,considering the complexity of realistic multi-sensor CPS,a multi-layer distributed fusion structure is introduced and a multi-layer distributed data fusion algorithm based on the covariance intersection fusion criterion is proposed.Simulation results demonstrate that the algorithm can improve the fusion accuracy and reduce the impact of the attack on the system.Based on this structure,the detection function is designed to monitor the system state in real time by analyzing the statistical characteristics obeyed by the difference of the fusion estimation of different region nodes.Finally,the effectiveness of the detection scheme is demonstrated by simulation.