Research On Replay Attack Detection Methods Of Cyber Physical Systems Based On Coding

Cyber physical systems(CPSs)are complex systems integrated of sensing communication,calculation and control process.At present,cyber physical systems are widely used in many fields such as smart grids,medical monitoring,smart factories,intelligent transportation and so on.Different from the closed industrial internal network in the past,the access of communication network not only enhances the operating efficiency of the system dramatically,but also brings many security problems.In recent years,attacks on cyber physical systems have occurred frequently,causing tremendous economic and social losses,which have attracted great attention from domestic and foreign scholars and government agencies.Replay attacks are one of the most common and easily achievable attack methods in cyber physical systems.Replay attackers can successfully launch attacks without knowing any knowledge of the system model,and destroy the normal operation of the system.Therefore,it is of great practical and economic significance to accurately detect replay attacks against cyber physical systems.In this paper,the detection methods of replay attacks against cyber physical systems are studied based on the linear time-invariant system model,and the following research results are obtained:(1)Considering the controlled object of the cyber physical systems as a linear timeinvariant system model,the attack model of replay attacks and the stability of the system after being attacked are studied.Results have shown that when the model parameters of the system satisfy some conditions,the system is no longer stable after suffering from replay attacks,apparently,the control performance of the system will be seriously affected.It shows the necessity of studying the detection methods of replay attacks.The accuracy and reliability of the above analysis have been verified by simulation experiments conducted on the Matlab platform.(2)Some scholars have proposed that replay attacks can be detected by adding random Gaussian noise to the control signal.Since the control signal at this time is not optimal,this method is at the cost of certain system performance in exchange for the attack detection rate.On this basis,in order to reduce the sacrificed system performance,a control signal coding detection scheme based on auxiliary information is proposed: adding the corresponding auxiliary signal to the measured value when the control signal is encoded.First,the detectability of the replay attacks under this scheme is proved,and the relationship between the upper bound of the detection rate and the detection function threshold is derived.Second,it is proved that less control performance of the system will be sacrificed by using the proposed scheme.Third,an optimization problem is given to calculate the coding signal covariance matrix.The simulation experiment results of Matlab confirm the effectiveness of the proposed replay attack detection scheme based on auxiliary information compensation control signal coding.(3)Since the control signal coding scheme will inevitably cause system performance loss,how to detect replay attacks without sacrificing the control performance of the system is a problem worthy of study.In this paper,a measurement noise coding scheme based on system performance index is put forward.First,according to the given system performance index fluctuation threshold,the maximum allowable measurement noise can be obtained,and this noise is added to the sensor output to encode the measurement value,which is decoded at the state estimation end.At last,combining with the fading Kalman filter algorithm,the system performance can be estimated in real time.When the system is not under attack,it will not cause damage to the system performance.After the replay attack occurs,the system suffers from noise that exceeds the allowable range due to the asynchronous encoding and decoding,which causes the performance index to change and exceed a given range,so that the attack can be detected.The simulation experiment of Matlab verifies the effectiveness of the proposed scheme.