Analysis Of Deception Models And Detection Algorithms On CPS Control Layer

Cyber-physical system (CPS) is the next intelligent system based on embedded system, computer network, control theory and wireless sensor network, which is widely applied in industrial control and infrastructure area. Currently, the cyber threat to information system and networked infrastructures is keeping increasing and evolving. Because CPS combines network transmission and physical process tightly, cyber attacks against CPS will have direct impacts on the physical objects controlled by the system. For the purpose of protecting CPS control layer, this paper has done the work as follows:(1) Models of5types of deception attacks against CPS control layer implemented through network is established from the perspective of attack results;(2) Dynamic state estimation methods are introduced into change detection algorithm to detect false data injection attacks. We analyze the detection performance of change detection algorithms based on ideal states and dynamic recursive estimation and their suitable application scenarios;(3) State detection algorithm is proposed by combining residual detection and change detection as a complement in attack scenarios that change detections cannot cover;(4) Dictionary model is introduced into data matching, and3modeling methods of detection based on dictionary are proposed to detect replay attacks;(5) Tennessee-Eastman Process (TEP) is utilized as the experiment platform to verify the performance of proposed detection algorithms and models, and the results are compared with those of the original algorithms.