Research On Extraction Technology Of CPS Security Situation Elements Based On Multi-Source Information Fusion

Cyber-physical system(CPS)is a multi-dimensional complex system that integrates computing,network and physical environment.Compared with traditional physical systems,the operating environment of cyber-physical systems has changed from being closed and isolated to being more open and interconnected,which has also led to more and more prominent security issues of CPS.Once the CPS is attacked or abnormal,it will not only cause property losses,but even pose a threat to national security.Therefore,protecting the security of CPS becomes very important.Situational awareness technology is an important means of security protection,and situational element extraction is an important part of situational awareness.At present,most of the research targets on situational element extraction are network systems,while CPS combines network systems with physical systems to form an intelligent system.Only considering the network security information in the CPS environment cannot accurately extract the elements of the CPS security situation.Therefore,for the extraction of CPS security situation elements,it is necessary to integrate various security information.This thesis proposes a CPS security situation element extraction model with reference to the network situation awareness model,and extracts CPS security situation elements based on multi-source information fusion technology.In the CPS security situation element extraction model,this thesis divides CPS into network subsystem,physical subsystem and security subsystem according to the source of security threats faced by CPS.The network subsystem involves computer networks and communication facilities,the physical subsystem includes hardware components such as sensors and actuators,and the security subsystem is responsible for monitoring the CPS from abnormal personnel on-site intrusion.This thesis proposes different situational element extraction methods for the three subsystems,respectively extracting CPS network security situational elements,physical security situational elements and protection security situational elements.For the CPS network subsystem,a compound fusion model is proposed.In the simulation experiment of the KDD CUP99 dataset,it is verified that the composite fusion model can effectively avoid the problem that the characteristics of individual information sources are not obvious compared with the full information source feature level fusion,and further improve the effect and accuracy of element extraction.For the CPS physical subsystem,a physical security situation element extraction method based on spatio-temporal fusion is proposed.In the simulation experiment of SWa T dataset,the feasibility and effectiveness of the method model are verified.For the CPS security subsystem,elements of the security situation are extracted based on face recognition technology.In the simulation experiment,it can effectively identify multi-pose face information,and extract security situation elements according to the legitimacy of personnel.Finally,based on the improved D-S evidence theory,the situational elements of the three subsystems are integrated to realize the extraction of CPS overall situational elements and the extraction of CPS composite situational elements,which are used to evaluate the overall security situation of CPS,and also provide a basis for situational element understanding and situation prediction after situational awareness.