Analysis And Modeling Of Vulnerability In Industrial Control System

The meaning of security varies significantly from safety in Industrial Control System (ICS). Safety emphasizes protecting people from accident harm, referring mainly to property safety, personal safety and environmental safety; Security emphasizes protecting property (data) and services (e.g., denial of service) from malicious harm, referring mainly to communications security, data security, etc.This paper concern only ICS cyber security.ICS security faces severe threats due to the following factors:1) usage of standard protocols and technologies with known vulnerabilities;2) Increasing connectivity to external networks;3) Insecure and non-authentication access to control systems;4) Control systems being absent of security functionality; Attackers can easily exploit the vulnerabilities caused by these factors and attack the ICSs.The ICS security issues are imminent, in order to better carry out security defenses for control system, this thesis carries out the following works on control network vulnerability analysis and attack modeling.(1) Analysis of basic questions of control system vulnerability. For the question of the integration of control systems and corporate networks, this paper gives an ICS four layers network structure, selects the lower levels as the research object. For the vulnerabilities and attack methods existing in control network, this paper analyzes the vulnerabilities in the industrial Ethernet, fieldbus, industrial PC and industrial softwares in detail, and summarizes the attack methods; then analyzes two common network attack modeling technology, they are attack tree and attack graph.(2) Modeling network attack based on attack graph for vulnerabilities in control network. This paper gives attack graph generation process and system framework; then analyzes the applicability of the information network attack model to control network. On the basis of the above results, a security analysis-oriented network attack model is established; then gives definitions and formulas for some concepts, such as atomic attack gain, unit path gain,etc; finally, this paper proposes an attack objective function based attack graph generation algorithm.(3) To verify the effectiveness of the proposed model and algorithm, and find out the optimal attack goal, a control network experimental environment is constructed; then carrys out some experiments and results analysis.