Industrial control system vulnerability dynamic evaluation method based on vulnerability association attack graph
Through the vulnerability association-based attack graph generation method, combining vulnerability exploitation score, harm and device node importance, cross-domain and cross-layer attack graphs are generated, solving the problem that multi-step attack cannot be comprehensively analyzed in the existing technology and realizing industrial Dynamic real-time assessment and security enhancement of control system vulnerability.
Patent Information
- Application Number
- CN202410939283.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-12
AI Technical Summary
The existing technology cannot comprehensively analyze multi-step attacks in industrial control systems, and it is easy to ignore the relationship between equipment, vulnerabilities and attack paths, and the lack of timeliness of vulnerabilities, resulting in the incomplete and real-time security assessment of industrial control systems.
Using the attack graph generation method based on vulnerability association, establish industrial equipment situation information-industrial vulnerability information-attack behavior knowledge graph, build vulnerability quantitative indicators with vulnerability exploitation scores, vulnerability hazards and device node importance as the core, and generate cross-domain Cross-layer attack map, combining quantitative indicators and attack maps, realizes dynamic comprehensive real-time assessment of vulnerability to industrial control systems.
By considering vulnerability correlation and timeliness, a comprehensive analysis of multi-step attack paths in industrial control systems is improved scientific assessment of potential threats, and the system's security protection and rapid recovery capabilities are enhanced.
Smart Images

Figure CN118940275A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of Internet data technology, relates to security analysis research under industrial control systems, and specifically to a dynamic vulnerability assessment method for industrial control systems based on attack graphs. Background Art
[0002] Industrial Control Systems (ICS) are the "brains" of industrial production. In the fourth industrial revolution, the reliable operation of ICS is a prerequisite for the transformation of manufacturing to intelligence and service. At present, there are more and more information security attacks on industrial control systems, the number of industrial systems and equipment exposed to the Internet is increasing day by day, and the number of high-risk vulnerabilities is also high. The security situation of industrial control systems at home and abroad is very severe. Therefore, it is urgent to reveal the vulnerability assessment mechanism of industrial control systems, assist in discovering potential vulnerabilities, attack paths and hazards caused by different attack methods in ICS, so as to improve the security protection effect, reduce the protection cost and enhance the rapid recovery ability of the system.
[0003] Traditional assessments mainly rely on isolated vulnerability scanning tools, which cannot fully analyze the common multi-step attacks in ICS environments, tend to ignore the association between devices, vulnerabilities and attack paths, and lack consideration of the timeliness of vulnerabilities. However, attackers do not attack the vulnerabilities on a certain device in the IoT device system in isolation, but use multiple vulnerabilities to conduct multi-step attacks from different devices and different attack paths to achieve the purpose of attacking devices. The prior art proposes to use attack graph generation algorithms to perform security analysis on attack path predictions to infer the attack stage of vulnerabilities and realize system-level IoT system security analysis; or use system logs to detect threats and predict attack paths. However, the current public research on network security analysis using attack graphs does not take into account the impact of vulnerability associations and device importance on atomic attacks, and does not achieve dynamic, comprehensive and real-time reflection of network vulnerabilities. Summary of the invention
[0004] In view of this, in order to address the deficiencies of the above-mentioned prior art, the purpose of the present invention is to provide an industrial control system dynamic vulnerability assessment method based on vulnerability association attack graph, establish a formal representation of the knowledge graph, construct an ICS vulnerability assessment index system, generate an attack graph based on vulnerability association pre- and post-conditions, and combine quantitative indicators with the attack graph to scientifically and rationally evaluate the vulnerability of each attack path, thereby achieving dynamic, comprehensive and real-time reflection of network vulnerability.
[0005] To achieve the above object, the technical solution adopted by the present invention is: a dynamic vulnerability assessment method for industrial control systems based on vulnerability association attack graph, comprising the following steps: S1. Integrate data based on multi-channel and multi-source semi-structured data in industrial scenarios to establish a formal representation of the industrial equipment scenario information-industrial vulnerability information-attack behavior knowledge graph; S2. Taking attack expectation as the overall indicator, we build a vulnerability quantitative index by combining the vulnerability exploitability score, vulnerability harm, and device node importance. S3. Generate a cross-domain and cross-layer attack graph based on vulnerability association using a cross-domain and cross-layer attack graph generation algorithm, combine vulnerability quantification indicators, attack expectation values and attack graphs, and output an attack graph that integrates attack expectation values; S4. Calculate the atomic attack expectation of each step in the attack process, analyze all possible attack paths based on the attack graph that integrates the attack expectation value, and realize dynamic vulnerability quantitative assessment of the industrial Internet attack path.
[0006] Furthermore, the vulnerability quantitative indicators in step S2 include the following: a: Vulnerability exploitability score Vul exp , represents the possibility of successfully exploiting a vulnerability to implement an attack, and the vulnerability life cycle function and the vulnerability basic exploitability score are considered in the calculation; b. Vulnerability sev , which indicates the damage caused to ICS by attackers using vulnerabilities. The vulnerability attributes are graded comprehensively and the weights of related indicators are assigned; c. Device node importance Dev imp , which indicates the importance of each device node in the attack graph to the system security, and is evaluated based on factors including the importance and device value of the node in the topology.
[0007] Furthermore, the attack expectation is the product of the vulnerability exploitability score, the vulnerability severity, and the importance of the device node.
[0008] Furthermore, the cross-domain and cross-layer attack graph generation algorithm based on vulnerability association generates a cross-domain and cross-layer attack graph, which specifically includes the following contents: The industrial security knowledge graph in S1 is used as the input for automatically generating attack graphs. The postconditions of previous vulnerabilities are associated with the preconditions of subsequent vulnerabilities, and the attack paths are formed in combination with the quantified results of the nodes. The cross-domain and cross-layer attack graphs are generated with reference to the expected quantified values of the attacks. The attack graphs are used to reflect the possibility of ICS nodes being attacked and to characterize the intrusion paths of attackers within the system.
[0009] The beneficial effects of the present invention are: The industrial control system dynamic vulnerability assessment method of the present invention takes into account vulnerability correlation, integrates vulnerability attributes into assessment quantification values, generates cross-domain and cross-layer attack graphs based on vulnerability correlation pre- and post-conditions, pays attention to the heterogeneity of different vulnerability attributes, and solves the problem that the correlation between devices, vulnerabilities and attack paths is easily ignored in the prior art; Combine quantitative indicators with attack graphs, expand the vulnerability data dimension, propose a definition of attack expectation indicators, form an ICS vulnerability assessment indicator system with vulnerability exploitability score, vulnerability harm, and device node importance as the core, comprehensively analyze multi-step attacks in the ICS environment, comprehensively and dynamically analyze potential hidden dangers and threats in industrial control systems, and realize scientific assessment of the vulnerability of industrial Internet attack paths; The impact of vulnerability timeliness on ICS is taken into consideration, and the vulnerability patch date is incorporated into the vulnerability lifecycle function when calculating attack expectations, thus achieving real-time dynamic evaluation of industrial control systems. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0011] Figure 1 It is a framework diagram of the method for dynamic vulnerability assessment of industrial control systems of the present invention; Figure 2 It is a flow chart of the method for dynamic vulnerability assessment of industrial control systems of the present invention; Figure 3 It is a principle diagram of the attack expectation quantification index in the evaluation method of the present invention; Figure 4 A schematic diagram of vulnerability life cycle risk in the assessment method of the present invention; Figure 5 is the network topology structure of SWaT in Example 2 of the present invention; Figure 6 It is an attack graph based on vulnerability association in the SWaT scenario in Example 2 of the present invention; Figure 7 A schematic diagram of the device attack expectation calculated in Example 2 of the present invention; Figure 8 This is a comparison chart of the exploitability score of each device vulnerability-vulnerability harmfulness-device node importance index in Example 2 of the present invention. DETAILED DESCRIPTION
[0012] The following specific examples are given to further clearly, completely and in detail illustrate the technical solution of the present invention. This example is the best example based on the technical solution of the present invention, but the protection scope of the present invention is not limited to the following examples.
[0013] Example 1 Traditional assessments mainly rely on isolated vulnerability scanning tools, which cannot fully analyze the common multi-step attacks in ICS environments, easily overlook the relationship between devices, vulnerabilities and attack paths, and lack consideration of the timeliness of vulnerabilities. However, attackers do not attack the vulnerabilities of a certain device in the IoT device system in isolation, but use multiple vulnerabilities to conduct multi-step attacks from different devices and different attack paths to achieve the purpose of attacking devices.
[0014] The present invention proposes a method for dynamic vulnerability assessment of industrial control systems based on vulnerability association attack graphs, which specifically includes the following contents: Step 1: Integrate the data based on multi-channel and multi-source semi-structured data in industrial scenarios, and establish a formal representation of the industrial equipment scenario information-industrial vulnerability information-attack behavior knowledge graph.
[0015] Step 2: Take attack expectation as the overall indicator, combine the three dimensions of vulnerability exploitability score, vulnerability harm and device node importance to construct vulnerability quantitative indicators, and conduct quantitative assessment of ICS vulnerability based on the indicators.
[0016] Referring to the quantitative assignment of the CVSS standard, the atomic attack expectation of each step in the attack process is calculated to obtain the total attack expectation of each path in the attack graph, and the potential dangerous path of ICS is discovered; the attack expectation is the product of the vulnerability exploitability score, vulnerability severity, and device node importance. The specific definition of the vulnerability quantitative index is as follows: 1. Vulnerability exploitability score Vul exp , represents the possibility of successfully exploiting a vulnerability to implement an attack, and the vulnerability life cycle function and the vulnerability basic exploitability score are considered in the calculation; Vul exp =G(tv,tp)·Vul score ; A higher vulnerability exploitability score means that the vulnerability is more likely to be exploited by attackers, thus posing a more serious threat to the system or network. Regarding vulnerability exploitability scores, CVSS provides a general standard and defines the calculation formula as follows: Vul score =8.22×AV×AC×PR×UI; Among them, AV (Attack Vector, attack vector), AC (Attack Complexity, attack complexity), PR (Privileges Required, permission requirements), UI (User Interaction, user interaction) For specific grade assignment, please refer to CVSS; However, the above vulnerability exploitability scores only reflect the theoretical feasibility of the vulnerability and do not take into account the impact of the vulnerability's timeliness; however, ICS can usually maintain a long service life. As attackers deepen their research on vulnerabilities and the corresponding attack scripts mature, the probability of exploiting vulnerabilities will increase significantly. Figure 4 As shown in the figure, in the entire vulnerability life cycle, the vulnerability risk changes dynamically and is closely related to the cycle of the vulnerability. The characteristic values recorded in CVSS are static and discrete. In order to reveal the law of vulnerability changes over time, the life cycle function G(t v ,t p ): Among them, t p is the vulnerability patching time, t v is the effective survival time of the vulnerability, which is calculated as the time difference between the vulnerability exposure date and the vulnerability scoring date; constants α and constant k are defined as shape factors. In this embodiment, the values are 0.26 and 0.00161 respectively.
[0017] In order to fully understand the life cycle of the vulnerability, the vulnerability patching time t is introduced p , calculated as the time difference between the date the vulnerability is patched and the date the CVSS scores the vulnerability. When it is smaller, the original life cycle function value decreases, and vice versa. However, once the vulnerability is patched, its vulnerability exploitability score should decrease accordingly. Through the vulnerability life cycle function, ICS can be dynamically evaluated in real time based on the existence time of different vulnerabilities in the real world.
[0018] 2. Vulnerability sev , indicating the damage caused to ICS by attackers using vulnerabilities, usually through AV, AC, PR, UI and vulnerability impact value (Vul imp ) and a series of indicators are used to evaluate. Taking into account the classification of various attributes of vulnerabilities, referring to domestic and foreign vulnerability standards, the weights of relevant indicators are assigned as follows: In industrial control systems, confidentiality, integrity, and availability are of great significance for ensuring the normal operation, safety, and stability of industrial processes. The impact of vulnerability attacks is mainly reflected in confidentiality, integrity, and availability. For the impact value of a single vulnerability, CVSS provides relevant calculation standards, and the calculation formula is: Impact(v)=10.41[1-(1-C)(1-I)(1-A)]; Among them, C (Confidentiality), I (Integrity), A (Availability) For specific values, please refer to CVSS.
[0019] 3. Device node importance Devimp, which indicates the importance of each device node to the system security in the attack graph. It is evaluated based on factors including the importance and device value of the node in the topology structure: Dev imp =EC×Dev value ; EC is the eigenvector centrality, which is a measure of the global importance of a node in the network. It takes into account the importance of the node's neighbor nodes and the overall network structure. Therefore, nodes with higher eigenvector centrality may play a more critical role in the network and have a greater impact on risk propagation or the overall function of the network. The eigenvector centrality of a device node is calculated as follows: Where λ is the maximum eigenvalue of the adjacency matrix A in the industrial equipment topology diagram; x = (x1, x2, ..., x n ) T is the eigenvector corresponding to the maximum eigenvalue of the adjacency matrix A. In ICS, there are upper computers and lower computers according to the position of the components. They usually play different roles in the industrial control system. The upper computers mainly include user computers, workstations, HMI, SCADA, etc., and the lower computers mainly include PLC, single-chip microcomputer, remote control unit RTU, etc. Each component in ICS has different functions and values. As shown in Table 1, the value of each industrial control component Dev value Reference standard "GB / T 20984-2007 Information Security Technology Information Security Risk Assessment Specification". Table 1 Classification and meaning of industrial control components
[0020] Step 3: Generate a cross-domain and cross-layer attack graph based on the vulnerability association cross-domain and cross-layer attack graph generation algorithm, combine the vulnerability quantification index, attack expectation value and attack graph, and output the attack graph that integrates the attack expectation value.
[0021] In the attack graph, an atomic attack actually represents a vulnerability exploitation behavior, so the vulnerability is the core of the attack graph. On the one hand, the vulnerability exists in dependence on the hardware and software entities; on the other hand, the existence of the vulnerability is a necessary prerequisite for the attack to occur. The formation of the attack chain depends on the single-step attack. By evaluating the resources required and obtained by the single-step attack, the correlation between different single-step attacks can be determined. Therefore, during the attacker's attack, the demand and acquisition of related resources are set as the preconditions and postconditions of the attack process. When counting vulnerabilities, attention should be paid to the heterogeneity of different vulnerability attributes, so they are processed under different conditions to achieve the automatic generation of all vulnerability pre- and postconditions.
[0022] Generating a cross-domain and cross-layer attack graph includes the following: The industrial security knowledge graph in step 1 is used as the input for automatically generating an attack graph. The postconditions of the previous vulnerability are associated with the preconditions of the subsequent vulnerability, and the attack path is formed in combination with the quantified results of the nodes. The cross-domain and cross-layer attack graph is generated with reference to the expected quantified values of the attack. The attack graph is used to reflect the possibility of ICS nodes being attacked and to characterize the attacker's intrusion path within the system.
[0023] Step 4: Calculate the atomic attack expectation of each step in the attack process, and perform vulnerability analysis on all possible attack paths based on the generated attack graph of the fused attack expectation value.
[0024] Example 2 The present invention proposes a method for dynamic vulnerability assessment of industrial control systems based on vulnerability association attack graphs, which specifically includes the following contents: Firstly, based on the formal representation of industrial equipment context information, industrial vulnerability information and attack behavior knowledge graph, a series of vulnerability quantitative indicators are proposed with attack expectation as the overall indicator and combined with the three dimensions of vulnerability exploitability score, vulnerability harm and device node importance. Secondly, a method for calculating attack expectation considering vulnerability life cycle is designed. Combined with the attack expectation value, a cross-domain and cross-layer attack graph generation algorithm based on vulnerability association is proposed. Finally, the atomic attack expectations of each step in the attack process are used to conduct vulnerability analysis on all possible attack paths, comprehensively and dynamically analyze potential hidden dangers and threats in industrial control systems, and realize scientific assessment of the vulnerability of industrial Internet attack paths.
[0025] The following is a further explanation of the method in conjunction with the specific experimental content. The SWaT industrial water treatment plant proposed by iTrust, the Cyber Security Center of Singapore Polytechnic, is evaluated on the laboratory's internal industrial Internet platform enterprise security protection system. Figure 4 shown.
[0026] SWaT consists of a layered communication network, a programmable logic controller (PLC), a human-machine interface (HMI), a supervisory control and data acquisition (SCADA) workstation, and a historical control system (Historian). Data from sensors can be used by the SCADA system and recorded by the Historian for subsequent analysis. The vulnerability information on each component in the system is shown in Table 2. Table 2 Information device configuration information
[0027] For the above experimental scenarios, the attack graph is generated by the automatic generation algorithm based on vulnerability pre- and post-conditions and the automatic generation algorithm based on vulnerability association. Figure 5 As shown in the figure, the ellipse represents the attacker and the device, and the square represents the vulnerability model and the vulnerability pre- and post-conditions. The vulnerability node not only covers the pre- and post-conditions of the device vulnerability, but also uses these conditions to associate the order of the pre- and post-conditions to form an attack path. Taking the vulnerability life cycle into account in the calculation of the expected attributes of the attack and showing it in the attack relationship provides a specific reference for the quantitative assessment of the vulnerability.
[0028] According to the method of the present invention, after calculating the quantified values of vulnerability exploitation probability, vulnerability harm and device node importance, the attack expectation of each device is calculated, such as Figure 6 As shown. Taking PLC4 as an example, the device vulnerability model is CVE-2017-6034. According to the quantitative results of each attribute of the vulnerability in CVSS, the vulnerability exploitability score is 2.466, the vulnerability severity is 4.328, and the device node importance is 0.993. Therefore, the attack expectation of the device is 0.754 after normalization with the sigmoid function.
[0029] Thus, the total attack expectation on each attack path can be calculated, which is defined as the sum of the attack expectations of each device on the attack path. The top five attack paths with the largest total attack expectations are given, as shown in Table 3. Table 3 Attack paths and total expected attacks
[0030] As can be seen from Table 3, since the three factors of vulnerability exploitability score, vulnerability severity and device node importance jointly determine the expected value of attack, PLC has a greater impact in this ICS, so attacking PLC will obtain higher attack benefits and impacts, and it is necessary to focus on strengthening security protection measures. Figure 7The results show that the exploitability scores of Router1 and HMI are relatively high, and they are relatively easier to exploit. From the perspective of device node importance, Router2 is an important node connecting various devices in the topology structure. Once attacked, the impact range is large. PLC has the highest device value in the entire ICS, so both of them show high importance, and their security directly affects the stability and security of the entire control system.
[0031] In summary, the method for dynamic vulnerability assessment of industrial control systems based on vulnerability association attack graphs of the present invention can obtain a cross-domain and cross-layer attack graph based on vulnerability association through vulnerability pre- and post-conditions, and use the attack graph to reflect the possibility of ICS nodes being attacked, and characterize the intrusion path of attackers inside the system; using attack expectation as a total quantitative indicator, the nodes on the attack graph are quantitatively evaluated, so as to use the atomic attack expectation of each step in the attack process to perform vulnerability analysis on all possible attack paths, thereby realizing real-time dynamic assessment of industrial control systems.
[0032] The above shows and describes the main features, basic principles and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements according to actual conditions, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. Dynamic assessment method of industrial control system vulnerability based on vulnerability association attack graph, It is characterized in that The following steps are involved: S1. Integrate data based on multi-channel and multi-source semi-structured data in industrial scenarios to establish a formal representation of the industrial equipment scenario information-industrial vulnerability information-attack behavior knowledge graph; S2. Taking attack expectation as the overall indicator, we build a vulnerability quantitative index by combining the vulnerability exploitability score, vulnerability harm, and device node importance. S3. Generate a cross-domain and cross-layer attack graph based on vulnerability association using a cross-domain and cross-layer attack graph generation algorithm, combine vulnerability quantification indicators, attack expectation values and attack graphs, and output an attack graph that integrates attack expectation values; S4. Calculate the atomic attack expectation of each step in the attack process, analyze all possible attack paths based on the attack graph that integrates the attack expectation value, and realize dynamic quantitative assessment of the vulnerability of the industrial Internet attack path.
2. According to claim 1, the method for dynamic assessment of industrial control system vulnerability based on vulnerability association attack graph, It is characterized in that The vulnerability quantitative indicators in step S2 include the following: a. Vulnerability Exploitability Score , represents the possibility of successfully exploiting a vulnerability to implement an attack, and the vulnerability life cycle function and the vulnerability basic exploitability score are considered in the calculation; b. Danger of vulnerability , which indicates the damage caused to ICS by attackers using vulnerabilities. The weights of relevant indicators are assigned by comprehensively considering the classification of various attributes of the vulnerability; c. Importance of device nodes , which indicates the importance of each device node in the attack graph to the system security, and is evaluated based on factors including the importance and device value of the node in the topology.
3. According to claim 2, the method for dynamic assessment of industrial control system vulnerability based on vulnerability association attack graph, It is characterized in that When calculating the vulnerability exploitability score, the vulnerability patching time is taken into account in the vulnerability lifecycle function.
4. According to claim 3, the method for dynamic assessment of industrial control system vulnerability based on vulnerability association attack graph, It is characterized in that The vulnerability patching time is the time difference between the date when the vulnerability is patched and the date when the CVSS scores the vulnerability.
5. According to claim 1, the method for dynamic assessment of industrial control system vulnerability based on vulnerability association attack graph, It is characterized in that The attack expectation is the product of the vulnerability exploitability score, the vulnerability severity, and the importance of the device node.
6. The method for dynamic vulnerability assessment of industrial control systems based on vulnerability association attack graph according to claim 1, It is characterized in that The S3 specifically includes the following contents: The industrial security knowledge graph in S1 is used as the input for automatically generating attack graphs. The postconditions of previous vulnerabilities are associated with the preconditions of subsequent vulnerabilities, and the attack paths are formed in combination with the quantified results of the nodes. The cross-domain and cross-layer attack graphs are generated with reference to the expected quantified values of the attacks. The attack graphs are used to reflect the possibility of ICS nodes being attacked and to characterize the intrusion paths of attackers within the system.
Citation Information
Cited By
Saving insurance evaluation method and system
CN120896719A
Industrial internet multi-vulnerability association analysis method
CN122069111A
An industrial internet multi-vulnerability correlation analysis method
CN122069111B
A closed-loop adaptive security protection method and device for a power monitoring system and a storage medium
CN122697697A
