A Certification Testing Method for Automotive Cybersecurity Based on Attack Path Reasoning
By constructing a vehicle safety knowledge model and performing attack path reasoning and regulatory clause matching, structured penetration test cases are generated, solving the problems of incomplete testing and difficulty in evidence tracing in existing technologies, and realizing more efficient and reliable automotive cybersecurity regulatory certification testing.
Patent Information
- Application Number
- CN202611485031.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-22
- Publication Date
- 2026-11-17
AI Technical Summary
Existing automotive cybersecurity regulatory certification testing methods lack systematic attack path reasoning and automatic mapping of regulatory clauses, resulting in omissions in multi-stage attack chains and mismatches between test cases and vehicle models, affecting the integrity and repeatability of certification testing.
By constructing a vehicle safety knowledge model, inferring candidate attack paths, parsing regulatory clauses and performing rule, entity, and semantic matching, generating structured penetration test cases and constrained scripts, executing tests in an isolated environment, collecting and solidifying evidence, and generating regulatory compliance matrices and reports.
It improves the completeness and consistency of attack path coverage, enables automatic matching of regulatory clauses and test cases, enhances the adaptability of test cases to vehicle architecture, strengthens the traceability and anti-tampering capabilities of authentication evidence, and reduces the workload of repetitive analysis after architecture or regulatory changes.
Smart Images

Figure CN123122983A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automotive cybersecurity testing technology, and in particular to a method for automotive cybersecurity regulatory certification testing based on attack path reasoning. Background Technology
[0002] As a crucial component of modern transportation, intelligent connected vehicles undergo cybersecurity regulatory certification testing. This testing employs a collaborative approach involving attack path reasoning, regulatory clause analysis, and penetration testing, constructing a technical system encompassing everything from vehicle architecture to compliance verification. Specifically, this system covers key aspects such as attack surface identification, multi-step attack chain reasoning, regulatory element extraction, and test case generation. Among these, vehicle electronic and electrical architecture modeling, vulnerability knowledge bases, and descriptions of protective measures collectively support the automated reasoning of attack paths.
[0003] However, existing certification testing methods rely heavily on manual experience or single-point tools, lacking a systematic approach to attack path reasoning and automatic mapping of regulatory clauses. This can lead to omissions in multi-stage attack chains, mismatches between test cases and vehicle models, or difficulties in evidence tracing, thus affecting the completeness and repeatability of certification testing. As vehicle platforms and regulatory versions continue to evolve, this experience-based approach struggles to reliably cover cross-domain attack chains and maintain consistent traceability between clauses, paths, scripts, and evidence. Summary of the Invention
[0004] The present invention aims to at least partially solve one of the technical problems in the related art.
[0005] Therefore, the first objective of this invention is to propose a method for automotive cybersecurity regulatory certification testing based on attack path reasoning.
[0006] Another objective of this invention is to propose a vehicle cybersecurity regulatory certification testing device based on attack path reasoning.
[0007] The third objective of this invention is to provide a computer device.
[0008] A fourth objective of this invention is to provide a non-transitory computer-readable storage medium.
[0009] To achieve the above objectives, a first aspect of the present invention proposes a method for automotive cybersecurity regulatory certification testing based on attack path reasoning, comprising: S1. Obtain the electronic and electrical architecture information of the target vehicle model, construct a vehicle model security knowledge model including asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. S2 parses the target regulatory clauses, extracts test items and evidence requirements, associates regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identifies high-risk paths not covered by regulations; S3 generates structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configures monitoring points, judgment criteria and evidence collection items, and executes tests in an isolated authorized environment; S4 collects raw data during the testing process, determines the performance of security mechanisms, and uses unified traceability to identify related regulatory clauses, attack paths, test cases, execution environments, and judgment results, solidifying evidence and generating regulatory compliance matrices and test reports.
[0010] In one embodiment of the present invention, S1 includes: The heterogeneous data in the acquired electronic and electrical architecture information is normalized by name, bound by version and extracted by relationship. The heterogeneous data is converted into nodes containing attack entry points, functional modules, electronic control units, buses, services, data, keys and security-critical assets, as well as edges containing physical connections, logical communication, trust relationships, call relationships, data flows and exploitable relationships, to construct the vehicle safety knowledge model. The vehicle safety knowledge model is represented as a directed attribute graph. Starting from the set of attack entry points and targeting the set of safety-critical assets, path expansion is performed when the current state and the attack prerequisites of the candidate edges are met and the candidate edges are applicable to the target vehicle version. The set of candidate attack paths is obtained through reasoning using a constrained search algorithm.
[0011] In one embodiment of the present invention, it further includes: For any candidate edge, check whether the current state satisfies the attack prerequisites of the candidate edge and whether the candidate edge is applicable to the target vehicle version. Path expansion is allowed only if both are satisfied, and the permissions or capabilities generated by the postconditions of the candidate edge are added to the current state. Based on the maximum path length, prohibition of duplicate nodes, permission monotonicity, number of trust boundary crossings, version consistency, and mutual exclusion constraints of attack actions, the candidate attack path set is generated from the attack entry point set to the security critical asset set through depth-first search, breadth-first search, heuristic search, or graph neural network reasoning.
[0012] In one embodiment of the present invention, S2 includes: The target legal provisions are broken down into provisions identifiers, applicable objects, attack surfaces, attack methods, target assets, expected protective behaviors, test environment limitations, judgment criteria, and evidence requirements through rule templates, terminology dictionaries, or natural language models. The candidate attack paths are precisely matched with the elements of the disassembled legal provisions in terms of terms and entities. Then, semantic matching is performed based on the attack entry point, action sequence, target assets and harm. The candidate attack paths are verified to meet the prerequisites and monitoring capabilities required by the provisions. Candidate attack paths with a comprehensive matching score reaching the first threshold are marked as mandatory test candidates under regulations; candidate attack paths in the review period are marked as manually confirmed items; and candidate attack paths that do not match the clauses but have a risk score higher than the second threshold are marked as recommended test items.
[0013] In one embodiment of the present invention, it further includes: The attack entry point, action sequence, target asset, and harm of the candidate attack path are precisely matched at the entity level with the attack surface, attack method, and target asset in the legal clause elements to generate an entity matching score. Based on the entity matching score, the semantic similarity between the candidate attack path and the legal clause element is calculated to generate a semantic matching score. The rule matching score, entity matching score and semantic matching score are weighted and summed to obtain the comprehensive matching score.
[0014] In one embodiment of the present invention, S3 includes: Generate structured penetration test cases based on the associated attack paths and regulatory test items, including the pre-environment, test target, path steps, input parameters, tool adapters, monitoring points, expected protective behaviors, stopping conditions, recovery actions, and evidence lists. Select an operation from a validated library of action templates, bind parameters based on vehicle model and interface definition, and generate a constrained test script; Before test execution, verify the authorized scope, target resources, script signature, hazard level, rate limit, timeout and emergency stop conditions, and perform the test in a software-in-the-loop, hardware-in-the-loop or digital twin environment.
[0015] In one embodiment of the present invention, S4 includes: During the test execution, the monitoring module collects communication traffic, interface response, controller status, system logs, screenshots, and time information as raw data. Based on the expected protective behaviors and safety mechanisms of the regulations, the test results are judged as compliant, partially compliant, non-compliant, indeterminate, or environmentally abnormal. Each piece of evidence is bound to a unified traceability identifier, which is associated with the vehicle model and software version, regulatory version and clauses, attack path version, use case and script version, tool and execution environment, executor or service identity, start and end time and judgment result. A summary value is generated for the original evidence, and together with the operation log, timestamp and storage location, it is written into an evidence list that can only be appended or has anti-tampering capabilities, generating a regulatory compliance matrix and test report.
[0016] To achieve the above objectives, a second aspect of the present invention provides a vehicle cybersecurity regulatory certification testing device based on attack path reasoning, comprising: The vehicle safety knowledge modeling module is used to obtain the electronic and electrical architecture information of the target vehicle, construct a vehicle safety knowledge model that includes asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. The regulatory path association analysis module is used to parse target regulatory clauses, extract test items and evidence requirements, associate regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identify high-risk paths not covered by regulations. The penetration test case generation and execution module is used to generate structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configure monitoring points, judgment criteria and evidence collection items, and execute tests in an isolated authorized environment; The test evidence compliance archiving module is used to collect raw data during the testing process, determine the performance of security mechanisms, and uniformly trace and identify related regulatory clauses, attack paths, test cases, execution environments and judgment results, solidify evidence and generate regulatory compliance matrices and test reports.
[0017] The present invention provides a method and apparatus for automotive cybersecurity regulatory certification testing based on attack path reasoning, which can improve the completeness and consistency of attack path coverage, achieve automatic matching of regulatory clauses and test cases, improve the adaptability of test cases to vehicle architecture, enhance the traceability and anti-tampering capability of certification evidence, and reduce the workload of repetitive analysis after architecture or regulatory changes.
[0018] To achieve the above objectives, a third aspect of this application provides a computer device, including a processor and a memory; wherein the processor runs a program corresponding to the executable program code by reading executable program code stored in the memory, for implementing the method described in the first aspect embodiment.
[0019] To achieve the above objectives, a fourth aspect of this application provides a non-transitory computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described in the first aspect.
[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0021] Figure 1 This is a flowchart of a vehicle cybersecurity regulatory certification testing method based on attack path reasoning according to an embodiment of the present invention; Figure 2 This is a system module architecture and data flow diagram according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a regulatory certification testing method based on attack path reasoning according to an embodiment of the present invention; Figure 4 This is a structural diagram of an automotive cybersecurity regulatory certification testing device based on attack path reasoning according to an embodiment of the present invention; Figure 5 It is a computer device according to an embodiment of the present invention. Detailed Implementation
[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0024] The following description, with reference to the accompanying drawings, describes a method and apparatus for automotive cybersecurity regulatory certification testing based on attack path reasoning, according to an embodiment of the present invention.
[0025] Figure 1 This is a flowchart of a vehicle cybersecurity regulatory certification testing method based on attack path reasoning according to an embodiment of the present invention, such as... Figure 1 As shown, it includes: S1. Obtain the electronic and electrical architecture information of the target vehicle model, construct a vehicle model security knowledge model including asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. S2 parses the target regulatory clauses, extracts test items and evidence requirements, associates regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identifies high-risk paths not covered by regulations; S3 generates structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configures monitoring points, judgment criteria and evidence collection items, and executes tests in an isolated authorized environment; S4 collects raw data during the testing process, determines the performance of security mechanisms, and uses unified traceability to identify related regulatory clauses, attack paths, test cases, execution environments, and judgment results, solidifying evidence and generating regulatory compliance matrices and test reports.
[0026] In one embodiment of the present invention, such as Figure 2 As shown, the system includes at least a data access and standardization module, a vehicle safety knowledge model, an attack path inference engine, a regulatory element parser, a regulatory path matching module, a risk assessment and ranking module, a test case generation module, an execution scheduling and security control module, a monitoring and result judgment module, an evidence solidification module, a report generation module, and a change impact analysis module. These modules can be deployed on the same computing device or distributed through service interfaces.
[0027] Furthermore, the construction of the vehicle safety knowledge model first involves receiving the target vehicle's electronic and electrical architecture, network topology, communication matrix, interface specifications, software bill of materials, asset list, diagnostic service definitions, external connectivity capabilities, known vulnerabilities, and security mechanism information. Through name normalization, version binding, and relationship extraction, heterogeneous data is converted into nodes and edges. Nodes include at least attack entry points, functional modules, electronic control units, buses, services, data, keys, and security-critical assets; edges include at least physical connections, logical communication, trust relationships, call relationships, data flows, and exploitable relationships. Each exploitable relationship includes preconditions, required permissions, attack actions, capabilities gained upon success, applicable versions, detection points, and confidence levels. For example, an edge might indicate that "when local physical access is available and a diagnostic session has been established, a restricted diagnostic service can be initiated to the target ECU via the OBD-II interface." The system only stores and uses this information within authorized testing scopes.
[0028] Further, in attack path reasoning, let the vehicle model security model be a directed attribute graph G=(V,E), the entry set be A, and the set of security-critical assets be T. For any candidate edge e, expansion is allowed only if the current state satisfies its precondition Pre(e) and the edge is applicable to the target vehicle version. After expansion, the permissions or capabilities generated by Post(e) are added to the state. Through constrained depth-first, breadth-first, heuristic search, or graph neural network reasoning, a candidate path set P from A to T is obtained. To suppress meaningless combinations, maximum path length, prohibition of duplicate nodes, permission monotonicity, trust boundary crossing count, version consistency, and mutual exclusion constraints for attack actions can be set. The path risk score can be obtained by weighting the target asset hazard, entry exposure, exploitation difficulty, precondition permissions, path length, and existing protection effectiveness. The weights can be configured by organizational policies and do not constitute a limitation of this invention.
[0029] Furthermore, the regulatory clauses are parsed and mapped. The regulatory parser breaks down the clauses into clause identifiers, applicable objects, attack surfaces, attack methods, target assets, expected protective behaviors, test environment limitations, judgment criteria, and evidentiary requirements. Parsing can use rule templates, terminology dictionaries, natural language models, or a combination of both, and the original clause summary, parsing results, and scope of application are stored in a version repository. The mapping module first performs precise matching of terms and entities, then performs semantic matching based on attack entry points, action sequences, target assets, and hazards, and finally verifies whether the path meets the preconditions and monitoring capabilities required by the clause. Paths with a comprehensive matching score reaching the first threshold are marked as mandatory regulatory test candidates; paths in the review range enter manual confirmation; paths that do not match clauses but have a risk higher than the second threshold are marked as suggested test items. This hierarchical mechanism takes into account both regulatory coverage and the discovery of unknown high-risk paths.
[0030] Furthermore, in test case generation, execution, and evaluation, the system converts confirmed paths into structured test cases, including the pre-existing environment, test objectives, path steps, input parameters, tool adapters, monitoring points, expected protective behaviors, stop conditions, recovery actions, and a list of evidence. The script generator selects operations from a verified action template library, avoiding the direct generation of dangerous instructions outside the whitelist; parameters are bound according to vehicle model version and interface definition. Before execution, the safety control module verifies the authorized scope, target resources, script signature, dangerous action level, rate limit, timeout, and emergency stop conditions. Testing is preferentially conducted in software-in-the-loop, hardware-in-the-loop, or digital twin environments; physical isolation, personnel monitoring, and failover protection must be enabled when real-vehicle verification is required. The monitoring module collects communication traffic, interface responses, controller status, system logs, screenshots, and time information, and makes conclusions such as compliance, partial compliance, non-compliance, undeterminable, or environmental anomalies based on the expected regulatory behavior and safety mechanism performance.
[0031] Furthermore, the authentication evidence is solidified and reports are generated. Each piece of evidence is bound to a unique traceability identifier, which is associated with the vehicle model and software version, regulatory version and clauses, attack path version, use case and script version, tool and execution environment, executor or service identity, start and end times, and judgment results. Original evidence generates a summary value, which, along with operation logs, timestamps, and storage locations, is written into an append-only or tamper-proof evidence list. The reporting module generates a compliance matrix according to regulatory clauses, displaying clause elements, covered attack paths, executed use cases, results, evidence index, risk level, and remediation recommendations; gap lists are generated for uncovered clauses, non-executable paths, and high-risk recommendations. Each conclusion in the report can be traced back to the original evidence.
[0032] Furthermore, the system performs change impact analysis and regression. When vehicle model topology, interface definitions, software versions, vulnerability information, or regulatory versions change, the system compares the old and new models, locates affected nodes, edges, and clauses, retrieves affected attack paths, test cases, and evidence along the relationships, and automatically generates an incremental regression test set. Evidence that is unaffected and still meets the version and validity period conditions can be reused, thereby reducing the workload of repeated verification.
[0033] This invention proposes another automotive cybersecurity regulatory certification testing method based on attack path reasoning, the method flow of which is as follows: Figure 3 As shown: S1 acquires and standardizes vehicle architecture, assets, interfaces, vulnerabilities, and protection information, establishes a vehicle security knowledge model with version and preconditions, and infers a set of candidate attack paths.
[0034] S2 parses the test items and evidence requirements of the target regulations, uses rule, entity and semantic joint matching to associate the regulatory test items with the attack path, and identifies high-risk paths not covered by the regulations.
[0035] S3 generates structured penetration test cases and constrained scripts based on the confirmed mappings, and automatically configures monitoring points, judgment criteria, stopping conditions, and evidence collection items.
[0036] S4 performs tests in an isolated and authorized environment, collects raw data, determines the performance of security mechanisms, and solidifies evidence with traceability identifiers, digest values, and time information.
[0037] S5 generates a regulatory compliance matrix, test report, gap list, and rectification suggestions; it calculates the scope of impact and forms a regression test set when there are changes to vehicle models or regulations.
[0038] In one embodiment of the present invention, regulatory certification testing for a CAN bus denial-of-service scenario involves importing information from a vehicle model's T-Box, central gateway, body domain controller, chassis domain controller, IVI, OBD-II interface, and onboard bus. The model shows that external communication entry points can reach the chassis domain CAN network via the T-Box and gateway; braking control communication is marked as a safety-critical asset. The path inference engine generates a candidate path of "external entry point—T-Box—gateway—chassis CAN—braking communication interfered with" when interface reachability and gateway policy conditions are met. The regulatory parser extracts attack surface, flooding techniques, target assets, functional degradation hazards, expected protection, and traffic evidence requirements from CAN bus denial-of-service related test requirements. Paths become mandatory test candidates after entity and semantic joint matching. The system generates bench test cases constrained by rate, duration, and emergency stop, executes them on an isolated simulation bus, and collects bus load, frame latency, loss rate, gateway alarms, and controller status before, during, and after the attack. If the protection mechanism limits abnormal traffic and maintains a safe state within the threshold, it is deemed compliant; otherwise, non-compliance evidence and remediation recommendations are generated.
[0039] In one embodiment of the present invention, the on-board diagnostic interface security access test model forms an attack path of "physical contactor — OBD-II — diagnostic service — ECU protected data or configuration". Regulatory clause parsing results require verification of whether unauthorized diagnostic access is denied and the saving of request, response, security session state, and access control logs. The system selects a diagnostic security access test template, binds the target ECU address, allowed service set, number of attempts, and lockout time, and prohibits reading or modifying data beyond the authorized scope. During execution, the system records each request and response, ECU session state, counters or lockout behavior, and recovery results. If an unauthenticated request is denied and the protection policy meets expectations, compliance evidence is formed; if unauthorized access occurs, a stop condition is immediately triggered, the minimum necessary evidence is solidified, and the result is marked as non-compliance.
[0040] In one embodiment of the invention, a high-risk path not explicitly covered by regulations is identified. The path inference engine discovers a multi-stage path: "Bluetooth entry point—IVI third-party application—backend permissions—gateway diagnostic interface—chassis control network." This path does not meet the mandatory testing threshold of current regulatory test items, but its risk score exceeds the recommended testing threshold due to entry point exposure, target asset hazard, and the number of trust boundary crossings. Therefore, the system includes it in the recommended test items. The system generates verification test cases that only run in a digital twin or hardware-in-the-loop environment to confirm the effectiveness of permission boundaries, interface access control, and gateway isolation. The report clearly states that this path has not yet been directly mapped to the specified clauses, the basis for risk judgment, and the recommended verification results, thereby avoiding the misrepresentation of value-added discoveries as mandatory regulatory conclusions.
[0041] The embodiments of this invention also have the following technical effects: Systematic attack path reasoning is performed through vehicle topology, permissions, and prerequisite constraints, reducing omissions in multi-stage attack chains caused by relying solely on personal experience. Regulatory clauses are parsed into computable elements and automatically associated with attack paths, enabling the testing scope to simultaneously reflect regulatory requirements and the actual vehicle architecture. Use cases and monitoring points are generated by the joint drive of paths and regulatory constraints, reducing the problem of fixed templates being incompatible with target vehicles. Unified traceability identifiers are used to associate clauses, paths, scripts, environments, results, and evidence, improving the consistency, verifiability, and tamper-proof capabilities of certification materials. Affected paths are automatically located and incremental regression sets are formed based on differences in model and regulatory versions, reducing redundant analysis after software upgrades and regulatory updates. Independent recommendations are generated for paths not explicitly covered by current regulations but with high risk, improving the foresight of security verification while maintaining clear boundaries for regulatory conclusions.
[0042] To achieve the above embodiments, such as Figure 4 As shown, this embodiment also provides a vehicle cybersecurity regulatory certification testing device 10 based on attack path reasoning, including: The vehicle safety knowledge modeling module 100 is used to obtain the electronic and electrical architecture information of the target vehicle, construct a vehicle safety knowledge model that includes asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. The regulatory path association analysis module 200 is used to parse the target regulatory clauses, extract test items and evidence requirements, associate regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identify high-risk paths not covered by regulations. The penetration test case generation and execution module 300 is used to generate structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configure monitoring points, judgment criteria and evidence collection items, and execute tests in an isolated authorized environment. The Test Evidence Compliance Archiving Module 400 is used to collect raw data during the testing process, determine the performance of security mechanisms, and uniformly trace and identify related regulatory clauses, attack paths, test cases, execution environments, and judgment results. It also solidifies evidence and generates a regulatory compliance matrix and test reports.
[0043] An embodiment of the present invention provides an automotive cybersecurity regulatory certification testing device based on attack path reasoning, which can improve the completeness and consistency of attack path coverage, realize automatic matching of regulatory clauses and test cases, improve the adaptability of test cases to vehicle architecture, enhance the traceability and anti-tampering capability of certification evidence, and reduce the workload of repetitive analysis after architecture or regulatory changes.
[0044] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 5 As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the method described above.
[0045] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.
[0046] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0047] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.
Claims
1. A method for automotive cybersecurity regulatory certification testing based on attack path reasoning, characterized in that, include: S1. Obtain the electronic and electrical architecture information of the target vehicle model, construct a vehicle model security knowledge model including asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. S2 parses the target regulatory clauses, extracts test items and evidence requirements, associates regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identifies high-risk paths not covered by regulations; S3 generates structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configures monitoring points, judgment criteria and evidence collection items, and executes tests in an isolated authorized environment; S4 collects raw data during the testing process, determines the performance of security mechanisms, and uses unified traceability to identify related regulatory clauses, attack paths, test cases, execution environments, and judgment results, solidifying evidence and generating regulatory compliance matrices and test reports.
2. The method as described in claim 1, characterized in that, S1 includes: The heterogeneous data in the acquired electronic and electrical architecture information is normalized by name, bound by version and extracted by relationship. The heterogeneous data is converted into nodes containing attack entry points, functional modules, electronic control units, buses, services, data, keys and security-critical assets, as well as edges containing physical connections, logical communication, trust relationships, call relationships, data flows and exploitable relationships, to construct the vehicle safety knowledge model. The vehicle safety knowledge model is represented as a directed attribute graph. Starting from the set of attack entry points and targeting the set of safety-critical assets, path expansion is performed when the current state and the attack prerequisites of the candidate edges are met and the candidate edges are applicable to the target vehicle version. The set of candidate attack paths is obtained through reasoning using a constrained search algorithm.
3. The method as described in claim 2, characterized in that, The method further includes: For any candidate edge, check whether the current state satisfies the attack prerequisites of the candidate edge and whether the candidate edge is applicable to the target vehicle version. Path expansion is allowed only if both are satisfied, and the permissions or capabilities generated by the postconditions of the candidate edge are added to the current state. Based on the maximum path length, prohibition of duplicate nodes, permission monotonicity, number of trust boundary crossings, version consistency, and mutual exclusion constraints of attack actions, the candidate attack path set is generated from the attack entry point set to the security critical asset set through depth-first search, breadth-first search, heuristic search, or graph neural network reasoning.
4. The method as described in claim 1, characterized in that, S2 includes: The target legal provisions are broken down into provisions identifiers, applicable objects, attack surfaces, attack methods, target assets, expected protective behaviors, test environment limitations, judgment criteria, and evidence requirements through rule templates, terminology dictionaries, or natural language models. The candidate attack paths are precisely matched with the elements of the disassembled legal provisions in terms of terms and entities. Then, semantic matching is performed based on the attack entry point, action sequence, target assets and harm. The candidate attack paths are verified to meet the prerequisites and monitoring capabilities required by the provisions. Candidate attack paths with a comprehensive matching score reaching the first threshold are marked as mandatory test candidates under regulations; candidate attack paths in the review period are marked as manually confirmed items; and candidate attack paths that do not match the clauses but have a risk score higher than the second threshold are marked as recommended test items.
5. The method as described in claim 4, characterized in that, The method further includes: The attack entry point, action sequence, target asset, and harm of the candidate attack path are precisely matched at the entity level with the attack surface, attack method, and target asset in the legal clause elements to generate an entity matching score. Based on the entity matching score, the semantic similarity between the candidate attack path and the legal clause element is calculated to generate a semantic matching score. The rule matching score, entity matching score and semantic matching score are weighted and summed to obtain the comprehensive matching score.
6. The method as described in claim 1, characterized in that, The S3 includes: Generate structured penetration test cases based on the associated attack paths and regulatory test items, including the pre-environment, test target, path steps, input parameters, tool adapters, monitoring points, expected protective behaviors, stopping conditions, recovery actions, and evidence lists. Select an operation from a validated library of action templates, bind parameters based on vehicle model and interface definition, and generate a constrained test script; Before test execution, verify the authorized scope, target resources, script signature, hazard level, rate limit, timeout and emergency stop conditions, and perform the test in a software-in-the-loop, hardware-in-the-loop or digital twin environment.
7. The method as described in claim 1, characterized in that, The S4 includes: During the test execution, the monitoring module collects communication traffic, interface response, controller status, system logs, screenshots, and time information as raw data. Based on the expected protective behaviors and safety mechanisms of the regulations, the test results are judged as compliant, partially compliant, non-compliant, indeterminate, or environmentally abnormal. Each piece of evidence is bound to a unified traceability identifier, which is associated with the vehicle model and software version, regulatory version and clauses, attack path version, use case and script version, tool and execution environment, executor or service identity, start and end time and judgment result. A summary value is generated for the original evidence, and together with the operation log, timestamp and storage location, it is written into an evidence list that can only be appended or has anti-tampering capabilities, generating a regulatory compliance matrix and test report.
8. A vehicle cybersecurity regulatory certification testing device based on attack path reasoning, characterized in that, include: The vehicle safety knowledge modeling module is used to obtain the electronic and electrical architecture information of the target vehicle, construct a vehicle safety knowledge model that includes asset nodes, communication interfaces, trust boundaries and attack prerequisites, and infer a set of candidate attack paths from the attack entry point to the security-critical assets based on the model. The regulatory path association analysis module is used to parse target regulatory clauses, extract test items and evidence requirements, associate regulatory test items with candidate attack paths through rule matching, entity matching and semantic matching, and identify high-risk paths not covered by regulations. The penetration test case generation and execution module is used to generate structured penetration test cases and constrained test scripts based on the associated attack paths and regulatory test items, configure monitoring points, judgment criteria and evidence collection items, and execute tests in an isolated authorized environment; The test evidence compliance archiving module is used to collect raw data during the testing process, determine the performance of security mechanisms, and uniformly trace and identify related regulatory clauses, attack paths, test cases, execution environments and judgment results, solidify evidence and generate regulatory compliance matrices and test reports.
9. A computer device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the method as described in any one of claims 1-7.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.
