A method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology
By constructing an IoT attack chain model and a network security knowledge extraction model based on an attack chain knowledge graph, dynamic anomaly detection of edge devices is achieved, solving the problem that existing power distribution IoT security protection solutions cannot meet the requirements and improving the system's security.
Patent Information
- Application Number
- CN202311319856.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-12
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-10-12
AI Technical Summary
Existing security protection solutions for the distribution IoT are mainly passive defense-oriented and cannot meet the security protection needs of the distribution IoT.
By employing attack chain knowledge graph technology, an IoT attack chain model is constructed, a hierarchical network security knowledge extraction model is established, and a knowledge graph is automatically built to achieve dynamic anomaly detection of edge devices.
It improves the security of the power distribution Internet of Things (IoT), effectively detects dynamic anomalies in edge devices, and meets the security protection requirements of the power distribution IoT.
Smart Images

Figure CN117278302B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power distribution Internet of Things (IoT), specifically to a method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology. Background Technology
[0002] Under the "cloud, pipe, edge, and terminal" architecture of the distribution Internet of Things (IoT), the functional positioning, deployment mode, and working method of each link have undergone tremendous changes compared with the "master station-communication network-terminal" architecture of the traditional distribution monitoring system. The current security protection scheme of the distribution monitoring system, which is mainly based on passive defense, can no longer meet the security protection needs of the distribution IoT. Summary of the Invention
[0003] The purpose of this invention is to provide a method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology. This method can effectively detect dynamic anomalies in edge devices, thereby improving the security of the power distribution Internet of Things.
[0004] To achieve the above objectives, the technical solution adopted by this invention is: a method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology, comprising the following steps: 1) Analyze the attack paths of the power distribution Internet of Things (IoT) and construct an IoT attack chain model; 2) Based on the constructed IoT attack chain model, a hierarchical network security knowledge extraction model is established to achieve bottom-up network security knowledge extraction; 3) Based on the established network security knowledge extraction model, automatically construct a knowledge graph and instantiate the network security knowledge graph based on the attack chain; 4) Implement dynamic anomaly detection of edge devices based on the constructed knowledge graph.
[0005] Furthermore, regarding the attack exposure surface of the perception layer, the attack path is as follows: the attacker gains control of the edge IoT agent or IoT terminal by attacking the perception layer, and then attacks upwards to the platform layer; For network layer attack exposure surfaces, the attack path is as follows: attackers tamper with, listen to, or block replay data packets by attacking the communication network between the perception layer and the platform layer. For the application layer attack exposure surface, the attack path is as follows: the attacker attacks the business application server through the IoT business application, thereby gaining access to the internal network; Based on the analysis of the attack exposure surface and the corresponding attack path, an IoT attack chain model is constructed.
[0006] Furthermore, the established hierarchical cybersecurity knowledge extraction model includes a raw data layer, a unified data model, a security foundation data domain, and a cybersecurity knowledge graph.
[0007] Furthermore, for the raw data layer, network security raw data originates from terminal devices, security equipment, workstations, control systems, network flows, and external intelligence in the ubiquitous Internet of Things (IoT) network. It is collected and acquired at network boundaries and key nodes at all levels of the network. The acquired network security raw data is transformed through a unified data model to complete data cleaning, alignment, and correlation calculations, forming seven major themes of data in the security basic data domain and storing them. Among them, event, intelligence, and network flow data are semi-structured data and stored in the unstructured platform, while asset, alarm, status, and behavior data are structured data and stored in the internal database. Network security knowledge is directly extracted from the security basic data domain and mapped to the network security knowledge graph through the security basic data domain, and the network security knowledge graph based on the attack chain is instantiated.
[0008] Furthermore, the process of building a knowledge graph includes data extraction, association and fusion, and graph structure storage.
[0009] Furthermore, during data extraction, for structured data, a template-based approach is directly used to extract the attributes and attribute values related to entities and relationships in the graph based on the mapping relationship; for semi-structured data, a combination of template-based and feature recognition is used. After the template data extraction is completed, regular expression matching rules are used to identify the key elements of relevant attributes, thereby completing the data extraction. Data extracted from different themes are used to establish relationships based on the original primary and foreign keys, forming entities and relational objects that are identical to those in the knowledge graph. After using graph structures for attack knowledge extraction and entity relationship storage, rapid attack detection and threat tracing based on graph retrieval and graph algorithms can be achieved. Perform graph similarity calculations and correlate unknown attacks through graph similarity; Perform critical path analysis on the attack graph to correlate unknown attacks through the same attack paths.
[0010] Compared with the prior art, the present invention has the following beneficial effects: it provides a method for detecting dynamic anomalies of edge devices based on attack chain knowledge graph technology. This method, based on attack chain analysis and knowledge graph technology, can effectively detect dynamic anomalies of edge devices in the power distribution Internet of Things (IoT), thereby improving the security of the power distribution IoT and meeting the security protection requirements of the power distribution IoT. Attached Figure Description
[0011] Figure 1 This is a flowchart illustrating the method implementation of an embodiment of the present invention; Figure 2 This is a schematic diagram of the attack exposure surface and attack path of the power distribution Internet of Things in this embodiment of the invention; Figure 3 This is a diagram of the hierarchical network security knowledge extraction model architecture in this embodiment of the invention; Figure 4 This is a flowchart of the knowledge graph construction process in an embodiment of the present invention. Detailed Implementation
[0012] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0013] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.
[0014] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0015] like Figure 1 As shown, this embodiment provides a method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology, including the following steps: 1. Analyze the attack paths of the power distribution Internet of Things (IoT) and construct an IoT attack chain model.
[0016] In a power distribution IoT environment, interconnection between edge devices has been achieved, increasing the attack surface. In a typical power distribution IoT scenario, the existing attack surfaces and attack paths are as follows: Figure 2 As shown.
[0017] For the attack exposure surface of the perception layer, the attack path is as follows: the attacker gains control of the edge IoT agent or IoT terminal by attacking the perception layer, and then attacks upward to the platform layer.
[0018] For network layer attack exposure surfaces, the attack path is as follows: attackers tamper with, listen to, or block replay data packets by attacking the communication network between the perception layer and the platform layer.
[0019] For the application layer attack exposure surface, the attack path is as follows: attackers use IoT business applications to attack the business application servers, thereby gaining access to the company's internal network.
[0020] Based on the analysis of the attack exposure surface and the corresponding attack path, an IoT attack chain model is constructed.
[0021] 2. Based on the constructed IoT attack chain model, establish a hierarchical network security knowledge extraction model to achieve bottom-up network security knowledge extraction.
[0022] The established hierarchical network security knowledge extraction model, such as Figure 3 As shown, it includes the raw data layer, unified data model, security basic data domain, and network security knowledge graph.
[0023] For the raw data layer, network security raw data originates from terminal devices, security equipment, workstations, control systems, network flows, and external intelligence in the ubiquitous Internet of Things (IoT) network, and is collected at network boundaries and key nodes at all levels of the network. The acquired network security raw data undergoes a unified data model transformation to complete data cleaning, alignment, and correlation calculations, forming seven major themes in the security foundation data domain and storing them. Among these, event, intelligence, and network flow data are semi-structured data, stored in an unstructured platform, while asset, alarm, status, and behavior data are structured data, stored in an internal database. Network security knowledge is directly extracted from the security foundation data domain and, based on the methods proposed in this paper, such as... Figure 3 The cybersecurity knowledge system shown completes the construction of a cybersecurity knowledge graph based on the attack chain. Data mapping from the security foundation data domain to the cybersecurity knowledge graph is performed to instantiate the attack chain-based cybersecurity knowledge graph.
[0024] 3. Based on the established network security knowledge extraction model, automatically construct a knowledge graph.
[0025] like Figure 4 As shown, the process of building a knowledge graph includes data extraction, association fusion, and graph structure storage.
[0026] During data extraction, for structured data, a template-based approach is directly used to extract the attributes and attribute values related to entities and relationships in the graph based on the mapping relationship (table); for semi-structured data, a combination of template-based and feature recognition is used. After the template data extraction is completed, regular expression matching rules are used to identify the key elements of relevant attributes, thereby completing the data extraction.
[0027] Data extracted from different thematic datasets establishes relationships based on existing primary and foreign keys, forming entities and relational objects identical to those in the knowledge graph. For example, alarm-related attributes obtained from the alarm topic are {alarm id, alarm time, alarm level, alarm source device, attack source, attack target, alarm type, alarm name}, while attributes related to the ports of the attacked system components are obtained from the status topic, {server id, server IP, open port, time}. The relationship between the two is established through server id, server IP, and time.
[0028] By using graph structures for attack knowledge extraction and entity relationship storage, rapid attack detection and threat tracing based on graph retrieval and graph algorithms can be achieved.
[0029] Graph similarity calculation: This involves associating unknown attacks through graph similarity. Entity similarity calculation in knowledge graphs has been implemented in numerous industrial applications, such as similar product recommendations on e-commerce platforms and similar patient groups in medical efficacy analysis. In cybersecurity, user behavior on critical assets can be recorded in real time, and each behavioral process and its involved device information can be mapped to an ontology model, thus obtaining a "behavioral graph." A graph similarity algorithm is applied to calculate the similarity between this "behavioral graph" and the attack graph in a security knowledge base. When the similarity exceeds a set threshold, the behavior is deemed to pose an attack risk.
[0030] Critical path analysis is performed to correlate unknown attacks through similar attack paths. Based on various alarm information, system logs, and user operation history on the device (e.g., when the attack occurred, where the vulnerability was successfully exploited, how the server was logged in, whether a Trojan backdoor was left, and whether logs were cleared), feature vectors are generated to build a feature model of the attack process. Unsupervised machine learning clustering algorithms are then used to analyze and identify the critical paths in the attack process. Critical path analysis not only helps determine the attacker's identity and behavioral characteristics during the attack but also helps identify vulnerabilities in compromised devices. Based on the critical paths and newly collected information, supervised machine learning is performed. New attack behaviors are correlated with the critical paths by similarity matching.
[0031] 4. Implement dynamic anomaly detection for edge devices based on the constructed knowledge graph.
[0032] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0033] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.
[0034] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0035] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0036] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A method for detecting dynamic anomalies in edge devices based on attack chain knowledge graph technology, characterized in that, Includes the following steps: 1) Analyze the attack paths of the power distribution Internet of Things (IoT) and construct an IoT attack chain model; 2) Based on the constructed IoT attack chain model, a hierarchical network security knowledge extraction model is established to achieve bottom-up network security knowledge extraction; 3) Based on the established network security knowledge extraction model, automatically construct a knowledge graph and instantiate the network security knowledge graph based on the attack chain; 4) Implement dynamic anomaly detection for edge devices based on the constructed knowledge graph; For the attack exposure surface of the perception layer, the attack path is as follows: the attacker gains control of the edge IoT agent or IoT terminal by attacking the perception layer, and then attacks upward to the platform layer. For network layer attack exposure surfaces, the attack path is as follows: attackers tamper with, listen to, or block replay data packets by attacking the communication network between the perception layer and the platform layer. For the application layer attack exposure surface, the attack path is as follows: the attacker attacks the business application server through the IoT business application, thereby gaining access to the internal network; Based on the analysis of the attack exposure surface and the corresponding attack path, an IoT attack chain model is constructed. The established hierarchical cybersecurity knowledge extraction model includes a raw data layer, a unified data model, a security foundation data domain, and a cybersecurity knowledge graph. For the raw data layer, cybersecurity raw data originates from terminal devices, security equipment, workstations, control systems, network flows, and external intelligence in the ubiquitous Internet of Things (IoT) network, collected at network boundaries and key nodes at all levels. The acquired cybersecurity raw data undergoes a unified data model transformation to complete data cleaning, alignment, and correlation calculations, forming seven major themes within the security foundation data domain, which are then stored. Among these, event, intelligence, and network flow data are semi-structured data stored in an unstructured platform, while asset, alarm, status, and behavior data are structured data stored in an internal database. Cybersecurity knowledge is directly extracted from the security foundation data domain and mapped to the cybersecurity knowledge graph through the security foundation data domain, instantiating the attack chain-based cybersecurity knowledge graph. The process of building a knowledge graph includes data extraction, association fusion, and graph structure storage. During data extraction, for structured data, a template-based approach is used to extract the attributes and attribute values related to entities and relationships in the graph based on the mapping relationship. For semi-structured data, a combination of template-based and feature recognition is used. After the template data extraction is completed, regular expression matching rules are used to identify the key elements of related attributes, thereby completing the data extraction. Data extracted from different themes are used to establish relationships based on the original primary and foreign keys, forming entities and relational objects that are identical to those in the knowledge graph. After using graph structures for attack knowledge extraction and entity relationship storage, rapid attack detection and threat tracing based on graph retrieval and graph algorithms can be achieved. Perform graph similarity calculations and correlate unknown attacks through graph similarity; Perform critical path analysis on the attack graph to correlate unknown attacks through the same attack paths.
Citation Information
Patent Citations
Security event association method and system based on network security knowledge graph, and medium
CN111177417A
