A method and device for automatic deployment and revocation of massive terminal honeypots

By using Docker containers and feature extraction models to automatically identify abnormal traffic in network security, the automated deployment and management of massive terminal honeypots is realized, and the problems of cumbersome deployment and resource consumption in the existing technology are solved, and network protection is improved Efficiency and safety.

CN115396233BActive Publication Date: 2024-11-08JINJING YUNHUA TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211233127.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-10
Publication Date
2024-11-08
Estimated Expiration
2042-10-10

AI Technical Summary

Technical Problem

Existing network security protection technologies cannot effectively deal with complex and changeable network attacks. The honeypot deployment and undoing process relies on manual labor, which is cumbersome and difficult to achieve automation, resulting in large resource consumption and complex configuration, making it difficult to meet the needs of large-scale network environments.

Method used

The virtual operating system Docker deploys containers, automatically identify abnormal traffic through packet capture and feature extraction, establish and undo honeypots, and use PCA feature extraction model and SVM classification training to achieve automated deployment and management, and determine the honeypots based on the frequency of API calls. Retain or revoke.

Benefits of technology

It realizes the automated deployment and cancellation of massive terminal honeypots, saves manpower and resources, simplifies the configuration process, improves network protection efficiency and security, reduces resource consumption, is highly adaptable, and is suitable for large-scale network environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115396233B_ABST
    Figure CN115396233B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of honeypot deployment and protection against network attacks, and in particular provides a method and device for automatic deployment and revocation of massive terminal honeypots, comprising the following steps: in a Docker environment, capturing and classifying communication traffic data packets; extracting features of the classified data packets, dividing them into normal data packets and abnormal data packets, sending the normal data packets to the destination, and processing the features of the abnormal data packets; performing intrusion detection on the features of the processed abnormal data packets, automatically establishing a honeypot, introducing the malicious data packets into the corresponding honeypot, and further performing detection and analysis; within a certain period of time, counting the relationship between the number of API calls and time in each honeypot to determine the retention or revocation of the honeypot. The present invention realizes the automation of massive honeypot deployment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of honeypot deployment and protection against network attacks, and in particular provides a method and device for automatic deployment and revocation of honeypots for massive terminals. Background Art

[0002] With the deepening of computer information system applications, computer information systems are gradually developing from single machines to local area networks and wide area networks. In particular, with the rapid development of the Internet, networks have become more complex, and computer information system security faces new and more severe challenges. At the same time, existing network security protection technologies, including firewall technology, intrusion detection technology, virus protection technology, data encryption and authentication technology, are mainly based on known facts and attack patterns. When the attacker attacks, the system can only be passively protected, and it is often unable to cope with complex and changeable network attacks.

[0003] Traditional network security concepts focus on building a secure network environment, aiming to keep security threats out of the network. With the development of IDS evasion and firewall penetration technology, network security faces severe challenges. In recent years, honeypots have become a hot topic of concern and research in the security community. It takes an active approach, using false services or information to confuse attackers, divert and accommodate limited threats, and divert attacks from more important machines in the network. At the same time, during the period when hackers attack honeypots, they conduct in-depth analysis and research on their behaviors and processes, discover new attacks, retrieve new hacker tools, and understand the background, purpose, and activity patterns of hackers and hacker groups. Honeypots not only have accurate threat and judgment capabilities, but also can discover new attack methods and tools. In this way, when an attacker invades, the honeypot delays the attacker's attack on the real target, making the attacker waste time on the honeypot, effectively protecting the business network, while understanding the opponent and studying the threats to itself.

[0004] Similar scheme: An automated honeypot construction and threat perception method based on Docker. The method in the similar scheme simply uses Docker technology to build a honeypot, but still requires manual judgment of the construction time, location and operation. The present invention truly realizes automation and performs automated construction based on the results of the intrusion detection system. The method in the similar scheme analyzes the data in the honeypot and then perceives the threat. The present invention first determines the traffic anomaly and introduces the abnormal traffic into the honeypot, which has a faster response speed and higher security.

[0005] The honeypot system is a system that simulates a real system but has certain vulnerabilities. When facing attackers, the characteristics of the system are not significantly different from those of the real system; when facing users, the honeypot system is customizable, and users can deploy similar honeypot systems according to their own system types. The deployment of the honeypot system can divert most of the attacker's attack traffic to the honeypot system, thereby greatly reducing the burden on the user's main system. Therefore, while using various prevention technologies, it is also necessary to create a large number of honeypots to confuse known and unknown malicious attacks and protect network information. The behavioral data in the honeypot can also be used for later traceability analysis of malicious behaviors. However, the deployment and deletion process is overly dependent on manual labor, the layout is cumbersome, and the automatic configuration of honeypots is somewhat difficult. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a method and device for automatic deployment and revocation of massive terminal honeypots.

[0007] The present invention is implemented by providing a method for automatic deployment and revocation of massive terminal honeypots, comprising the following steps:

[0008] 1) Deploy the virtual operating system Docker on the host to form different containers;

[0009] 2) Capture the communication traffic packets between containers and classify them according to the destination of the packets;

[0010] 3) Extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets based on the extracted features. Send the normal data packets to the destination and process the features of the abnormal data packets.

[0011] 4) The features of the processed abnormal data packets are used for intrusion detection. When the abnormal data packets are identified as normal data packets, they are sent to the destination. When the abnormal data packets are identified as malicious data packets or cannot be accurately judged, a honeypot is automatically established and the malicious data packets are introduced into the corresponding honeypot for further detection and analysis.

[0012] 5) Within a certain period of time, the relationship between the number of API calls in each honeypot and time is counted. If the frequency of API calls in the honeypot is high and higher than the set threshold, the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been visited less frequently or has been identified as a honeypot, and the honeypot will be revoked.

[0013] Preferably, in step 2), the communication traffic data packets between containers are captured by using existing software to monitor the host network card interface.

[0014] Further preferably, in step 3), feature extraction is performed on the communication traffic data packet according to a feature extraction model based on PCA, specifically comprising the following steps:

[0015] 1) Perform PCA feature extraction on the classified communication traffic data packets: Calculate the covariance matrix S of the original sample space X; Calculate its orthogonal matrix Q and eigenvalue λ through the covariance matrix S 1 ≥λ 2 ≥…≥λ n ; Set the threshold of the cumulative contribution rate t and calculate the standard orthogonal vector u i And the sample space after dimensionality reduction U={u 1 ,u 2 ,…u d}, and obtain the projected principal component feature y={u 1 ,u 2 ,…u d} T x i , forming a new candidate feature subset F1, n, d, i all represent constants, and T represents transposition;

[0016] 2) Adaptively select the secondary feature selection module of PCA:

[0017] 201) When it is necessary to determine the key features of the feature subset F1 after dimensionality reduction, feature selection based on association filters is performed to check the relevance of each feature, that is, to maximize the correlation between attribute features and class attributes and minimize the redundancy between attributes, combined with the heuristic sequential backward algorithm search strategy, to obtain the candidate key feature subset F2;

[0018] 202) When it is not necessary to determine the key features of feature subset F1, feature extraction is skipped to obtain a relatively high classification accuracy;

[0019] 3) Based on the optimal feature subsets of the PCA feature extraction model and the PCA secondary feature selection model, a data set is formed and SVM classification training is performed. The classification results of communication traffic data packets are obtained on the test set, that is, they are classified into normal data packets and abnormal data packets.

[0020] Further preferably, in the step 4), after further detecting and analyzing the malicious data packet through the honeypot, if it is still identified as a malicious data packet with intrusion behavior, a blacklist strategy is executed to discard the corresponding malicious data packet, or an alarm strategy is executed.

[0021] The present invention also provides a device for automatically deploying and revoking massive terminal honeypots, comprising the following units:

[0022] The virtual operating system Docker is used to deploy on the host;

[0023] A data capture module is used to capture communication traffic packets between containers and classify them according to the destination of the packets;

[0024] A feature extraction module is used to extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets according to the extracted features, and send the normal data packets to the destination and process the features of the abnormal data packets;

[0025] An intrusion detection module is used to perform intrusion detection based on the features of the processed abnormal data packets;

[0026] Honeypots are used to further detect and analyze data packets that are identified as malicious or cannot be accurately judged;

[0027] The frequency calculation module is used to count the relationship between the number of API calls in each honeypot and time within a certain period of time. If the frequency of API calls in the honeypot is high and exceeds the set threshold , then the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been visited less frequently, or has been identified as a honeypot, then the honeypot will be revoked.

[0028] Compared with the prior art, the advantages of the present invention are:

[0029] The present invention can automatically deploy a large number of terminal honeypots, save manpower and cost, make the honeypot deployment method simple, highly operable, and effectively protect the network.

[0030] (1) Reduce resource consumption and facilitate deployment:

[0031] Traditional intrusion detection systems need to be configured according to different requirements, but in the present invention, in order to simplify the deployment of the intrusion detection system, the container in the Docker technology is used. Different intrusion detection systems can be run in the container, and the intrusion detection system running in the container only needs some configuration scripts to start when it is started, which greatly simplifies the configuration operation. At the same time, the intrusion detection system in this design does not depend on the environment and has good portability. As long as the relevant environment of Docker is configured in the host environment, the relevant scripts can be run to implement intrusion detection.

[0032] Traditional honeypots are configured based on the same hardware environment of the host. During the configuration process, not only is the configuration process cumbersome and complicated, but the same hardware environment is also required. Compared with traditional honeypot technology, the configuration process is reduced because multiple container instances can be established based on the image, which greatly simplifies the configuration process. After the environment is established, the attacker can be directly lured into the established honeypot to further analyze and identify the attacker's behavior.

[0033] (2) Establish multiple adaptive honeypots simultaneously with less resource consumption:

[0034] There is no need to manually design and deploy honeypots. Honeypots can be deployed in time when needed and revoked in time when not needed. One host can realize the entire data control, which is a multi-functional multi-honeypot high-interaction honeynet architecture that integrates data capture and data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flow chart of the method for automatic deployment and revocation of massive terminal honeypots provided by the present invention;

[0036] Figure 2 A connection diagram of device modules for automatic deployment and revocation of massive terminal honeypots provided by the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0038] At present, a single honeypot system cannot be well applied to today's Internet system. There are mainly the following problems:

[0039] 1) The system architecture deployment and maintenance are relatively complex, batch automated deployment is difficult, risks cannot be effectively controlled, and the system cannot meet the needs of large-scale network environments.

[0040] 2) The simulated service has a low level of interaction and is easy to identify;

[0041] 3) If a third-party network is attacked by hackers and causes damage, the victim will file a lawsuit, which will lead to legal issues;

[0042] 4) The level of interaction of simulated services is low, making it difficult to achieve adaptive network changes.

[0043] Traditional honeypots are configured based on the same hardware environment of the host. During the configuration process, not only is the configuration process cumbersome and complicated, it cannot be automatically deployed, and the same hardware environment is also required. In view of these problems, the present invention uses the virtual operating system Docker and integrates the intrusion detection system to automatically build and run the honeypot; based on the frequency of API calls, the honeypot is revoked to achieve the automatic deployment of massive terminal honeypots.

[0044] refer to Figure 1 The present invention provides a method for automatically deploying and revoking massive terminal honeypots, comprising the following steps:

[0045] 1) Deploy the virtual operating system Docker on the host to form different containers;

[0046] 2) Capture the communication traffic data packets between containers and classify them according to the destination of the data packets; in this step, the communication traffic data packets between containers are captured by using existing software to monitor the host network card interface.

[0047] 3) Extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets based on the extracted features. Send the normal data packets to the destination and process the features of the abnormal data packets.

[0048] In this step, feature extraction is performed on the communication traffic data packet according to the PCA-based feature extraction model, which specifically includes the following steps:

[0049] 1) Perform PCA feature extraction on the classified communication traffic data packets: Calculate the covariance matrix S of the original sample space X; Calculate its orthogonal matrix Q and eigenvalue λ through the covariance matrix S 1 ≥λ 2 ≥…≥λ n ; Set the threshold of the cumulative contribution rate t and calculate the standard orthogonal vector u i And the sample space after dimensionality reduction U={u 1 ,u 2 ,…u d}, and obtain the projected principal component feature y={u 1 ,u 2 ,…u d} T x i , forming a new candidate feature subset F1, n, d, i all represent constants, and T represents transposition;

[0050] 2) Adaptively select the secondary feature selection module of PCA:

[0051] 201) When it is necessary to determine the key features of the feature subset F1 after dimensionality reduction, feature selection based on association filters is performed to check the relevance of each feature, that is, to maximize the correlation between attribute features and class attributes and minimize the redundancy between attributes, combined with the heuristic sequential backward algorithm search strategy, to obtain the candidate key feature subset F2;

[0052] 202) When it is not necessary to determine the key features of feature subset F1, feature extraction is skipped to obtain a relatively high classification accuracy;

[0053] 3) Based on the optimal feature subsets of the PCA feature extraction model and the PCA secondary feature selection model, a data set is formed and SVM classification training is performed. The classification results of communication traffic data packets are obtained on the test set, that is, they are classified into normal data packets and abnormal data packets.

[0054] 4) The features of the processed abnormal data packets are used for intrusion detection. When the abnormal data packets are identified as normal data packets, they are sent to the destination. When the abnormal data packets are identified as malicious data packets or cannot be accurately judged, a honeypot is automatically established and the malicious data packets are introduced into the corresponding honeypot for further detection and analysis.

[0055] In this step, after further detecting and analyzing the malicious data packet through the honeypot, if it is still identified as a malicious data packet with intrusion behavior, the blacklist strategy is executed to discard the corresponding malicious data packet, or the alarm strategy is executed.

[0056] 5) In a certain period of time, the relationship between the number of API calls in each honeypot and time is counted. If the frequency of API calls in the honeypot is high and higher than the set threshold, the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been accessed less frequently or has been identified as a honeypot, and the honeypot will be revoked. That is:

[0057]

[0058] in, For the calling frequency, is the number of API calls within time period t, is a fixed detection time period. If t is greater than the threshold, detection will continue at intervals; if t is less than the threshold, the revocation script will be started to revoke the honeypot.

[0059] refer to Figure 2 The present invention also provides a device for automatically deploying and revoking massive terminal honeypots, comprising the following units:

[0060] The virtual operating system Docker is used to deploy on the host;

[0061] A data capture module is used to capture communication traffic packets between containers and classify them according to the destination of the packets;

[0062] A feature extraction module is used to extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets according to the extracted features, and send the normal data packets to the destination and process the features of the abnormal data packets;

[0063] An intrusion detection module is used to perform intrusion detection based on the features of the processed abnormal data packets;

[0064] Honeypots are used to further detect and analyze data packets that are identified as malicious or cannot be accurately judged;

[0065] The frequency calculation module is used to count the relationship between the number of API calls in each honeypot and time within a certain period of time. If the frequency of API calls in the honeypot is high and exceeds the set threshold , then the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been visited less frequently, or has been identified as a honeypot, then the honeypot will be revoked.

[0066] In summary, the present invention designs a method for automatically deploying honeypots for massive terminals, which can automatically complete the construction, operation, and cancellation of honeypots.

[0067] In terms of construction, the results of the intrusion detection system determine whether to automatically start the honeypot generation script. If malicious traffic is detected, the script will be automatically triggered to generate the honeypot. The generated honeypot can interact with the traffic to varying degrees.

[0068] In terms of revocation, the revocation script can be triggered according to the frequency of API calls to automatically revoke the honeypot, saving costs. The frequency threshold can be set according to the actual situation. If resources are sufficient, the threshold can be raised to retain some honeypots with infrequent interactions. If resources are limited, only the most vulnerable honeypots can be retained to ensure the highest efficiency.

[0069] The embodiments of the present invention are described in detail above, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A method for automatic deployment and revocation of massive terminal honeypots, characterized in that: The steps include: 1) Deploy the virtual operating system Docker on the host to form different containers; 2) Capture the communication traffic packets between containers and classify them according to the destination of the packets; 3) Extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets based on the extracted features. Send the normal data packets to the destination and process the features of the abnormal data packets. In the step 3), feature extraction is performed on the communication traffic data packet according to the PCA-based feature extraction model, which specifically includes the following steps: 301) Perform PCA feature extraction on the classified communication traffic data packets: calculate the covariance matrix S of the original sample space X; calculate its orthogonal matrix Q and eigenvalues ​​λ1≥λ2≥…≥λ through the covariance matrix S n ; Set the threshold of the cumulative contribution rate t and calculate the standard orthogonal vector u i And the sample space after dimensionality reduction U={u1,u2,…u d }, and obtain the projected principal component features y={u1,u2,…u d } T x i , forming a new candidate feature subset F1, n, d, i all represent constants, and T represents transposition; 302) Adaptive selection of PCA secondary feature selection module: 302-1) When it is necessary to determine the key features of the feature subset F1 after dimensionality reduction, feature selection based on association filter is performed to check the relevance of each feature, that is, to maximize the correlation between attribute features and class attributes and minimize the redundancy between attributes, and combine the heuristic sequential backward algorithm search strategy to obtain the candidate key feature subset F2; 302-2) When it is not necessary to determine the key features of the feature subset F1, feature extraction is skipped to obtain a relatively high classification accuracy; 303) On the optimal feature subsets of the PCA feature extraction model and the PCA secondary feature selection model, a data set is formed and SVM classification training is performed, and the classification results of the communication traffic data packets are obtained on the test set, that is, they are classified into normal data packets and abnormal data packets; 4) The features of the processed abnormal data packets are used for intrusion detection. When the abnormal data packets are identified as normal data packets, they are sent to the destination. When the abnormal data packets are identified as malicious data packets or cannot be accurately judged, a honeypot is automatically established and the malicious data packets are introduced into the corresponding honeypot for further detection and analysis. 5) Within a certain period of time, the relationship between the number of API calls in each honeypot and time is counted. If the frequency of API calls in the honeypot is high and higher than the set threshold, the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been visited less frequently or has been identified as a honeypot, and the honeypot will be revoked.

2. The method for automatic deployment and revocation of massive terminal honeypots according to claim 1, characterized in that: In the step 2), the communication traffic data packets between containers are captured by using existing software to monitor the host network card interface.

3. The method for automatic deployment and revocation of massive terminal honeypots according to claim 1, characterized in that: In the step 4), after further detecting and analyzing the malicious data packet through the honeypot, if it is still identified as a malicious data packet with intrusion behavior, a blacklist strategy is executed to discard the corresponding malicious data packet, or an alarm strategy is executed.

4. A device for automatic deployment and revocation of massive terminal honeypots, characterized in that: Based on the method for automatic deployment and revocation of massive terminal honeypots according to claim 1, the device includes the following units: The virtual operating system Docker is used to deploy on the host; A data capture module is used to capture communication traffic packets between containers and classify them according to the destination of the packets; A feature extraction module is used to extract features from the classified data packets, and classify the communication traffic data packets into normal data packets and abnormal data packets according to the extracted features, and send the normal data packets to the destination and process the features of the abnormal data packets; An intrusion detection module is used to perform intrusion detection based on the features of the processed abnormal data packets; Honeypots are used to further detect and analyze data packets that are identified as malicious or cannot be accurately judged; The frequency calculation module is used to count the relationship between the number of API calls in each honeypot and time within a certain period of time. If the frequency of API calls in the honeypot is high and exceeds the set threshold , then the honeypot will continue to be retained; if it is lower than the set threshold, the honeypot has been visited less frequently, or has been identified as a honeypot, then the honeypot will be revoked.