Docker attack traffic detection method based on ChatGPT model

Through the Docker attack traffic detection method based on the ChatGPT model, the abnormal detection is performed using virtual bridge and NLP technology, the problem of low detection efficiency of new threats and adversarial attacks in the existing technology is solved, and efficient identification and response to new threats is achieved. .

CN118900201APending Publication Date: 2024-11-05GUANGZHOU UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411154713.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-05

AI Technical Summary

Technical Problem

The existing Docker application traffic detection methods are less efficient when facing new threats and confrontational attacks, making it difficult to effectively deal with unknown threats.

Method used

The Docker attack traffic detection method based on the ChatGPT model is adopted, and the ChatGPT detection model is trained and optimized by acquiring and preprocessing traffic data, and a virtual bridge is used to obtain network data packets for abnormal detection. The detection capability is improved by combining NLP technology and integrated learning algorithms.

Benefits of technology

It significantly improves the detection success rate in the face of new threats and confrontational attacks, and can promptly detect and respond to new threat modes, improving detection efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118900201A_ABST
    Figure CN118900201A_ABST
Patent Text Reader

Abstract

The invention provides a Docker attack traffic detection method based on a ChatGPT model, and relates to the technical field of Docker container security. The Docker attack traffic detection method based on the ChatGPT model comprises the following steps: acquiring and preprocessing traffic data to obtain a network security data set, and training and optimizing the ChatGPT model based on the network security data set to obtain a ChatGPT detection model; and obtaining all network data packets accessing the Docker container application, and performing anomaly detection on the network data packets based on the ChatGPT detection model to obtain a detection result. According to the method, new threat and attack modes can be learned and adapted by researching the ChatGPT model, the collected flow data are processed, the ChatGPT model is trained and optimized to obtain the ChatGPT detection model, and the ChatGPT detection model is utilized to perform anomaly detection on the network data packet accessing the Docker container application, so that the detection efficiency for novel threats and antagonistic attacks is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of network security technology, and in particular to the field of Docker container security technology. Background Art

[0002] Docker is an open source software container platform. Developers can use Docker technology to package applications and their dependencies into a lightweight, portable container, and then publish it to any popular Linux machine. Docker container technology can greatly improve development efficiency and predictability, simplify deployment and operation and maintenance, and improve system performance and security.

[0003] In the prior art, the detection methods of Docker application traffic are usually based on rules, statistics and signatures. These methods usually rely on predefined features, so they are effective for known attack patterns. However, they are less efficient when facing new threats and adversarial attacks. Therefore, it is urgent to provide a solution to improve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide a Docker attack traffic detection method based on the ChatGPT model, which improves the problem of low detection efficiency in the prior art when facing new threats and adversarial attacks.

[0005] The present invention provides a Docker attack traffic detection method based on the ChatGPT model, which adopts the following technical solutions:

[0006] In the first aspect, a Docker attack traffic detection method based on the ChatGPT model includes:

[0007] Acquire and preprocess traffic data to obtain a network security data set, and train and optimize the ChatGPT model based on the network security data set to obtain a ChatGPT detection model;

[0008] Based on the virtual bridge created by the Docker server, all network data packets accessing the Docker container application are obtained, and the ChatGPT detection model reads the network data packets, performs anomaly detection and obtains the detection results.

[0009] In the above-mentioned Docker attack traffic detection method based on the ChatGPT model, the process of acquiring and preprocessing traffic data to obtain a network security data set includes:

[0010] The traffic data is cleaned, labeled, restored, characterized, calculated and stored in real time, and data noise is filtered and information is extracted to obtain a network security data set; the traffic data includes a vulnerability analysis data set, a traffic analysis data set, an attack assessment data set and a log analysis data set.

[0011] As described above, the Docker attack traffic detection method based on the ChatGPT model, wherein the process of data noise filtering and information extraction includes: performing noise filtering on the text information in the traffic data based on NLP technology, and capturing and analyzing the time series changes of the traffic data.

[0012] The Docker attack traffic detection method based on the ChatGPT model as described above, wherein, before training and optimizing the ChatGPT model based on the network security data set to obtain the ChatGPT detection model, it includes: fine-tuning the network security scenario adaptability of the ChatGPT initial model based on the model adaptability adjustment technology to obtain the ChatGPT model.

[0013] The Docker attack traffic detection method based on the ChatGPT model as described above, wherein the process of training and optimizing the ChatGPT model includes:

[0014] The ChatGPT model is trained in a targeted manner based on the network security data set and multiple training strategies, and a ChatGPT detection model is obtained by introducing multiple learning algorithms into the ChatGPT model based on an integrated learning approach.

[0015] The Docker attack traffic detection method based on the ChatGPT model described above, wherein the process of obtaining all network data packets accessing the Docker container application based on the virtual bridge created by the Docker server includes:

[0016] A TCPdump collector is generated, network data packets passing through the virtual bridge are captured based on the TCPdump collector, and the network data packets are stored in a memory.

[0017] As described above, the Docker attack traffic detection method based on the ChatGPT model, wherein the process of the ChatGPT detection model reading the network data packet includes: the ChatGPT detection model reads the network data packet in the memory through the API interface.

[0018] As described above, the Docker attack traffic detection method based on the ChatGPT model, wherein the process of performing anomaly detection and obtaining detection results includes: performing anomaly detection on the network data packet based on the ChatGPT detection model, obtaining the detection result and returning the detection result to the storage.

[0019] In a second aspect, a computer-readable storage medium stores a computer program, wherein the program is used to execute any of the above-mentioned Docker attack traffic detection methods.

[0020] In a third aspect, a terminal device includes a processor and a computer-readable storage medium, wherein the processor is used to execute a computer program, and the computer-readable storage medium is used to store a computer program, and the program is used to execute any of the above-mentioned Docker attack traffic detection methods.

[0021] The beneficial effects of the second and third aspects can refer to the description of the first aspect.

[0022] The beneficial effects of the Docker attack traffic detection method based on the ChatGPT model proposed in the present invention are:

[0023] 1. The present invention can timely discover and respond to new threat patterns by acquiring and preprocessing traffic data.

[0024] 2. The ChatGPT detection model of the present invention can significantly improve the detection success rate when facing new threats and adversarial attacks. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flow chart of a Docker attack traffic detection method based on the ChatGPT model provided in an embodiment of the present invention;

[0026] Figure 2 The present invention provides a virtual bridge schematic diagram of a Docker attack traffic detection method based on a ChatGPT model. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with general skills in the field to which the present invention belongs. "Including" and similar words used in this article mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0028] The embodiment of the present invention provides a Docker attack traffic detection method based on the ChatGPT model, see Figure 1 ,include:

[0029] S1. Acquire and preprocess traffic data to obtain a network security data set, and train and optimize the ChatGPT model based on the network security data set to obtain a ChatGPT detection model;

[0030] S2. Based on the virtual bridge created by the Docker server, all network data packets accessing the Docker container application are obtained, and the ChatGPT detection model reads the network data packets and performs anomaly detection to obtain detection results.

[0031] In some embodiments, the process of executing step S1 includes:

[0032] S1.1. Obtain and preprocess traffic data to obtain a network security dataset;

[0033] S1.2. Based on the network security dataset, the ChatGPT model is trained and optimized to obtain a ChatGPT detection model.

[0034] Specifically, when executing step S1.1, acquiring and preprocessing the traffic data to obtain a network security data set includes:

[0035] S1.1.1. Obtain flow data;

[0036] S1.1.2. Preprocess the traffic data to obtain the network security dataset.

[0037] Specifically, when executing step S1.1.1 and obtaining traffic data, the traffic data integrates diversified traffic data from network security companies, equipment manufacturing companies, application industries, and third-party data service providers to ensure that the data covers various scenarios and devices.

[0038] Furthermore, the traffic data includes a vulnerability analysis data set, a traffic analysis data set, an attack analysis data set, and a log analysis data set.

[0039] In fact, a dynamically updated data pool will be established to regularly obtain the latest traffic data based on the traffic data sources to achieve automatic data collection and integration.

[0040] Specifically, when executing step S1.1.2, it includes:

[0041] S1.1.2.1, perform data cleaning, data identification, word form restoration, feature extraction, real-time calculation and real-time storage on the traffic data;

[0042] S1.1.2.2. After data noise filtering and information extraction, the network security dataset is obtained.

[0043] Specifically, when executing step S1.1.2.1, it includes:

[0044] The data cleaning is to clean the traffic data by deleting redundant, invalid or noisy data, and use automated data cleaning tools to detect and correct abnormal values ​​and missing values ​​in the data;

[0045] The data identification is to identify and classify the collected data, and mark the data with standard labels (such as attack type, traffic characteristics, source IP, target IP, etc.) to facilitate rapid retrieval and analysis in subsequent training;

[0046] The word form restoration is to restore the word form of the text part in the traffic data to reduce the redundancy of vocabulary;

[0047] The feature extraction is to extract key traffic data features (such as packet size, traffic direction, communication protocol, etc.) to improve the recognition accuracy of the model.

[0048] Specifically, when executing step S1.1.2.2, it includes:

[0049] Based on NLP technology, noise is filtered out of text information in the traffic data, and time series changes of the traffic data are captured and analyzed.

[0050] In fact, using NLP technology to filter noise on the network security dataset can automatically highlight key threat information; capturing and analyzing time series changes in traffic data within the network security dataset can be used to discover and respond to new threat patterns.

[0051] Specifically, when executing step S1.2, it includes:

[0052] S1.2.1. Fine-tune the adaptability of the ChatGPT initial model to network security scenarios based on the model adaptability adjustment technology, and fine-tune the data drive of the ChatGPT initial model based on network security data to obtain the ChatGPT model;

[0053] S1.2.2. Train and optimize the ChatGPT model to obtain a ChatGPT detection model.

[0054] Specifically, when executing step S1.2.1, the model adaptive adjustment technology includes Low-Rank Adaptation and Adapter technologies, and fine-tuning the ChatGPT initial model can ensure that while maintaining the original capabilities of the model, it improves its adaptability to specific network security scenarios and improves the detection accuracy of the model.

[0055] Specifically, when executing step S1.2.2, the ChatGPT model is trained in a targeted manner based on the network security data set and multiple training strategies, and multiple learning algorithms are introduced into the ChatGPT model based on an integrated learning method to obtain a ChatGPT detection model.

[0056] Among them, the training strategies include adversarial training and transfer learning.

[0057] In fact, in the process of training and optimizing the ChatGPT model, data generated by threat intelligence, open source platforms and custom traffic will be integrated into the network security dataset. By enriching the training data samples, the generalization ability of the model will be improved. By applying adversarial training technology, adversarial samples will be generated and injected to enhance the model's robustness and resistance to unknown attacks.

[0058] In some embodiments, the process of executing step S2 includes:

[0059] S2.1, based on the virtual bridge created by the Docker server, obtain all network data packets accessing the Docker container application;

[0060] S2.2. The ChatGPT detection model reads the network data packet and performs anomaly detection to obtain a detection result.

[0061] Specifically, when executing step S2.1, it includes:

[0062] S2.1.1. Deploy Docker container applications;

[0063] S2.1.2. Obtain all network data packets accessing the Docker container application based on the virtual bridge.

[0064] Specifically, when executing step S2.1.1, deploying the Docker container application, it includes:

[0065] Write application code, create a Dockerfile and execute a Docker image creation command on the Dockerfile, run the Docker image, and obtain a Docker container that runs the application.

[0066] Specifically, when executing step S2.1.2, it includes:

[0067] S2.1.2.1, Virtual bridge created based on Docker server;

[0068] S2.1.2.2. Obtain all network data packets accessing the Docker container application.

[0069] Specifically, when executing step S2.1.2.1, it includes: starting the Docker server, creating a virtual bridge on the host, the Docker container application started by the host will be connected to the virtual bridge, Docker will select an IP address and subnet different from the host machine to assign to the virtual bridge, and the Docker container application connected to the virtual bridge will select an unoccupied IP address from the subnet for use.

[0070] In fact, see Figure 2 , Docker uses veth pair technology to create two virtual network interface devices on the host, namely Ethernet interface and virtual Ethernet interface. The characteristics of Veth pair technology can ensure that no matter which interface receives the network message, it will transmit the message to the other interface. Docker attaches the virtual Ethernet interface to the virtual bridge created by Docker to ensure that the host's network messages can be sent to the virtual Ethernet interface, and adds the Ethernet interface to the namespace to which the Docker container application belongs to ensure that the host's network messages sent to the virtual Ethernet interface will be immediately received by the Ethernet interface, thereby achieving connectivity from the host to the Docker container application network, and at the same time ensuring that the Docker container application uses the Ethernet interface alone to achieve isolation of the container network environment.

[0071] Specifically, when executing step S2.1.2.2, it includes: generating a TCPdump collector, capturing network data packets passing through the virtual bridge based on the TCPdump collector, and storing the network data packets in a memory.

[0072] Specifically, when executing step S2.2, the ChatGPT detection model reads the network data packet and performs anomaly detection to obtain a detection result, including:

[0073] The ChatGPT detection model reads the network data packet in the memory through the API interface;

[0074] Based on the ChatGPT detection model, anomaly detection is performed on the network data packet to obtain a detection result and return the detection result to the memory.

[0075] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, wherein the program is used to execute any of the above-mentioned Docker attack traffic detection methods.

[0076] An embodiment of the present invention also provides a terminal device, including a processor and a computer-readable storage medium, wherein the processor is used to execute a computer program, and the computer-readable storage medium is used to store a computer program, and the program is used to execute any of the above-mentioned Docker attack traffic detection methods.

[0077] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways.

Claims

1. A Docker attack traffic detection method based on the ChatGPT model, characterized in that: The following steps are involved: Acquire and preprocess traffic data to obtain a network security data set, and train and optimize the ChatGPT model based on the network security data set to obtain a ChatGPT detection model; Based on the virtual bridge created by the Docker server, all network data packets accessing the Docker container application are obtained, and the ChatGPT detection model reads the network data packets, performs anomaly detection and obtains the detection results.

2. According to a ChatGPT model-based Docker attack traffic detection method according to claim 1, it is characterized in that: The process of acquiring and preprocessing traffic data to obtain a network security dataset includes: The traffic data is cleaned, labeled, restored, characterized, calculated and stored in real time, and data noise is filtered and information is extracted to obtain a network security data set; the traffic data includes a vulnerability analysis data set, a traffic analysis data set, an attack assessment data set and a log analysis data set.

3. According to a ChatGPT model-based Docker attack traffic detection method according to claim 2, it is characterized in that: The process of data noise filtering and information extraction includes: filtering the text information in the traffic data based on NLP technology, and capturing and analyzing the time series changes of the traffic data.

4. According to a ChatGPT model-based Docker attack traffic detection method according to claim 1, it is characterized in that: Before training and optimizing the ChatGPT model based on the network security dataset to obtain the ChatGPT detection model, the following steps are included: The ChatGPT model is obtained by fine-tuning the adaptability of the ChatGPT initial model to network security scenarios based on the model adaptability adjustment technology.

5. According to a ChatGPT model-based Docker attack traffic detection method according to claim 4, it is characterized in that: The process of training and optimizing the ChatGPT model includes: The ChatGPT model is trained in a targeted manner based on the network security data set and multiple training strategies, and a ChatGPT detection model is obtained by introducing multiple learning algorithms into the ChatGPT model based on an integrated learning approach.

6. According to a ChatGPT model-based Docker attack traffic detection method according to claim 1, it is characterized in that: The process of obtaining all network data packets accessing the Docker container application based on the virtual bridge created by the Docker server includes: A TCPdump collector is generated, network data packets passing through the virtual bridge are captured based on the TCPdump collector, and the network data packets are stored in a memory.

7. According to a ChatGPT model-based Docker attack traffic detection method according to claim 6, it is characterized in that: The process of the ChatGPT detection model reading the network data packet includes: the ChatGPT detection model reads the network data packet in the memory through the API interface.

8. According to a ChatGPT model-based Docker attack traffic detection method according to claim 7, it is characterized in that: The process of performing anomaly detection and obtaining detection results includes: The ChatGPT detection model performs anomaly detection on the network data packet, obtains the detection result and returns the detection result to the memory.

9. A computer-readable storage medium storing a computer program, characterized in that: The program is used to execute the Docker attack traffic detection method as described in any one of claims 1 to 8.

10. A terminal device comprising a processor and a computer-readable storage medium, characterized in that: The processor is used to execute a computer program, the computer-readable storage medium is used to store a computer program, and the program is used to execute the Docker attack traffic detection method according to any one of claims 1 to 8.