Network sensitive application communication behavior detection method and device, equipment and medium

By extracting the five-tuple data of the connecting traffic data from the encrypted connecting traffic data and inputting it into the target detection model, the problem of general detection conclusions in the existing technology is solved, the malicious detection efficiency of network sensitive applications is improved, and network communication security is ensured.

CN119011272AInactive Publication Date: 2024-11-22NO 15 INST OF CHINA ELECTRONICS TECH GRP
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411222249.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing network-sensitive application communication behavior detection methods cannot fully characterize the true intention of encrypted traffic data, resulting in a general detection conclusion and judgment deviation, which makes the detection efficiency poor.

Method used

By extracting the five-tuple data of the connecting traffic data (source IP address, destination IP address, encryption protocol, source port number and destination port number) from the encrypted connecting traffic data, enter the target detection model to determine the maliciousness of the connecting behavior of sensitive applications, adopt multiple The combined model of the base classifier and the second network branch performs the output of the initial and final detection results.

Benefits of technology

It effectively improves the malicious detection efficiency of sensitive application communication behavior, ensures network communication security, and reduces detection bias by extracting five-tuple information to cover the real behavioral intention of encrypted traffic data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119011272A_ABST
    Figure CN119011272A_ABST
Patent Text Reader

Abstract

The invention provides a network sensitive application communication behavior detection method and device, equipment and a medium. The method comprises the steps that whether a sensitive application communication behavior exists in a communication network or not is detected; if it is detected that the sensitive application communication behavior exists in the communication network, obtaining encrypted communication flow data of the sensitive application communication behavior; acquiring corresponding communication quintuple data from the encrypted communication flow data of the sensitive application communication behavior; and inputting the communication quintuple data into the target detection model, and determining a malicious detection result of the sensitive application communication behavior based on an output result of the target detection model. Therefore, by extracting the quintuple information content in the encrypted traffic data to cover the real behavior intention reflected by the encrypted traffic data, the malicious detection efficiency of the sensitive application communication behavior can be effectively improved, and the network communication security is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the technical field of malicious behavior detection, and in particular, to a method, apparatus, device and medium for detecting communication behavior of network-sensitive applications. Background Art

[0002] With the continuous development of Internet technology, the number of Internet users is also increasing, and cyberspace security issues are also receiving more and more attention. At present, as more and more network communications adopt traffic encryption technology, especially encryption applications represented by Tor, SSL, VPN, etc. have been widely used, the rapid application of encrypted traffic technology is also changing the situation of network security threats. Network attackers have begun to use encryption technology to hide malicious network application attack paths and attack traces. Many malware are hidden in the transport layer security protocol traffic to transmit malicious messages, posing a serious threat to network security. Therefore, it is of great significance to maintain network security and stability by detecting and identifying the communication behavior of network-sensitive encryption applications.

[0003] In the related technology, the network sensitive application communication behavior detection model is mainly built based on the common machine learning algorithm. By randomly extracting part of the traffic data of the sensitive application behavior, the detection results output by the machine learning algorithm model are used to directly judge whether the network sensitive application communication behavior has certain maliciousness or harmfulness.

[0004] However, the sensitive application behavior traffic data extracted by the existing methods cannot fully characterize the true intention of the communication behavior of network sensitive applications. The detection conclusions drawn are relatively general and have certain judgment biases, resulting in low detection efficiency. Summary of the invention

[0005] The embodiments described herein provide a method, apparatus, device and medium for detecting communication behavior of network-sensitive applications, which overcome the above-mentioned problems.

[0006] In a first aspect, according to the content of the present disclosure, a method for detecting communication behavior of network-sensitive applications is provided, comprising:

[0007] Detect whether there is any sensitive application communication behavior in the communication network;

[0008] If the sensitive application communication behavior is detected in the communication network, obtaining encrypted communication flow data of the sensitive application communication behavior;

[0009] Obtain corresponding communication five-tuple data from the encrypted communication traffic data of the communication behavior of the sensitive application, wherein the communication five-tuple data is used to describe the traffic data associated with the source IP address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number;

[0010] Inputting the communication five-tuple data into a target detection model, and determining a malicious detection result of the communication behavior of the sensitive application based on an output result of the target detection model;

[0011] The malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior is a malicious communication behavior.

[0012] Optionally, the detecting whether there is any sensitive application communication behavior in the communication network includes:

[0013] Acquire network application communication information, wherein the network application communication information includes: a communication node application and an application communication state, wherein the application communication state is used to describe whether the communication node application currently has a communication behavior;

[0014] If the application communication state of the communication node application indicates that the communication node application has a communication behavior, then obtaining the application type of the communication node application;

[0015] If the application type of the communication node application is a sensitive application class, determining that the sensitive application communication behavior exists in the communication network;

[0016] If the application type of the communication node application is a non-sensitive application class, it is determined that the sensitive application communication behavior does not exist in the communication network.

[0017] Optionally, before inputting the communication five-tuple data into a target detection model and determining the malicious detection result of the communication behavior of the sensitive application based on the output result of the target detection model, the method further includes:

[0018] Obtain sample communication traffic data of historical application communication behaviors;

[0019] Performing data preprocessing on the sample communication flow data of the historical application communication behavior to obtain target communication flow data;

[0020] Training a preset network model based on the target communication flow data to obtain the target detection model;

[0021] Among them, the preset network model includes: a first network branch and a second network branch, the first network branch includes multiple base classifiers, each of the base classifiers outputs an initial detection result of the application communication behavior corresponding to the application communication traffic data based on the input application communication traffic data, and the second network branch is used to determine the final detection result of the application communication behavior from the multiple initial detection results of the application communication behavior.

[0022] Optionally, the performing data preprocessing on the sample communication flow data of the historical application communication behavior to obtain target communication flow data includes:

[0023] Acquire sample encrypted traffic data from the sample communication traffic data of the historical application communication behavior;

[0024] Performing a five-tuple data extraction operation on the sample encrypted traffic data to obtain sample five-tuple data, wherein the five-tuple data extraction operation is used to extract traffic data associated with the source IP address, traffic data associated with the target IP address, traffic data associated with the encryption protocol, traffic data associated with the source port number, and traffic data associated with the target port number from the sample encrypted traffic data;

[0025] Data anomaly processing is performed on the sample quintuple data to obtain the target communication flow data.

[0026] Optionally, performing data anomaly processing on the sample quintuple data to obtain the target communication flow data includes:

[0027] If it is determined that there is preset abnormal data in the sample quintuple data, updating the preset abnormal data based on adjacent traffic data of the preset abnormal data within a first preset data range;

[0028] If it is determined that there is missing data at a preset data position in the sample quintuple data, the preset data position is filled with data based on the adjacent traffic data within the second data range corresponding to the preset data position.

[0029] Optionally, the training of a preset network model based on the target communication flow data to obtain the target detection model includes:

[0030] Inputting the target communication flow data into each of the base classifiers in the first network branch respectively to obtain a plurality of initial detection results of the historical application communication behaviors;

[0031] Inputting a plurality of initial detection results of the historical application communication behavior into the second network branch to obtain a final detection result of the historical application communication behavior;

[0032] Based on the loss value between the final detection result of the historical application communication behavior and the expected detection result of the historical application communication behavior, each base classifier in the first network branch is trained to obtain the target detection model.

[0033] Optionally, also include:

[0034] Obtain malicious detection results of the communication behavior of the sensitive application;

[0035] If the malicious detection result of the sensitive application communication behavior indicates that the sensitive application communication behavior is a malicious communication behavior, a behavior warning prompt is sent to the application device corresponding to the sensitive application communication behavior.

[0036] In a second aspect, according to the content of the present disclosure, a network-sensitive application communication behavior detection device is provided, comprising:

[0037] A detection module is used to detect whether there is any sensitive application communication behavior in the communication network;

[0038] A first acquisition module is used to acquire encrypted communication flow data of the sensitive application communication behavior if the sensitive application communication behavior is detected in the communication network;

[0039] A second acquisition module is used to acquire corresponding communication five-tuple data from the encrypted communication traffic data of the communication behavior of the sensitive application, wherein the communication five-tuple data is used to describe the traffic data associated with the source IP address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number;

[0040] A determination module, used for inputting the communication five-tuple data into a target detection model, and determining a malicious detection result of the communication behavior of the sensitive application based on an output result of the target detection model;

[0041] The malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior is a malicious communication behavior.

[0042] In a third aspect, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method for detecting communication behavior of network-sensitive applications in any of the above embodiments are implemented.

[0043] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for detecting communication behavior of network-sensitive applications in any of the above embodiments are implemented.

[0044] The network sensitive application communication behavior detection method provided in the embodiment of the present application detects whether there is sensitive application communication behavior in the communication network; if the sensitive application communication behavior is detected in the communication network, the encrypted communication flow data of the sensitive application communication behavior is obtained; the corresponding communication five-tuple data is obtained from the encrypted communication flow data of the sensitive application communication behavior, and the communication five-tuple data is used to describe the flow data associated with the source IP address, the flow data associated with the target IP address, the flow data associated with the encryption protocol, the flow data associated with the source port number, and the flow data associated with the target port number; the communication five-tuple data is input into the target detection model, and the malicious detection result of the sensitive application communication behavior is determined based on the output result of the target detection model; wherein, the malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior belongs to malicious communication behavior. In this way, by extracting the five-tuple information content in the encrypted flow data to cover the real behavior intention reflected by the encrypted flow data, it is convenient to effectively improve the malicious detection efficiency of the sensitive application communication behavior and ensure the security of network communication.

[0045] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments will be briefly described below. It should be noted that the drawings described below only relate to some embodiments of the present disclosure, but are not intended to limit the present disclosure, wherein:

[0047] Figure 1 It is a flow chart of a method for detecting communication behavior of network-sensitive applications provided by the present invention.

[0048] Figure 2 It is a structural schematic diagram of a network sensitive application communication behavior detection device provided by the present invention.

[0049] Figure 3 It is a structural schematic diagram of a computer device provided by the present disclosure.

[0050] It should be noted that the elements in the drawings are schematic and not drawn to scale. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work also fall within the scope of protection of the present disclosure.

[0052] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by a person skilled in the art to which the subject matter of the present disclosure belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the specification and the relevant art, and will not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, a statement that two or more parts are "connected" or "coupled" together shall mean that the parts are joined together directly or through one or more intermediate components.

[0053] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiments" in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0054] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists, A and B exist at the same time, and B exists. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. Terms such as "first" and "second" are only used to distinguish one component (or a part of a component) from another component (or another part of a component).

[0055] In the description of the present application, unless otherwise specified, "plurality" means more than two (including two), and similarly, "multiple groups" means more than two groups (including two).

[0056] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0057] Figure 1 is a flow chart of a method for detecting communication behavior of network-sensitive applications provided by an embodiment of the present disclosure, such as Figure 1As shown, the specific process of the network sensitive application communication behavior detection method includes:

[0058] S110, detecting whether there is any sensitive application communication behavior in the communication network.

[0059] Among them, the communication network may have multiple application communication behaviors in the current communication state, and the multiple application communication behaviors may include sensitive application communication behaviors and non-sensitive application communication behaviors. Sensitive application communication behaviors may be communication behaviors between sensitive applications and non-sensitive applications, or communication behaviors between sensitive applications and sensitive applications. Non-sensitive application communication behaviors may be communication behaviors between non-sensitive applications and non-sensitive applications.

[0060] In some embodiments, detecting whether there is any sensitive application communication behavior in the communication network includes:

[0061] Obtain network application communication information, the network application communication information including: communication node application and application communication status; if the application communication status of the communication node application indicates that the communication node application has communication behavior, obtain the application type of the communication node application; if the application type of the communication node application is a sensitive application class, determine that sensitive application communication behavior exists in the communication network; if the application type of the communication node application is a non-sensitive application class, determine that no sensitive application communication behavior exists in the communication network.

[0062] Among them, the application communication state is used to describe whether the communication node application currently has communication behavior. For example, when the application communication state of the communication node application is the preset state, it means that the communication node application has communication behavior; when the application communication state of the communication node application is the non-preset state, it means that the communication node application does not have communication behavior; the preset state is used to describe that a communication node application is in data communication.

[0063] The application type of the communication node application may include: sensitive application class and non-sensitive application class. The application type of the communication node application may be determined based on the data transmission type of the communication node application. For example, if the data transmission type of a communication node application is privacy class / non-public class / advanced encryption class, then the application type of the communication node application is determined to be a sensitive application class; the application type of the communication node application may also be derived based on pre-defined.

[0064] Thus, communication node applications with communication behaviors are determined from the communication network, and through the application types of the communication node applications, it is effectively determined whether there is sensitive application communication behavior in the current state of the communication network.

[0065] S120: If sensitive application communication behavior is detected in the communication network, the encrypted communication flow data of the sensitive application communication behavior is obtained.

[0066] The encrypted communication traffic data of sensitive application communication behavior is the encrypted traffic data generated by sensitive application behavior. Since sensitive application communication behavior will generate encrypted traffic data and non-encrypted traffic data, the encrypted traffic data can be filtered out through traffic filtering as the encrypted communication traffic data of sensitive application communication behavior.

[0067] S130. Obtain corresponding communication five-tuple data from the encrypted communication flow data of the sensitive application communication behavior.

[0068] Among them, the communication five-tuple data is used to describe the traffic data associated with the source IP (Internet Protocol) address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number.

[0069] Obtaining corresponding communication five-tuple data from the encrypted communication traffic data of sensitive application communication behavior may include: extracting traffic data associated with the source IP address, traffic data associated with the target IP address, traffic data associated with the encryption protocol, traffic data associated with the source port number, and traffic data associated with the target port number from the encrypted communication traffic data of sensitive application communication behavior.

[0070] S140: Input the communication five-tuple data into the target detection model, and determine the malicious detection result of the communication behavior of the sensitive application based on the output result of the target detection model.

[0071] Among them, the malicious detection results of sensitive application communication behaviors are used to describe whether the sensitive application communication behaviors are malicious communication behaviors.

[0072] The malicious detection result of the sensitive application communication behavior includes: a first detection result and a second detection result. When the malicious detection result of the sensitive application communication behavior is the first detection result, it means that the sensitive application communication behavior is a malicious communication behavior; when the malicious detection result of the sensitive application communication behavior is the second detection result, it means that the sensitive application communication behavior is not a malicious communication behavior.

[0073] In some embodiments, before inputting the communication five-tuple data into the target detection model and determining the malicious detection result of the communication behavior of the sensitive application based on the output result of the target detection model, the method of this embodiment further includes:

[0074] Acquire sample communication traffic data of historical application communication behaviors; perform data preprocessing on the sample communication traffic data of historical application communication behaviors to obtain target communication traffic data; and train a preset network model based on the target communication traffic data to obtain a target detection model.

[0075] Among them, the preset network model includes: a first network branch and a second network branch, the first network branch includes multiple base classifiers, each base classifier outputs an initial detection result of the application communication behavior corresponding to the application communication traffic data based on the input application communication traffic data, and the second network branch is used to determine the final detection result of the application communication behavior from the multiple initial detection results of the application communication behavior.

[0076] The first network branch can be a coarse detection branch, which is used to output multiple initial detection results, and the second network branch can be a fine detection score, which is used to evaluate the final detection result from multiple initial detection results. Thus, by combining multiple base classifiers to perform unified detection on the model input, it is convenient to determine a more accurate detection result from multiple detection results, effectively improving the detection efficiency.

[0077] In some embodiments, data preprocessing is performed on sample communication flow data of historical application communication behaviors to obtain target communication flow data, including:

[0078] Obtain sample encrypted traffic data from sample communication traffic data of historical application communication behaviors; perform a five-tuple data extraction operation on the sample encrypted traffic data to obtain sample five-tuple data; perform data anomaly processing on the sample five-tuple data to obtain target communication traffic data.

[0079] Among them, since the captured sample communication traffic data may contain some traffic data of other network applications, and may also include some traffic data packets of non-encrypted protocols, it is necessary to filter the captured sample communication traffic data to obtain a purer and more complete experimental data set (i.e., model input). The specific filtering rules can be based on the first 5 bytes of the sample communication traffic data packet to determine whether it is a data packet using the SSL (Secure Sockets Layer) protocol.

[0080] Generally, the basic structure of the SSL protocol is mainly composed of the record layer and the handshake layer, in which the record layer is the SSL record protocol, and the handshake layer mainly includes the application data protocol, the SSL handshake protocol, the SSL alarm protocol, etc. By selecting the first 5 data packet bytes of the traffic data and testing them, it is confirmed whether there is data of SSL-related protocols inside them, so as to determine whether the traffic data packet of this part adopts the SSL protocol, and then the sample encrypted traffic data in the sample communication traffic data can be screened out.

[0081] After obtaining the sample encrypted traffic data, data labeling can also be performed. For example, the manual labeling method is used to complete the centralized labeling of all data, and the IP ownership information is determined based on the source IP address of the captured data packet, and the corresponding destination server information is determined by the elimination method. Manual labeling mainly marks whether the data packet is malicious encrypted traffic data based on the source IP address of the captured data packet and its corresponding destination server information, and divides the data set into different malicious types based on the IP information it carries. The labeling result is: the data set has been labeled with information, so that it can directly distinguish whether it is malicious encrypted traffic, which is convenient as the expected detection result of historical application communication behavior.

[0082] After formulating filtering rules to screen out data packets that meet the requirements (i.e., sample encrypted traffic data), the quintuple of the data packet is extracted and stored in the corresponding database. At the same time, the interactive protocol fields of SSL are sorted, connected and deduplicated. Finally, an experimental data set that meets the requirements is generated for model training.

[0083] The five-tuple data extraction operation is used to extract traffic data associated with the source IP address, traffic data associated with the target IP address, traffic data associated with the encryption protocol, traffic data associated with the source port number, and traffic data associated with the target port number from the sample encrypted traffic data.

[0084] It should be noted that during the quintuple data extraction process, if a data packet with missing quintuple pattern content is encountered, it will be discarded and the remaining data packets will continue to be filtered to ensure that the filtered data set fully meets the experimental requirements.

[0085] According to the structural characteristics of the SSL interactive protocol field, this embodiment extracts the interactive protocol field content of the SSL protocol corresponding to each data packet, removes the repeated field content and reconnects the filtered field content to complete the processing of the experimental data set and effectively obtain the target communication traffic data.

[0086] In some embodiments, data anomaly processing is performed on the sample quintuple data to obtain target communication flow data, including:

[0087] If it is determined that there is preset abnormal data in the sample quintuple data, the preset abnormal data is updated based on the adjacent traffic data of the preset abnormal data within the first preset data range; if it is determined that there is missing data at the preset data position in the sample quintuple data, the preset data position is filled with data based on the adjacent traffic data corresponding to the preset data position within the second data range.

[0088] The first preset data range may be the same as the second preset data range, or the first preset data range may be different from the second preset data range.

[0089] For example, when it is determined that there is preset abnormal data in the sample five-tuple data, the preset abnormal data can be replaced according to the mean of the 10 adjacent data items before and after the preset abnormal data; when it is determined that there is missing data at the preset data position in the sample five-tuple data, the missing data can be filled according to the mean of the 10 adjacent data items before and after the preset data position. This is convenient for effectively improving the integrity of the target communication flow data, thereby improving the training accuracy of the model.

[0090] After performing data anomaly processing on the sample quintuple data and obtaining the target communication flow data, the target communication flow data can also be normalized to transform data values ​​of different sizes into the same size range to speed up data operations. For example, formula (1) can be used to convert the range of data value variables to between [0,1], which helps to ensure that the ranges of values ​​of different sizes can be on a similar scale, avoid possible data difference risks, and facilitate model training and optimization.

[0091]

[0092] In formula (1), X is the target communication flow data before normalization, and X' is the target communication flow data after normalization.

[0093] In some embodiments, a preset network model is trained based on target communication traffic data to obtain a target detection model, including:

[0094] The target communication traffic data is respectively input into each base classifier in the first network branch to obtain multiple initial detection results of the historical application communication behavior; the multiple initial detection results of the historical application communication behavior are input into the second network branch to obtain the final detection result of the historical application communication behavior; based on the loss value between the final detection result of the historical application communication behavior and the expected detection result of the historical application communication behavior, each base classifier in the first network branch is trained to obtain a target detection model.

[0095] Among them, when the loss value between the final detection result of the historical application communication behavior and the expected detection result of the historical application communication behavior is greater than the preset loss threshold, the target communication traffic data can be divided into a training set and a test set, and each base classifier in the first network branch is trained. The training method can adopt a six-fold cross-validation method.

[0096] Taking the first network branch including 5 base classifiers as an example, the target communication traffic data is divided into two parts: training set T and test set D, and the training set T is divided into 6 subsets of equal size, namely: T1, T2, T3, T4, T5, T6, T1 is the verification subset, T2, T3, T4, T5, T6 are training subsets, which serve as the input data of each base classifier.

[0097] When training the five base classifiers, each base classifier is first tested on the test set D and the validation subset T1, and the prediction result r1 of the validation subset T1 and the prediction result p1 of the test set D are obtained according to the output of the second network branch; then T2 is set as the validation subset, T1, T3, T4, T5, and T6 are the training subsets, and each base classifier is tested on the test set D and the validation subset T2, and the prediction result r2 of the validation subset T2 and the prediction result p1 of the test set D are obtained according to the output of the second network branch. The prediction result p2 of the test set D is obtained by using the output of the second network branch. T3 is set as the validation subset, T1, T2, T4, T5, and T6 are training subsets, and each base classifier is tested on the test set D and the validation subset T3. The prediction result r3 of the validation subset T3 and the prediction result p3 of the test set D are obtained by using the output of the second network branch. T4 is set as the validation subset, T1, T2, T3, T5, and T6 are training subsets, and each base classifier is tested on the test set D and the validation subset T4. The output is the prediction result r4 of the validation subset T4 and the prediction result p4 of the test set D; T5 is set as the validation subset, T1, T2, T3, T4, and T6 are training subsets, and each base classifier is tested on the test set D and the validation subset T5. According to the output of the second network branch, the prediction result r5 of the validation subset T5 and the prediction result p5 of the test set D are obtained; T6 is set as the validation subset, T1, T2, T3, T4, and T5 are training subsets, and each base classifier is tested on the test set D and the validation subset T5. The network is tested on the validation subset T6, and the prediction result r6 of the validation subset T6 and the prediction result p6 of the test set D are obtained according to the output of the second network branch; in this way, the prediction results r1, r2, r3, r4, r5, r6 of the six validation subsets and the prediction results p1, p2, p3, p4, p5, p6 of the six test sets are obtained. Finally, r1, r2, r3, r4, r5, r6 are merged into R1, and the average of p1, p2, p3, p4, p5, p6 is obtained to obtain P1.

[0098] Repeat the above steps three times to obtain the prediction results R2, R3, R4 of the validation subset; the prediction results P2, P3, P4 of the test set. Use R1, R2, R3, R4 as the input of the meta-classifier to verify the detection results of the base classifier, and use P1, P2, P3, P4 for testing to confirm the validity of the target detection model. The target detection model in this embodiment mainly refers to a Stacking ensemble learning detection model constructed by using the random forest algorithm and the XGBoost algorithm model as learners.

[0099] In this embodiment, by detecting whether there is a sensitive application communication behavior in the communication network; if the sensitive application communication behavior is detected in the communication network, the encrypted communication flow data of the sensitive application communication behavior is obtained; the corresponding communication five-tuple data is obtained from the encrypted communication flow data of the sensitive application communication behavior, and the communication five-tuple data is used to describe the flow data associated with the source IP address, the flow data associated with the target IP address, the flow data associated with the encryption protocol, the flow data associated with the source port number, and the flow data associated with the target port number; the communication five-tuple data is input into the target detection model, and the malicious detection result of the sensitive application communication behavior is determined based on the output result of the target detection model; wherein, the malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior belongs to malicious communication behavior. In this way, by extracting the five-tuple information content in the encrypted flow data to cover the real behavior intention reflected by the encrypted flow data, it is convenient to effectively improve the malicious detection efficiency of the sensitive application communication behavior and ensure the security of network communication.

[0100] In some embodiments, the method of this embodiment further includes:

[0101] Obtain malicious detection results of sensitive application communication behaviors; if the malicious detection results of sensitive application communication behaviors indicate that the sensitive application communication behaviors are malicious communication behaviors, send behavior warning prompts to application devices corresponding to the sensitive application communication behaviors.

[0102] Among them, the early warning prompt can be used to instruct the application device corresponding to the sensitive application communication behavior to stop the sensitive application communication behavior, so as to effectively avoid communication network security problems caused by malicious sensitive application communication behavior.

[0103] In addition, this embodiment also provides an experimental verification example, the experimental data set comes from part of the encrypted traffic of common sensitive encryption applications in the captured network, specifically from encryption applications represented by Tor, Twitter, VPN, etc., and Wireshark is mainly used to capture the traffic data packets of these three encryption applications. The specific capture results are shown in Table 1 below.

[0104] Table 1 Experimental dataset

[0105]

[0106] The experimental environment in this embodiment is configured under Windows 10 operating system, 64G memory, Python 3.7 and other environments.

[0107] Experimental results analysis: Four different detection models were designed and compared on the test set, namely the target detection model in this embodiment, the detection model based on the K nearest neighbor algorithm, the detection model based on the convolutional neural network algorithm, and the detection model based on the XGBost algorithm. The final detection effects and detection efficiencies of different detection models were compared, as shown in Table 2.

[0108] Table 2 Experimental results

[0109]

[0110] It can be concluded that the overall detection effect of the target detection model in this embodiment is better than that of other detection models. The model has achieved the best results in accuracy, precision, recall, and F1 score. Compared with the other three detection models, it has improved overall, indicating that the target detection model can achieve better detection results in general behavior detection of network sensitive encryption applications.

[0111] The implementation of the method in this embodiment mainly includes: flow data preprocessing and detection model construction.

[0112] Traffic data preprocessing process: filter the captured data traffic and remove the traffic packets of non-encrypted protocols to obtain a purer and more complete experimental data set. The specific filtering rules mainly judge whether it is a data packet using the SSL protocol based on the content of the first 5 bytes of the traffic data packet; the manual marking method is mainly used to complete the centralized marking of all data sets. According to the source IP address of the captured data packet, its IP ownership information is judged, and the corresponding destination server information is determined by the elimination method; the five-tuple information of the data packets that meet the requirements are extracted and stored in the corresponding database, and the SSL interactive protocol fields are sorted, connected and deduplicated, and finally a data set that meets the requirements is generated; in the captured network sensitive application traffic data set, some feature values ​​may be too large or too small, mainly by using the average value of the 10 adjacent data before and after the abnormal value to replace the abnormal value; analysis of the existing data set found that the data set may have some missing values, mainly by using the average value of the 10 adjacent data before and after the missing data to fill the corresponding missing values; and combined with the idea of ​​data normalization, data values ​​of different sizes are transformed into the same size range to speed up data calculation. Specifically, the range of the data value variable is converted to [0,1], which helps ensure that the ranges of values ​​of different sizes are on a similar scale, avoids possible data discrepancy risks, and is beneficial to model training and optimization.

[0113] Detection model construction process: The feature set obtained after traffic data preprocessing is divided into two parts: training set and test set. The training set is divided into 6 equal-sized subsets according to the collected data; all base classifiers are trained using a six-fold cross-validation method, and the test is completed in combination with the validation set to obtain the prediction results of all base classifiers; the prediction values ​​obtained by all base learners are combined to obtain a new training set and a new test set; the training set is used as the input feature of the meta-learner to complete the training of the meta-learner, and the test set is tested to finally obtain the corresponding prediction results.

[0114] In summary, this embodiment is based on SSL protocol traffic detection technology, and constructs a Stacking ensemble learning detection model for network-sensitive encryption applications (such as Tor, VPN, etc.). By integrating multiple different types of base models together, the overall prediction accuracy and generalization ability are improved, and the base learners are selected to determine an excellent set of base learners. The accuracy of the prediction of the communication behavior of network-sensitive applications is further improved, and the effectiveness of this algorithm is verified through experiments to achieve effective detection of communication behaviors for network-sensitive applications.

[0115] Figure 2 A schematic diagram of the structure of a network-sensitive application communication behavior detection device provided in this embodiment. The network-sensitive application communication behavior detection device may include: a detection module 210, a first acquisition module 220, a second acquisition module 230 and a determination module 240.

[0116] The detection module 210 is used to detect whether there is any sensitive application communication behavior in the communication network.

[0117] The first acquisition module 220 is used to acquire encrypted communication flow data of the sensitive application communication behavior if sensitive application communication behavior is detected in the communication network.

[0118] The second acquisition module 230 is used to obtain corresponding communication five-tuple data from the encrypted communication traffic data of the sensitive application communication behavior, and the communication five-tuple data is used to describe the traffic data associated with the source IP address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number.

[0119] The determination module 240 is used to input the communication five-tuple data into the target detection model, and determine the malicious detection result of the sensitive application communication behavior based on the output result of the target detection model.

[0120] Among them, the malicious detection results of sensitive application communication behaviors are used to describe whether the sensitive application communication behaviors are malicious communication behaviors.

[0121] In this embodiment, optionally, the detection module 210 is specifically configured to:

[0122] Obtain network application communication information, the network application communication information includes: communication node application and application communication status, the application communication status is used to describe whether the communication node application currently has communication behavior; if the application communication status of the communication node application indicates that the communication node application has communication behavior, then obtain the application type of the communication node application; if the application type of the communication node application is a sensitive application class, then determine that sensitive application communication behavior exists in the communication network; if the application type of the communication node application is a non-sensitive application class, then determine that no sensitive application communication behavior exists in the communication network.

[0123] In this embodiment, optionally, it also includes: a third acquisition module, a processing module and a training module.

[0124] The third acquisition module is used to obtain sample communication flow data of historical application communication behaviors.

[0125] The processing module is used to pre-process the sample communication flow data of the historical application communication behavior to obtain the target communication flow data.

[0126] The training module is used to train the preset network model based on the target communication traffic data to obtain the target detection model.

[0127] Among them, the preset network model includes: a first network branch and a second network branch, the first network branch includes multiple base classifiers, each base classifier outputs an initial detection result of the application communication behavior corresponding to the application communication traffic data based on the input application communication traffic data, and the second network branch is used to determine the final detection result of the application communication behavior from the multiple initial detection results of the application communication behavior.

[0128] In this embodiment, the processing module is optionally used to:

[0129] Obtain sample encrypted traffic data from sample communication traffic data of historical application communication behaviors; perform a five-tuple data extraction operation on the sample encrypted traffic data to obtain sample five-tuple data, and the five-tuple data extraction operation is used to extract traffic data associated with the source IP address, traffic data associated with the target IP address, traffic data associated with the encryption protocol, traffic data associated with the source port number, and traffic data associated with the target port number from the sample encrypted traffic data; perform data anomaly processing on the sample five-tuple data to obtain target communication traffic data.

[0130] In this embodiment, the processing module is optionally used to:

[0131] If it is determined that there is preset abnormal data in the sample quintuple data, the preset abnormal data is updated based on the adjacent traffic data of the preset abnormal data within the first preset data range; if it is determined that there is missing data at the preset data position in the sample quintuple data, the preset data position is filled with data based on the adjacent traffic data corresponding to the preset data position within the second data range.

[0132] In this embodiment, the training module is optionally used to:

[0133] The target communication traffic data is respectively input into each base classifier in the first network branch to obtain multiple initial detection results of the historical application communication behavior; the multiple initial detection results of the historical application communication behavior are input into the second network branch to obtain the final detection result of the historical application communication behavior; based on the loss value between the final detection result of the historical application communication behavior and the expected detection result of the historical application communication behavior, each base classifier in the first network branch is trained to obtain a target detection model.

[0134] In this embodiment, optionally, it also includes: a fourth acquisition module and a sending module.

[0135] The fourth acquisition module is used to obtain malicious detection results of sensitive application communication behaviors.

[0136] The sending module is used to send a behavior warning prompt to the application device corresponding to the sensitive application communication behavior if the malicious detection result of the sensitive application communication behavior indicates that the sensitive application communication behavior is a malicious communication behavior.

[0137] The network sensitive application communication behavior detection device provided in the present disclosure can execute the above method embodiments. Its specific implementation principles and technical effects can be found in the above method embodiments, and the present disclosure will not repeat them here.

[0138] The present application also provides a computer device. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0139] The computer device includes a memory 310 and a processor 320 that are connected to each other through a system bus. It should be noted that the figure only shows a computer device with a memory 310 and a processor 320, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0140] Computer devices can be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. Computer devices can interact with users through keyboards, mice, remote controls, touch pads, or voice control devices.

[0141] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes a non-volatile memory or a volatile memory, for example, a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc., and the RAM may include a static RAM or a dynamic RAM. In some embodiments, the memory 310 may be an internal storage unit of a computer device, for example, a hard disk or a memory of the computer device. In other embodiments, the memory 310 may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, or a flash card (FlashCard) equipped on the computer device. Of course, the memory 310 may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device, such as the program code of the above method. In addition, the memory 310 may also be used to temporarily store various data that have been output or are to be output.

[0142] The processor 320 is generally used to perform the overall operation of the computer device. In this embodiment, the memory 310 is used to store program codes or instructions, the program code includes computer operation instructions, and the processor 320 is used to execute the program codes or instructions stored in the memory 310 or process data, such as running the program code of the above method.

[0143] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus system can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0144] Another embodiment of the present application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads a computer-readable program code stored in the computer-readable medium, so that the processor can execute the functional actions specified in each step or a combination of steps in the above method; and generate a device for implementing the functional actions specified in each block or a combination of blocks in the block diagram.

[0145] Computer-readable media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any appropriate combination of the foregoing, the memory is used to store program codes or instructions, the program codes include computer operating instructions, and the processor is used to execute the program codes or instructions of the above methods stored in the memory.

[0146] For the definitions of memory and processor, please refer to the description of the aforementioned computer device embodiment and will not be repeated here.

[0147] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0148] Each functional unit or module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0149] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program code.

[0150] In the claims, any reference symbols placed between brackets shall not be construed as limiting the claims. The word "comprising" described in the present application does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application can be implemented with the aid of hardware comprising several different elements and with the aid of a suitably programmed computer. In a unit claim that lists a number of devices, several units of these devices may be embodied by the same hardware item. The use of first, second, and third, etc. does not indicate any order, and these words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

[0151] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting communication behavior of network sensitive applications, characterized in that: include: Detect whether there is any sensitive application communication behavior in the communication network; If the sensitive application communication behavior is detected in the communication network, obtaining encrypted communication flow data of the sensitive application communication behavior; Obtain corresponding communication five-tuple data from the encrypted communication traffic data of the communication behavior of the sensitive application, wherein the communication five-tuple data is used to describe the traffic data associated with the source IP address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number; Inputting the communication five-tuple data into a target detection model, and determining a malicious detection result of the communication behavior of the sensitive application based on an output result of the target detection model; The malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior is a malicious communication behavior.

2. The method according to claim 1, characterized in that: The detecting whether there is a sensitive application communication behavior in the communication network includes: Acquire network application communication information, wherein the network application communication information includes: a communication node application and an application communication state, wherein the application communication state is used to describe whether the communication node application currently has a communication behavior; If the application communication state of the communication node application indicates that the communication node application has a communication behavior, then obtaining the application type of the communication node application; If the application type of the communication node application is a sensitive application class, determining that the sensitive application communication behavior exists in the communication network; If the application type of the communication node application is a non-sensitive application class, it is determined that the sensitive application communication behavior does not exist in the communication network.

3. The method according to claim 1, characterized in that Before inputting the communication five-tuple data into a target detection model and determining the malicious detection result of the communication behavior of the sensitive application based on the output result of the target detection model, the method further includes: Obtain sample communication traffic data of historical application communication behaviors; Performing data preprocessing on the sample communication flow data of the historical application communication behavior to obtain target communication flow data; Training a preset network model based on the target communication flow data to obtain the target detection model; Among them, the preset network model includes: a first network branch and a second network branch, the first network branch includes multiple base classifiers, each of the base classifiers outputs an initial detection result of the application communication behavior corresponding to the application communication traffic data based on the input application communication traffic data, and the second network branch is used to determine the final detection result of the application communication behavior from the multiple initial detection results of the application communication behavior.

4. The method according to claim 3, characterized in that The performing data preprocessing on the sample communication flow data of the historical application communication behavior to obtain target communication flow data includes: Acquire sample encrypted traffic data from the sample communication traffic data of the historical application communication behavior; Performing a five-tuple data extraction operation on the sample encrypted traffic data to obtain sample five-tuple data, wherein the five-tuple data extraction operation is used to extract traffic data associated with the source IP address, traffic data associated with the target IP address, traffic data associated with the encryption protocol, traffic data associated with the source port number, and traffic data associated with the target port number from the sample encrypted traffic data; Data anomaly processing is performed on the sample quintuple data to obtain the target communication flow data.

5. The method according to claim 4, characterized in that The performing data anomaly processing on the sample quintuple data to obtain the target communication flow data includes: If it is determined that there is preset abnormal data in the sample quintuple data, updating the preset abnormal data based on adjacent traffic data of the preset abnormal data within a first preset data range; If it is determined that there is missing data at a preset data position in the sample quintuple data, the preset data position is filled with data based on the adjacent traffic data within the second data range corresponding to the preset data position.

6. The method according to claim 3, characterized in that The training of a preset network model based on the target communication flow data to obtain the target detection model includes: Inputting the target communication flow data into each of the base classifiers in the first network branch respectively to obtain a plurality of initial detection results of the historical application communication behaviors; Inputting a plurality of initial detection results of the historical application communication behavior into the second network branch to obtain a final detection result of the historical application communication behavior; Based on the loss value between the final detection result of the historical application communication behavior and the expected detection result of the historical application communication behavior, each base classifier in the first network branch is trained to obtain the target detection model.

7. The method according to claim 1, characterized in that Also includes: Obtain malicious detection results of the communication behavior of the sensitive application; If the malicious detection result of the sensitive application communication behavior indicates that the sensitive application communication behavior is a malicious communication behavior, a behavior warning prompt is sent to the application device corresponding to the sensitive application communication behavior.

8. A network sensitive application communication behavior detection device, characterized in that: include: A detection module is used to detect whether there is any sensitive application communication behavior in the communication network; A first acquisition module is used to acquire encrypted communication flow data of the sensitive application communication behavior if the sensitive application communication behavior is detected in the communication network; A second acquisition module is used to acquire corresponding communication five-tuple data from the encrypted communication traffic data of the communication behavior of the sensitive application, wherein the communication five-tuple data is used to describe the traffic data associated with the source IP address, the traffic data associated with the target IP address, the traffic data associated with the encryption protocol, the traffic data associated with the source port number, and the traffic data associated with the target port number; A determination module, used for inputting the communication five-tuple data into a target detection model, and determining a malicious detection result of the communication behavior of the sensitive application based on an output result of the target detection model; The malicious detection result of the sensitive application communication behavior is used to describe whether the sensitive application communication behavior is a malicious communication behavior.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, a method for detecting communication behavior of network-sensitive applications as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting network-sensitive application communication behavior as described in any one of claims 1 to 7 is implemented.

Citation Information

Patent Citations

  • Method, device and system for detecting malicious software on mobile terminal

    CN103400076A

  • Abnormal behavior detection system of Android platform software

    CN105956474A

  • Encrypted malicious traffic detection method

    CN114172748A

  • Internet of Things terminal risk assessment method and device

    CN115065547A

  • Encrypted network flow detection method and device and electronic equipment

    CN117439763A