A computing power Internet traffic processing method and system

By building a malicious node detection model and risk assessment system, identifying and isolating malicious nodes in the computing power network, the problem of insufficient security of computing power network is solved, and the security and monitoring efficiency of user data and facilities are improved.

CN118740518BActive Publication Date: 2024-11-22SHENZHEN HUMENG TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411203276.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-22
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

Computing power networks are vulnerable to malicious node attacks during the traffic billing and business identification stages, resulting in insufficient security of user data and facilities, and traditional security means are difficult to meet new needs.

Method used

By obtaining the historical traffic data of computing power nodes, building a malicious node detection model, detecting node types, and conducting risk assessment based on topological structure and evaluation results, classifying the traffic of nodes to be dispatched, conducting business identification and traffic usage statistics to avoid malicious node access , improve safety.

Benefits of technology

Effectively identify and isolate malicious nodes, improve the security of computing power network, reduce resource consumption, improve monitoring efficiency, and avoid attacks on traffic billing and business identification types.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118740518B_ABST
    Figure CN118740518B_ABST
Patent Text Reader

Abstract

The present invention discloses a computing power Internet traffic processing method and system, which detects each network node, determines each malicious node and the associated risk node, and performs classified risk assessment according to different node types when allocating computing power tasks, so as to monitor the computing power tasks according to the risk assessment results, avoid malicious node access, attack traffic billing and business identification types, improve the security of user data and computing power network facilities, and at the same time perform classified monitoring according to the node risk level, improve the computing power task monitoring efficiency, and reduce resource consumption.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computing power network technology, and in particular to a computing power Internet traffic processing method and system. Background Art

[0002] At present, with the gradual popularization of personal and family smart terminals, cloud computing, edge computing, end computing and artificial intelligence based on network connections will be deeply integrated into the lives of individuals and families. The computing network meets users' needs for intelligent and immersive experience in scenarios with surging connection volume and data volume through the coordinated supply of multi-level computing power and network security. As a new business model, the computing network redefines the relationship between cloud, pipe, edge and end in the computing system. However, due to the wide distribution of computing nodes, complex environment and large number, many applications failed to fully consider security risks at the beginning of design, including the security of computing user data and the security of computing provider infrastructure. Traditional security protection methods can no longer fully meet the security protection needs of ubiquitous computing. Once controlled by an attack, it may bring greater security risks, affect user data and hinder the development of the industry.

[0003] The existing computing power network is accessed through different nodes and is vulnerable to attacks from malicious nodes during both the traffic billing stage and the service identification stage, resulting in the inability to guarantee the security of user data and computing power network facilities. Summary of the invention

[0004] The present invention provides a computing power Internet traffic processing method and system to solve the technical problem of low security of user data and computing power network facilities when the computing power network performs task calculations.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a computing power Internet traffic processing method, including:

[0006] Obtain historical traffic data of each computing power node, build a malicious node detection model based on the historical traffic data, and detect the node type of the network node based on the malicious node detection model, wherein the node type includes malicious nodes and normal nodes;

[0007] Obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, using a preset risk radius and the network node topology structure; if there is an associated node in the risk circle of the current normal node that is a malicious node, change the node type of the current normal node to a risk node;

[0008] Obtain the traffic of the node to be scheduled, and classify and evaluate the traffic of the node to be scheduled according to the node type of the traffic of the node to be scheduled to obtain an evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the traffic of the node to be scheduled is a normal node, perform a primary evaluation; if the node type of the traffic of the node to be scheduled is a risk node, perform a secondary evaluation; if the node type of the traffic of the node to be scheduled is a malicious node, perform a tertiary evaluation;

[0009] The computing power tasks of the node traffic to be scheduled are matched with computing power resources according to the risk level of the evaluation result, and business identification and traffic usage statistics are performed on the computing power tasks of the node traffic to be scheduled.

[0010] The present invention detects each network node, determines each malicious node and the associated risk node, and thus performs classified risk assessment according to different node types when allocating computing power tasks. The computing power tasks are monitored according to the risk assessment results to avoid malicious node access, attack traffic billing and business identification types, thereby improving the security of user data and computing power network facilities. At the same time, classified monitoring is performed according to the node risk level, thereby improving the efficiency of computing power task monitoring and reducing resource consumption.

[0011] Furthermore, the historical traffic data of each computing power node is obtained, and a malicious node detection model is constructed according to the historical traffic data, specifically:

[0012] Obtaining historical traffic data of each computing power node, and classifying the historical traffic data to determine the data type of the historical traffic data, wherein the data type includes abnormal data and normal data;

[0013] Calculate the abnormal sample size according to the normal data and the preset abnormal sample data ratio; if the abnormal data is less than the abnormal sample size, perform data enhancement processing on the abnormal data to generate training samples;

[0014] An initial malicious node detection model is constructed according to the neural network model, and the initial malicious node detection model is trained according to the training samples until the model accuracy reaches a preset condition, and the training is stopped to obtain the malicious node detection model.

[0015] Further, the node type of the network node is detected according to the malicious node detection model, specifically:

[0016] Obtaining a malicious node detection model and local traffic data of a computing power node, and optimizing and updating the malicious node detection model according to the local traffic data;

[0017] The node type of the network node is detected according to the updated malicious node detection model.

[0018] Furthermore, if the node type of the node traffic to be scheduled is a normal node, a first-level evaluation is performed, specifically:

[0019] If the node type of the node traffic to be scheduled is a normal node, obtaining a node associated table of the normal node;

[0020] The risk nodes are queried in the associated node table, and the node score is calculated according to the number of associated risk nodes. If the node score is greater than or equal to the preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0021] Furthermore, if the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed, specifically:

[0022] If the node type of the node traffic to be scheduled is a risk node, then obtaining traffic data information and an associated node table of the node traffic to be scheduled, wherein the traffic data information includes traffic access information and traffic source;

[0023] Conduct risk screening based on the traffic access information and traffic sources, and conduct screening and scoring based on the screening results;

[0024] Acquire the number of malicious nodes associated with the current risk node according to the associated node table, and generate an associated score according to the number of malicious nodes;

[0025] A weighted calculation is performed based on the association score and the screening score to generate a node score for the current risk node. If the node score is greater than or equal to a preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0026] Furthermore, if the node type of the node traffic to be scheduled is a malicious node, a three-level evaluation is performed, specifically:

[0027] If the node type of the node traffic to be scheduled is a malicious node, then the traffic data information and the number of historical malicious attacks of the node traffic to be scheduled are obtained; the traffic data information includes traffic access information and traffic source;

[0028] If the number of historical malicious attacks is greater than or equal to a preset attack number threshold, node isolation processing is performed on the traffic of the node to be scheduled;

[0029] If the number of historical malicious attacks is less than the preset attack number threshold, a risk check is performed based on the traffic access information and traffic source, and a score is assigned based on the check result;

[0030] A weighted calculation is performed based on the number of historical malicious attacks and the screening score to obtain a node score. If the node score is greater than or equal to a preset trust threshold, the evaluation result is to be monitored, otherwise node isolation processing is performed on the node traffic to be scheduled.

[0031] Furthermore, after performing node isolation processing on the to-be-scheduled node traffic, the method further includes:

[0032] Performing node type detection on the network nodes for node isolation according to the malicious node detection model and the traffic data information;

[0033] If the detection result is a normal node, the traffic access information and traffic source are checked and a check score is generated;

[0034] A weighted calculation is performed based on the detection results and the screening scores to generate a node score, and the node traffic to be scheduled of the current node is de-isolated based on the node score.

[0035] Furthermore, the computing power task of the node traffic to be scheduled is matched with computing power resources according to the risk level of the evaluation result, and business identification and traffic usage statistics are performed on the computing power task of the node traffic to be scheduled, specifically:

[0036] If the evaluation result is trustworthy, the computing power task of the node traffic to be scheduled is placed in the first-level matching team; if the evaluation result is to be monitored, the computing power task of the node traffic to be scheduled is placed in the second-level matching team;

[0037] Classify the computing tasks in the first-level matching team and the second-level matching team and perform traffic usage statistics.

[0038] Furthermore, the classified business identification and traffic usage statistics of the computing tasks in the first-level matching team and the second-level matching team are specifically as follows:

[0039] Divide the computing tasks in the first-level matching team into several groups according to the association relationship, and monitor the business identification and traffic statistics in each group as a whole;

[0040] A monitoring network is allocated to each computing task in the secondary matching team, and the business identification and traffic statistics of each computing task are monitored separately according to the monitoring network.

[0041] In a second aspect, the present invention provides a computing power Internet traffic processing system, including: a node detection module, a node update module, an evaluation module and a business monitoring module;

[0042] The node detection module is used to obtain historical traffic data of each computing power node, build a malicious node detection model based on the historical traffic data, and detect the node type of the network node according to the malicious node detection model, wherein the node type includes malicious nodes and normal nodes;

[0043] The node update module is used to obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, the preset risk radius and the network node topology structure, and if there is a malicious node in the risk circle of the current normal node, the node type of the current normal node is changed to a risk node;

[0044] The evaluation module is used to obtain the node traffic to be scheduled, and classify and evaluate the node traffic to be scheduled according to the node type of the node traffic to be scheduled to obtain the evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the node traffic to be scheduled is a normal node, a primary evaluation is performed; if the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed; if the node type of the node traffic to be scheduled is a malicious node, a tertiary evaluation is performed;

[0045] The business monitoring module is used to match computing resources to the computing tasks of the node traffic to be scheduled according to the risk level of the evaluation result, and to perform business identification and traffic usage statistics on the computing tasks of the node traffic to be scheduled. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flowchart of a computing power Internet traffic processing method provided by an embodiment of the present invention;

[0047] Figure 2 A structural schematic diagram of a computing power Internet traffic processing system provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0049] Embodiment 1

[0050] Please refer to Figure 1 , Figure 1 A flowchart of a computing power Internet traffic processing method provided by an embodiment of the present invention includes steps 101 to 104, which are specifically as follows:

[0051] Step 101: Obtain historical traffic data of each computing power node, build a malicious node detection model based on the historical traffic data, and detect the node type of the network node based on the malicious node detection model, where the node type includes malicious nodes and normal nodes;

[0052] In this embodiment, the acquisition of historical traffic data of each computing power node and the construction of a malicious node detection model based on the historical traffic data are specifically as follows:

[0053] Obtaining historical traffic data of each computing power node, and classifying the historical traffic data to determine the data type of the historical traffic data, wherein the data type includes abnormal data and normal data;

[0054] Calculate the abnormal sample size according to the normal data and the preset abnormal sample data ratio; if the abnormal data is less than the abnormal sample size, perform data enhancement processing on the abnormal data to generate training samples;

[0055] An initial malicious node detection model is constructed according to the neural network model, and the initial malicious node detection model is trained according to the training samples until the model accuracy reaches a preset condition, and the training is stopped to obtain the malicious node detection model.

[0056] In this embodiment, an initial malicious node detection model is constructed based on a neural network, the initial malicious node detection model is sent to each computing power node, and historical traffic data of each computing power node is obtained. The historical traffic data includes traffic access information, business type, traffic expectation value, traffic statistics and other data information.

[0057] In this embodiment, in each computing power node, the historical traffic data is classified and the total number of samples of each type and the number of abnormal samples are determined. If the total number of samples does not reach the preset first sample size value, sample expansion processing is performed on the samples in the current computing power node; if the number of abnormal samples does not reach the preset abnormal sample size, data enhancement processing is performed on the abnormal data.

[0058] In an optional embodiment, the sample expansion processing is performed on the samples in the current computing power node, specifically: obtaining the associated topological map of the current computing power node, and taking the current computing power node as the starting point, searching for the computing power node in the associated topological map that is shortest distance from the current computing power node, and adding the historical traffic data of the computing power node to the sample data of the current computing power node. If the total number of samples still does not meet the preset first sample value, continue to search for the next closest associated computing power node until the total number of samples of the current computing power node reaches the first sample value.

[0059] In this embodiment, after obtaining a sufficient number of samples, the proportion of abnormal data in the samples to the total number of samples is calculated. If the proportion of abnormal data is less than the abnormal sample data ratio, data enhancement processing is performed on the abnormal data.

[0060] In this embodiment, under ideal conditions, the abnormal data and normal data are set to 1:1. In most cases, there are fewer abnormal data, which easily leads to sample imbalance, causing the model to overfit. Therefore, data expansion is used to reduce sample imbalance.

[0061] In an optional embodiment, the data enhancement processing of the abnormal data is specifically: obtaining historical traffic data of malicious nodes in the Internet, and detecting the historical traffic data according to the initial malicious node detection model, generating corresponding pseudo-label data, and adding the abnormal data in the pseudo-label data to the abnormal data of the current node.

[0062] As a specific example of an embodiment of the present invention, the historical traffic data of the current computing power node is 9,000 copies, the first sample value is 10,000, and the abnormal sample data ratio is set to 0.3, then the computing power node closest to the current computing power node is found, and 1,000 sample data are randomly selected from the computing power node and added to the sample data of the current computing power node. If the number of abnormal samples is 500 after the total number of samples is expanded, and the sample data ratio is 0.05 at this time, which does not reach the preset sample data ratio, then the historical traffic data of the malicious node is detected according to the initial malicious node detection model, and the corresponding pseudo-label data is generated, and the data labeled as abnormal is added to the training sample until the abnormal data volume in the training sample accounts for 0.3.

[0063] In this embodiment, the initial malicious node detection model is trained according to the expanded training samples. During the training process, the importance scores of the model parameters are calculated according to the gradients, and the parameters are pruned according to the importance scores of each parameter. At the same time, the model performance is evaluated according to the model accuracy until the model performance reaches the preset conditions. The training is stopped, the trained malicious detection model is output, and the malicious detection model is transmitted back to the server so that the malicious detection models of each computing node of the server are aggregated to generate the final malicious node detection model.

[0064] In this embodiment, the node type of the network node is detected according to the malicious node detection model, specifically:

[0065] Obtaining a malicious node detection model and local traffic data of a computing power node, and optimizing and updating the malicious node detection model according to the local traffic data;

[0066] The node type of the network node is detected according to the updated malicious node detection model.

[0067] In this embodiment, the malicious node detection model aggregated by the server is obtained, and the malicious node detection model is optimized according to local traffic data, and the type of each network node is detected according to the optimized malicious node detection model.

[0068] Step 102: Obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, using a preset risk radius and the network node topology structure; if there is a node associated with a malicious node in the risk circle of the current normal node, change the node type of the current normal node to a risk node;

[0069] In this embodiment, a topological structure diagram of each network node is obtained, and with each normal node as the center, the risk circle of each normal node is determined according to a preset risk radius and the network node topological structure, and a node association table of associated nodes in each network node risk circle is constructed.

[0070] In this embodiment, a network node that is registered for the first time needs to be subject to risk assessment. If there is a malicious node associated with the network node in the risk circle of the network node that is registered for the first time, the node type of the current normal node is changed to a risk node.

[0071] In this embodiment, if there is an associated node in the risk circle of the current normal node that is a malicious node, the node type of the current normal node is changed to a risk node to prevent some nodes from not launching an attack when the node is accessed, but attacking the service type identification and traffic billing statistics after the service is completed.

[0072] Step 103: Obtain the traffic of the node to be scheduled, and classify and evaluate the traffic of the node to be scheduled according to the node type of the traffic of the node to be scheduled to obtain an evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the traffic of the node to be scheduled is a normal node, perform a primary evaluation; if the node type of the traffic of the node to be scheduled is a risk node, perform a secondary evaluation; if the node type of the traffic of the node to be scheduled is a malicious node, perform a tertiary evaluation;

[0073] In this embodiment, the node traffic to be scheduled is obtained, and the node type of the node traffic to be scheduled is determined, and a hierarchical evaluation is performed according to the node type.

[0074] In this embodiment, if the node type of the node traffic to be scheduled is a normal node, a first-level evaluation is performed, specifically:

[0075] If the node type of the node traffic to be scheduled is a normal node, obtaining a node associated table of the normal node;

[0076] The risk nodes are queried in the associated node table, and the node score is calculated according to the number of associated risk nodes. If the node score is greater than or equal to the preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0077] In this embodiment, the risk of normal nodes is relatively low, but it is not ruled out that some nodes are the first malicious attack. Therefore, it is necessary to query the associated nodes of the normal nodes and calculate the node scores based on the risk nodes in the associated nodes. The node score calculation is related to the proportion of risk nodes in the associated nodes. If the node score is greater than or equal to the preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0078] As a specific example of an embodiment of the present invention, the preset trust threshold is 0.95. If the number of associated nodes of the normal node is 1000 and the number of risk nodes is 30, the node score is 0.97, which is greater than the trust threshold. The evaluation result of the current node traffic to be scheduled is trustworthy; if the total number of associated nodes of the normal node is 900 and the number of risk nodes is 90, the node score is 0.9, which is less than the trust threshold. The evaluation result of the current node traffic to be scheduled is to be monitored.

[0079] In this embodiment, if the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed, specifically:

[0080] If the node type of the node traffic to be scheduled is a risk node, then obtaining traffic data information and an associated node table of the node traffic to be scheduled, wherein the traffic data information includes traffic access information and traffic source;

[0081] Conduct risk screening based on the traffic access information and traffic sources, and conduct screening and scoring based on the screening results;

[0082] Acquire the number of malicious nodes associated with the current risk node according to the associated node table, and generate an associated score according to the number of malicious nodes;

[0083] A weighted calculation is performed based on the association score and the screening score to generate a node score for the current risk node. If the node score is greater than or equal to a preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0084] In this embodiment, since the risk node is associated with the malicious node and the degree of association is strong, it is a medium-risk node.

[0085] In this embodiment, risk screening is performed based on the traffic access information and traffic source, and screening and scoring are performed based on the screening results; and the number of malicious nodes associated with the current risk node and the correlation degree of each malicious node are determined based on the associated node table, and the correlation score is calculated, and the node score is weighted based on the screening score and the association score.

[0086] In this embodiment, the investigation score is related to the inter-domain access data, the extra-domain access data and the credibility of the traffic source of the traffic access information.

[0087] In this embodiment, the node scoring is specifically:

[0088]

[0089] Among them, s is the node score, A is the screening score, B is the association score, n is the number of malicious nodes associated with the current node, m is the total number of nodes associated with the current node, w1 is the preset screening weight, w2 is the preset association weight, and w1+w2=1.

[0090] In this embodiment, if the node score is greater than or equal to a preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0091] In this embodiment, if the node type of the node traffic to be scheduled is a malicious node, a three-level evaluation is performed, specifically:

[0092] If the node type of the node traffic to be scheduled is a malicious node, then the traffic data information and the number of historical malicious attacks of the node traffic to be scheduled are obtained; the traffic data information includes traffic access information and traffic source;

[0093] If the number of historical malicious attacks is greater than or equal to a preset attack number threshold, node isolation processing is performed on the traffic of the node to be scheduled;

[0094] If the number of historical malicious attacks is less than the preset attack number threshold, a risk check is performed based on the traffic access information and traffic source, and a score is assigned based on the check result;

[0095] A weighted calculation is performed based on the number of historical malicious attacks and the screening score to obtain a node score. If the node score is greater than or equal to a preset trust threshold, the evaluation result is to be monitored, otherwise node isolation processing is performed on the node traffic to be scheduled.

[0096] In this embodiment, due to the high risk of malicious nodes, if the malicious node is marked as a malicious attack multiple times, the malicious node is isolated, and the computing resources are not matched for the scheduled traffic of the isolated malicious node, and its computing power tasks are stored in the malicious attack team.

[0097] In this embodiment, if the number of historical malicious attacks of the malicious node is less than a preset attack number threshold, a risk check is performed on the malicious node, and a check score is obtained, and a node score is generated based on the check score and the number of historical malicious attacks.

[0098] In this embodiment, the node score of the malicious node is specifically:

[0099]

[0100] Among them, s is the node score, A is the investigation score, x is the number of malicious attacks, y is the total number of accesses, w3 is the preset investigation weight, w4 is the preset attack weight, and w3+w4=1.

[0101] In this embodiment, due to the high risk of malicious nodes, even if the node score is greater than or equal to the preset trust threshold, its evaluation result is still set to be monitored to prevent malicious nodes from conducting malicious attacks. If the node score is less than the preset trust threshold, the node traffic to be scheduled is isolated and stored in the malicious attack queue.

[0102] In this embodiment, after performing node isolation processing on the to-be-scheduled node traffic, the method further includes:

[0103] Performing node type detection on the network nodes for node isolation according to the malicious node detection model and the traffic data information;

[0104] If the detection result is a normal node, the traffic access information and traffic source are checked and a check score is generated;

[0105] A weighted calculation is performed based on the detection results and the screening scores to generate a node score, and the node traffic to be scheduled of the current node is de-isolated based on the node score.

[0106] In this embodiment, for the isolated malicious nodes in the malicious attack queue, node type detection is performed on the network nodes for node isolation regularly based on the malicious node detection model and the traffic data information, and the nodes are scored based on the detection results and the investigation scores. If the node score is greater than or equal to the preset trust threshold, the node traffic to be scheduled of the current node is un-isolated.

[0107] In this embodiment, for malicious nodes that have been removed from isolation, risk reassessment still needs to be performed based on the three-level assessment.

[0108] Step 104: Match computing resources to the computing tasks of the node traffic to be scheduled according to the risk level of the assessment result, and perform business identification and traffic usage statistics on the computing tasks of the node traffic to be scheduled.

[0109] In this embodiment, the computing power task of the node traffic to be scheduled is matched with computing power resources according to the risk level of the evaluation result, and the computing power task of the node traffic to be scheduled is identified and traffic usage statistics are performed, specifically:

[0110] If the evaluation result is trustworthy, the computing power task of the node traffic to be scheduled is placed in the first-level matching team; if the evaluation result is to be monitored, the computing power task of the node traffic to be scheduled is placed in the second-level matching team;

[0111] Classify the computing tasks in the first-level matching team and the second-level matching team and perform traffic usage statistics.

[0112] In this embodiment, if the evaluation result is trustworthy, the computing power task of the node traffic to be scheduled is assigned to the first-level matching team; since the nodes in the first-level matching team have a higher degree of trust, the computing power resources are matched preferentially to the computing power tasks of the first-level matching team.

[0113] In this embodiment, if the evaluation result is to be monitored, the computing power task of the node traffic to be scheduled is assigned to the secondary matching team; at this time, the trust of the nodes in the secondary matching team is average, so it is necessary to perform task monitoring on the nodes in the secondary matching team.

[0114] In this embodiment, the classified business identification and traffic usage statistics of the computing tasks in the first-level matching team and the second-level matching team are specifically as follows:

[0115] Divide the computing tasks in the first-level matching team into several groups according to the association relationship, and monitor the business identification and traffic statistics in each group as a whole;

[0116] A monitoring network is allocated to each computing task in the secondary matching team, and the business identification and traffic statistics of each computing task are monitored separately according to the monitoring network.

[0117] In this embodiment, since the trust level of nodes in the first-level matching team is relatively high, the computing tasks in the first-level matching team are divided into several groups according to the relationship between nodes within a certain period of time, and the computing tasks in each group are monitored as a whole to reduce resource consumption.

[0118] In this embodiment, since the trust level of nodes in the secondary matching team is average, each computing task in the secondary matching team needs to be monitored separately. By allocating a monitoring network to each computing task, the business identification status and traffic billing statistics of the computing task are monitored; the monitoring network includes a business identification module and a traffic monitoring module.

[0119] In this embodiment, each network node is detected to determine each malicious node and the associated risk node, so that classified risk assessment is performed according to different node types when allocating computing power tasks. The computing power tasks are monitored according to the risk assessment results to avoid malicious node access, attack traffic billing and business identification types, thereby improving the security of user data and computing power network facilities. At the same time, classified monitoring is performed according to the node risk level to improve the computing power task monitoring efficiency and reduce resource consumption.

[0120] Please refer to Figure 2 , Figure 2 A schematic diagram of a computing power Internet traffic processing system provided by an embodiment of the present invention includes: a node detection module 201, a node update module 202, an evaluation module 203 and a business monitoring module 204;

[0121] The node detection module 201 is used to obtain historical traffic data of each computing power node, build a malicious node detection model according to the historical traffic data, and detect the node type of the network node according to the malicious node detection model, wherein the node type includes malicious nodes and normal nodes;

[0122] The node updating module 202 is used to obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, the preset risk radius and the network node topology structure, and if there is a malicious node in the risk circle of the current normal node, the node type of the current normal node is changed to a risk node;

[0123] The evaluation module 203 is used to obtain the node traffic to be scheduled, and classify and evaluate the node traffic to be scheduled according to the node type of the node traffic to be scheduled to obtain the evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the node traffic to be scheduled is a normal node, a primary evaluation is performed; if the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed; if the node type of the node traffic to be scheduled is a malicious node, a tertiary evaluation is performed;

[0124] The business monitoring module 204 is used to match computing resources to the computing tasks of the node traffic to be scheduled according to the risk level of the evaluation result, and to perform business identification and traffic usage statistics on the computing tasks of the node traffic to be scheduled.

[0125] In this embodiment, the node detection module is specifically used to:

[0126] Obtaining historical traffic data of each computing power node, and classifying the historical traffic data to determine the data type of the historical traffic data, wherein the data type includes abnormal data and normal data;

[0127] Calculate the abnormal sample size according to the normal data and the preset abnormal sample data ratio; if the abnormal data is less than the abnormal sample size, perform data enhancement processing on the abnormal data to generate training samples;

[0128] An initial malicious node detection model is constructed according to the neural network model, and the initial malicious node detection model is trained according to the training samples until the model accuracy reaches a preset condition, and the training is stopped to obtain the malicious node detection model.

[0129] In this embodiment, the node detection module is further used to:

[0130] Obtaining a malicious node detection model and local traffic data of a computing power node, and optimizing and updating the malicious node detection model according to the local traffic data;

[0131] The node type of the network node is detected according to the updated malicious node detection model.

[0132] In this embodiment, the evaluation module is specifically used for:

[0133] If the node type of the node traffic to be scheduled is a normal node, obtaining a node associated table of the normal node;

[0134] The risk nodes are queried in the associated node table, and the node score is calculated according to the number of associated risk nodes. If the node score is greater than or equal to the preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0135] In this embodiment, the evaluation module is specifically used for:

[0136] If the node type of the node traffic to be scheduled is a risk node, then obtaining traffic data information and an associated node table of the node traffic to be scheduled, wherein the traffic data information includes traffic access information and traffic source;

[0137] Conduct risk screening based on the traffic access information and traffic sources, and conduct screening and scoring based on the screening results;

[0138] Acquire the number of malicious nodes associated with the current risk node according to the associated node table, and generate an associated score according to the number of malicious nodes;

[0139] A weighted calculation is performed based on the association score and the screening score to generate a node score for the current risk node. If the node score is greater than or equal to a preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

[0140] In this embodiment, the evaluation module is specifically used for:

[0141] If the node type of the node traffic to be scheduled is a malicious node, then the traffic data information and the number of historical malicious attacks of the node traffic to be scheduled are obtained; the traffic data information includes traffic access information and traffic source;

[0142] If the number of historical malicious attacks is greater than or equal to a preset attack number threshold, node isolation processing is performed on the traffic of the node to be scheduled;

[0143] If the number of historical malicious attacks is less than the preset attack number threshold, a risk check is performed based on the traffic access information and traffic source, and a score is assigned based on the check result;

[0144] A weighted calculation is performed based on the number of historical malicious attacks and the screening score to obtain a node score. If the node score is greater than or equal to a preset trust threshold, the evaluation result is to be monitored, otherwise node isolation processing is performed on the node traffic to be scheduled.

[0145] In this embodiment, the evaluation module is further used for:

[0146] Performing node type detection on the network nodes for node isolation according to the malicious node detection model and the traffic data information;

[0147] If the detection result is a normal node, the traffic access information and traffic source are checked and a check score is generated;

[0148] A weighted calculation is performed based on the detection results and the screening scores to generate a node score, and the node traffic to be scheduled of the current node is de-isolated based on the node score.

[0149] In this embodiment, the service monitoring module is specifically used to:

[0150] If the evaluation result is trustworthy, the computing power task of the node traffic to be scheduled is placed in the first-level matching team; if the evaluation result is to be monitored, the computing power task of the node traffic to be scheduled is placed in the second-level matching team;

[0151] Classify the computing tasks in the first-level matching team and the second-level matching team and perform traffic usage statistics.

[0152] In this embodiment, the service monitoring module is specifically used to:

[0153] Divide the computing tasks in the first-level matching team into several groups according to the association relationship, and monitor the business identification and traffic statistics in each group as a whole;

[0154] A monitoring network is allocated to each computing task in the secondary matching team, and the business identification and traffic statistics of each computing task are monitored separately according to the monitoring network.

[0155] In this embodiment, the present invention detects each network node, determines each malicious node and the associated risk node, and thus performs classified risk assessment according to different node types when allocating computing power tasks, so as to monitor the computing power tasks according to the risk assessment results, avoid malicious node access, attack traffic billing and business identification types, improve the security of user data and computing power network facilities, and at the same time perform classified monitoring according to the node risk level, improve the computing power task monitoring efficiency, and reduce resource consumption.

[0156] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for processing computing power Internet traffic, It is characterized in that include: Obtain historical traffic data of each computing power node, build a malicious node detection model based on the historical traffic data, and detect the node type of the network node based on the malicious node detection model, wherein the node type includes malicious nodes and normal nodes; Obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, using a preset risk radius and the network node topology structure; if there is an associated node in the risk circle of the current normal node that is a malicious node, change the node type of the current normal node to a risk node; Obtain the traffic of the node to be scheduled, and classify and evaluate the traffic of the node to be scheduled according to the node type of the traffic of the node to be scheduled, and obtain the evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the traffic of the node to be scheduled is a normal node, perform a primary evaluation; if the node type of the traffic of the node to be scheduled is a risk node, perform a secondary evaluation; If the node type of the node traffic to be scheduled is a malicious node, a three-level evaluation is performed; The computing power tasks of the node traffic to be scheduled are matched with computing power resources according to the risk level of the evaluation result, and business identification and traffic usage statistics are performed on the computing power tasks of the node traffic to be scheduled.

2. A method for processing computing power Internet traffic as claimed in claim 1, It is characterized in that The acquisition of historical traffic data of each computing power node and the construction of a malicious node detection model based on the historical traffic data are specifically as follows: Obtaining historical traffic data of each computing power node, and classifying the historical traffic data to determine the data type of the historical traffic data, wherein the data type includes abnormal data and normal data; Calculate the abnormal sample size according to the normal data and the preset abnormal sample data ratio; if the abnormal data is less than the abnormal sample size, perform data enhancement processing on the abnormal data to generate training samples; An initial malicious node detection model is constructed according to the neural network model, and the initial malicious node detection model is trained according to the training samples until the model accuracy reaches a preset condition, and the training is stopped to obtain the malicious node detection model.

3. A method for processing computing power Internet traffic as claimed in claim 1, It is characterized in that The node type of the network node is detected according to the malicious node detection model, specifically: Obtaining a malicious node detection model and local traffic data of a computing power node, and optimizing and updating the malicious node detection model according to the local traffic data; The node type of the network node is detected according to the updated malicious node detection model.

4. A method for processing computing power Internet traffic as claimed in claim 3, It is characterized in that If the node type of the node traffic to be scheduled is a normal node, a first-level evaluation is performed, specifically: If the node type of the node traffic to be scheduled is a normal node, obtaining a node associated table of the normal node; The risk nodes are queried in the associated node table, and the node score is calculated according to the number of associated risk nodes. If the node score is greater than or equal to the preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

5. A method for processing computing power Internet traffic as claimed in claim 4, It is characterized in that If the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed, specifically: If the node type of the node traffic to be scheduled is a risk node, then obtaining traffic data information and an associated node table of the node traffic to be scheduled, wherein the traffic data information includes traffic access information and traffic source; Conduct risk screening based on the traffic access information and traffic sources, and conduct screening and scoring based on the screening results; Acquire the number of malicious nodes associated with the current risk node according to the associated node table, and generate an associated score according to the number of malicious nodes; A weighted calculation is performed based on the association score and the screening score to generate a node score for the current risk node. If the node score is greater than or equal to a preset trust threshold, the evaluation result of the current node traffic to be scheduled is trustworthy, otherwise it is to be monitored.

6. A method for processing computing power Internet traffic as claimed in claim 5, It is characterized in that If the node type of the node traffic to be scheduled is a malicious node, a three-level evaluation is performed, specifically: If the node type of the node traffic to be scheduled is a malicious node, then the traffic data information and the number of historical malicious attacks of the node traffic to be scheduled are obtained; the traffic data information includes traffic access information and traffic source; If the number of historical malicious attacks is greater than or equal to a preset attack number threshold, node isolation processing is performed on the traffic of the node to be scheduled; If the number of historical malicious attacks is less than the preset attack number threshold, a risk check is performed based on the traffic access information and traffic source, and a score is assigned based on the check result; A weighted calculation is performed based on the number of historical malicious attacks and the screening score to obtain a node score. If the node score is greater than or equal to a preset trust threshold, the evaluation result is to be monitored, otherwise node isolation processing is performed on the node traffic to be scheduled.

7. A method for processing computing power Internet traffic as claimed in claim 6, It is characterized in that After performing node isolation processing on the to-be-scheduled node traffic, the method further includes: Performing node type detection on the network nodes for node isolation according to the malicious node detection model and the traffic data information; If the detection result is a normal node, the traffic access information and traffic source are checked and a check score is generated; A weighted calculation is performed based on the detection results and the screening scores to generate a node score, and the node traffic to be scheduled of the current node is de-isolated based on the node score.

8. A method for processing computing power Internet traffic according to any one of claims 1 to 7, It is characterized in that The computing power task of the node traffic to be scheduled is matched with computing power resources according to the risk level of the evaluation result, and business identification and traffic usage statistics are performed on the computing power task of the node traffic to be scheduled, specifically: If the evaluation result is trustworthy, the computing power task of the node traffic to be scheduled is placed in the first-level matching team; if the evaluation result is to be monitored, the computing power task of the node traffic to be scheduled is placed in the second-level matching team; Classify the computing tasks in the first-level matching team and the second-level matching team and perform traffic usage statistics.

9. A method for processing computing power Internet traffic as claimed in claim 8, It is characterized in that The classified business identification and traffic usage statistics of the computing tasks in the first-level matching team and the second-level matching team are specifically as follows: Divide the computing tasks in the first-level matching team into several groups according to the association relationship, and monitor the business identification and traffic statistics in each group as a whole; A monitoring network is allocated to each computing task in the secondary matching team, and the business identification and traffic statistics of each computing task are monitored separately according to the monitoring network.

10. A computing power Internet traffic processing system, It is characterized in that include: Node detection module, node update module, evaluation module and business monitoring module; The node detection module is used to obtain historical traffic data of each computing power node, build a malicious node detection model based on the historical traffic data, and detect the node type of the network node according to the malicious node detection model, wherein the node type includes malicious nodes and normal nodes; The node update module is used to obtain the network node topology structure, and determine the risk circle of each normal node with each normal node as the center, the preset risk radius and the network node topology structure, and if there is a malicious node in the risk circle of the current normal node, the node type of the current normal node is changed to a risk node; The evaluation module is used to obtain the node traffic to be scheduled, and classify and evaluate the node traffic to be scheduled according to the node type of the node traffic to be scheduled to obtain the evaluation result; the classification evaluation includes primary evaluation, secondary evaluation and tertiary evaluation; if the node type of the node traffic to be scheduled is a normal node, a primary evaluation is performed; if the node type of the node traffic to be scheduled is a risk node, a secondary evaluation is performed; If the node type of the node traffic to be scheduled is a malicious node, a three-level evaluation is performed; The business monitoring module is used to match computing resources to the computing tasks of the node traffic to be scheduled according to the risk level of the evaluation result, and to perform business identification and traffic usage statistics on the computing tasks of the node traffic to be scheduled.

Citation Information

Patent Citations

  • Mobile network malicious node detection method and system

    CN110401955A

  • Hybrid computing power network resource scheduling optimization method and follow-up control device

    CN116471333A

  • Computing power network security arrangement method, device and equipment and storage medium

    CN116915432A

  • Method and system for evaluating safety capability level of computing power service and computer equipment

    CN118378268A

  • Learning maliciousness in cybersecurity graphs

    US20190182273A1