CN117527622A

CN117527622A - Data processing method and system of network switch

Info

Publication number: CN117527622A
Application number: CN202410014285.8A
Authority: CN
Inventors: 王小梅; 韩云鹏; 周宗平; 黄志勇; 林宋佳; 李泽涵
Original assignee: Shenzhen Yijia Technology Co ltd; Shenzhen Science Service Information Technology Co ltd
Current assignee: Shenzhen Yijia Technology Co ltd; Shenzhen Science Service Information Technology Co ltd
Priority date: 2024-01-05
Filing date: 2024-01-05
Publication date: 2024-02-06
Anticipated expiration: 2044-01-05
Also published as: CN117527622B

Abstract

The application relates to the technical field of data processing and discloses a data processing method and system of a network switch. The method comprises the following steps: performing multidimensional transmission data monitoring on the target network switch cluster; creating a target transmission relation network according to the data sent by the transmission end, the load data of the switches and the acquired data of the receiving ends; extracting transmission risk influence factors to obtain a plurality of transmission risk influence factors; performing numerical feature analysis and vector mapping to generate a risk influence feature vector; creating a transmission risk analysis model, and carrying out transmission risk prediction and analysis to obtain an initial transmission risk analysis result; and generating a target data exchange risk execution scheme of the target network switch cluster according to the initial transmission risk analysis result, thereby improving the data processing accuracy of the network switch.

Description

Data processing method and system of network switch

Technical Field

The present disclosure relates to the field of data processing technologies, and in particular, to a data processing method and system for a network switch.

Background

Network switches play a vital role in modern communication networks, responsible for the transmission and exchange of data. However, with the ever-increasing scale and increasing complexity of networks, conventional network switches face a number of challenges in handling large-scale data transmissions. In order to effectively manage data transmission, optimize network performance, and identify and cope with potential transmission risks in advance, it is important to conduct intensive research on a data processing method of a network switch.

In the past research, data processing methods for network switches have focused mainly on improving transmission efficiency and reducing network congestion. However, with the increasing size and complexity of networks, previous approaches have been struggled to cope with large-scale data transmissions and complex network topologies. One of the problems to be solved is how to identify and evaluate the risk factors faced by the network switch during transmission more accurately in order to formulate a more efficient data exchange implementation.

Disclosure of Invention

The application provides a data processing method and a data processing system for a network switch, and further improves the data processing accuracy of the network switch.

The first aspect of the present application provides a data processing method of a network switch, where the data processing method of the network switch includes:

performing multidimensional transmission data monitoring on the target network switch cluster to obtain transmission data of a transmission end, load data of a plurality of switches and acquisition data of a plurality of receiving ends;

creating an initial transmission relation network according to the data sent by the transmission end, the load data of the switches and the acquired data of the receiving ends, and performing network weight change optimization on the initial transmission relation network to obtain a target transmission relation network;

Based on the target transmission relation network, respectively extracting transmission risk influence factors of a plurality of target network switches in the target network switch cluster to obtain a plurality of transmission risk influence factors of each target network switch;

performing numerical feature analysis and vector mapping on the plurality of transmission risk influence factors to generate a risk influence feature vector of each target network switch;

creating a transmission risk analysis model corresponding to each target network switch, and carrying out transmission risk prediction and analysis on the risk influence feature vector through the transmission risk analysis model to obtain an initial transmission risk analysis result corresponding to each target network switch;

and generating a target data exchange risk execution scheme corresponding to the target network switch cluster according to the initial transmission risk analysis result.

A second aspect of the present application provides a data processing system of a network switch, the data processing system of the network switch comprising:

the monitoring module is used for carrying out multidimensional transmission data monitoring on the target network switch cluster to obtain transmission data of a transmission end, load data of a plurality of switches and acquisition data of a plurality of receiving ends;

The creation module is used for creating an initial transmission relation network according to the data sent by the transmission end, the load data of the switches and the acquired data of the receiving ends, and carrying out network weight changing optimization on the initial transmission relation network to obtain a target transmission relation network;

the extraction module is used for extracting transmission risk influence factors of a plurality of target network switches in the target network switch cluster based on the target transmission relation network to obtain a plurality of transmission risk influence factors of each target network switch;

the mapping module is used for carrying out numerical feature analysis and vector mapping on the plurality of transmission risk influence factors to generate a risk influence feature vector of each target network switch;

the prediction module is used for creating a transmission risk analysis model corresponding to each target network switch, and carrying out transmission risk prediction and analysis on the risk influence feature vector through the transmission risk analysis model to obtain an initial transmission risk analysis result corresponding to each target network switch;

and the generation module is used for generating a target data exchange risk execution scheme corresponding to the target network switch cluster according to the initial transmission risk analysis result.

According to the technical scheme, the system can obtain the overall information of the data sent by the transmission end, the load data of the plurality of switches and the acquired data of the plurality of receiving ends by carrying out multidimensional transmission data monitoring on the target network switch cluster. The network weight change optimization is carried out on the initial transmission relation network, so that the transmission efficiency is improved, and the potential risk is reduced. The dynamic time warping algorithm is adopted to align the sequence of the target multidimensional transmission data, and the SBD measurement algorithm is used to calculate the distance, so that the relevance and the similarity between the transmission sequences can be evaluated more accurately, and the accuracy of data analysis is improved. By using a decision tree algorithm to perform relationship analysis and weight optimization, a target transmission relationship network can be effectively established, and the adaptability and response speed of a network structure are improved, so that the whole transmission system is optimized. And the graph neural network is utilized to carry out node traversal and feature clustering on the target transmission relation network, so that transmission risk influence factors are extracted from a complex network structure, and a more comprehensive basis is provided for risk analysis. The single-factor cloud model is adopted to carry out digital feature calculation and vector mapping, so that a plurality of transmission risk influence factors are converted into comparable numerical features, complex information is simplified, and analysis efficiency is improved. The risk influence feature vectors are processed by using deep learning models such as a long-short-term memory network, a threshold circulation network and the like, so that transmission risk prediction and analysis can be more accurately carried out, and the sensitivity and accuracy of the system to risks are improved. The global execution scheme is initialized and optimized through the genetic algorithm, so that the optimal solution can be found in a plurality of execution schemes, the overall performance of the system is improved, the risk of data exchange is reduced, and the data processing accuracy of the network switch is further improved.

Drawings

FIG. 1 is a schematic diagram of one embodiment of a data processing method of a network switch according to an embodiment of the present application;

figure 2 is a schematic diagram of one embodiment of a data processing system of a network switch in an embodiment of the present application.

Detailed Description

The embodiment of the application provides a data processing method and a data processing system for a network switch, and further improves the data processing accuracy of the network switch.

The terms "first," "second," "third," "fourth" and the like in the description and in the claims of this application and in the above-described figures, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For ease of understanding, the following describes a specific flow of an embodiment of the present application, referring to fig. 1, and one embodiment of a data processing method of a network switch in the embodiment of the present application includes:

step 101, performing multidimensional transmission data monitoring on a target network switch cluster to obtain transmission data of a transmission end, load data of a plurality of switches and acquisition data of a plurality of receiving ends;

it will be appreciated that the execution subject of the present application may be a data processing system of a network switch, and may also be a terminal or a server, which is not limited herein. The embodiment of the present application will be described by taking a server as an execution body.

Specifically, multidimensional transmission data monitoring is carried out on the target network switch cluster through a preset cloud monitoring platform. The powerful processing capacity and wide data access capacity of cloud computing are utilized, so that various transmission behaviors and states of the switch cluster are captured and recorded in real time, including data traffic, port states, network delays and the like. The raw multidimensional transmission data collected is diverse and contains problems of noise or inconsistent formats, so that data cleaning and data standardization processing are required. Data cleansing is mainly to reject invalid, erroneous or repeated data, while data normalization is to convert data of different formats or metrics into a unified format for subsequent processing. A dynamic time warping (Dynamic Time Warping, DTW) algorithm is used to sequence align the target multidimensional transmission data. The DTW algorithm is a technique for measuring the similarity between two time series, which can flexibly stretch or compress the time series to find the best alignment between the two sequences. And performing Distance calculation on the first transmission sequence data by using an SBD (Shape-Based Distance) measurement algorithm. The SBD algorithm is a shape-based sequence distance measurement method that calculates the similarity between sequences by considering the overall shape of the sequences rather than just single point values. This approach helps to reveal the inherent association and variability between different transmitted data sequences. After the first transmission sequence data and the distance between the first transmission sequence data and the first transmission sequence data are combined, more comprehensive and fine target combined transmission data can be generated. The combined data can show the transmission characteristics of the network switch cluster from multiple dimensions, and provides rich information sources for further data classification. The target combined transmission data are finely classified so as to accurately distinguish the data sent by the transmission end, the load data of each switch and the data acquired by a plurality of receiving ends.

102, creating an initial transmission relation network according to data sent by a transmission end, load data of a plurality of switches and acquired data of a plurality of receiving ends, and performing network weight change optimization on the initial transmission relation network to obtain a target transmission relation network;

specifically, according to the data sent by the transmission end, a plurality of target transmission end nodes corresponding to the data are determined. These nodes represent the origin of data transmissions in the network, the identification of which is based on the source and nature of the transmitted data. By analyzing the load data of the plurality of switches, a corresponding plurality of target switch nodes are determined, which nodes represent the midpoints of data transmission in the network. Similarly, according to the data acquired by the plurality of receiving ends, a plurality of corresponding target receiving end nodes are determined, and the nodes represent the end points of data transmission. And carrying out relation analysis through a preset decision tree algorithm. The decision tree algorithm plays a key role in analyzing potential relationships among nodes by its efficient classification and predictive capabilities. The algorithm is used for analyzing the relation between the target transmission end node and the target switch node, so as to generate a plurality of first transmission relations. Similarly, the algorithm is also used to analyze the relationship between the target switch node and the target receiving end node to generate a plurality of second transmission relationships. Analysis and identification of these relationships provides a basis for the establishment of subsequent network structures. Based on these first and second transmission relationships, a corresponding relationship edge is created. The relationship edges here refer to paths or links connecting different nodes, which play a role in data transmission in the network. The first relationship edge connects the target transmitting end node and the target switch node, and the second relationship edge connects the target switch node and the target receiving end node. With such a connection, an initial transport relationship network is formed that presents the entire path and flow of data from transmission to reception. And carrying out network weight change optimization on the initial transmission relation network. And carrying out dynamic weight changing calculation on the initially set weights to generate a plurality of target weights. The purpose of the network weight change optimization is to adjust and optimize the importance and priority of each relationship edge in the network, thereby making the whole network more efficient and stable. Based on the calculated target weights, the initial transmission relation network is adjusted to obtain a final target transmission relation network. The optimized network can reflect the actual data transmission condition and network state, and provides a more accurate and effective basis for the management and optimization of the network.

Step 103, based on a target transmission relation network, respectively extracting transmission risk influence factors from a plurality of target network switches in a target network switch cluster to obtain a plurality of transmission risk influence factors of each target network switch;

it should be noted that, node traversal is performed on the target network switch cluster in the target transport relationship network, each target network switch is checked one by one, and their location and roles in the network are determined. Each target network switch corresponds to a target switch node, which is the basis for subsequent analysis. In order to understand the characteristics and functions of each target switch node in depth, each node is taken as a characteristic network center, and the characteristic centers are subjected to characteristic clustering by using a preset Graph Neural Network (GNN) model. A graph neural network is a neural network that specifically processes graph structure data and is capable of effectively capturing complex relationships and features between nodes. Through the feature clustering, target clustering results of each target switch node can be obtained, and the results reveal the relative positions and the importance of each node in the network. And based on the target clustering result, adopting a shortest path algorithm to perform subordinate relation calculation on the target transmission relation network. The shortest path algorithm is an algorithm commonly used in graph theory that can effectively identify and calculate the shortest links between nodes in the graph. By the calculation, a tree structure of target affiliations can be obtained, the dependency and the affiliation among all nodes in the network are clearly shown by the structure, and a structural view is provided for identifying potential transmission risks. And according to the target subordinate relation tree structure, identifying and sequencing the transmission risk influence factors for each target switch node. Factors that affect network stability and security, such as data congestion, hardware failures, security threats, and the like, are identified. These risk factors are ranked after they are identified to determine which factors have the greatest impact on the network. And screening the influence intensities of the identified initial risk influence factors, so as to obtain the key transmission risk influence factors of each target network switch. This screening process is based on a comprehensive assessment of risk factor impact and probability of occurrence. Through the screening, the focus can be ensured to be focused on risk factors which have the most negative influence on the network performance, and scientific and accurate basis is provided for subsequent risk management and countermeasure.

104, carrying out numerical feature analysis and vector mapping on a plurality of transmission risk influence factors to generate a risk influence feature vector of each target network switch;

specifically, the target digital characteristics of each transmission risk influence factor are calculated through a preset single factor cloud model. The single factor cloud model is an effective data processing method for converting qualitative concepts into quantitative descriptions, and is suitable for processing data with uncertainty and ambiguity. In this process, each transmission risk impact factor is quantified as a specific digital value that represents the potential impact of each factor on network stability and security. To make these target digital features easier to analyze and compare, they are subjected to a digital feature normalization process. The normalization process is to convert values of different ranges into a uniform standard or range (typically between 0 and 1), which helps to eliminate deviations between different features due to dimensions or ranges of values, thereby ensuring accuracy and consistency of subsequent analysis. Through the normalization process, each transmission risk influencing factor is converted into a standardized numerical value, and the normalized numerical characteristics can more accurately reflect the relative importance among different risk factors. Then, the order of the influence intensities of the transmission risk influence factors is obtained, and the normalized digital features are converted in sequence based on the order. According to the importance of the risk influencing factors, the normalized numerical characteristics are arranged according to a certain sequence to form a target characteristic sequence of each target network switch. This serialization transformation not only reflects the relative importance of the individual risk factors, but also provides a basis for constructing feature vectors. Vector mapping is performed on the target feature sequences to generate risk impact feature vectors for each target network switch. Vector mapping is the conversion of serialized feature data into vector form in high-dimensional space, which helps to convert complex risk-influencing factors into a form that can be processed by machine learning models and algorithms. Through this transformation, the risk influencing factors of each target network switch are effectively encoded into a multi-dimensional feature vector that contains detailed information about the risk influencing factors and can be used for further data analysis and model training.

Step 105, creating a transmission risk analysis model corresponding to each target network switch, and carrying out transmission risk prediction and analysis on the risk influence feature vector through the transmission risk analysis model to obtain an initial transmission risk analysis result corresponding to each target network switch;

specifically, a transmission risk analysis model corresponding to each target network switch is created. The core components of this model include an input layer, a long short memory network (LSTM), a threshold cycle network (GRU), a full connection layer, and an output layer. And carrying out vector normalization processing on the risk influence characteristic vector through an input layer in the model. The original feature vectors are converted into a format that can be processed more efficiently by the model to ensure accuracy and efficiency of subsequent calculations. And extracting time sequence characteristics of the standard influence characteristic vector through a long-short time memory network in the model. Long and short term memory networks are adept at processing and analyzing time series data, which can effectively capture time-varying features and patterns. In this link, the LSTM network analyzes time correlation and dynamic changes in the feature vectors, thereby generating timing-affected feature vectors. And further extracting hidden state characteristics from the time sequence influence characteristic vectors through a threshold circulation network in the model. The GRU is a highly efficient recurrent neural network structure that has highly efficient and accurate characteristics in processing complex sequence data. The GRU generates a hidden state feature vector capable of reflecting transmission risk characteristics more comprehensively by extracting and analyzing hidden state information in the feature vector. And carrying out transmission risk probability prediction on the hidden state feature vectors through a full connection layer in the model. The full connection layer plays a role of integrating the integrated information in the neural network, and can quantitatively evaluate risks based on the extracted features, and the model calculates the target transmission risk probability of each target network switch based on the hidden state feature vector. And comparing the target transmission risk probability with a preset target probability threshold through an output layer in the model, so as to generate an initial transmission risk analysis result corresponding to each target network switch. The output layer not only provides the final output of risk prediction, but also helps determine which switches are at higher risk by comparison to preset thresholds.

And 106, generating a target data exchange risk execution scheme corresponding to the target network switch cluster according to the initial transmission risk analysis result.

Specifically, according to the initial transmission risk analysis result, performing execution scheme analysis on each target network switch, and formulating an initial data exchange risk execution scheme for each switch. These initial schemes reflect understanding and coping strategies of each switch's current risk status, tailored based on specific risk assessment and requirements of individual network switches. And carrying out global scheme initialization on the whole target network switch cluster according to the initial data exchange risk execution schemes of the individuals through a preset genetic algorithm, thereby generating an initial global execution scheme group. By utilizing the optimizing capability of the genetic algorithm, an optimal execution scheme is searched by simulating mechanisms such as selection, crossover, variation and the like in the biological evolution process. The initial global execution scheme population contains a plurality of first candidate risk execution schemes, each representing a network risk management policy. These first candidate risk executions are fitness calculated to evaluate their respective applicability and efficiency in the actual network environment. The fitness directly determines the merits of the scheme, and is a key index for scheme selection and optimization in genetic algorithm. Based on the evaluation results of the first fitness, the first candidate risk execution schemes are subjected to group division to form a plurality of target global execution scheme groups. This partitioning helps to identify the most potential schemes and provides the basis for further optimization. The target global execution scheme groups are subjected to scheme generation, and a plurality of second candidate risk execution schemes are generated. The genetic algorithm generates new schemes through operations such as crossover, mutation and the like, and the second candidate risk execution schemes represent improvements and innovations of the original schemes. Each of the second candidate risk executions also requires fitness calculations to evaluate their actual performance and effectiveness in the network environment. And (3) carrying out optimization analysis according to the second adaptability of the second candidate risk execution schemes, and screening the optimal schemes from all the candidate schemes to ensure the maximum efficiency and optimal performance of the selected schemes in the aspect of network risk management. Through such optimization analysis, a target data exchange risk execution scheme for the target network switch cluster can be generated. The implementation scheme comprehensively considers the actual running environment of the network, the risk condition of each switch and the safety and efficiency requirements of the whole network system, thereby providing a scientific, systematic and effective solution for the risk management of the network switch.

In the embodiment of the application, the system can obtain the overall information of the data sent by the transmission end, the load data of a plurality of switches and the acquired data of a plurality of receiving ends by carrying out multidimensional transmission data monitoring on the target network switch cluster. The network weight change optimization is carried out on the initial transmission relation network, so that the transmission efficiency is improved, and the potential risk is reduced. The dynamic time warping algorithm is adopted to align the sequence of the target multidimensional transmission data, and the SBD measurement algorithm is used to calculate the distance, so that the relevance and the similarity between the transmission sequences can be evaluated more accurately, and the accuracy of data analysis is improved. By using a decision tree algorithm to perform relationship analysis and weight optimization, a target transmission relationship network can be effectively established, and the adaptability and response speed of a network structure are improved, so that the whole transmission system is optimized. And the graph neural network is utilized to carry out node traversal and feature clustering on the target transmission relation network, so that transmission risk influence factors are extracted from a complex network structure, and a more comprehensive basis is provided for risk analysis. The single-factor cloud model is adopted to carry out digital feature calculation and vector mapping, so that a plurality of transmission risk influence factors are converted into comparable numerical features, complex information is simplified, and analysis efficiency is improved. The risk influence feature vectors are processed by using deep learning models such as a long-short-term memory network, a threshold circulation network and the like, so that transmission risk prediction and analysis can be more accurately carried out, and the sensitivity and accuracy of the system to risks are improved. The global execution scheme is initialized and optimized through the genetic algorithm, so that the optimal solution can be found in a plurality of execution schemes, the overall performance of the system is improved, the risk of data exchange is reduced, and the data processing accuracy of the network switch is further improved.

In a specific embodiment, the process of executing step 101 may specifically include the following steps:

(1) The method comprises the steps of monitoring multi-dimensional transmission data of a target network switch cluster through a preset cloud monitoring platform to obtain original multi-dimensional transmission data;

(2) Performing data cleaning and data standardization processing on the original multidimensional transmission data to obtain target multidimensional transmission data;

(3) Performing sequence alignment on target multidimensional transmission data by adopting a dynamic time warping algorithm to obtain a plurality of first transmission sequence data;

(4) Performing distance calculation on a plurality of first transmission sequence data by adopting an SBD (sequence based digital) measurement algorithm to obtain the distance between sequence pairs;

(5) Data combination is carried out on a plurality of first transmission sequence data and the distances between sequence pairs, and target combination transmission data is generated;

(6) And carrying out data classification on the target combined transmission data to obtain transmission data of a transmission end, load data of a plurality of switches and acquired data of a plurality of receiving ends.

Specifically, the target network switch cluster is subjected to multidimensional transmission data monitoring through a preset cloud monitoring platform. The cloud monitoring platform can capture and record the data flow condition in the switch cluster in real time, wherein the data flow condition comprises multidimensional information such as the sending amount, the receiving amount, the delay, the packet loss rate and the like of data. The original multidimensional transmission data is subjected to data cleaning and standardization. Data cleansing primarily refers to the removal of invalid, erroneous or duplicate data, which helps ensure data quality and accuracy of analysis. For example, abnormal data points generated due to sensor failure or repeatedly recorded data need to be removed. Data standardization is to convert data in different formats or metrics into a unified format, for example, unify data traffic reported by different switches into Mbps units, or unify timestamps into UTC standard time. This step helps to eliminate inconsistencies between the data, making the subsequent analysis more accurate and efficient. Thereafter, a Dynamic Time Warping (DTW) algorithm is used to sequence align the target multidimensional transmission data. The DTW algorithm is adapted to process time series data by warping the time axis such that the two sequences are aligned in the time dimension, thereby enabling comparison of the data collected at different points in time. For example, if one switch delays reporting data for some reason, the DTW algorithm may align the data with the data of the other switch at the same point in time, thereby ensuring fairness of comparison and analysis. After the DTW algorithm is applied, the obtained first transmission sequence data is time-adjusted and aligned, and is more suitable for further analysis. These first transmission sequence data are Distance calculated using a Shape-Based Distance (SBD) metric algorithm. The SBD metric algorithm takes into account the overall shape of the sequence when calculating the distance between sequences, not just the values of the points in the sequence. This allows the algorithm to capture a deeper level of similarity or variability between sequences. For example, even if the data traffic curves of two switches differ in value, the SBD algorithm can identify the similarity between them by comparing the overall shape of the curves. By this means, the distance between pairs of sequences can be obtained, which information is helpful for understanding the data transfer relationship between the different switches. And carrying out data combination on the plurality of first transmission sequence data and the distance between the first transmission sequence data and the first transmission sequence data to generate target combined transmission data. The time-aligned transmission data and the distance information between the sequences are combined to form a comprehensive data set. This combined data set contains not only the transmission data of each switch, but also the interrelationship between these data. And carrying out data classification on the target combined transmission data to obtain transmission data of a transmission end, load data of a plurality of switches and data acquired by a plurality of receiving ends. The information in the combined dataset is analyzed and interpreted to distinguish between different types of data. For example, the transmission end transmitting data includes information such as the size, the transmitting frequency, the transmitting time, etc. of the data packet; the load data of the exchanger comprises data traffic, processing delay, cache service condition and other information; the data acquired by the receiving end comprises the information such as the received data quantity, the receiving quality, the receiving delay and the like.

In a specific embodiment, the process of executing step 102 may specifically include the following steps:

(1) Determining a plurality of corresponding target transmission end nodes according to the transmission end sending data, determining a plurality of corresponding target switch nodes according to the switch load data, and determining a plurality of corresponding target receiving end nodes according to the receiving end obtaining data;

(2) Performing relationship analysis on a plurality of target transmission end nodes and a plurality of target switch nodes through a preset decision tree algorithm to obtain a plurality of first transmission relationships, and performing relationship analysis on a plurality of target switch nodes and a plurality of target receiving end nodes through a decision tree algorithm to obtain a plurality of second transmission relationships;

(3) Creating a plurality of first relationship edges between a plurality of target transmission end nodes and a plurality of target switch nodes according to a plurality of first transmission relationships, and creating a plurality of second relationship edges between a plurality of target switch nodes and a plurality of target receiving end nodes according to a plurality of second transmission relationships;

(4) Network connection is carried out on the first relationship edges and the second relationship edges based on preset initial weights, and a corresponding initial transmission relationship network is generated;

(5) And carrying out dynamic variable weight calculation on the initial weights, generating a plurality of corresponding target weights, and carrying out network variable weight optimization on the initial transmission relation network based on the plurality of target weights to obtain a target transmission relation network.

Specifically, based on the data sent by the transmission end, a plurality of corresponding target transmission end nodes are determined. These nodes represent the origin of data transmissions in the network. For example, it is possible to determine which devices are the primary sources of data transmission by analyzing parameters such as data transmission frequency, packet size, and transmission delay of different devices. Likewise, a corresponding plurality of target switch nodes is determined from the load data of the plurality of switches. Load data herein includes, but is not limited to, per-switch data throughput, traffic allocation, buffer usage, etc., which helps identify key switch nodes in the network. And determining a plurality of corresponding target receiving end nodes according to the data acquired by the plurality of receiving ends. The data of the receiving end, such as data receiving amount, receiving frequency, data integrity and the like, can be used for identifying key nodes for data receiving. And carrying out relation analysis on the nodes through a preset decision tree algorithm. The decision tree algorithm is a data mining tool that reveals potential relationships between data by creating a tree structure based on attribute selection criteria. And analyzing the relation between the target transmission end node and the target switch node by using a decision tree algorithm to obtain a plurality of first transmission relations. For example, algorithms may find that a transmitting end node often sends large amounts of data to a particular switch node, which pattern forms a first transmission relationship. And similarly, analyzing the relation between the target switch node and the target receiving end node by using a decision tree algorithm to obtain a plurality of second transmission relations. For example, a certain switch node may preferentially route data to a particular receiving end node, forming a second transmission relationship. Based on these first and second transmission relationships, a first relationship edge between a plurality of target transmission end nodes and target switch nodes, and a second relationship edge between target switch nodes and target receiving end nodes are created. These relationship edges represent paths for data transmission in the network. For example, a first relationship edge connects a transmitting end node and a switch node, representing the transmission route of data. Likewise, the second relationship edge connects the switch node and the receiving end node, completing the last segment of the data transmission. And carrying out network connection on the relationship edges based on preset initial weights to generate an initial transmission relationship network. The network is a preliminary network model constructed based on the results of the current data analysis, reflecting the basic structure and pattern of data transmission in the network. And carrying out dynamic variable weight calculation on the initial weights to generate a plurality of target weights. Dynamic weighting computation is a complex process that adjusts the weight of each relationship edge based on a variety of factors, such as network traffic variation, node performance, historical data, etc. For example, if the load of a certain switch node suddenly increases, the weight of the relationship edge connected to that node needs to be increased to reflect its increasing importance in the network. And based on the target weights, carrying out network weight change optimization on the initial transmission relation network to obtain the target transmission relation network. The optimized network reflects the actual importance of each node and path in the network more accurately, and provides an important basis for management and optimization of the network. For example, the optimized network may highlight certain critical data transmission paths or identify potential bottleneck nodes, thereby guiding the network administrator for efficient network adjustments and optimization.

In a specific embodiment, the process of executing step 103 may specifically include the following steps:

(1) Traversing nodes of a plurality of target network switches in a target network switch cluster in a target transmission relation network to obtain target switch nodes corresponding to each target network switch;

(2) Taking each target switch node as a characteristic network center, and carrying out characteristic clustering on the characteristic network center through a preset graph neural network model to obtain a target clustering result of each target switch node;

(3) According to the target clustering result, performing dependency calculation on the target transmission relation network by adopting a shortest path algorithm to obtain a target dependency tree structure;

(4) According to the target subordinate relation tree structure, identifying and sequencing transmission risk influence factors of each target switch node to obtain a plurality of initial risk influence factors;

(5) And screening the influence intensities of the initial risk influence factors to obtain a plurality of transmission risk influence factors of each target network switch.

Specifically, node traversal is performed on a target network switch cluster in a target transport relationship network. All target network switches are identified by systematically inspecting each switch node in the network. For example, a large network of tens of switches, the node traversal process examines the running state, connection, traffic data, etc. of the switches one by one to determine which switches belong to the critical node, i.e., the target network switch. These target switches are because they carry high traffic, connect critical devices, or are in a critical location on the network. And taking each target switch node as a characteristic network center, and carrying out characteristic clustering through a preset Graph Neural Network (GNN) model. The graph neural network is a tool for processing graph structure data, and can effectively capture complex relations and attributes among nodes. During feature clustering, the GNN analyzes the connection patterns, data traffic, and other relevant features of each target switch node with other nodes, thereby aggregating nodes with similar features. For example, if certain switch nodes exhibit similar patterns of behavior when handling high traffic data, the GNNs will cluster the nodes together to form a feature group. And according to the clustering results, performing membership calculation on the target transmission relation network by adopting a shortest path algorithm, thereby obtaining a tree structure of target membership. The shortest path algorithm here functions to reveal dependencies and communication paths between nodes. For example, if data is typically flowing from a particular source switch to a particular destination switch, a dependent path exists between the two nodes. Through the shortest path algorithm, a tree structure reflecting the whole network data flow and node dependency relationship can be constructed. And identifying and sequencing transmission risk influence factors for each target switch node according to the target subordinate relation tree structure. At this stage, the location of each node, the number and type of connections, and past performance and fault records, etc., are analyzed to identify factors that pose a risk to transmissions. For example, a switch node that connects multiple critical devices is identified as a high risk node if its performance is unstable or fails frequently in the past. At the same time, these risk factors are ranked according to their degree of impact on the network, thereby determining the most critical risk points. And screening the influence intensity of the initial risk influence factors to obtain key transmission risk influence factors of each target network switch. The most important risk points are screened by evaluating the potential impact and occurrence probability of each risk factor. For example, even if the performance of a certain switch is not stable, if it is connected to only non-critical devices or rarely processes data, then it poses less serious risks than those at the network core that process large amounts of sensitive data. By such screening, it is ensured that important concerns are focused on risk factors that have the greatest impact on network stability and security.

In a specific embodiment, the process of executing step 104 may specifically include the following steps:

(1) Respectively calculating the target digital characteristics of each transmission risk influence factor through a preset single-factor cloud model;

(2) Carrying out digital feature normalization on the target digital features of each transmission risk influence factor to generate normalized digital features of each transmission risk influence factor;

(3) Acquiring the influence intensity sequence of a plurality of transmission risk influence factors, and carrying out serialization conversion on the normalized digital characteristics based on the influence intensity sequence to generate a target characteristic sequence of each target network switch;

(4) And carrying out vector mapping on the target feature sequences to generate risk influence feature vectors of each target network switch.

Specifically, the target digital characteristics of each transmission risk influence factor are calculated through a preset single factor cloud model. The one-factor cloud model is used to convert qualitative concepts (e.g., risk ratings) to quantitative descriptions. For example, one risk influencing factor is the failure rate of a particular switch. Using a one-factor cloud model, this qualitative risk factor (e.g., "high failure rate") can be translated into a specific digital value that reflects the potential impact of the failure rate on the overall network performance. And carrying out digital characteristic normalization processing on the target digital characteristic of each transmission risk influence factor to generate normalized digital characteristics of each transmission risk influence factor. Normalization converts values of different ranges into a uniform standard or range (typically between 0 and 1), eliminating deviations between different features due to dimensional or numerical range differences, thereby ensuring the accuracy and consistency of subsequent analysis. For example, if the failure rate of one switch is 0.02 and the packet loss rate of the other is 0.1, these two numbers are difficult to directly compare on the original scale, but after normalization they can be compared under the same standard. And acquiring the influence intensity sequence of a plurality of transmission risk influence factors, and carrying out serialization conversion on the normalized digital characteristics based on the sequence to generate a target characteristic sequence of each target network switch. The normalized risk impact factors are ranked according to their importance or impact. For example, if the failure rate affects network performance more than the data packet loss rate, then the failure rate should be characterized in the sequence before the data packet loss rate. This serialized feature list provides a risk overview for each switch that is ordered by impact. Vector mapping is carried out on the target feature sequences, and risk influence feature vectors of each target network switch are generated. Vector mapping is the conversion of these serialized features into vector form in high dimensional space for further analysis and processing using machine learning algorithms. For example, a feature sequence of a switch is mapped into a multidimensional vector, where each dimension represents a particular risk factor whose value reflects the importance and impact of that factor. This vectorized representation allows advanced data analysis techniques, such as cluster analysis or neural networks, to be used to identify risk patterns or predict potential problems.

In a specific embodiment, the process of executing step 105 may specifically include the following steps:

(1) Creating a transmission risk analysis model corresponding to each target network switch, wherein the transmission risk analysis model comprises: an input layer, a long and short time memory network, a threshold circulation network, a full connection layer and an output layer;

(2) Carrying out vector standardization processing on the risk influence characteristic vector through an input layer in the transmission risk analysis model to obtain a standard influence characteristic vector;

(3) Performing time sequence feature extraction on the standard influence feature vector through a long-short-time memory network in the transmission risk analysis model to obtain a time sequence influence feature vector;

(4) Carrying out hidden state feature extraction on the time sequence influence feature vector through a threshold circulation network in the transmission risk analysis model to obtain a hidden state feature vector;

(5) Carrying out transmission risk probability prediction on the hidden state feature vector through a full connection layer in the transmission risk analysis model to obtain target transmission risk probability;

(6) And comparing the target transmission risk probability with a preset target probability threshold value through an output layer in the transmission risk analysis model, and generating an initial transmission risk analysis result corresponding to each target network switch.

Specifically, a transmission risk analysis model corresponding to each target network switch is created, wherein the transmission risk analysis model comprises: an input layer, a long and short time memory network, a threshold circulation network, a full connection layer and an output layer. And carrying out vector normalization processing on the risk influence characteristic vector through an input layer in the model. Normalization is the process of converting feature vectors into a format that can be efficiently handled by the model. For example, a target network switch has a series of original risk-influencing feature vectors, such as failure rate, traffic congestion, etc., which, after normalization, are converted into a standard-influencing feature vector in a unified format, so that the model can more easily recognize and process the data. The long and short term memory network (LSTM) in the transmission risk analysis model performs time series feature extraction on these standard impact feature vectors. LSTM is good at processing time series data and is able to capture changes in data over time. For example, LSTM may analyze the trend of packet loss rate over time, or periodic fluctuations in network traffic, which timing features help predict future risk potential. By processing the LSTM layer, time sequence influence characteristic vectors are obtained, and the time sequence influence characteristic vectors contain key information of time change of the original data. And then, extracting hidden state features of the sequence influence feature vectors through a threshold cyclic network (GRU) in the transmission risk analysis model. The GRU is an efficient recurrent neural network architecture capable of handling more complex data sequences. The model may further analyze hidden layer information in the timing-affected feature vector. For example, a GRU may identify a hidden pattern that is prone to failure under certain network loading conditions, or a security breach that is prone to occurrence in data transmission. These hidden state feature vectors provide a deeper level of understanding of network behavior. And carrying out transmission risk probability prediction on the hidden state feature vector through a full connection layer in the transmission risk analysis model. The role of the fully connected layer in the neural network is to integrate all features extracted at the previous layer level and make the final predictions based on these features. For example, the layer may calculate the transmission risk probability for each target network switch in combination with consideration of network traffic patterns, failure rate history, and other factors. These probabilities reflect the degree of risk each switch encounters over a period of time in the future. And comparing the target transmission risk probability with a preset target probability threshold value through an output layer in the transmission risk analysis model, and generating an initial transmission risk analysis result corresponding to each target network switch. The output layer compares the risk probabilities to a pre-set threshold to determine which switches have risk levels that exceed an acceptable range. For example, if the risk probability of a switch exceeds a set threshold, this indicates that it requires immediate attention and intervention to reduce the potential risk.

In a specific embodiment, the process of executing step 106 may specifically include the following steps:

(1) According to the initial transmission risk analysis result, performing execution scheme analysis on each target network switch to obtain an initial data exchange risk execution scheme of each target network switch;

(2) Carrying out global scheme initialization on the target network switch cluster according to the initial data exchange risk execution scheme of each target network switch through a preset genetic algorithm to generate an initial global execution scheme group, wherein the initial global execution scheme group comprises a plurality of first candidate risk execution schemes;

(3) Respectively calculating first fitness of each first candidate risk execution scheme, and dividing scheme groups of the first candidate risk execution schemes according to the first fitness to obtain a plurality of target global execution scheme groups;

(4) Generating schemes of the target global execution scheme groups to obtain a plurality of second candidate risk execution schemes, and respectively calculating the second fitness of each second candidate risk execution scheme;

(5) And carrying out scheme optimization analysis on the plurality of second candidate risk execution schemes according to the second fitness, and generating a target data exchange risk execution scheme corresponding to the target network switch cluster.

Specifically, according to the initial transmission risk analysis result, performing an execution scheme analysis on each target network switch to obtain an initial data exchange risk execution scheme of each target network switch. And analyzing the risk level and the characteristics of each switch, and formulating a corresponding risk management strategy according to the analysis results. For example, if a certain switch is identified as a high risk node due to its location and traffic pattern, its initial implementation includes actions such as adding redundant connections, increasing the data encryption level, or adjusting traffic routing policies. And carrying out global scheme initialization on the whole target network switch cluster according to the initial data exchange risk execution scheme of each target network switch through a preset genetic algorithm, and generating an initial global execution scheme group. Genetic algorithms are optimization algorithms that simulate natural selection and genetic mechanisms and that can efficiently search a wide range of solution spaces to find optimal solutions. The initial execution plan for each switch is considered a candidate solution and the entire initial global execution plan population contains a plurality of such first candidate risk execution plans. And respectively calculating the first fitness of each first candidate risk execution scheme, and carrying out scheme group division on a plurality of first candidate risk execution schemes according to the first fitness so as to obtain a plurality of target global execution scheme groups. Fitness is a standard for measuring the quality of a solution, and is generally based on factors such as efficiency, cost, feasibility and the like of solving the problem. For example, a solution would be highly adaptable if it could significantly reduce the risk of a high risk switch while not unduly increasing the overall operating cost of the network. And then, generating schemes of the target global execution scheme groups to obtain a plurality of second candidate risk execution schemes, and respectively calculating the second fitness of each second candidate risk execution scheme. New candidate solutions are generated from existing solutions by crossover and mutation operations of genetic algorithms. These new schemes combine different elements of the previous schemes or introduce new strategies. For example, two original schemes can be respectively aimed at different risk points, and a new scheme for comprehensively considering the two risk points can be generated through intersection and mutation. And carrying out scheme optimization analysis according to the second adaptability of the second candidate risk execution schemes so as to generate a target data exchange risk execution scheme corresponding to the target network switch cluster. The most effective risk-reducing solution is selected from all candidate solutions. For example, one implementation of optimization includes hardware upgrades to specific switches, adjustments to network topology, or improvements to data encryption policies to minimize overall network risk.

The foregoing describes a data processing method of a network switch in an embodiment of the present application, and the following describes a data processing system of a network switch in an embodiment of the present application, referring to fig. 2, an embodiment of the data processing system of a network switch in an embodiment of the present application includes:

the monitoring module 201 is configured to monitor the multi-dimensional transmission data of the target network switch cluster, so as to obtain transmission data of a transmission end, load data of a plurality of switches, and acquisition data of a plurality of receiving ends;

the creation module 202 is configured to create an initial transmission relationship network according to the data sent by the transmission end, the load data of the switches, and the acquired data of the receiving ends, and perform network weight change optimization on the initial transmission relationship network to obtain a target transmission relationship network;

the extracting module 203 is configured to extract transmission risk influence factors of a plurality of target network switches in the target network switch cluster based on the target transmission relation network, so as to obtain a plurality of transmission risk influence factors of each target network switch;

the mapping module 204 is configured to perform numerical feature analysis and vector mapping on the multiple transmission risk impact factors, and generate a risk impact feature vector of each target network switch;

The prediction module 205 is configured to create a transmission risk analysis model corresponding to each target network switch, and perform transmission risk prediction and analysis on the risk influence feature vector through the transmission risk analysis model to obtain an initial transmission risk analysis result corresponding to each target network switch;

and the generating module 206 is configured to generate a target data exchange risk execution scheme corresponding to the target network switch cluster according to the initial transmission risk analysis result.

Through the cooperation of the components, the system can obtain the overall information of the data sent by the transmission end, the load data of a plurality of switches and the acquired data of a plurality of receiving ends by carrying out multidimensional transmission data monitoring on the target network switch cluster. The network weight change optimization is carried out on the initial transmission relation network, so that the transmission efficiency is improved, and the potential risk is reduced. The dynamic time warping algorithm is adopted to align the sequence of the target multidimensional transmission data, and the SBD measurement algorithm is used to calculate the distance, so that the relevance and the similarity between the transmission sequences can be evaluated more accurately, and the accuracy of data analysis is improved. By using a decision tree algorithm to perform relationship analysis and weight optimization, a target transmission relationship network can be effectively established, and the adaptability and response speed of a network structure are improved, so that the whole transmission system is optimized. And the graph neural network is utilized to carry out node traversal and feature clustering on the target transmission relation network, so that transmission risk influence factors are extracted from a complex network structure, and a more comprehensive basis is provided for risk analysis. The single-factor cloud model is adopted to carry out digital feature calculation and vector mapping, so that a plurality of transmission risk influence factors are converted into comparable numerical features, complex information is simplified, and analysis efficiency is improved. The risk influence feature vectors are processed by using deep learning models such as a long-short-term memory network, a threshold circulation network and the like, so that transmission risk prediction and analysis can be more accurately carried out, and the sensitivity and accuracy of the system to risks are improved. The global execution scheme is initialized and optimized through the genetic algorithm, so that the optimal solution can be found in a plurality of execution schemes, the overall performance of the system is improved, the risk of data exchange is reduced, and the data processing accuracy of the network switch is further improved.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are merely for illustrating the technical solution of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. A data processing method of a network switch, the data processing method of the network switch comprising:

2. The method for processing data of a network switch according to claim 1, wherein the performing multidimensional transmission data monitoring on the target network switch cluster to obtain transmission data of a transmission end, load data of a plurality of switches, and acquisition data of a plurality of receiving ends includes:

the method comprises the steps of monitoring multi-dimensional transmission data of a target network switch cluster through a preset cloud monitoring platform to obtain original multi-dimensional transmission data;

performing data cleaning and data standardization processing on the original multidimensional transmission data to obtain target multidimensional transmission data;

Performing sequence alignment on the target multidimensional transmission data by adopting a dynamic time warping algorithm to obtain a plurality of first transmission sequence data;

performing distance calculation on the plurality of first transmission sequence data by adopting an SBD (sequence based digital) measurement algorithm to obtain the distance between the sequence pairs;

data combination is carried out on the plurality of first transmission sequence data and the distances between the sequence pairs, and target combined transmission data is generated;

and carrying out data classification on the target combined transmission data to obtain transmission data of a transmission end, load data of a plurality of switches and acquisition data of a plurality of receiving ends.

3. The method for processing data of a network switch according to claim 1, wherein creating an initial transmission relationship network according to the data sent by the transmission end, the load data of the switches, and the acquired data of the receiving ends, and performing network weight change optimization on the initial transmission relationship network to obtain a target transmission relationship network, includes:

determining a plurality of corresponding target transmission end nodes according to the transmission end sending data, determining a plurality of corresponding target switch nodes according to the switch load data, and determining a plurality of corresponding target receiving end nodes according to the receiving end obtaining data;

Performing relationship analysis on the plurality of target transmission end nodes and the plurality of target switch nodes through a preset decision tree algorithm to obtain a plurality of first transmission relationships, and performing relationship analysis on the plurality of target switch nodes and the plurality of target receiving end nodes through the decision tree algorithm to obtain a plurality of second transmission relationships;

creating a plurality of first relationship edges between the plurality of target transmission end nodes and the plurality of target switch nodes according to the plurality of first transmission relationships, and creating a plurality of second relationship edges between the plurality of target switch nodes and the plurality of target receiving end nodes according to the plurality of second transmission relationships;

network connection is carried out on the plurality of first relation edges and the plurality of second relation edges based on preset initial weights, and a corresponding initial transmission relation network is generated;

and carrying out dynamic weight change calculation on the initial weight, generating a plurality of corresponding target weights, and carrying out network weight change optimization on the initial transmission relation network based on the plurality of target weights to obtain a target transmission relation network.

4. The method for processing data of a network switch according to claim 3, wherein the extracting the transmission risk influencing factors of the plurality of target network switches in the target network switch cluster based on the target transmission relation network to obtain the plurality of transmission risk influencing factors of each target network switch includes:

Performing node traversal on a plurality of target network switches in a target network switch cluster in the target transmission relation network to obtain target switch nodes corresponding to each target network switch;

taking each target switch node as a characteristic network center, and carrying out characteristic clustering on the characteristic network center through a preset graph neural network model to obtain a target clustering result of each target switch node;

according to the target clustering result, performing membership calculation on the target transmission relation network by adopting a shortest path algorithm to obtain a target membership tree structure;

according to the target subordinate relation tree structure, identifying and sequencing transmission risk influence factors of each target switch node to obtain a plurality of initial risk influence factors;

and screening the influence intensities of the initial risk influence factors to obtain a plurality of transmission risk influence factors of each target network switch.

5. The method for processing data of a network switch according to claim 1, wherein the performing numerical feature analysis and vector mapping on the plurality of transmission risk impact factors to generate a risk impact feature vector of each target network switch includes:

Respectively calculating the target digital characteristics of each transmission risk influence factor through a preset single-factor cloud model;

carrying out digital feature normalization on the target digital features of each transmission risk influence factor to generate normalized digital features of each transmission risk influence factor;

acquiring the influence intensity sequence of the transmission risk influence factors, and carrying out serialization conversion on the normalized digital features based on the influence intensity sequence to generate a target feature sequence of each target network switch;

and carrying out vector mapping on the target feature sequences to generate risk influence feature vectors of each target network switch.

6. The method for processing data of network switches according to claim 1, wherein creating a transmission risk analysis model corresponding to each target network switch, and performing transmission risk prediction and analysis on the risk influence feature vector through the transmission risk analysis model, to obtain an initial transmission risk analysis result corresponding to each target network switch, includes:

creating a transmission risk analysis model corresponding to each target network switch, wherein the transmission risk analysis model comprises: an input layer, a long and short time memory network, a threshold circulation network, a full connection layer and an output layer;

Carrying out vector standardization processing on the risk influence characteristic vector through an input layer in the transmission risk analysis model to obtain a standard influence characteristic vector;

performing time sequence feature extraction on the standard influence feature vector through a long-short-time memory network in the transmission risk analysis model to obtain a time sequence influence feature vector;

extracting hidden state features of the time sequence influence feature vectors through a threshold circulation network in the transmission risk analysis model to obtain hidden state feature vectors;

carrying out transmission risk probability prediction on the hidden state feature vector through a full connection layer in the transmission risk analysis model to obtain target transmission risk probability;

and comparing the target transmission risk probability with a preset target probability threshold value through an output layer in the transmission risk analysis model to generate an initial transmission risk analysis result corresponding to each target network switch.

7. The method for processing data of a network switch according to claim 1, wherein the generating a target data exchange risk execution scheme corresponding to the target network switch cluster according to the initial transmission risk analysis result includes:

Performing execution scheme analysis on each target network switch according to the initial transmission risk analysis result to obtain an initial data exchange risk execution scheme of each target network switch;

carrying out global scheme initialization on the target network switch cluster according to an initial data exchange risk execution scheme of each target network switch through a preset genetic algorithm to generate an initial global execution scheme group, wherein the initial global execution scheme group comprises a plurality of first candidate risk execution schemes;

respectively calculating first fitness of each first candidate risk execution scheme, and dividing scheme groups of the first candidate risk execution schemes according to the first fitness to obtain a plurality of target global execution scheme groups;

generating schemes of the target global execution scheme groups to obtain a plurality of second candidate risk execution schemes, and respectively calculating second fitness of each second candidate risk execution scheme;

and carrying out scheme optimization analysis on the plurality of second candidate risk execution schemes according to the second fitness to generate a target data exchange risk execution scheme corresponding to the target network switch cluster.

8. A data processing system of a network switch, the data processing system of the network switch comprising: