WO2021068489A1 - Routing path intelligent selection method and apparatus, device, and readable storage medium - Google Patents

Abstract

The present application relates to the technical field of artificial intelligence and discloses a routing path intelligent selection method, comprising the following steps: mining, according to a preset transaction type and routing path, current network state information data by means of a data mining model, so as to obtain transaction type data and routing path data; determining whether the transaction type data is currently acquired; if so, obtaining the routing path data according to a correlation, and predicting the routing path data by means of a neural network model, so as to obtain a routing path set; determining whether there is a routing path where the transaction type data congestion occurs in the routing path set; and if so, forwarding the transaction type data by means of other routing paths where no transaction type data congestion occurs. The present application further discloses a routing path intelligent selection apparatus, a device, and a computer-readable storage medium. The routing path intelligent selection method provided in the present application solves the technical problem of low degree of intelligence of path selection in a routing path selection scenario.

Description

Intelligent routing path selection method, device, equipment and readable storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 12, 2019, the application number is 201910967899.7, and the invention title is "Routing Path Intelligent Selection Method, Device, Equipment, and Readable Storage Medium", and its entire content Incorporated in this application by reference.

Technical field

This application relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and computer-readable storage medium for intelligent routing path selection.

Background technique

Due to the rapid increase in business requirements in the prior art, data of different business types often require different routing paths to transmit, while the routing path provided by the existing routing scheme is relatively single, and can only match the routing path for the data of the existing business. When introducing new business type data, it is often necessary to manually select the routing path manually. The inventor realizes that the existing routing path cannot meet the data transmission requirements, and the data selection of the routing path is not intelligent enough. The existing router can only match the routing path for the existing service type data. When introducing new service type data , Often need to manually select the routing path manually. Therefore, how to intelligently select a routing path according to the data transaction type and transaction volume is a technical problem to be solved urgently in this field.

Summary of the invention

The main purpose of this application is to provide a method, device, device, and computer-readable storage medium for intelligent routing path selection, aiming to solve the technical problem of low degree of intelligentization of path selection in routing path selection scenarios.

In order to achieve the above objective, the present application provides an intelligent routing path selection method. The routing path intelligent selection method includes the following steps:

Acquiring a training sample set in real time, the training sample set including network state information data;

The training sample data in the training sample set is sequentially calculated by the following formula to obtain the network state information data feature set:

Wherein, the network state information data feature set has two different types of training sample data, W ^T U ₀ and W ^T U ₁ represent the projection of the centers of the two different types of training sample data on a straight line, W ^T ∑ ₀ W and W ^T ∑ ₁ W represent the covariance of the two types of training sample data after projection;

Training a first data mining model using the network state information data feature set to obtain a second data mining model, wherein the first data mining model includes a plurality of initial data mining models to be selected;

According to the preset transaction type and routing path, mining the current network state information data through the second data mining model to obtain transaction type data and routing path data;

A hash algorithm is used to establish the correspondence between the transaction type data and the routing path data, and the correspondence is a one-to-many correspondence;

Determine whether the transaction type data is currently obtained;

If transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model to obtain a routing path set, wherein the routing path set includes at least two A routing path, if the transaction type data is not currently obtained, it is judged whether the transaction type data is currently obtained;

If the transaction type data is not currently obtained, judge whether the transaction type data is currently obtained;

Judging whether there is a routing path where transaction type data congestion occurs in the routing path set;

If there is a routing path that has transaction type data congestion in the routing path set, the transaction type data is forwarded through other routing paths in the routing path set that do not have transaction type data congestion. If there is no transaction type data congestion in the routing path set, If there is a routing path with transaction type data congestion, it is determined whether there is a routing path with transaction type data congestion in the routing path set, wherein the forwarding follows a preset routing path adjustment strategy.

Further, in order to achieve the above object, the present application also provides a first routing path intelligent selection device, the routing path intelligent selection device includes:

An obtaining module, configured to obtain a training sample set in real time, the training sample set including network state information data;

The first calculation module is configured to sequentially calculate the training sample data in the training sample set by the following formula to obtain a network state information data feature set;

A training module for training a first data mining model using the network state information data feature set to obtain a second data mining model;

The mining module is used to mine the current network state information data through the second data mining model according to preset transaction types and routing paths to obtain transaction type data and routing path data;

The establishment module is configured to establish a correspondence between the transaction type data and the routing path data by using a hash algorithm, and the correspondence is a one-to-many correspondence;

The first judgment module is used to judge whether the transaction type data is currently acquired;

The prediction module is configured to obtain the routing path data according to the corresponding relationship if the transaction type data is currently obtained, and predict the routing path data through a neural network model to obtain a routing path set;

The second judgment module is used for judging whether the transaction type is currently acquired if the transaction type data is not currently acquired;

The third judgment module is used to judge whether there is a routing path where transaction type data congestion occurs in the routing path set;

The first forwarding module is configured to forward the transaction type data through other routing paths in the routing path set without transaction type data congestion if there is a routing path in the routing path set that has transaction type data congestion.

The fourth judging module is configured to determine whether there is a routing path where transaction type data congestion occurs in the routing path set if there is no routing path where transaction type data congestion occurs in the routing path set, wherein the forwarding follows a preset Routing path adjustment strategy.

Optionally, the training module includes the following units:

An extraction unit, configured to extract n training sets from the N network state information data features of the network state information data feature set by a bagging method, where the N is greater than or equal to n;

A training unit, configured to use the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

The screening unit is configured to screen out a plurality of initial data mining models from the plurality of preset initial data mining models to be selected by a preset voting method according to the plurality of initial classification results to obtain a second data mining model.

Optionally, the intelligent routing path selection device further includes the following modules:

The fifth judgment module is configured to monitor the first service message data in real time through in-band network telemetry technology to obtain the monitoring result, and judge whether the first service message data reaches the preset important level based on the monitoring result;

The mirroring module is configured to, if the first service message data reaches a preset important level, use in-band network telemetry technology to mirror the second service message data according to the first service message data;

The sixth judgment module is configured to, if the first service message data does not reach the preset important level, monitor the first service message data in real time through in-band network telemetry technology to obtain the monitoring result, and determine the location based on the monitoring result. Whether the first service message data reaches the preset important level;

Further, in order to achieve the above object, the present application also provides an intelligent routing path selection device, the routing path intelligent selection device includes a memory, a processor, and a router that is stored in the memory and can run on the processor. A path intelligent selection program, when the routing path intelligent selection program is executed by the processor, the steps of the routing path intelligent selection method as described in any one of the above are implemented.

Further, in order to achieve the above-mentioned object, the present application also provides a computer-readable storage medium with a routing path intelligent selection program stored on the computer-readable storage medium, and the routing path intelligent selection program is executed by the processor to achieve The steps of the intelligent routing path selection method described in any one of the above.

Beneficial effects of this application: This application is to solve the technical problem of manually selecting a routing path according to the real-time newly added service type in the prior art. Provide an intelligent routing path selection method. The implementation process of this application is: mining network state information data through a data mining model to obtain mining data, and establishing a one-to-many correspondence between the transaction type data and routing path data through a hash algorithm, and then through the neural network model The routing path set is predicted and obtained based on the routing path data, and different routing paths are selected according to the status of each routing path forwarding transaction type data in the routing path set. Realize the intelligent routing path selection based on the data of different transaction types.

Description of the drawings

FIG. 1 is a schematic structural diagram of an operating environment of a routing path intelligent selection device involved in a solution according to an embodiment of the application;

2 is a schematic flowchart of a first embodiment of a method for intelligently selecting a routing path according to this application;

FIG. 3 is a schematic diagram of the detailed flow of step S30 in FIG. 2;

4 is a schematic flowchart of a second embodiment of a method for intelligently selecting a routing path according to this application;

FIG. 5 is a schematic flowchart of a third embodiment of a routing path intelligent selection method according to this application;

6 is a schematic flowchart of a fourth embodiment of a routing path intelligent selection method according to this application;

FIG. 7 is a detailed flowchart of step S80 in FIG. 2;

FIG. 8 is a schematic flowchart of a fifth embodiment of a method for intelligently selecting a routing path according to this application;

FIG. 9 is a schematic diagram of functional modules of an embodiment of a device for intelligent routing path selection according to the present application.

Detailed ways

It should be understood that the specific embodiments described here are only used to explain the present application, and are not used to limit the present application.

This application provides an intelligent routing path selection device.

Referring to FIG. 1, FIG. 1 is a schematic structural diagram of an operating environment of a routing path intelligent selection device involved in a solution of an embodiment of the application.

As shown in FIG. 1, the intelligent routing path selection device includes: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to implement connection and communication between these components. The user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard), and the network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as a magnetic disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

Those skilled in the art can understand that the hardware structure of the routing path intelligent selection device shown in FIG. 1 does not constitute a limitation on the routing path intelligent selection device, and may include more or less components than shown in the figure, or a combination of certain components. Components, or different component arrangements.

As shown in FIG. 1, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a routing path intelligent selection program. Among them, the operating system is a program that manages and controls the routing path intelligent selection equipment and software resources, and supports the routing path intelligent selection program and the operation of other software and/or programs.

In the hardware structure of the routing path intelligent selection device shown in FIG. 1, the network interface 1004 is mainly used to access the network; the user interface 1003 is mainly used to detect and confirm instructions and edit instructions. The processor 1001 may be used to call the routing path intelligent selection program stored in the memory 1005, and execute the operations of the following embodiments of the routing path intelligent selection method.

Based on the foregoing hardware structure of the routing path intelligent selection device, various embodiments of the routing path intelligent selection method of the present application are proposed.

Referring to Fig. 2, Fig. 2 is a schematic flowchart of a first embodiment of a method for intelligently selecting a routing path according to the present application. In this embodiment, the intelligent routing path selection method includes the following steps:

Step S10, obtaining a training sample set in real time, the training sample set including network state information data;

In this embodiment, the network status information data can be obtained in real time through the preset API interface, and the total number of training sample data in the training sample set is:

The network status information data includes at least: device status information data and flow status information data. The operating status of the device can be judged by the acquired device status information data, and the acquired flow status information data can be used to monitor whether the data flow exceeds the current The load range of the routing path.

In step S20, the training sample data in the training sample set is sequentially calculated by the following formula to obtain a network state information data feature set:

Wherein, the network state information data feature set has two different types of training sample data, W ^T U ₀ and W ^T U ₁ represent the projection of the centers of the two different types of training sample data on a straight line, W ^T ∑ ₀ W and W ^T ∑ ₁ W represent the covariance of the two types of training sample data after projection;

In this embodiment, the training sample data in the training sample set is sequentially calculated by the following formula to obtain the network state information data feature set:

After the multi-dimensional training sample data in the training sample set is projected by this formula, they are all projected on a one-dimensional straight line, that is, the dimensionality reduction of the multi-dimensional data is realized. In addition, the massive training sample data in the training sample set is divided into two Therefore, this embodiment can not only achieve the purpose of accelerating the subsequent calculation speed through dimensionality reduction, but also divide the data into two categories to prepare for data segmentation and data mining.

Project the data onto a straight line w, the projections of the centers of the two types of samples on the straight line are W ^T U ₀ and W ^T U ₁ , respectively. If all the sample points are projected onto the straight line, the two types of sample projections The covariances are W ^T ∑ ₀ W and W ^T ∑ ₁ W, respectively.

Among them, u refers to the mean vector of all samples

Find the projected covariance of different types of training samples according to the following formula, where X refers to the vector of the current sample:

Divide the samples into closer classes. From the above formula, we can see that if the covariance matrix before projection is ∑, then the one after projection is W ^T ∑W.

The goal now is: the projection points of similar samples are as close as possible, and the projection points of heterogeneous samples are as far away as possible, so the variance of the projection points of similar samples should be as small as possible, that is, W ^T ∑ ₀ W+W ^T ∑ ₁ W Small, it is necessary to make the center projection of heterogeneous samples as far as possible, namely

As large as possible, then get the maximum goal J:

Different data can be divided into different categories by maximizing the goal J.

Step S30: Use the network state information data feature set to train a first data mining model to obtain a second data mining model, where the first data mining model includes multiple initial data mining models to be selected;

In this embodiment, the first data mining model includes multiple initial data mining models, and the first data mining model refers to a model set. If the first data mining model has effective data mining capabilities, it is also necessary to train the first data mining model through the network state information data feature set until the first data mining model can output according to the current network state information data feature set. The data of the scene, if the output data is in line with the current scene, it means that the first data mining model has been trained. In order to distinguish it from the first data mining model, the trained first data mining model is named second data mining model.

Step S40: According to the preset transaction type and routing path, mining the current network state information data through the second data mining model to obtain transaction type data and routing path data;

In this embodiment, the preset data mining model includes a data mining model constructed based on a network state information data feature set and a clustering algorithm. In order to mine the corresponding type of data that meets the current application scenario from a large amount of network state information data, it is necessary to mine through a pre-built data mining model. The data mining model includes the following steps. First, extract features from a large amount of network state information data. Data, and then classify the extracted feature data to obtain transaction type data and routing path data.

Step S50, using a hash algorithm to establish a correspondence between the transaction type data and the routing path data, and the correspondence is a one-to-many correspondence;

In this embodiment, the hash value of each transaction type data and the hash value of the routing path data are obtained through the calculation of the hash algorithm, and the hash value of each transaction type data and the hash value of the routing path data are established. The corresponding relationship between each transaction type data can correspond to multiple routing path data.

Step S60, it is judged whether the transaction type data is currently acquired;

In this embodiment, a preset monitoring device can be used to detect whether new transaction type data is added to the service message in real time. The new transaction type data can be existing transaction type data or transaction type data of a new service.

Step S70: If the transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through the neural network model to obtain the routing path set. If the transaction type is not currently acquired Data, return to step S60, where the routing path set includes at least two routing paths.

In this embodiment, the routing path training sample data is used to train the neural network model constructed by the neural network algorithm and the routing path training sample data, and each prediction result output by the neural network model is normalized to obtain the routing path set prediction As a result, the prediction result is compared with the preset prediction result to determine whether the preset threshold is satisfied. If it is satisfied, it means that the neural network model has the ability to predict the preset accuracy rate. And output routing path set through neural network model.

In this embodiment, if the transaction type data is currently acquired, the routing path data is obtained according to the correspondence between the transaction type data and the routing path data, and the routing path data is predicted through the neural network model to obtain the routing path set For example, the current data are public transaction data and private transaction data, and because there is a one-to-many relationship between transaction type data and routing path data, data for each transaction type will be multiple For routing path data, the routing path data is predicted through a neural network model to obtain a routing path set for forwarding public transaction data and a routing path set for forwarding private transaction data.

Step S80, judging whether there is a routing path where transaction type data congestion occurs in the routing path set;

In this embodiment, the in-band network detection technology can be used to detect whether transaction type data congestion occurs at the first routing path node in the routing path set. The reason for judging whether there is a routing path where transaction type data congestion occurs in the routing path set is to prevent Data is lost due to data congestion.

Step S90: If there is a routing path that has transaction type data congestion in the routing path set, the transaction type data is forwarded through other routing paths in the routing path set that do not have transaction type data congestion. If there is no routing path where transaction type data congestion occurs, return to step S80, where the forwarding follows the preset routing path adjustment strategy.

In this embodiment, when data transaction type data congestion occurs in the current routing path, other data cannot continue to pass through the current routing path. In order to achieve effective data forwarding, it is necessary to use the routing path to centralize other routes that have not occurred transaction type data congestion. The path forwards transaction type data. The path adjustment strategy refers to the routing path that has transaction type data congestion in the routing path set, and when the transaction type data is forwarded through other routing paths in the routing path set without transaction type data congestion, other transaction type data congestion does not occur. There are different priority levels between routing paths, and other routing paths with higher priority levels that are not subject to transaction type data congestion can be used to forward transaction type data.

This solution uses the data mining model to mine the network status information data to obtain the mining data. The one-to-many correspondence between the transaction type data and the routing path data is established through the hash algorithm, and the neural network model is used to predict the routing path data And get the routing path set, and select different routing paths according to the status of each routing path forwarding transaction type data in the routing path set. Realize the intelligent routing path selection based on the data of different transaction types.

Referring to FIG. 3, FIG. 3 is a detailed flowchart of step S30 in FIG. 2. In this embodiment, the above step S30 specifically includes the following steps:

Step S301, extracting n training sets from the N network state information data features of the network state information data feature set by the bagging method, where the N is greater than or equal to n;

In this embodiment, n training sets are extracted from the N data of the network state information data feature set. Each round uses the bagging method to extract n training samples from the network state information data feature set. In the training set, some samples may be drawn multiple times, while some samples may not be drawn at once. A total of k rounds of extraction are performed, and k training sets are obtained, among which the k training sets are independent of each other. Each time one training set is used to obtain an initial data mining model that has been trained, and k training sets are used to obtain k initial data mining models that have been trained. The k initial data mining models that have been trained in the previous step are used to obtain the classification results by voting. According to the preset classification results, it is judged whether the classification results meet the preset conditions. If they are satisfied, the trained data mining model is obtained. Among the data mining models that have been trained, there are k initial data mining models that have completed training. The classification results of each initial data mining model that have completed training may be different, but the weight of each initial data mining model is the same of.

Step S302, using the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

In this embodiment, the n training sets are used to train multiple preset initial data mining models to be selected, and multiple initial classification results are obtained.

Step S303: According to the multiple initial classification results, multiple initial data mining models are selected from the multiple preset initial data mining models to be selected by a preset voting method to obtain a second data mining model.

In this embodiment, according to the plurality of initial classification results, a plurality of initial data mining models are selected from the plurality of preset candidate initial data mining models through a preset voting method to obtain a trained data mining model. The voting method means that the initial classification result is compared with the pre-classified result. If the difference between the initial classification result output by the current candidate initial data mining model and the pre-classified result meets the preset threshold, the selected The initial data mining model to be selected.

Referring to FIG. 4, FIG. 4 is a schematic flowchart of a second embodiment of a method for intelligent routing path selection according to this application. In this embodiment, before step S60 in FIG. 2, the following steps are further included:

In step S100, the first service message data is monitored in real time through the in-band network telemetry technology to obtain the monitoring result, and based on the monitoring result, it is judged whether the first service message data reaches a preset important level, wherein the first service The message data includes network device status information data and traffic status information data;

In this embodiment, before real-time monitoring of the first service message data through the in-band network telemetry technology, the importance level is set for the data in the first service message data, for example, the data of some important events needs to be set to the important level , For non-important data set to non-important level, the classification of the level is divided according to the importance of the current business. When the first business message data is monitored in real time through in-band network telemetry technology, and the monitoring result is obtained, it is judged Whether the currently monitored data reaches the important level, and if so, perform mirroring processing on the first service message data.

Step S110: If the first service message data reaches the preset importance level, the second service message data is mirrored according to the first service message data through the in-band network telemetry technology. If the first service message data is If the document data does not reach the preset importance level, return to step S100.

In this embodiment, the first service message data is monitored in real time through the in-band network telemetry technology, and the second service message data is mirrored according to the first service message data, for example, a routing path for forwarding important level data If a failure occurs, data loss may occur. In order to avoid the loss of important level data when this situation occurs, when the first service message data is detected, the first service message data is mirrored to obtain Output the second service message data.

Referring to FIG. 5, FIG. 5 is a schematic flowchart of a third embodiment of a routing path intelligent selection method according to this application. In this embodiment, before step S70 in FIG. 2, the following steps are further included:

Step S120, judging whether each routing path in the routing path set is greater than a preset minimum routing path;

In this embodiment, since whether it is the smallest routing path determines the speed of obtaining data through the routing path, if data is forwarded through the largest routing path each time, there may be a waste of resources, so it is necessary to determine the routing Whether the path is greater than the preset minimum routing path.

Step S130: If each routing path in the routing path set is greater than a preset minimum routing path, adopt a back propagation algorithm to adjust the parameter value of the neural network model until each routing path in the routing path set is less than or equal to The minimum routing path is preset, if not, it will not be processed.

In this embodiment, if the actual routing path is greater than the preset minimum routing path, it means that the routing path obtained in step S70 does not meet the requirements, so the parameter values of the neural network model need to be adjusted, for example, the parameter values of the neural network model They are w ₁ and w _{2 respectively} . _{The sum of w 1} and w ₂ is 1, and w _{1 is} greater than w _2. If the output routing path is larger than the preset routing path, you can adjust w ₁ and w ₂ , until the current routing path is less than or equal to the preset minimum routing path.

Referring to FIG. 6, FIG. 6 is a schematic flowchart of a fourth embodiment of a routing path intelligent selection method according to this application. In this embodiment, before step S80 in FIG. 2, the following steps are further included:

Step S140, judging whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, wherein the preset load value is less than the maximum load value of the current routing path;

In this embodiment, if the preset load value is equal to or greater than the maximum load value of the current routing path, data congestion will occur in the current routing path. Therefore, in this embodiment, the preset load value is less than the maximum load value of the current routing path. Load value. When the number of transaction type data reaches the preset load value of the current routing path, other routing paths are used. For example, the maximum load value of the routing path is 200, and the preset load value is 100. When the transaction type data quantity value is 1 to 100, the first routing path is used. When the transaction type data quantity value is greater than 100, 1 to 100 The transaction type data selects the first routing path, and the transaction type data greater than 100 and less than 200 uses the second routing path. The second path refers to the routing path in the path set except the first routing path.

Step S150: If the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, it is determined whether there is a routing path that does not forward transaction type data in the routing path set. If the number of transaction type data forwarded by the current routing path in the routing path set does not reach the preset load value of the current routing path, return to step S140;

In this embodiment, if the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, it is determined whether there is a routing path that does not forward transaction type data in the routing path set, for example, a routing path There are three routing paths in the set, namely, A, B, and C. The preset load values for A, B, and C are all 50, and the maximum load value of A, B, and C is 100. If the data volume is 150 at this time, and there are data in the follow-up, the routing path can be known after judgment. There is no routing path that does not forward transaction type data in the centralized. If the data volume is 90 at this time, and there are subsequent data, there is a routing path that does not forward transaction type data. At this time, it is necessary to continue to determine the transaction forwarded by the current routing path in the routing path set Whether the number of type data reaches the preset load value of the current routing path.

Step S160: If there is a routing path that does not forward transaction type data in the routing path set, forward the transaction type data through the routing path that does not forward transaction type data, if there is no transaction type that is not forwarded in the routing path set For the data routing path, the transaction type data is forwarded through the routing path that reaches the preset load value.

In this embodiment, when the number of transaction type data is greater than the maximum load value of the current routing path, the transaction type data that exceeds the preset load value of the current routing path is forwarded through other routing paths in the routing path set. Preferentially, the shortest route is selected.

Referring to FIG. 7, FIG. 7 is a detailed flowchart of step S80 in FIG. 2. In this embodiment, the above step S80 specifically includes the following steps:

Step 801: Send a congestion detection request to each routing path in the routing path set;

In this embodiment, a congestion detection request is sent to the first routing path node, and the channel load at the first routing path node in the channel is monitored through the preset channel load whether the channel load at the first routing path node is higher than a preset threshold, and if it is higher than the preset threshold, then It shows that data congestion occurs at the first routing path node. For example, in an actual scenario, if data congestion occurs at the first routing path node, an alarm is issued.

Step S802: According to the congestion response messages received from the respective routing paths, it is determined whether there is a routing path where transaction type data congestion occurs in the routing path set.

In this embodiment, the in-band network telemetry technology can be used to detect whether congestion occurs at the first routing path node, and if it occurs, a congestion response message is sent. A congestion detection request is sent to the first routing path node, where the congestion detection request is used to request the first routing path node to perform congestion detection on the priority routing path, according to the congestion detection request from the first routing path node The received congestion response message is used to determine whether data congestion occurs in the priority routing path.

Referring to FIG. 8, FIG. 8 is a schematic flowchart of a fifth embodiment of a method for intelligently selecting a routing path according to this application. In this embodiment, after step S80 in FIG. 2, the following steps are further included:

Step S170: Calculate the frequency of occurrence of data of each transaction type through a summation formula;

In this embodiment, through the summation formula:

Calculate the frequency of each transaction type data, where i represents the transaction type data of the i type, P represents the frequency of the transaction type data of the i type, and N represents the Nth occurrence of the transaction type data of the i type. The trained neural network model can predict and obtain transaction type data based on the previous network state information data, for example, public transaction data and private transaction data. If the frequency of public transaction data is 20%, private transaction data The frequency of occurrence of class data is 80%.

Step S180, determining the priority level of each routing path in the routing path set to forward the transaction type data according to the frequency;

In this embodiment, if the transaction type data appears frequently, it can be known from the corresponding relationship that the frequency of the routing path data corresponding to the transaction type data will also increase accordingly. The neural network model is based on routing path data with different occurrence frequencies. Output different numbers of routing paths. The larger the number of routing paths, the greater the demand for the current transaction type data for this routing path. Therefore, the priority level between each routing path can be determined according to the number of routing paths. For example, if the frequency of the private transaction type data is high, the routing paths in the path concentration are preferentially provided for the use of the private transaction type data.

Step S190: Determine a preset routing path adjustment strategy according to the priority level.

In this embodiment, since the preset routing path adjustment strategy is determined according to the priority level of each routing path in the routing path set to forward transaction type data, when there is an instruction to use the routing path to forward data, it can be adjusted according to the preset routing path Strategy to forward.

Referring to FIG. 9, FIG. 9 is a schematic diagram of functional modules of an embodiment of an apparatus for intelligent routing path selection according to the present application. In this embodiment, the intelligent routing path selection device includes:

The obtaining module 10 is configured to obtain a training sample set in real time, and the training sample set includes network state information data;

The calculation module 20 is configured to sequentially calculate the training sample data in the training sample set by the following formula to obtain a network state information data feature set:

The training module 30 is configured to use the network state information data feature set to train a first data mining model to obtain a second data mining model;

The mining module 40 is used to mine current network state information data through the second data mining model according to preset transaction types and routing paths to obtain transaction type data and routing path data;

The establishment module 50 is configured to establish a correspondence between the transaction type data and the routing path data by using a hash algorithm, and the correspondence is a one-to-many correspondence;

The first judgment module 60 is used to judge whether the transaction type data is currently acquired;

The prediction module 70 is configured to obtain the routing path data according to the corresponding relationship if the transaction type data is currently obtained, and predict the routing path data through a neural network model to obtain a routing path set;

The second judgment module 80 is used for judging whether the transaction type is currently acquired if the transaction type data is not currently acquired;

The third judgment module 90 is configured to judge whether there is a routing path in which transaction type data congestion occurs in the routing path set;

The forwarding module 100 is configured to forward the transaction type data through other routing paths in the routing path set without transaction type data congestion if there is a routing path where transaction type data congestion occurs in the routing path set;

The fourth judging module 110 is configured to determine whether there is a routing path where transaction type data congestion occurs in the routing path set if there is no routing path where transaction type data congestion occurs in the routing path set, wherein the forwarding follows a predetermined Set routing path adjustment strategy.

The application also provides a computer-readable storage medium.

In this embodiment, a routing path intelligent selection program is stored on the computer-readable storage medium. The computer-readable storage medium may be non-volatile or volatile. The routing path intelligent selection program is executed by the processor. The steps of the intelligent routing path selection method as described in any of the above embodiments are implemented during execution.

Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。 Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM), including Several instructions are used to make a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) execute the methods described in the various embodiments of the present application.

The embodiments of the application are described above with reference to the accompanying drawings, but the application is not limited to the above-mentioned specific embodiments. The above-mentioned specific embodiments are only illustrative and not restrictive. Those of ordinary skill in the art are Under the enlightenment of this application, without departing from the purpose of this application and the scope of protection of the claims, many forms can be made, any equivalent structure or equivalent process transformation made by using the content of the description and drawings of this application, or It is directly or indirectly used in other related technical fields, and these all fall within the protection of this application.

Claims (19)

1. A method for intelligently selecting a routing path, the method for intelligently selecting a routing path includes the following steps:

   Acquiring a training sample set in real time, the training sample set including network state information data;

   The training sample data in the training sample set is sequentially calculated by the following formula to obtain the network state information data feature set:

   

   Wherein, the network state information data feature set has two different types of training sample data, W ^T U ₀ and W ^T U ₁ represent the projection of the centers of the two different types of training sample data on a straight line, W ^T ∑ ₀ W and W ^T ∑ ₁ W represent the covariance of the two types of training sample data after projection;

   Training a first data mining model using the network state information data feature set to obtain a second data mining model, wherein the first data mining model includes a plurality of initial data mining models to be selected;

   According to the preset transaction type and routing path, mining the current network state information data through the second data mining model to obtain transaction type data and routing path data;

   A hash algorithm is used to establish the correspondence between the transaction type data and the routing path data, and the correspondence is a one-to-many correspondence;

   Determine whether the transaction type data is currently obtained;

   If transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model to obtain a routing path set, wherein the routing path set includes at least two A routing path, if the transaction type data is not currently obtained, it is judged whether the transaction type is currently obtained;

   Judging whether there is a routing path where transaction type data congestion occurs in the routing path set;

   If yes, forward the transaction type data through other routing paths in the routing path set that are not congested by transaction type data, where the forwarding follows a preset routing path adjustment strategy.
2. The intelligent routing path selection method according to claim 1, wherein said training a first data mining model using said network state information data feature set to obtain a second data mining model comprises the following steps:

   Extracting n training sets from the N network state information data features of the network state information data feature set by a bagging method, where the N is greater than or equal to n;

   Using the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

   According to the multiple initial classification results, multiple initial data mining models are selected from the multiple preset initial data mining models to be selected by a preset voting method to obtain a second data mining model.
3. The intelligent routing path selection method according to claim 1 or 2, before the step of judging whether the transaction type data is currently obtained, the method further comprises the following steps:

   The first service message data is monitored in real time through in-band network telemetry technology to obtain a monitoring result, and based on the monitoring result, it is determined whether the first service message data reaches a preset important level, wherein the first service message data Including network equipment status information data and traffic status information data;

   If it is, the second service message data is mirrored according to the first service message data through the in-band network telemetry technology; if not, the first service message data is monitored in real time through the in-band network telemetry technology to obtain the monitoring result, Judging whether the data of the first service message reaches a preset importance level based on the monitoring result.
4. The intelligent routing path selection method of claim 1, wherein if the transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model Before the step of obtaining the routing path set, it also includes the following steps:

   Judging whether each routing path in the routing path set is greater than a preset minimum routing path;

   If it is, the back propagation algorithm is used to adjust the parameter value of the neural network model until each routing path in the routing path set is less than or equal to the preset minimum routing path.
5. The intelligent routing path selection method according to claim 1, before the step of judging whether there is a routing path with transaction type data congestion in the routing path set, the method further comprises the following steps:

   Judging whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, where the preset load value is less than the maximum load value of the current routing path;

   If yes, determine whether there is a routing path that does not forward transaction type data in the routing path set; if not, determine whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load of the current routing path value.
6. The intelligent routing path selection method according to claim 1, wherein the judging whether there is a routing path that has transaction type data congestion in the routing path set comprises the following steps:

   Sending a congestion detection request to each routing path in the routing path set;

   According to the congestion response messages received from the respective routing paths, it is determined whether there is a routing path where transaction type data congestion occurs in the routing path set.
7. The intelligent routing path selection method according to claim 1, after the step of judging whether there is a routing path with transaction type data congestion in the routing path set, the method further comprises the following steps:

   Calculate the frequency of each transaction type data through the summation formula;

   Determine the priority level of each routing path in the routing path set to forward the transaction type data according to the frequency;

   The preset routing path adjustment strategy is determined according to the priority level.
8. An intelligent routing path selection device, the routing path intelligent selection device includes:

   An obtaining module, configured to obtain a training sample set in real time, the training sample set including network state information data;

   The calculation module is used to sequentially calculate the training sample data in the training sample set by the following formula to obtain the network state information data feature set;

   A training module for training a first data mining model using the network state information data feature set to obtain a second data mining model;

   The mining module is used to mine the current network state information data through the second data mining model according to preset transaction types and routing paths to obtain transaction type data and routing path data;

   The establishment module is configured to establish a correspondence between the transaction type data and the routing path data by using a hash algorithm, and the correspondence is a one-to-many correspondence;

   The first judgment module is used to judge whether the transaction type data is currently acquired;

   The prediction module is configured to obtain the routing path data according to the corresponding relationship if the transaction type data is currently obtained, and predict the routing path data through a neural network model to obtain a routing path set;

   The second judgment module is used for judging whether the transaction type is currently acquired if the transaction type data is not currently acquired;

   The third judgment module is used to judge whether there is a routing path where transaction type data congestion occurs in the routing path set;

   The forwarding module is configured to forward the transaction type data through other routing paths in the routing path set without transaction type data congestion if there is a routing path in the routing path set where transaction type data congestion occurs, wherein the Forwarding follows the preset routing path adjustment strategy;

   The fourth determining module is configured to determine whether there is a routing path where transaction type data congestion occurs in the routing path set if there is no routing path where transaction type data congestion occurs in the routing path set.
9. The intelligent routing path selection device according to claim 8, wherein the training module comprises:

   An extraction unit, configured to extract n training sets from the N network state information data features of the network state information data feature set by a bagging method, where the N is greater than or equal to n;

   A training unit, configured to use the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

   The screening unit is configured to screen out a plurality of initial data mining models from the plurality of preset initial data mining models to be selected by a preset voting method according to the plurality of initial classification results to obtain a second data mining model.
10. The intelligent routing path selection device according to claim 8 or 9, wherein the routing path intelligent selection device further comprises:

    The detection module is used to monitor the data of the first service message in real time through the in-band network telemetry technology to obtain the monitoring result, and determine whether the data of the first service message reaches the preset important level based on the monitoring result. 1. Service message data includes network device status information data and traffic status information data;

    The mirroring module is used to mirror the second service packet data according to the first service packet data through the in-band network telemetry technology when the first service packet data reaches the preset importance level. When the service message data does not reach the preset important level, the first service message data is monitored in real time through the in-band network telemetry technology to obtain the monitoring result, and based on the monitoring result, it is determined whether the first service message data reaches the preset important level. level. 11. A routing path intelligent selection device, characterized in that the routing path intelligent selection device includes a memory, a processor, and a routing path intelligent selection program stored in the memory and running on the processor, so The routing path intelligent selection program is executed by the processor to realize the routing path intelligent selection method:

    Wherein, the intelligent routing path selection method includes:

    Acquiring a training sample set in real time, the training sample set including network state information data;

    The training sample data in the training sample set is sequentially calculated by the following formula to obtain the network state information data feature set:

    

    Wherein, the network state information data feature set has two different types of training sample data, W ^T U ₀ and W ^T U ₁ represent the projection of the centers of the two different types of training sample data on a straight line, W ^T ∑ ₀ W and W ^T ∑ ₁ W represent the covariance of the two types of training sample data after projection;

    Training a first data mining model using the network state information data feature set to obtain a second data mining model, wherein the first data mining model includes a plurality of initial data mining models to be selected;

    According to the preset transaction type and routing path, mining the current network state information data through the second data mining model to obtain transaction type data and routing path data;

    A hash algorithm is used to establish the correspondence between the transaction type data and the routing path data, and the correspondence is a one-to-many correspondence;

    Determine whether the transaction type data is currently obtained;

    If transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model to obtain a routing path set, wherein the routing path set includes at least two A routing path, if the transaction type data is not currently obtained, it is judged whether the transaction type is currently obtained;

    Judging whether there is a routing path where transaction type data congestion occurs in the routing path set;

    If yes, forward the transaction type data through other routing paths in the routing path set that are not congested by transaction type data, where the forwarding follows a preset routing path adjustment strategy.
11. The intelligent routing path selection device according to claim 11, wherein the step of using the network state information data feature set to train a first data mining model to obtain a second data mining model comprises:

    Extracting n training sets from the N network state information data features of the network state information data feature set by a bagging method, where the N is greater than or equal to n;

    Using the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

    According to the multiple initial classification results, multiple initial data mining models are selected from the multiple preset initial data mining models to be selected by a preset voting method to obtain a second data mining model.
12. The intelligent routing path selection device according to claim 11 or 12, before the step of judging whether transaction type data is currently obtained, further comprising:

    The first service message data is monitored in real time through in-band network telemetry technology to obtain a monitoring result, and based on the monitoring result, it is determined whether the first service message data reaches a preset important level, wherein the first service message data Including network equipment status information data and traffic status information data;

    If it is, the second service message data is mirrored according to the first service message data through the in-band network telemetry technology; if not, the first service message data is monitored in real time through the in-band network telemetry technology to obtain the monitoring result, Judging whether the data of the first service message reaches a preset importance level based on the monitoring result.
13. The intelligent routing path selection device of claim 11, wherein if the transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model Before the step of obtaining the routing path set, it also includes the following steps:

    Judging whether each routing path in the routing path set is greater than a preset minimum routing path;

    If it is, the back propagation algorithm is used to adjust the parameter value of the neural network model until each routing path in the routing path set is less than or equal to the preset minimum routing path.
14. 11. The routing path intelligent selection device according to claim 11, before the step of determining whether there is a routing path where transaction type data congestion occurs in the routing path set, further comprising the following steps:

    Judging whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, where the preset load value is less than the maximum load value of the current routing path;

    If yes, determine whether there is a routing path that does not forward transaction type data in the routing path set; if not, determine whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load of the current routing path value.
15. A computer-readable storage medium, characterized in that a routing path intelligent selection program is stored on the computer-readable storage medium, and the routing path intelligent selection program is executed by a processor to realize a routing path intelligent selection method:

    Wherein, the intelligent routing path selection method includes:

    Acquiring a training sample set in real time, the training sample set including network state information data;

    The training sample data in the training sample set is sequentially calculated by the following formula to obtain the network state information data feature set:

    

    Wherein, the network state information data feature set has two different types of training sample data, W ^T U ₀ and W ^T U ₁ represent the projection of the centers of the two different types of training sample data on a straight line, W ^T ∑ ₀ W and W ^T ∑ ₁ W represent the covariance of the two types of training sample data after projection;

    Training a first data mining model using the network state information data feature set to obtain a second data mining model, wherein the first data mining model includes a plurality of initial data mining models to be selected;

    According to the preset transaction type and routing path, mining the current network state information data through the second data mining model to obtain transaction type data and routing path data;

    A hash algorithm is used to establish the correspondence between the transaction type data and the routing path data, and the correspondence is a one-to-many correspondence;

    Determine whether the transaction type data is currently obtained;

    If transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model to obtain a routing path set, wherein the routing path set includes at least two A routing path, if the transaction type data is not currently obtained, it is judged whether the transaction type is currently obtained;

    Judging whether there is a routing path where transaction type data congestion occurs in the routing path set;

    If yes, forward the transaction type data through other routing paths in the routing path set that are not congested by transaction type data, where the forwarding follows a preset routing path adjustment strategy.
16. 15. The computer-readable storage medium of claim 16, wherein the step of using the network state information data feature set to train a first data mining model to obtain a second data mining model comprises:

    Extracting n training sets from the N network state information data features of the network state information data feature set by a bagging method, where the N is greater than or equal to n;

    Using the n training sets to train the multiple preset to-be-selected initial data mining models to obtain multiple initial classification results;

    According to the multiple initial classification results, multiple initial data mining models are selected from the multiple preset initial data mining models to be selected by a preset voting method to obtain a second data mining model.
17. The computer-readable storage medium according to claim 16 or 17, before the step of determining whether transaction type data is currently obtained, further comprising:

    The first service message data is monitored in real time through in-band network telemetry technology to obtain a monitoring result, and based on the monitoring result, it is determined whether the first service message data reaches a preset important level, wherein the first service message data Including network equipment status information data and traffic status information data;

    If yes, use the in-band network telemetry technology to mirror the second service message data according to the first service message data; if not, use the in-band network telemetry technology to monitor the first service message data in real time to obtain the monitoring result, Judging whether the data of the first service message reaches a preset importance level based on the monitoring result.
18. The computer-readable storage medium according to claim 16, wherein if the transaction type data is currently acquired, the routing path data is obtained according to the corresponding relationship, and the routing path data is predicted through a neural network model ,Before the step of obtaining the routing path set, it also includes:

    Judging whether each routing path in the routing path set is greater than a preset minimum routing path;

    If it is, the back propagation algorithm is used to adjust the parameter value of the neural network model until each routing path in the routing path set is less than or equal to the preset minimum routing path.
19. 16. The computer-readable storage medium according to claim 16, before the step of judging whether there is a routing path with transaction type data congestion in the routing path set, further comprising the following steps:

    Judging whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load value of the current routing path, where the preset load value is less than the maximum load value of the current routing path;

    If yes, determine whether there is a routing path that does not forward transaction type data in the routing path set; if not, determine whether the number of transaction type data forwarded by the current routing path in the routing path set reaches the preset load of the current routing path value.

Images