Software-defined network performance prediction method, device, electronic device and storage medium

By building routing hypergraphs and using hypergraph neural networks, the problems of network performance measurement and prediction in software-defined networks are solved, and accurate prediction and optimization of network operation quality is achieved, improving network operation quality and QoS guarantee.

CN115333958BActive Publication Date: 2024-11-05CHINA TELECOM CORP LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210975635.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-15
Publication Date
2024-11-05
Estimated Expiration
2042-08-15

AI Technical Summary

Technical Problem

The active passive network performance measurement technology in existing software-defined networks cannot obtain network global performance information in real time, it is difficult to quickly respond to traffic adjustments and optimize network structures, and the intelligent network performance estimation method based on deep learning fails to accurately predict network operation quality.

Method used

By obtaining network topology information and routing configuration information, a routing hypergraph is constructed, and a pre-trained hypergraph neural network is used to mine spatial features from the end-to-end traffic matrix, output network performance prediction information, and optimize network configuration with target policy information.

Benefits of technology

It realizes accurate prediction and optimization of network operation quality, improves network operation quality, can quickly respond to traffic adjustments and optimize network structure, and provides better QoS guarantees.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115333958B_ABST
    Figure CN115333958B_ABST
Patent Text Reader

Abstract

The present disclosure provides a software-defined network performance prediction method, device, electronic device and storage medium, and relates to the field of software-defined network technology. The method includes obtaining network topology information and routing configuration information of the network to be predicted, and determining the routing hypergraph of the network to be predicted based on the network topology information and routing configuration information; obtaining network traffic information of the network to be predicted, and determining the end-to-end traffic matrix information in the network to be predicted based on the network traffic information; inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and outputting the performance prediction information of the network to be predicted. The present disclosure constructs a routing hypergraph based on network topology information and routing configuration information, and uses a hypergraph neural network to mine spatial features from the end-to-end traffic matrix of the routing hypergraph and learn to output the prediction of end-to-end network performance, which can improve the quality of network operation.
Need to check novelty before this filing date? Find Prior Art

Description

Background Art

[0002] In software-defined networks, active and passive network performance measurement technologies can be used to monitor network operation quality and understand network behavior, thereby helping to optimize network structure and improve network service quality. However, active and passive network performance measurement technologies require injecting a large number of measurement packets that repeatedly traverse network paths into the network for a long time, which will increase network load and cause changes in network operation status; in addition, active and passive network performance measurement technologies can only measure certain links or end-to-end paths, making it difficult to obtain global network performance information, and evaluating path performance is also passive and lagging, unable to quickly respond to traffic adjustments and assist in real-time QoS deployment.

[0003] In the prior art, the intelligent network performance estimation method based on deep learning technology takes the network end-to-end traffic matrix and network topology as input, and outputs the end-to-end performance estimation value after learning the network spatial characteristics and traffic characteristics through deep models such as codec neural networks and graph neural networks. However, this method does not consider the impact of routing strategies on spatial characteristics and network performance, and cannot accurately predict the network operation quality.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The present disclosure provides a method, device, electronic device and storage medium for predicting software-defined network performance, which at least to a certain extent overcome the problem that the related art cannot accurately predict the network operation quality.

[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0007] According to one aspect of the present disclosure, a method for predicting performance of a software-defined network is provided, comprising: obtaining network topology information and routing configuration information of a network to be predicted, and determining a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes hypergraph adjacency matrix information of the network to be predicted; obtaining network traffic information of the network to be predicted, and determining end-to-end traffic matrix information in the network to be predicted based on the network traffic information; inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and outputting performance prediction information of the network to be predicted.

[0008] In one embodiment of the present disclosure, the hypergraph adjacency matrix information and the end-to-end traffic matrix information are input into a pre-trained hypergraph neural network, and performance prediction information of the network to be predicted is output, including: inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information as input information into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyper-edge feature learning on the input information to obtain hyper-edge features, and inputting the hyper-edge features into the activation layer and the linear output layer in sequence for calculation and then outputting an end-to-end network performance indicator numerical matrix; determining the network performance indicator numerical matrix as the performance prediction information of the network to be predicted.

[0009] In one embodiment of the present disclosure, the method further includes: obtaining target policy information of the network to be predicted, wherein the target policy information represents preset network topology information and routing configuration information; and determining the network topology information and routing configuration information of the network to be predicted based on the target policy information of the network to be predicted.

[0010] In one embodiment of the present disclosure, in obtaining network topology information and routing configuration information of a network to be predicted, and determining a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, the method further includes: obtaining network topology information and routing configuration information of the network to be predicted; based on the network topology information and routing configuration information, connecting the network nodes in the network topology information and routing configuration information in a set manner, and determining the routes in the routing configuration information as hyperedges to constitute a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

[0011] In one embodiment of the present disclosure, after inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network and outputting the performance prediction information of the network to be predicted, the method further includes: sending the performance prediction information of the network to be predicted to the network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

[0012] In one embodiment of the present disclosure, the method also includes: obtaining a network modeling open source dataset, wherein the network modeling open source dataset includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated by a simulation process; using the network modeling open source dataset as a training set, performing model training on the hypergraph convolutional network, and determining the hypergraph neural network.

[0013] In one embodiment of the present disclosure, the method further includes: using a mean square error loss function in model training of the hypergraph convolutional network to reduce the difference between the real data and the predicted estimation results.

[0014] According to another aspect of the present disclosure, a software-defined network performance prediction device is provided, including: a routing hypergraph configuration module, used to obtain network topology information and routing configuration information of a network to be predicted, and determine the routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes hypergraph adjacency matrix information of the network to be predicted; a matrix information configuration module, used to obtain network traffic information of the network to be predicted, and determine end-to-end traffic matrix information in the network to be predicted based on the network traffic information; a performance prediction module, used to input the hypergraph adjacency matrix information and end-to-end traffic matrix information into a pre-trained hypergraph neural network, and output performance prediction information of the network to be predicted.

[0015] According to another aspect of the present disclosure, there is provided an electronic device, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the above-mentioned software-defined network performance prediction method by executing the executable instructions.

[0016] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned software-defined network performance prediction method is implemented.

[0017] According to another aspect of the present disclosure, a computer program product is provided, including computer instructions, wherein the computer instructions are stored in a computer-readable storage medium, and when the computer instructions are executed by a processor, the operating instructions of any of the above-mentioned software-defined network performance prediction methods are implemented.

[0018] The embodiment of the present disclosure provides a method for predicting the performance of a software-defined network, which obtains network topology information and routing configuration information of a network to be predicted, determines a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes hypergraph adjacency matrix information of the network to be predicted; obtains network traffic information of the network to be predicted, determines end-to-end traffic matrix information in the network to be predicted based on the network traffic information; inputs the hypergraph adjacency matrix information and end-to-end traffic matrix information into a pre-trained hypergraph neural network, and outputs performance prediction information of the network to be predicted. The present disclosure constructs a routing hypergraph based on network topology information and routing configuration information, uses a hypergraph neural network to mine spatial features from the end-to-end traffic matrix of the routing hypergraph, and learns to output predictions of end-to-end network performance, which can improve the quality of network operation.

[0019] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0021] Figure 1 A schematic diagram showing the structure of a software-defined network performance prediction system in an embodiment of the present disclosure;

[0022] Figure 2 A flow chart of a method for predicting software-defined network performance in an embodiment of the present disclosure is shown;

[0023] Figure 3 A flow chart of another software defined network performance prediction method according to an embodiment of the present disclosure is shown;

[0024] Figure 4 A flow chart of another software defined network performance prediction method according to an embodiment of the present disclosure is shown;

[0025] Figure 5 Another flow chart of a software defined network performance prediction method according to an embodiment of the present disclosure is shown;

[0026] Figure 6 A flowchart showing a specific example of a method for predicting software-defined network performance in an embodiment of the present disclosure;

[0027] Figure 7 A flowchart showing a specific example of a method for predicting software-defined network performance in an embodiment of the present disclosure;

[0028] Figure 8 A schematic diagram showing a specific example of a software-defined network performance prediction system structure in an embodiment of the present disclosure;

[0029] Fig. 9 A schematic diagram of a software-defined network performance prediction device in an embodiment of the present disclosure is shown;

[0030] Fig.10 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown;

[0031] Fig.11 A schematic diagram showing a computer-readable storage medium in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0032] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0033] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0034] For ease of understanding, several terms involved in the present disclosure are explained below:

[0035] QoS: Quality of Service, service quality;

[0036] SDN: Software Defined Network, software defined network;

[0037] HGNN: Hypergraph Neural Network, hypergraph neural network;

[0038] CPU: Central Processing Unit, central processing unit;

[0039] ReLU function: Linear rectification function, linear rectification function;

[0040] OpenFlow, a network communication protocol, belongs to the data link layer, and can control the forwarding plane of switches or routers on the Internet, thereby changing the network path taken by network data packets;

[0041] OpenSketch, a general, abstract measurement framework that uses a sampled, compressed counting method and stores the resulting traffic information in a data structure called Sketch;

[0042] Telemetry is a technology for remotely collecting data from physical or virtual devices at high speed. The device periodically and actively sends information such as interface traffic statistics, CPU or memory data to the collector through push mode. Compared with the traditional question-and-answer interaction of pull mode, it provides more real-time and high-speed data collection function.

[0043] Figure 1 A schematic diagram showing an exemplary system architecture that can be applied to a software-defined network performance prediction method or a software-defined network performance prediction device according to an embodiment of the present disclosure.

[0044] like Figure 1 As shown, the software defined network system architecture 100 may include an SDN control component 101, a routing hypergraph neural network model 102 and a network device 103.

[0045] The network device 103 may be any electronic device connected to the network, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, a wearable device, an augmented reality device, a virtual reality device, a network printer, a network camera, etc.

[0046] When the system is running, first, the network operation and maintenance personnel 105 sends the target policy intent to the SDN control component 101, and the SDN control component inputs the network preset configurations such as the network topology and routing strategy in the target policy intent into the routing hypergraph neural network model 102; the control plane of the software-defined network can obtain the network traffic status through measurement technologies such as OpenFlow, OpenSketch or Telemetry, and input it into the routing hypergraph neural network model; after the routing hypergraph neural network model obtains the network preset configuration and the network traffic matrix (equivalent to the above-mentioned network traffic status) input, it calculates the network end-to-end delay, jitter and other performance conditions in real time, and outputs them to the SDN control component; the SDN control component judges whether to actually send the optimized configuration to the network device based on the preset optimization configuration (equivalent to the preset configurations such as the network topology and routing strategy in the above-mentioned target policy intent) and the estimated network operation quality.

[0047] Those skilled in the art will know that Figure 1 The SDN control components, routing hypergraph neural network model, and the number of network devices are only illustrative, and any number of terminal devices, networks, and servers may be provided according to actual needs. This disclosure embodiment does not limit this.

[0048] The present exemplary implementation is described in detail below with reference to the accompanying drawings and embodiments.

[0049] Figure 2A flow chart of a method for predicting software defined network performance in an embodiment of the present disclosure is shown as follows: Figure 2 As shown, the software defined network performance prediction method provided in the embodiment of the present disclosure includes the following steps:

[0050] S202, obtaining network topology information and routing configuration information of the network to be predicted, and determining a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes hypergraph adjacency matrix information of the network to be predicted.

[0051] It should be noted that the above network can be composed of several nodes and links connecting these nodes, representing many objects and their mutual connections. The above network topology information can be the physical layout of various devices interconnected by transmission media, which refers to the specific physical, i.e. real, or logical, i.e. virtual arrangement between the members constituting the network. For example, if the connection structure of two networks is the same, we say that their network topologies are the same, although their internal physical wiring and node distances may be different. The above routing configuration information can be a routing strategy, a data packet routing forwarding mechanism that is more flexible than routing based on the target network. For example, when a routing strategy is applied, the router will decide how to process the data packet that needs to be routed through the routing map. The routing map determines the next hop forwarding router for a data packet.

[0052] S204, obtaining network traffic information of the network to be predicted, and determining end-to-end traffic matrix information in the network to be predicted according to the network traffic information.

[0053] It should be noted that the network traffic can be the amount of data transmitted on the network. The end-to-end can be a network connection. For example, for a network to communicate, a connection must be established between the two ends (source and destination). Once the connection is established, it is an end-to-end connection, that is, the end-to-end is a logical link. This path may pass through a very complex physical route, but the hosts at both ends only think that there is a connection between the two ends. Once the communication is completed, the connection is released, and the physical line may be used by other applications to establish a connection.

[0054] S206, inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and outputting the performance prediction information of the network to be predicted.

[0055] It should be noted that the above-mentioned hypergraph can be a hypergraph with authors as nodes and achievements as edge sets to describe the characteristics of this type of network. For example, for an ordinary graph, an edge can only be connected to two vertices; while for a hypergraph, it is generalized to a generalized edge, that is, a hyperedge that can be connected to any number of vertices.

[0056] The present invention constructs a routing hypergraph based on network topology information and routing configuration information, uses a hypergraph neural network to mine spatial features from the end-to-end traffic matrix of the routing hypergraph and learns to output predictions of end-to-end network performance, thereby improving the quality of network operation.

[0057] In one embodiment of the present disclosure, Figure 3 As shown, the software-defined network performance prediction method provided in the embodiment of the present disclosure can determine the performance prediction information of the network to be predicted through the following steps, and can accurately predict the performance prediction information of the network to be predicted:

[0058] S302, taking the hypergraph adjacency matrix information and the end-to-end traffic matrix information as input information, and inputting them into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information, obtains hyperedge features, and sequentially inputs the hyperedge features into the activation layer and the linear output layer for calculation, and then outputs an end-to-end network performance indicator numerical matrix;

[0059] S304: Determine the network performance indicator value matrix as performance prediction information of the network to be predicted.

[0060] In one embodiment of the present disclosure, Figure 4 As shown, the software-defined network performance prediction method provided in the embodiment of the present disclosure can determine the network topology information and routing configuration information of the network to be predicted through the following steps, and can predict the performance prediction information of different networks to be predicted according to the target strategy information of different requirements:

[0061] S402, obtaining target strategy information of the network to be predicted, wherein the target strategy information represents preset network topology information and routing configuration information;

[0062] S404: Determine network topology information and routing configuration information of the network to be predicted according to target policy information of the network to be predicted.

[0063] In one embodiment of the present disclosure, Figure 5 As shown, the software-defined network performance prediction method provided in the embodiment of the present disclosure can determine the routing hypergraph of the network to be predicted through the following steps, and can convert the network information into a hypergraph form:

[0064] S502, obtaining network topology information and routing configuration information of the network to be predicted;

[0065] S504, according to the network topology information and the routing configuration information, the network nodes in the network topology information and the routing configuration information are connected in a set manner, and the routes in the routing configuration information are determined as hyperedges to form a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

[0066] In one embodiment of the present disclosure, Figure 6 As shown, the software-defined network performance prediction method provided in the embodiment of the present disclosure can send the performance prediction information of the network to be predicted through the following steps, and can configure the performance prediction information of the network to be predicted to the network node to improve the network operation quality of the network node:

[0067] S602: Send performance prediction information of the network to be predicted to the network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

[0068] In one embodiment of the present disclosure, Figure 7 As shown, the software-defined network performance prediction method provided in the embodiment of the present disclosure can determine the hypergraph neural network through the following steps, and can establish the impact of routing strategies on spatial characteristics and network performance in the model:

[0069] S702, obtaining a network modeling open source data set, wherein the network modeling open source data set includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated during a simulation process;

[0070] S704, using the network modeling open source dataset as a training set, performing model training on the hypergraph convolutional network, and determining the hypergraph neural network.

[0071] In one embodiment, a mean square error loss function is used in model training of a hypergraph convolutional network to reduce the difference between real data and predicted estimation results.

[0072] The present invention can calculate and output end-to-end network performance indicators including delay, jitter and packet loss rate based on input elements such as network topology, routing strategy and end-to-end traffic matrix.

[0073] Figure 8 A schematic diagram showing a specific example of another exemplary system architecture that can be applied to the software-defined network performance prediction method or software-defined network performance prediction device of the embodiments of the present disclosure.

[0074] like Figure 8 As shown, the software-defined network system architecture 800 may include a routing hypergraph construction module 801 and a hypergraph convolutional network module 802.

[0075] The hypergraph construction module 801 is used for the routing hypergraph construction process, and the hypergraph convolutional network module 802 is used for the hypergraph convolutional neural network performance estimation process.

[0076] In the process of constructing the routing hypergraph, the hypergraph construction module regards each route in the routing mechanism as a hyperedge to form a routing hypergraph based on the main elements of the IP network such as network topology and routing mechanism. At this time, the network nodes are connected in a set manner, rather than in a pairwise manner as in a simple graph. The hyperedge connection method based on routing correlation can effectively express the complex spatial correlation between the IP network topology, routing mechanism and network nodes in a high-order manner. In addition, an adjacency matrix with nodes as rows and hyperedges as columns is used to represent the hypergraph structure.

[0077] In the hypergraph construction module 801, an end-to-end flow of the IP network is transmitted and forwarded through the C->B->A->E route (from C through B, then through A, and reaches E). The routing hypergraph uses hyperedges to connect the C, B, A and E nodes involved in the route, and the hypergraph adjacency matrix represents the hypergraph structure.

[0078] The mathematical representation of the routing hypergraph is HG=(V, E, W), where V represents the node set, E represents the hyperedge set (ie, the routing set), and W represents the hyperedge weight diagonal matrix.

[0079] The routing hypergraph adjacency matrix is ​​denoted as H, H ve Each element is expressed by the following formula (1):

[0080]

[0081] In addition, the node degree matrix D and the hyperedge degree matrix B are expressed by the following formula (2) and formula (3) respectively:

[0082]

[0083]

[0084] Where W represents the hyperedge weight matrix, e represents an item in the hyperedge set, e represents the set of hyperedges, v Represents a collection of nodes. v Represents an item in a node collection.

[0085] In the process of estimating the performance of the hypergraph convolutional neural network, the hypergraph convolutional network module 802 is used for a hypergraph convolutional network consisting of two layers of hypergraph convolutional layers. First, the routing hypergraph adjacency matrix and the end-to-end traffic matrix are used as inputs. After learning the hyperedge features through the hypergraph convolution process, the hyperedge features are successively input into the ReLU activation layer and the linear output layer to calculate the end-to-end network performance index numerical matrix. The two layers of hypergraph convolutional layers are mathematically expressed by the following formulas (4) and (5):

[0086]

[0087]

[0088] Among them, f represents the number of two hypergraph convolution layers, σ represents the ReLU activation layer, X represents the end-to-end traffic matrix, H represents the hypergraph adjacency matrix, represents the standard normalized hypergraph matrix, θ (1) and θ (2) They represent the trainable parameters of the first and second convolution layers respectively.

[0089] In addition, the hypergraph convolution process in the hypergraph convolutional neural network is a node-to-edge-to-node feature conversion process, which is very suitable for exploring the complex spatial co-occurrence characteristics between routing and network nodes, and can capture the nonlinear mapping relationship between network traffic and network performance.

[0090] In the hyperedge feature aggregation process of the hypergraph convolution network module 802, each hyperedge aggregates features from the connected nodes and forms a new hyperedge feature; then, after the node feature aggregation process, each node aggregates features from the involved hyperedges to form a new node feature. When traffic occurs in the network, each network node will converge and forward the traffic passing through the node according to the routing instructions. At the same time, the performance status of each end-to-end route will also be affected by the network node conditions contained. It can be seen that the hypergraph convolution process is consistent with the network traffic operation process, which is natively suitable for mining complex high-order spatial features in the routing hypergraph, and perfectly supports the calculation of delay features of end-to-end paths.

[0091] In addition, the model uses the mean square error loss function during the training process to reduce the difference between the real data and the predicted estimation results. The loss function is mathematically expressed as follows:

[0092]

[0093] Among them, Loss represents the loss function, Y represents the real data value, represents the predicted value, L reg is the L2 regular function, and λ represents the parameter.

[0094] In one embodiment, a network modeling open source dataset generated by simulation on OMNET++ based on NSFNet network topology is used for model training. The dataset mainly includes input configuration information such as network topology, routing configuration, and traffic distribution, as well as network performance indicator measurement values ​​such as delay and jitter generated by the simulation process.

[0095] The present invention uses a hypergraph to perform high-level modeling of spatial features such as network topology and routing strategies. The constructed routing hypergraph can fully represent the complex spatial correlation between network nodes and routing configurations. The prior art simply uses a simple graph to model network topology as a feature input for a network performance estimation model, but does not consider the impact of routing configuration information and the complex relationship between network nodes and routes on network performance.

[0096] The present disclosure uses a hypergraph neural network to explore and mine the rich features contained in the routing hypergraph, and outputs the estimated end-to-end network performance indicator value after modeling and learning; while the existing technology uses deep learning models such as codec neural networks and graph neural networks to learn spatial features in topology. Such models are limited by mathematical principles and model architectures and have relatively weak feature learning capabilities.

[0097] The present invention uses a hypergraph structure rather than a simple graph structure to perform high-level modeling of node connection relationships in cyberspace based on information such as routing configuration. The routing hypergraph formed contains rich and complex spatial feature information, and can perform high-level modeling of cyberspace information and routing information, thereby more effectively representing cyberspace features.

[0098] The present invention uses a hypergraph neural network instead of a graph neural network to mine the dynamic point-edge interaction information and complex spatial co-occurrence in the routing hypergraph, and outputs network performance estimation. It has stronger spatial feature learning capabilities and can achieve accurate network end-to-end performance estimation.

[0099] The present disclosure can be used for advanced service level agreements for application services. The use of hypergraph neural network models can help find routing solutions with low latency, high reliability, and low packet loss rate, improve QoS availability, and provide clear and quantifiable service guarantees for QoS-sensitive services such as real-time monitoring, video conferencing, and live video broadcasting. Network optimization configuration simulation verification. The configuration information such as the network topology, routing strategy, and traffic distribution to be adjusted is input into the trained network performance estimation model, and the calculated network operation quality can be used as a reference for prediction by operation and maintenance personnel. The present disclosure can enable operators to provide satisfactory dedicated line services to government and enterprise customers more quickly and efficiently; it enables operators to optimize and adjust the network structure more quickly and at a low cost to improve the quality of network operation.

[0100] Based on the same inventive concept, the present disclosure also provides a software-defined network performance prediction device, as described in the following embodiments. Since the principle of solving the problem in the device embodiment is similar to that in the above method embodiment, the implementation of the device embodiment can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.

[0101] Fig. 9 A schematic diagram of a software defined network performance prediction device according to an embodiment of the present disclosure is shown. Fig. 9 As shown, the device includes: a routing hypergraph configuration module 91, a matrix information configuration module 92, a performance prediction module 93, a target strategy information acquisition module 94, an information sending module 95 and a hypergraph neural network training module 96.

[0102] Among them, the routing hypergraph configuration module 91 is used to obtain the network topology information and routing configuration information of the network to be predicted, and determine the routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes the hypergraph adjacency matrix information of the network to be predicted; the matrix information configuration module 92 is used to obtain the network traffic information of the network to be predicted, and determine the end-to-end traffic matrix information in the network to be predicted based on the network traffic information; the performance prediction module 93 is used to input the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and output the performance prediction information of the network to be predicted.

[0103] In one embodiment of the present disclosure, the above-mentioned performance prediction module is also used to: input the hypergraph adjacency matrix information and the end-to-end traffic matrix information as input information into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information to obtain the hyperedge feature, and input the hyperedge feature into the activation layer and the linear output layer in turn for calculation and then output the end-to-end network performance indicator numerical matrix; determine the network performance indicator numerical matrix as the performance prediction information of the network to be predicted.

[0104] In one embodiment of the present disclosure, the above-mentioned software-defined network performance prediction device also includes a target policy information acquisition module 94, which is used to obtain target policy information of the network to be predicted, wherein the target policy information represents preset network topology information and routing configuration information; based on the target policy information of the network to be predicted, the network topology information and routing configuration information of the network to be predicted are determined.

[0105] In one embodiment of the present disclosure, the above-mentioned routing hypergraph configuration module is also used to: obtain network topology information and routing configuration information of the network to be predicted; according to the network topology information and routing configuration information, connect the network nodes in the network topology information and routing configuration information in a set manner, and determine the routes in the routing configuration information as hyperedges to form a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

[0106] In one embodiment of the present disclosure, the software-defined network performance prediction device further includes an information sending module 95 for sending the performance prediction information of the network to be predicted to the network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

[0107] In one embodiment of the present disclosure, the above-mentioned software-defined network performance prediction device also includes a hypergraph neural network training module 96, which is used to obtain a network modeling open source data set, wherein the network modeling open source data set includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated by the simulation process; the network modeling open source data set is used as a training set to perform model training on the hypergraph convolutional network to determine the hypergraph neural network.

[0108] In one embodiment of the present disclosure, the above-mentioned hypergraph neural network training module is also used to: use the mean square error loss function in the model training of the hypergraph convolutional network to reduce the difference between the real data and the predicted estimation results.

[0109] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0110] Refer to the following Fig.10 The electronic device 1000 according to this embodiment of the present disclosure is described. Fig.10 The electronic device 1000 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0111] like Fig.10 As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device 1000 may include but are not limited to: the at least one processing unit 1010, the at least one storage unit 1020, and a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010).

[0112] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 executes the steps described in the above “exemplary method” section of this specification according to various exemplary embodiments of the present disclosure.

[0113] For example, the processing unit 1010 can execute the following steps of the above method embodiment: obtain network topology information and routing configuration information of the network to be predicted, and determine the routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes the hypergraph adjacency matrix information of the network to be predicted; obtain network traffic information of the network to be predicted, and determine the end-to-end traffic matrix information in the network to be predicted based on the network traffic information; input the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and output the performance prediction information of the network to be predicted.

[0114] For example, the processing unit 1010 can execute the following steps of the above-mentioned method embodiment: use the hypergraph adjacency matrix information and the end-to-end traffic matrix information as input information, and input them into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information to obtain the hyperedge features, and input the hyperedge features into the activation layer and the linear output layer in turn for calculation and then output an end-to-end network performance indicator numerical matrix; determine the network performance indicator numerical matrix as the performance prediction information of the network to be predicted.

[0115] For example, the processing unit 1010 can execute the following steps of the above method embodiment: obtain target policy information of the network to be predicted, wherein the target policy information represents preset network topology information and routing configuration information; determine the network topology information and routing configuration information of the network to be predicted based on the target policy information of the network to be predicted.

[0116] For example, the processing unit 1010 can execute the following steps of the above method embodiment: obtain network topology information and routing configuration information of the network to be predicted; according to the network topology information and routing configuration information, connect the network nodes in the network topology information and routing configuration information in a set manner, and determine the routes in the routing configuration information as hyperedges to form a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

[0117] For example, the processing unit 1010 may execute the following steps of the above method embodiment: sending the performance prediction information of the network to be predicted to the network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

[0118] For example, the processing unit 1010 can execute the following steps of the above-mentioned method embodiment: obtain a network modeling open source data set, wherein the network modeling open source data set includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated by the simulation process; use the network modeling open source data set as a training set, perform model training on the hypergraph convolutional network, and determine the hypergraph neural network.

[0119] For example, the processing unit 1010 may perform the following steps of the above method embodiment: using a mean square error loss function in model training of the hypergraph convolutional network to reduce the difference between the real data and the predicted estimation results.

[0120] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 10201 and / or a cache storage unit 10202 , and may further include a read-only storage unit (ROM) 10203 .

[0121] The storage unit 1020 may also include a program / utility 10204 having a set (at least one) of program modules 10205, such program modules 10205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0122] Bus 1030 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0123] The electronic device 1000 may also communicate with one or more external devices 1040 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1000, and / or communicate with any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1060. As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0124] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0125] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, and the computer-readable storage medium may be a readable signal medium or a readable storage medium. Fig.11 A schematic diagram showing a computer-readable storage medium in an embodiment of the present disclosure is shown as follows: Fig.11 As shown, a program product capable of implementing the above method of the present disclosure is stored on the computer-readable storage medium 1100. In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary implementations of the present disclosure described in the above “Exemplary Method” section of this specification.

[0126] For example, when the program product in the embodiment of the present disclosure is executed by a processor, the following steps are implemented: obtaining network topology information and routing configuration information of the network to be predicted, and determining the routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes the hypergraph adjacency matrix information of the network to be predicted; obtaining network traffic information of the network to be predicted, and determining the end-to-end traffic matrix information in the network to be predicted based on the network traffic information; inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network, and outputting performance prediction information of the network to be predicted.

[0127] In some embodiments, the program product in the embodiments of the present disclosure implements the following steps when executed by a processor: inputting hypergraph adjacency matrix information and end-to-end traffic matrix information as input information into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information to obtain hyperedge features, and inputting the hyperedge features into the activation layer and the linear output layer in turn for calculation and then outputting an end-to-end network performance indicator numerical matrix; determining the network performance indicator numerical matrix as the performance prediction information of the network to be predicted.

[0128] In some embodiments, the program product in the embodiments of the present disclosure implements the following method steps when executed by a processor: obtaining target policy information of the network to be predicted, wherein the target policy information represents preset network topology information and routing configuration information; determining the network topology information and routing configuration information of the network to be predicted based on the target policy information of the network to be predicted.

[0129] In some embodiments, the program product in the embodiments of the present disclosure implements the following method steps when executed by a processor: obtaining network topology information and routing configuration information of the network to be predicted; connecting the network nodes in the network topology information and routing configuration information in a set manner based on the network topology information and routing configuration information, and determining the routes in the routing configuration information as hyperedges to form a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

[0130] In some embodiments, the program product in the embodiments of the present disclosure implements the following method when executed by a processor: sending performance prediction information of the network to be predicted to a network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

[0131] In some embodiments, the program product in the embodiments of the present disclosure implements the following steps when executed by a processor: obtaining a network modeling open source dataset, wherein the network modeling open source dataset includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated by a simulation process; using the network modeling open source dataset as a training set, performing model training on the hypergraph convolutional network, and determining the hypergraph neural network.

[0132] In some embodiments, the program product in the embodiments of the present disclosure implements the following method when executed by a processor: using a mean square error loss function in model training of a hypergraph convolutional network to reduce the difference between real data and predicted estimation results.

[0133] More specific examples of computer-readable storage media in the present disclosure may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0134] In the present disclosure, a computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0135] Alternatively, the program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0136] In a specific implementation, the program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0137] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0138] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0139] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0140] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.

Claims

1. A software-defined network performance prediction method, It is characterized in that include: Acquire network topology information and routing configuration information of the network to be predicted, and determine a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, wherein the routing hypergraph includes hypergraph adjacency matrix information of the network to be predicted, and the routing hypergraph is composed of each route in the routing mechanism being regarded as a hyperedge; Acquire network traffic information of the network to be predicted, and determine end-to-end traffic matrix information in the network to be predicted based on the network traffic information; The hypergraph adjacency matrix information and the end-to-end traffic matrix information are input as input information into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information to obtain hyperedge features, and the hyperedge features are sequentially input into the activation layer and the linear output layer for calculation, and then an end-to-end network performance indicator numerical matrix is ​​output; The network performance indicator value matrix is ​​determined as the performance prediction information of the network to be predicted.

2. The software defined network performance prediction method according to claim 1, It is characterized in that The method further comprises: Acquire target strategy information of the network to be predicted, wherein the target strategy information represents preset network topology information and routing configuration information; The network topology information and routing configuration information of the network to be predicted are determined according to the target strategy information of the network to be predicted.

3. The software defined network performance prediction method according to claim 1, It is characterized in that In obtaining network topology information and routing configuration information of the network to be predicted, and determining a routing hypergraph of the network to be predicted based on the network topology information and routing configuration information, the method further includes: Obtain network topology information and routing configuration information of the network to be predicted; According to the network topology information and routing configuration information, the network nodes in the network topology information and routing configuration information are connected in a set manner, and the routes in the routing configuration information are determined as hyperedges to form a routing hypergraph, wherein the routing hypergraph includes hypergraph adjacency matrix information with nodes as rows and hyperedges as columns.

4. The software defined network performance prediction method according to claim 1, It is characterized in that After inputting the hypergraph adjacency matrix information and the end-to-end traffic matrix information into a pre-trained hypergraph neural network and outputting performance prediction information of the network to be predicted, the method further includes: The performance prediction information of the network to be predicted is sent to a network node, so that the network node is configured according to the performance prediction information of the network to be predicted.

5. The software defined network performance prediction method according to claim 1, It is characterized in that The method further comprises: Acquire a network modeling open source data set, wherein the network modeling open source data set includes network topology information, routing configuration information, traffic distribution information, and delay information and jitter information generated during a simulation process; The network modeling open source dataset is used as a training set to perform model training on the hypergraph convolutional network to determine the hypergraph neural network.

6. The software defined network performance prediction method according to claim 5, It is characterized in that The method further comprises: The mean square error loss function is used in model training of the hypergraph convolutional network to reduce the difference between the real data and the predicted estimation results.

7. A software-defined network performance prediction device, It is characterized in that include: A routing hypergraph configuration module, used to obtain network topology information and routing configuration information of the network to be predicted, and determine the routing hypergraph of the network to be predicted according to the network topology information and routing configuration information, wherein the routing hypergraph includes the hypergraph adjacency matrix information of the network to be predicted, and the routing hypergraph is composed of each route in the routing mechanism being regarded as a hyperedge; A matrix information configuration module, used to obtain network traffic information of the network to be predicted, and determine end-to-end traffic matrix information in the network to be predicted based on the network traffic information; A performance prediction module is used to input the hypergraph adjacency matrix information and end-to-end traffic matrix information as input information into a pre-trained hypergraph neural network, so that the convolution layer of the hypergraph neural network performs hyperedge feature learning on the input information to obtain hyperedge features, and the hyperedge features are sequentially input into the activation layer and the linear output layer for calculation, and then an end-to-end network performance indicator numerical matrix is ​​output; the network performance indicator numerical matrix is ​​determined as the performance prediction information of the network to be predicted.

8. An electronic device, It is characterized in that include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the software-defined network performance prediction method according to any one of claims 1 to 6 by executing the executable instructions.

9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the software-defined network performance prediction method described in any one of claims 1 to 6 is implemented.