Network element management method and device of local area network, electronic equipment and storage medium

Abstract

The embodiment of the invention relates to a network element management method and device of a local area network, electronic equipment and a storage medium, and is applied to control equipment in the local area network, and the local area network further comprises a plurality of network elements. The method comprises the following steps: acquiring network element resource information respectively corresponding to a plurality of network elements and communication information respectively corresponding to the plurality of network elements; extracting features of the network element resource information corresponding to each network element to obtain resource features corresponding to each network element; generating a target feature matrix corresponding to the local area network according to the resource features corresponding to the network elements; generating a target adjacent matrix corresponding to the local area network according to the communication information corresponding to the plurality of network elements; based on a graph convolutional neural network, generating a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacent matrix; and managing the flow corresponding to each network element according to the resource matrix. According to the network element management method and device of the local area network, the control equipment and the storage medium, the intelligent level of network element management is improved.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a network element management method, device, electronic device and storage medium for a local area network. Background Art

[0002] Currently, network asset management, especially the management of network element resources in the enterprise's internal LAN, almost all relies on manual approval, manual management and manual investigation. Therefore, in the management of each network element in the LAN, manual judgment is dominant, the management efficiency is low, and subjective judgment errors are prone to occur. Summary of the invention

[0003] The embodiments of the present application disclose a network element management method, device, electronic device and storage medium for a local area network, which can improve the efficiency and accuracy of network element management to improve the intelligence level of network element management.

[0004] In a first aspect, an embodiment of the present application discloses a network element management method for a local area network, which is applied to a control device in the local area network, wherein the local area network further includes a plurality of network elements; the method includes:

[0005] Acquire network element resource information corresponding to the multiple network elements respectively, and communication information corresponding to the multiple network elements respectively; wherein the communication information corresponding to the first network element is used to indicate each other network element with which the first network element can communicate, and the first network element is any network element among the multiple network elements;

[0006] Extracting characteristics of network element resource information corresponding to each of the network elements to obtain resource characteristics corresponding to each of the network elements;

[0007] Generating a target feature matrix corresponding to the local area network according to resource features corresponding to each of the network elements;

[0008] Generate a target adjacency matrix corresponding to the local area network according to the communication information respectively corresponding to the multiple network elements;

[0009] Based on a graph convolutional neural network, generating a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix, wherein the resource matrix is ​​used to indicate the association relationship between the multiple network elements in the local area network;

[0010] The traffic corresponding to each of the network elements is managed according to the resource matrix.

[0011] As an optional implementation, in the first aspect of the embodiment of the present application, the communication information corresponding to the first network element includes the communication information of the first network element accessing the second network element, and the communication information of the first network element accessing the second network element includes the number of first ports through which the first network element can access the second network element, the number of second ports through which the second network element can be accessed by the first network element, and the communication protocol type supported by the first network element and the second network element; the second network element is any other network element with which the first network element can communicate; generating the target adjacency matrix corresponding to the local area network according to the communication information corresponding to the multiple network elements respectively, includes: generating the matrix elements corresponding to the first network element accessing the second network element according to the first port number, the second port number and the communication protocol type; generating the target adjacency matrix corresponding to the local area network according to the matrix elements corresponding to any one network element in the local area network accessing another network element.

[0012] As an optional implementation, in the first aspect of the embodiment of the present application, the graph convolutional neural network generates a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix, including: using the target feature matrix and the target adjacency matrix as the input matrix of the first layer of the graph convolutional neural network; encoding the input matrix of the current layer based on the graph convolution formula corresponding to the current layer to obtain the output matrix of the current layer; using the output matrix of the current layer and the target adjacency matrix as the input matrix of the next layer, and continuing to execute the step of encoding the input matrix of the current layer based on the graph convolution formula corresponding to the current layer to obtain the output matrix of the current layer, until the output matrix of the last layer is obtained; using the output matrix of the last layer as the resource matrix corresponding to the local area network.

[0013] As an optional implementation, in the first aspect of the embodiment of the present application, the training process of the graph convolutional neural network includes: obtaining a sample set and constructing a pre-trained graph convolutional neural network according to preset parameters; the sample set includes multiple sample feature matrices, respectively corresponding sample adjacency matrices and respectively corresponding target resource matrices; based on the pre-trained graph convolutional neural network, generating a sample resource matrix corresponding to the first sample feature matrix according to the first sample feature matrix and the corresponding sample adjacency matrix; the first sample feature matrix is ​​any sample feature matrix of the sample set; according to the sample resource matrix corresponding to each of the sample feature matrices and the corresponding target resource matrix, calculating the loss error corresponding to the pre-trained graph convolutional neural network; if the loss error is greater than the preset loss value, adjusting the graph convolution formula corresponding to each layer in the pre-trained graph convolutional neural network to obtain a new pre-trained graph convolutional neural network; repeatedly executing the step of generating a sample resource matrix corresponding to the first sample feature matrix according to the first sample feature matrix and the corresponding sample adjacency matrix based on the pre-trained graph convolutional neural network, until the loss error is less than or equal to the preset loss value, and obtaining a trained graph convolutional neural network.

[0014] As an optional implementation, in a first aspect of an embodiment of the present application, the resource matrix includes network element vectors corresponding to the multiple network elements respectively, and the network element vector corresponding to the first network element is used to indicate the association relationship between the first network element and each network element in the local area network; the traffic corresponding to each network element is managed according to the resource matrix, including: receiving an access request sent by the first network element, the access request including network element information corresponding to a third network element to be accessed by the first network element; the third network element is any network element in the local area network that is different from the first network element; according to the access request, determining from the resource matrix a first network element vector corresponding to the first network element and a second network element vector corresponding to the third network element; calculating the vector distance between the first network element vector and the second network element vector to obtain a first distance; if the first distance is less than or equal to a preset distance threshold, forwarding the access request to the third network element.

[0015] As an optional implementation, in the first aspect of the embodiment of the present application, the traffic corresponding to each of the network elements is managed according to the resource matrix, including: if the network element resource information sent by the first network element is not received within a preset time length, determining at least one accessible network element that can access the first network element and at least one visited network element that can be accessed by the first network element according to the resource matrix; adjusting the traffic transmission direction between the at least one accessible network element and the at least one visited network element so that the at least one accessible network element can access the at least one visited network element.

[0016] As an optional implementation, in the first aspect of the embodiment of the present application, the network element resource information corresponding to the first network element includes multiple target network element resources corresponding to the first network element; the extraction of characteristics of the network element resource information corresponding to each of the network elements to obtain the resource characteristics corresponding to each of the network elements includes: according to the resource mapping table, determining the mapping data corresponding to the multiple target network element resources corresponding to the first network element respectively to obtain the resource characteristics corresponding to the first network element; the resource mapping table contains each network element resource and the corresponding mapping value.

[0017] In a second aspect, an embodiment of the present application discloses a network element management device for a local area network, which is applied to a control device in the local area network, wherein the local area network further includes a plurality of network elements; the device includes:

[0018] An information acquisition module, used to acquire network element resource information corresponding to the multiple network elements respectively, and communication information corresponding to the multiple network elements respectively; wherein the communication information corresponding to the first network element is used to indicate each other network element with which the first network element can communicate, and the first network element is any network element among the multiple network elements;

[0019] A feature extraction module, used to extract features of network element resource information corresponding to each network element, and obtain resource features corresponding to each network element;

[0020] A matrix generation module, used to generate a target feature matrix corresponding to the local area network according to resource features corresponding to each of the network elements;

[0021] The matrix generation module is further used to generate a target adjacency matrix corresponding to the local area network according to the communication information respectively corresponding to the multiple network elements;

[0022] A graph convolution module, configured to generate a resource matrix corresponding to the local area network based on a graph convolutional neural network and according to the target feature matrix and the target adjacency matrix, wherein the resource matrix is ​​used to indicate an association relationship between the multiple network elements in the local area network;

[0023] The network element management module is used to manage the traffic corresponding to each of the network elements according to the resource matrix.

[0024] In a third aspect, an embodiment of the present application discloses an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor implements the method described in any of the above embodiments.

[0025] In a fourth aspect, an embodiment of the present application discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any of the above embodiments is implemented.

[0026] In a fifth aspect, an embodiment of the present application discloses a computer program product, including a computer program, and when the computer program is executed by a processor, the method described in any of the above embodiments is implemented.

[0027] The network element management method, device, control device and storage medium of the local area network disclosed in the embodiments of the present application obtain network element resource information corresponding to multiple network elements, and communication information corresponding to multiple network elements; wherein the communication information corresponding to the first network element is used to indicate each other network element with which the first network element can communicate, and the first network element is any network element among the multiple network elements; the characteristics of the network element resource information corresponding to each network element are extracted to obtain the resource characteristics corresponding to each network element; according to the resource characteristics corresponding to each network element, a target feature matrix corresponding to the local area network is generated; according to the communication information corresponding to the multiple network elements, a target adjacency matrix corresponding to the local area network is generated; based on the graph convolutional neural network, a resource matrix corresponding to the local area network is generated according to the target feature matrix and the target adjacency matrix, and the resource matrix is ​​used to indicate the association relationship between multiple network elements in the local area network; according to the resource matrix, the traffic corresponding to each network element is managed. In an embodiment of the present application, a control device in a local area network generates a target feature matrix by extracting resource features of network element resource information of each network element, and generates a target adjacency matrix according to the communication information corresponding to each network element, and processes the target feature matrix and the target adjacency matrix through a graph convolutional neural network to obtain a resource matrix. Since the target feature matrix and the target adjacency matrix can accurately characterize the resource status of each network element in the local area network and the communication relationship between different network elements, the obtained resource matrix can accurately reflect the association relationship between different network elements in the local area network. Based on the resource matrix, the control device can more accurately manage the traffic of each network element in the local area network according to the association relationship between different network elements in the local area network, thereby improving the efficiency and accuracy of network element management and improving the intelligence level of network element management. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1 An application scenario diagram of a network element management method for a local area network in one embodiment;

[0030] Figure 2 is a flow chart of a network element management method for a local area network in one embodiment;

[0031] Figure 3 A flowchart for generating a target adjacency matrix in one embodiment;

[0032] Figure 4 A flowchart for generating a resource matrix in one embodiment;

[0033] Figure 5 A training flow chart of a graph convolutional neural network in one embodiment;

[0034] Figure 6 is a flow chart of network element management in one embodiment;

[0035] Figure 7 is a flow chart of network element management in one embodiment;

[0036] Figure 8 is a block diagram of a network element management device for a local area network in one embodiment;

[0037] Fig. 9 FIG. 4 is a structural block diagram of a control device in an embodiment. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] It should be noted that the terms "including" and "having" and any variations thereof in the embodiments of the present application and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0040] It is understood that the terms "first", "second", etc. used in this application can be used in this article to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first partial sub-energy can be referred to as a second partial sub-energy, and similarly, a second partial sub-energy can be referred to as a first partial sub-energy. Both the first partial sub-energy and the second partial sub-energy contain at least one sub-energy, but they are not the same sub-energy.

[0041] In related technologies, the information of each network element is manually identified, and the status of each network element and the communication relationship with other network elements also need to be manually managed. Specifically, traffic management, policy approval, network communication, etc. all rely on manual judgment, and manual inspections of the network are required regularly or irregularly to discover potential problems. There is a lack of automated management methods, and it is impossible to respond to various emergencies and perform automated operations in a timely manner.

[0042] The embodiments of the present application disclose a network element management method, device, electronic device and storage medium for a local area network, which can improve the efficiency and accuracy of network element management and improve the intelligence level of network element management.

[0043] The following is an introduction to some of the terms involved in this application:

[0044] Local Area Network: In this application, a local area network, also known as an intranet, refers to a communication network that connects various computers, peripherals, databases and other devices within a preset range (such as an office, a building or a floor). For example, the internal network of an organization or enterprise.

[0045] Network element: refers to each network device and server in a local area network, which may include but is not limited to physical network elements (such as switches, routers, base stations, etc.) and virtual network elements (such as virtual machines, containers, etc.).

[0046] NE resource information: refers to detailed information about the NE itself, such as physical information and model parameters.

[0047] The communication information corresponding to the network element is used to describe the communication relationship between the network element and other network elements that can communicate in the local area network.

[0048] The target feature matrix contains the resource features of each network element in the LAN. Each row in the target feature matrix can represent a network element, and each column can represent a resource feature (such as CPU usage, memory usage, disk usage, etc.). The target feature matrix may include N*M elements, where N represents the number of network elements in the LAN, and M represents the number of resource features of each network element. For example, the element x in the target feature matrix 12 A second network element resource characteristic of a first network element in a local area network may be represented.

[0049] The target adjacency matrix is ​​used to characterize the communication relationship between different network elements in the local area network. The target adjacency matrix may include N*N elements, each of which may indicate whether a network element can access another network element, or the degree of access of a network element to another network element. For example, the element y in the target adjacency matrix is 12 A value of 1 indicates that the first network element in the LAN can access the second network element.

[0050] The resource matrix is ​​used to indicate the relationship between multiple network elements in the local area network. The resource matrix integrates the network element resource information and the corresponding communication information of each network element. Each element can represent resource characteristics or communication information. For example, the element z of the resource matrix 12 It can indicate the strength of association between a first network element and a second network element in a local area network in terms of specific resource characteristics or communication information.

[0051] Graph convolutional neural network: refers to a neural network model used to process graph data. Its core idea is to generalize the convolution operation from traditional data (such as images or grids) to graph data. In this application, the graph convolutional neural network can learn a graph convolution formula that can aggregate the features of the node (network element) itself and the features of the adjacent nodes in the graph data, thereby generating a relational representation of the node.

[0052] Figure 1 The network element management analysis method for a local area network provided in this application can be applied to the following scenarios: Figure 1 In the control device 100 shown. Figure 1 As shown, in the local area network where the control device 100 is located, the control device 100 can be connected to any network element 110 in the local area network. The control device 100 may include but is not limited to an intelligent terminal device, a tablet computer, a PC (Personal Computer), a computer, a laptop computer, a server or a server cluster, etc. Any network element 110 can communicate with other network elements 110 only through the control device 100.

[0053] It should be noted that each network element 110 connected to the local area network is managed by the local area network control device 100. The management includes access management, traffic management, security management and other aspects.

[0054] Access management is used to control whether each network element 110 in the local area network can access another network element 110. Through access management, it can be ensured that only network elements 110 authorized by the control device 100 can access specific network resources or services, thereby preventing unauthorized access and data leakage.

[0055] Traffic management is used to control the uplink speed and downlink speed of the network element 110 to obtain network services, as well as the allocation of network resources. It also involves the feasibility management of network element communication and the management of network element data packet transmission. Traffic management may include bandwidth limitation, traffic priority setting, traffic monitoring, etc. Through traffic management, it is possible to ensure that the network resources of the local area network are reasonably allocated to avoid network congestion and performance degradation.

[0056] Security management is used to detect whether the network element 110 is safe, so as to decide whether to provide the network element 110 with the network service of the local area network. Through security management, security risks such as network attacks and virus invasions can be discovered and dealt with in a timely manner, thus protecting the security of the network and data.

[0057] The control device 100 can obtain network element resource information from each network element 110, and obtain communication information corresponding to each network element 110 from each network element 110 or the control device 100 itself. The control device 100 extracts the characteristics of the network element resource information corresponding to each network element 110, and obtains the resource characteristics corresponding to each network element 110; and generates a target feature matrix corresponding to the local area network according to the resource characteristics corresponding to each network element 110. The control device 100 can generate a target adjacency matrix corresponding to the local area network according to the communication information corresponding to multiple network elements 110 respectively; and then based on the graph convolutional neural network, generate a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix. The control device 100 manages the traffic corresponding to each network element 110 according to the resource matrix to improve the efficiency and accuracy of network element management, so as to improve the intelligent level of the control device 100 to manage the network element 110.

[0058] like Figure 2 As shown, in one embodiment, a method for managing network elements of a local area network is provided, which can be applied to the above control device. The method may include the following steps:

[0059] Step 210: Acquire network element resource information corresponding to a plurality of network elements, and communication information corresponding to a plurality of network elements.

[0060] In some embodiments, the first network element is any network element among multiple network elements in the local area network.

[0061] In some embodiments, the network element resource information corresponding to the first network element may include multiple target network element resources corresponding to the first network element. The target network element resources refer to various information that can describe and define the characteristics and capabilities of the network element. Specifically, the multiple target network element resources of the network element resource information may include but are not limited to the model, serial number, hardware configuration (such as CPU, memory, disk, etc.), software version, interface type, etc. of the network element. For example, the network element resource information of a router may include its brand, model, CPU model, memory size, network interface type and quantity, etc.

[0062] In some embodiments, the communication information corresponding to the first network element may include, but is not limited to, one or more of the first network element's own IP address, port number, number of ports, MAC address, routing table, access control list, etc., and may also include one or more of the IP address (Internet Protocol Address), MAC (Media Access Control Address) address, port number, number of ports, routing table, access control list, etc. of other network elements with which the first network element can communicate. Among them, the routing table refers to a database or table structure in the network element used to determine how a data packet is transmitted from a source address to a destination address through the network; the access control list refers to a network security policy tool used to define which data packets can be received and which data packets are denied access; other network elements refer to any network element in the local area network that is different from the first network element.

[0063] Optionally, the communication information corresponding to the first network element includes the communication information of the first network element accessing the second network element, and the communication information of the first network element accessing the second network element includes the number of first ports that the first network element can access the second network element, the number of second ports that the first network element can access, and the types of communication protocols supported by the first network element and the second network element; the first network element is any network element in the local area network; the second network element is any other network element that the first network element can communicate with. For example, the communication information of network element A accessing network element B includes that the first port that network element A can access network element B is port a, the second port that network element B can access by network element A is port c, and the supported communication protocol type is TCP (Transmission Control Protocol), which means that port a of network element A can access port c of network element B through the CAN bus protocol and get a response.

[0064] In some embodiments, each network element in the local area network may send network element resource information to the control device at a first time interval. Optionally, the first time interval may be set according to the network service of the local area network. For example, when the network service changes frequently, the first time interval may be set to be shorter; and when the network service is relatively stable or the resource changes little, the first time interval may be set to be longer.

[0065] Optionally, the control device may send a resource acquisition request to each website, and each network element, upon receiving the resource acquisition request, sends network element resource information to the control device. The resource acquisition request is used to request each network element to obtain network element resource information. Specifically, the resource acquisition request may include specific network element resources of the network element resource information that the control device needs to obtain, and the network element may send the network element resource information required by the control device to the control device according to the resource acquisition request.

[0066] In some embodiments, the control device may obtain a preset permission table, and obtain communication information corresponding to each network element currently connected to the local area network from the preset permission table; the preset permission table is used to indicate the access permission relationship between each network element, and may include which other network elements a network element can access, or which other network elements can access the network element. Optionally, the preset permission table may be updated in real time.

[0067] In some embodiments, the control device may determine the communication level corresponding to each network element and the network element resource information corresponding to each network element, wherein each network element may access other network elements whose communication level is not higher than that of the network element. Optionally, the communication level corresponding to each network element may be determined based on factors such as the hardware configuration and software version of the network element.

[0068] Through the various methods for obtaining network element resource information introduced in the above embodiments, it is possible to accurately and flexibly obtain network element resource information corresponding to multiple network elements and communication information corresponding to multiple network elements, providing basic and accurate information for subsequent management. It should be noted that other methods can also be used to obtain network element resource information of each network element, and are not limited to the above methods.

[0069] Step 220: extract the characteristics of the network element resource information corresponding to each network element to obtain the resource characteristics corresponding to each network element.

[0070] The resource characteristics refer to key attributes or characteristics that can summarize and describe the network element resource information, so that the performance and status of the corresponding network element can be accurately understood through the resource characteristics. Optionally, the resource characteristics may include multiple target network element resources corresponding to the first network element and / or corresponding specific values.

[0071] Specifically, resource characteristics may include hardware characteristics (such as the CPU type, quantity, main frequency, memory size, disk type or capacity of the network element, etc.), software characteristics (such as the operating system version running the network element, supported network protocols, software modules and applications, etc.), network characteristics (such as the network element's IP address, MAC address, subnet mask, gateway address, DNS server address and other network configuration information), and security characteristics (such as the network element's firewall settings, encryption technology, identity authentication mechanism, security policy, etc.).

[0072] In some embodiments, the control device may determine the mapping data corresponding to the multiple target network element resources corresponding to the first network element according to the resource mapping table, so as to obtain the resource characteristics corresponding to the first network element; the resource mapping table includes each network element resource and the corresponding mapping value. Mapping the network element resources into resource characteristics in numerical form can help the control device better understand and manage the network element resources.

[0073] Optionally, for network element resources that can be converted into numerical values, the control device can directly convert the network element resources into numerical values; for network element resources that cannot be directly converted into numerical values, the control device can determine the numerical value mapped to the network element resource through the resource mapping table. For example, the IP address of the network element can be converted into int format (for example, the IPv4 address "192.168.1.1" can be converted into "3232301153"); for the communication protocol, the mapping value corresponding to TCP can be 1, and the mapping value corresponding to UDP (User Datagram Protocol) can be 2.

[0074] Step 230: Generate a target feature matrix corresponding to the local area network according to the resource features corresponding to each network element.

[0075] In some embodiments, the control device assigns a corresponding number to each network element when each network element accesses the local area network. The number corresponding to the network element indicates the order in which the network element joins the local area network. After obtaining the resource characteristics corresponding to each network element, the control device can generate a target characteristic matrix corresponding to the local area network according to the number and resource characteristics corresponding to each network element. Furthermore, the number of rows of the resource characteristics of each network element in the target characteristic matrix can be matched with the number corresponding to the network element. For example, if the number corresponding to network element A is 2, then each resource characteristic of network element A is in the second row of the target characteristic matrix corresponding to the local area network.

[0076] In some embodiments, the control device constructs a feature vector of each network element according to the resource features corresponding to each of the N network elements in the local area network. For example, the feature vector of the i-th network element is X i ={x i1 ,x i2 ,...,x iD}, i = 1, 2, ..., N, where N is a positive integer, D represents the number of resource features corresponding to each network element, x i1 Indicates the first resource feature of the i-th network element.

[0077] Optionally, the control device may construct a target feature matrix corresponding to the local area network based on the feature vectors of each network element. Among them, X N Represents the feature vector of the Nth network element.

[0078] Step 240: Generate a target adjacency matrix corresponding to the local area network according to the communication information corresponding to the plurality of network elements.

[0079] Specifically, if there are N network elements in the local area network, the target adjacency matrix corresponding to the local area network can be Among them, a 12 Indicates that the first NE in the LAN can access the second NE.

[0080] It should be noted that, since the network element cannot access the local network element, the diagonal of the target connection matrix corresponding to the local area network is 0.

[0081] In some embodiments, the control device may use a numerical value in the target adjacency matrix to indicate whether any network element can access other network elements. For example, a numerical value of 1 is used to indicate that the horizontal axis network element can access the vertical axis network element, and a numerical value of 0 is used to indicate that the horizontal axis network element cannot access the vertical axis network element. For example, a target adjacency matrix containing 5 network elements (network element 1, network element 2, network element 3, network element 4 and network element 5), assuming that network element 1 can only access network element 2 and network element 3, then the elements of the 1st row and 2nd column and the 1st row and 3rd column in the target adjacency matrix are 1, and the other elements of the 1st row are 0.

[0082] It should be noted that the fact that network element 1 can access network element 2 does not mean that network element 2 can access network element 1. The control device needs to obtain the communication information corresponding to network element 1 and / or the communication information corresponding to network element 2 to determine whether network element 2 can access network element 1.

[0083] Specifically, assume that there are four network elements (network element 1, network element 2, network element 3 and network element 4) in the local area network, among which network element 1 can access network element 2 and network element 3, network element 2 can access network element 1 and network element 4, network element 3 can access network element 1, and network element 4 can access network element 2. Therefore, the target adjacency matrix corresponding to the local area network can be

[0084] Step 250, based on the graph convolutional neural network, generates a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix.

[0085] In a local area network, the control device can use network elements as nodes and determine directed edges based on communication information (for example, network element 1 can access network element 2, indicating that there is a directed edge from the node of network element 1 to the node of network element 2), thereby forming a directed graph structure corresponding to the local area network. Through the graph convolutional neural network, the network element resource information contained in the target feature matrix and the communication information between network elements contained in the target adjacency matrix can be integrated to generate a resource matrix that reflects the association relationship between multiple network elements in the local area network.

[0086] Step 260: manage the traffic corresponding to each network element according to the resource matrix.

[0087] The resource matrix includes network element vectors corresponding to multiple network elements, and the network element vector corresponding to the first network element is used to indicate the association relationship between the first network element and each network element in the local area network. Specifically, the network element vector may include information of multiple dimensions such as the processing capability, storage capacity, network bandwidth, and geographical location of the network element.

[0088] It should be noted that, in the resource matrix, each element on the same horizontal axis or vertical axis represents an element in a network element vector corresponding to the same network element.

[0089] In some embodiments, the resource matrix may include a target flow threshold for each network element in the local area network, and the target flow threshold may include the average flow or maximum flow within a preset time period; in the local area network, the control device may obtain the real-time flow corresponding to each network element. If it is detected that the real-time flow corresponding to the first network element is greater than the target flow threshold corresponding to the first network element in the resource matrix, the control device may determine that the flow of the first network element is abnormal, and does not allow the first network element to obtain the network service of the local area network. For example, the network element vector corresponding to network element 3 in the resource matrix is ​​X3 = {0.1, 0.2, 10, 5, 0.1}, where the fourth element is the target flow threshold. If the control device detects that the real-time flow of network element 3 is greater than 5, it is considered that the flow of network element 3 is abnormal. This effectively monitors and manages the network flow of each network element in the local area network, ensuring that the network resources of the local area network can be reasonably allocated and used.

[0090] Optionally, when the control device determines that the traffic of the first network element is abnormal, an abnormal warning may be sent to the first network element, and the abnormal warning is used to remind the first network element that the real-time traffic is too high, which helps to promptly remind the administrator to pay attention to potential network problems and take corresponding measures to solve them.

[0091] Optionally, the control device may also detect the number of abnormal warnings sent to the first network element after sending an abnormal warning to the first network element and receiving a network acquisition request from the first network element, wherein the network acquisition request is a request from the network element to the control device to obtain the network service of the local area network; if the number is less than a preset number, the first network element is allowed to reacquire the network service of the local area network; if the number is greater than or equal to the preset number, the first network element is not allowed to reacquire the network service of the local area network. For example, assuming that the preset number is 5, the terminal device detects the network acquisition request of network element 4, and detects that 6 abnormal warnings have been sent to network element 4, then the control device does not allow network element 4 to reacquire the network service of the local area network.

[0092] In some embodiments, if the control device detects that the network element resource information corresponding to any network element changes, or detects that a new network element is connected to the local area network, or receives information that any network element leaves the local area network, or every second time interval, the control device recalculates a new resource matrix. By updating the resource matrix, the control device can more accurately understand the network element resources of each network element to ensure continuous monitoring and management of each network element.

[0093] In an embodiment of the present application, a control device in a local area network generates a target feature matrix by extracting resource features of network element resource information of each network element, and generates a target adjacency matrix according to the communication information corresponding to each network element, and processes the target feature matrix and the target adjacency matrix through a graph convolutional neural network to obtain a resource matrix. Since the target feature matrix and the target adjacency matrix can accurately characterize the resource status of each network element in the local area network and the communication relationship between different network elements, the obtained resource matrix can accurately reflect the association relationship between different network elements in the local area network. Based on the resource matrix, the control device can more accurately manage the traffic of each network element in the local area network according to the association relationship between different network elements in the local area network, thereby improving the efficiency and accuracy of network element management, and at the same time improving the intelligence level of network element management.

[0094] like Figure 3 As shown, in some embodiments, the step of generating a target adjacency matrix corresponding to the local area network according to communication information corresponding to a plurality of network elements may include the following steps 302 to 304.

[0095] Step 302: Generate a matrix element corresponding to the first network element accessing the second network element according to the first port quantity, the second port quantity and the communication protocol type.

[0096] The first port number refers to the number of ports on the first network element that can be used to access the second network element. The second port number refers to the number of ports on the second network element that can be used to be accessed by the first network element. The communication protocol type refers to the type of protocol used by the first network element to communicate with the second network element (e.g., TCP, UDP, HTTP (HyperText Transfer Protocol, Hypertext Transfer Protocol), etc.).

[0097] In some embodiments, the control device may generate a matrix element corresponding to the first network element accessing the second network element based on an element mapping table according to a first quantity interval to which the first port quantity belongs, a second quantity interval to which the second port quantity belongs, and a third quantity interval to which the type quantity of the communication protocol type belongs; the element mapping table includes each first quantity interval, each second quantity interval, each third quantity interval and a commonly corresponding preset matrix element. For example, the protocol types used by the first network element to access the second network element for communication include TCP and UDP, and the type quantity of the communication protocol type is 2.

[0098] Exemplarily, the preset first number intervals include interval 1, interval 2, and interval 3; the preset second number intervals include interval A, interval B, and interval C; the preset third number intervals include interval a, interval b, and interval c. Assuming that the communication information of the first network element accessing the second network element may include that the first number interval is interval 2, the second number interval is interval A, and the third number interval is interval a, the control device may determine the preset matrix element corresponding to interval 2, interval A, and interval a as the matrix element corresponding to the first network element accessing the second network element.

[0099] In some embodiments, the control device can determine the access degree corresponding to the first network element accessing the second network element according to the number of first ports, the number of second ports and the type of communication protocol, and use the access degree as the matrix element corresponding to the first network element accessing the second network element; the access degree is used to indicate the access capability of the first network element to the second network element. Specifically, the greater the number of first ports and the number of second ports, the greater the corresponding access degree; the greater the communication protocol type is within the preset protocol type range, the greater the corresponding access degree.

[0100] Exemplarily, if any port of network element 1 does not access any port of network element 2 through any protocol, the control device may consider that the matrix element corresponding to network element 1 accessing network element 2 is 0. If all ports of network element 1 can access port 80 of network element 2 through TCP protocol, the control device may consider that the matrix element corresponding to network element 1 accessing network element 2 is 0.02. If all ports of network element 1 can access port 80 of network element 2 through UDP protocol, the control device may consider that the matrix element corresponding to network element 1 accessing network element 2 is 0.01. If all ports of network element 1 can access ports 443, 22 and 80 of network element 2 through TCP protocol, the control device may consider that the matrix element corresponding to network element 1 accessing network element 2 is 0.3. If all ports of network element 1 can access any port of network element 2 through any protocol, the control device may consider that the matrix element corresponding to network element 1 accessing network element 2 is 1.

[0101] Step 304: Generate a target adjacency matrix corresponding to the local area network according to a matrix element corresponding to any network element accessing another network element in the local area network.

[0102] In some embodiments, the control device may fill the matrix elements corresponding to each network element accessing another network element into an N×N matrix in sequence to obtain a target adjacency matrix corresponding to the local area network. N is the number of network elements included in the local area network. Specifically, the same horizontal axis of the target adjacency matrix may represent the communication information of the same network element accessing other network elements, and the same vertical axis may represent the communication information of the same network element being accessed.

[0103] For example, there are three network elements (network element 1, network element 2 and network element 3) in the local area network, among which the matrix element corresponding to network element 1 accessing network element 2 is 0.1, and the matrix element corresponding to network element 3 is 0.05; the matrix element corresponding to network element 2 accessing network element 1 is 0, and the matrix element corresponding to network element 3 is 0.6; the matrix element corresponding to network element 3 accessing network element 1 is 1, and the matrix element corresponding to network element 2 is 0.02; therefore, the target adjacency matrix corresponding to the local area network can be

[0104] In an embodiment of the present application, by quantifying the communication information and using it as a matrix element, a more refined matrix element than the traditional binary (0 / 1) can be obtained, and the matrix elements are determined by comprehensively considering the number of first ports, the number of second ports and the type of communication protocol. The target adjacency matrix thus obtained can better reflect the communication information of each network element in the local area network accessing another network element.

[0105] like Figure 4 As shown, in some embodiments, the steps are based on a graph convolutional neural network, and a resource matrix corresponding to the local area network is generated according to a target feature matrix and a target adjacency matrix, and may also include the following steps 402 to 410.

[0106] Step 402: Use the target feature matrix and the target adjacency matrix as the input matrix of the first layer of the graph convolutional neural network.

[0107] Step 404, based on the graph convolution formula corresponding to the current layer, encode the input matrix of the current layer to obtain the output matrix of the current layer.

[0108] In some embodiments, the propagation formula of GCN (Graph Convolutional Networks) is Among them, A is the target adjacency matrix, is the sum of the target adjacency matrix and the identity matrix, For the matrix The degree matrix of H (l) is the feature matrix of the lth layer, σ(X) is a nonlinear activation function, which can include Softmax function, ReLU function, etc., W (l) is the parameter of the lth layer, that is, the weight multiplied by the feature matrix, also known as the weight matrix. When l is 1, H (l) is the target feature matrix X.

[0109] Specifically, the control device can obtain the graph convolution formula corresponding to each layer according to the propagation formula of GCN and the weight parameters corresponding to each layer.

[0110] Step 406, determine whether the current layer is the last layer. If yes, execute step 408; if no, execute step 410.

[0111] Step 408: Use the output matrix of the last layer as the resource matrix corresponding to the local area network.

[0112] If the control device detects that the current layer is the last layer, the output matrix of the last layer is used as the resource matrix corresponding to the local area network. The output matrix of the last layer has been processed by a multi-layer graph convolutional network, integrating the characteristic information and connection relationship (communication information) of the network element, so it can be directly used as a resource matrix.

[0113] Step 410 , using the output matrix of the current layer and the target adjacency matrix as the input matrix of the next layer, and executing step 404 .

[0114] In some embodiments, the control device uses the output matrix and the target adjacency matrix of the current layer as the input matrix of the next layer, and when it is determined that the current layer is not the last layer, the encoding process of step 404 can be re-executed until the current layer is the last layer.

[0115] In an embodiment of the present application, the control device gradually extracts and fuses the resource characteristics and communication information of each network element in the local area network through a multi-layer graph convolutional network, and finally more accurately generates a resource matrix that can reflect the network element resource distribution of each network element in the local area network and the communication relationship between each network element, which is conducive to the subsequent correct management of each network element in the local area network according to the resource matrix, thereby improving the efficiency and accuracy of network element management.

[0116] like Figure 5 As shown, in some embodiments, the training process of the graph convolutional neural network may include the following steps 502 to 512.

[0117] Step 502, obtain a sample set and construct a pre-trained graph convolutional neural network according to preset parameters.

[0118] The sample set includes multiple sample feature matrices, corresponding sample adjacency matrices and corresponding target resource matrices. The target resource matrix is ​​an actual resource matrix that can be obtained by the control device when the sample feature matrix and the corresponding sample adjacency matrix are obtained.

[0119] The preset parameters are used to define the structure of the neural network, training parameters and other related network components. The mapping parameters may include but are not limited to the number of neural network layers, the initial weight matrix corresponding to each layer, the feature dimension, the learning rate, the weight decay (regularization term), the optimizer type, the loss function type, etc.

[0120] In some embodiments, the control device can construct a pre-trained graph convolutional neural network based on the number of neural network layers in preset parameters and the initial weight matrix corresponding to each layer.

[0121] In some embodiments, the control device can obtain sample feature matrices corresponding to multiple preset historical time periods based on the historical network element resource information corresponding to each network element in multiple preset historical time periods; and obtain sample adjacency matrices corresponding to multiple preset historical time periods based on the historical communication information corresponding to each network element in multiple preset historical time periods; and obtain target resource matrices corresponding to multiple preset historical time periods.

[0122] For example, the control device can obtain the sample feature matrix corresponding to historical time A based on the historical network element resource information corresponding to multiple network elements contained in the local area network at historical time A; obtain the sample adjacency matrix corresponding to historical time A based on the historical communication information corresponding to multiple network elements contained in the local area network at historical time A; and obtain the target resource matrix corresponding to historical time A.

[0123] Step 504: Based on the pre-trained graph convolutional neural network, a sample resource matrix corresponding to the first sample feature matrix is ​​generated according to the first sample feature matrix and the corresponding sample adjacency matrix.

[0124] The first sample feature matrix is ​​any sample feature matrix in the sample set. The sample resource matrix refers to a predicted resource matrix obtained by the control device through a pre-trained graph convolutional neural network when the sample feature matrix and the corresponding sample adjacency matrix are obtained.

[0125] In some embodiments, the control device may use the first sample feature matrix and the corresponding sample adjacency matrix as the first layer input matrix of a pre-trained graph convolutional neural network; and based on the pre-trained graph convolutional neural network, encode the input matrix of the current layer to obtain the output matrix of the current layer, until the output matrix of the last layer is obtained as the sample resource matrix corresponding to the first sample feature matrix.

[0126] Step 506, calculate the loss error corresponding to the pre-trained graph convolutional neural network according to the sample resource matrix and the corresponding target resource matrix corresponding to each sample feature matrix.

[0127] In some embodiments, in order to evaluate the performance of a pre-trained graph convolutional neural network and thus optimize the graph convolutional neural network, the control device may calculate the loss error of the pre-trained graph convolutional neural network to evaluate the performance and thereby adjust the parameters of the neural network.

[0128] Optionally, the loss error calculation methods that can be used by the control device include but are not limited to mean square error loss, cross entropy loss, L1 norm loss, triple loss, etc.

[0129] Specifically, taking the cross entropy loss as an example, the control device can calculate the loss error corresponding to the pre-trained graph convolutional neural network based on the cross entropy loss function, according to the sample resource matrix corresponding to each sample feature matrix and the corresponding target resource matrix.

[0130] For example, the cross entropy loss function is Among them, Y is the sample resource matrix, is the target resource matrix, N is the number of network elements in the LAN, M is the feature dimension in the preset parameters, y ij is the element in the i-th row and j-th column of the sample resource matrix, is the element in the i-th row and j-th column of the target resource matrix.

[0131] Step 508, determine whether the loss error is greater than a preset loss value. If yes, go to step 510; if no, go to step 512.

[0132] Step 510: adjust the graph convolution formula corresponding to each layer in the pre-trained graph convolutional neural network to obtain a new pre-trained graph convolutional neural network. Repeat step 504.

[0133] If the control device detects that the loss error is greater than the preset loss value, it means that the current pre-trained graph convolutional neural network cannot fuse the feature matrix and the adjacency matrix well. Therefore, the control device can adjust the graph convolution formula corresponding to each layer in the pre-trained graph convolutional neural network to obtain a new pre-trained graph convolutional neural network, and based on the new pre-trained graph convolutional neural network, re-execute the step of generating a sample resource matrix corresponding to the first sample feature matrix according to the first sample feature matrix and the corresponding sample adjacency matrix until the loss error is less than or equal to the preset loss value.

[0134] Step 512, obtaining a trained graph convolutional neural network.

[0135] If the control device detects that the loss error is less than or equal to the preset loss value, it means that the current pre-trained graph convolutional neural network can integrate the resource features in the feature matrix and the communication information in the adjacency matrix, and the resource matrix obtained by the pre-trained graph convolutional neural network can well indicate the association relationship between multiple network elements in the local area network. Therefore, the current pre-trained graph convolutional neural network is used as a trained graph convolutional neural network.

[0136] In an embodiment of the present application, during the training process of the graph convolutional neural network, the performance of the graph convolutional neural network can be optimized by continuously adjusting the parameters of the neural network; so that the trained graph convolutional neural network can more accurately understand the relationship between multiple network elements in the local area network, thereby improving the accuracy and efficiency of obtaining the resource matrix, which helps to improve the intelligence of management.

[0137] like Figure 6 As shown, in some embodiments, the control device manages each network element in the local area network, which may include the following steps 602 to 612.

[0138] Step 602: Receive an access request sent by a first network element.

[0139] In some embodiments, the control device can manage the access and access of each network element in the local area network, including whether the first network element can access the third network element. The access request includes network element information corresponding to the third network element to be accessed by the first network element; the third network element is any network element in the local area network that is different from the first network element.

[0140] When the first network element needs to access the third network element, it will send a corresponding access request to the control device; the control device will determine whether to forward the access request to the third network element, that is, whether to allow the first network element to access the third network element, based on the access request and the current resource matrix.

[0141] Specifically, the network element information corresponding to the third network element to be accessed by the first network element may include but is not limited to which first ports the first network element requests to access the third network element, which protocols are provided for access, which second ports the first network element uses to access the third network element, identification information of the first network element and identification information of the third network element, etc.

[0142] Step 604: Determine, according to the access request, a first network element vector corresponding to the first network element and a second network element vector corresponding to the third network element from the resource matrix.

[0143] In some embodiments, the control device may obtain the first network element vector corresponding to the first network element and the second network element vector corresponding to the third network element from the resource matrix according to the identification information of the first network element and the identification information of the third network element in the access request. Specifically, the control device may obtain the numbers corresponding to the identification information of the first network element and the identification information of the third network element, respectively. The numbers corresponding to the network elements indicate the row of the target feature matrix in which the resource features of the network element are located. According to the characteristics of the graph convolutional neural network, the dimension of the resource matrix is ​​the same as the dimension of the target feature matrix. Then, the control device may determine the rows of the resource matrix in which the first network element vector and the second network element vector are located according to the numbers corresponding to the identification information of the first network element and the identification information of the third network element.

[0144] Step 606: Calculate the vector distance between the first network element vector and the second network element vector to obtain a first distance.

[0145] In some embodiments, the control device may use vector distance calculation methods including Euclidean distance, Manhattan distance, cosine similarity, and the like.

[0146] Taking the Euclidean distance as an example, the control device may calculate the first distance between the first network element vector and the second network element vector according to the Euclidean formula. Specifically, the Euclidean formula is: Where d represents the first distance, n represents the number of elements in the network element vector, and x i represents the i-th element of the first network element vector, y i Represents the i-th element of the second network element vector.

[0147] Step 608, determining whether the first distance is less than or equal to a preset distance threshold. If yes, executing step 610; if no, executing step 612.

[0148] Step 610: forward the access request to the third network element.

[0149] When the control device detects that the vector distance between the first network element vector and the second network element vector is less than or equal to a preset distance threshold, it indicates that the first network element and the third network element are relatively similar or close in resources (features), that is, the first network element can access the third network element. The control device forwards the access request to the third network element so that the first network element can successfully access the third network element.

[0150] Step 612: Do not forward the access request to the third network element.

[0151] When the control device detects that the vector distance between the first network element vector and the second network element vector is greater than a preset distance threshold, indicating that the first network element and the third network element have large differences in resources (characteristics), the control device does not forward the access request to the third network element, so that the first network element cannot access the third network element.

[0152] In an embodiment of the present application, by calculating the first distance between the network element vectors of two network elements and comparing it with a preset threshold, the control device can automatically determine whether a network element is allowed to access another network element without the need for manual analysis and approval, thereby improving the efficiency and accuracy of network element management of each network element.

[0153] like Figure 7 As shown, in some embodiments, the control device manages each network element in the local area network, which may include the following steps 702 to 704.

[0154] Step 702: If the network element resource information sent by the first network element is not received within the preset time length, at least one accessible network element that can access the first network element and at least one visited network element that can be accessed by the first network element are determined according to the resource matrix.

[0155] The accessible network elements refer to other network elements in the local area network that can access the first network element; the visited network elements refer to other network elements in the local area network that can be accessed by the first network element.

[0156] In some embodiments, each network element in the local area network may periodically send network element resource information to the control device. If the control device does not receive the network element resource information sent by the first network element within a preset time length, it indicates that the first network element may have problems such as failure, communication interruption, or resource information update delay, resulting in the first network element being unable to successfully access the local area network. In order to avoid the traffic transmission direction of other network elements in the local area network being affected by the first network element, it is necessary to adjust the traffic transmission direction of the accessible network element and the visited network element.

[0157] Step 704: adjust the traffic transmission direction of at least one accessible network element and at least one visited network element, so that at least one accessible network element can access at least one visited network element.

[0158] In some embodiments, the control device may determine, based on the resource matrix, an accessible vector corresponding to at least one accessible network element and an accessed vector corresponding to at least one visited network element; calculate the second distance between each accessible vector and each visited vector; and the control device adjusts the traffic transmission direction from the accessible network element to the first network element to from the accessible network element to the visited network element according to the order of each second distance from small to large, until at least one accessible network element can access at least one visited network element.

[0159] It should be noted that, in the local area network after the traffic transmission direction is adjusted, each accessible network element can access at least one visited network element, and each visited network element can be accessed by at least one accessible network element.

[0160] For example, assuming that there are three accessible network elements (network element A, network element B and network element C) and two visited network elements (network element a and network element b), the control module calculates the second distances between each accessible vector and each visited vector as (network element A, network element a, 6), (network element A, network element b, 3), (network element B, network element a, 5), (network element B, network element b, 2), (network element C, network element a, 7), (network element C, network element b, 4). The control module adjusts the traffic transmission direction of at least one accessible network element and at least one visited network element according to the order of the second distances from small to large, so that network element B accesses network element b, network element A accesses network element b, and network element C accesses network element b. At this time, the three accessible network elements have been adjusted, but network element a has not been visited. Then, according to the order of the second distances from small to large, network element B is directed to network element a, so that the two visited network elements have been adjusted.

[0161] In an embodiment of the present application, the control device can promptly discover and respond to possible failures, communication interruptions, or resource information update delays of the first network element by monitoring the network element resource information sent by each network element; and, in the event that a problem occurs with the first network element, the control device can quickly determine the accessible network elements and the accessed network elements based on the resource matrix, and adjust their traffic transmission directions to ensure that traffic transmission in the local area network is not affected by the failure of a single network element, thereby optimizing the traffic path and improving the network efficiency of the local area network.

[0162] like Figure 8 As shown, in one embodiment, a network element management device 800 for a local area network is provided, which can be applied to the above-mentioned control device. The network element management device 800 for a local area network may include an information acquisition module 810, a feature extraction module 820, a matrix generation module 830, a graph convolution module 840 and a network element management module 850.

[0163] The information acquisition module 810 is used to obtain network element resource information corresponding to multiple network elements and communication information corresponding to multiple network elements; wherein the communication information corresponding to the first network element is used to indicate the other network elements with which the first network element can communicate, and the first network element is any network element among the multiple network elements.

[0164] The feature extraction module 820 is used to extract the features of the network element resource information corresponding to each network element, and obtain the resource features corresponding to each network element.

[0165] The matrix generation module 830 is used to generate a target feature matrix corresponding to the local area network according to the resource features corresponding to each network element.

[0166] Optionally, the matrix generating module 830 is further configured to generate a target adjacency matrix corresponding to the local area network according to communication information corresponding to the plurality of network elements.

[0167] The graph convolution module 840 is used to generate a resource matrix corresponding to the local area network based on the graph convolutional neural network, according to the target feature matrix and the target adjacency matrix. The resource matrix is ​​used to indicate the association relationship between multiple network elements in the local area network.

[0168] The network element management module 850 is used to manage the traffic corresponding to each network element according to the resource matrix.

[0169] Optionally, the communication information corresponding to the first network element includes communication information of the first network element accessing the second network element, and the communication information of the first network element accessing the second network element includes the number of first ports through which the first network element can access the second network element, the number of second ports through which the second network element can be accessed by the first network element, and the types of communication protocols supported by the first network element and the second network element; the second network element is any other network element with which the first network element can communicate.

[0170] In some embodiments, the matrix generation module 830 is also used to generate matrix elements corresponding to the first network element accessing the second network element based on the first port number, the second port number and the communication protocol type; and to generate a target adjacency matrix corresponding to the local area network based on the matrix elements corresponding to any network element in the local area network accessing another network element.

[0171] Optionally, the graph convolution module 840 is also used to use the target feature matrix and the target adjacency matrix as the input matrix of the first layer of the graph convolution neural network; based on the graph convolution formula corresponding to the current layer, encode the input matrix of the current layer to obtain the output matrix of the current layer; use the output matrix and the target adjacency matrix of the current layer as the input matrix of the next layer, and continue to execute the steps of encoding the input matrix of the current layer based on the graph convolution formula corresponding to the current layer to obtain the output matrix of the current layer, until the output matrix of the last layer is obtained; use the output matrix of the last layer as the resource matrix corresponding to the local area network.

[0172] In some embodiments, the network element management device 800 of the local area network may also include a neural network training module.

[0173] A neural network training module is used to obtain a sample set and construct a pre-trained graph convolutional neural network according to preset parameters; the sample set includes multiple sample feature matrices, respectively corresponding sample adjacency matrices and respectively corresponding target resource matrices; based on the pre-trained graph convolutional neural network, a sample resource matrix corresponding to the first sample feature matrix is ​​generated according to the first sample feature matrix and the corresponding sample adjacency matrix; the first sample feature matrix is ​​any sample feature matrix of the sample set; according to the sample resource matrix corresponding to each sample feature matrix and the corresponding target resource matrix, the loss error corresponding to the pre-trained graph convolutional neural network is calculated; if the loss error is greater than the preset loss value, the graph convolution formula corresponding to each layer in the pre-trained graph convolutional neural network is adjusted to obtain a new pre-trained graph convolutional neural network; repeatedly executing the steps of generating a sample resource matrix corresponding to the first sample feature matrix according to the first sample feature matrix and the corresponding sample adjacency matrix based on the pre-trained graph convolutional neural network, until the loss error is less than or equal to the preset loss value, and a trained graph convolutional neural network is obtained.

[0174] Optionally, the resource matrix includes network element vectors corresponding to a plurality of network elements respectively, and the network element vector corresponding to the first network element is used to indicate an association relationship between the first network element and each network element in the local area network.

[0175] In some embodiments, the network element management module 850 is also used to receive an access request sent by the first network element, the access request including network element information corresponding to a third network element to be accessed by the first network element; the third network element is any network element in the local area network that is different from the first network element; according to the access request, determine the first network element vector corresponding to the first network element and the second network element vector corresponding to the third network element from the resource matrix; calculate the vector distance between the first network element vector and the second network element vector to obtain a first distance; if the first distance is less than or equal to a preset distance threshold, forward the access request to the third network element.

[0176] In some embodiments, the network element management module 850 is also used to determine, based on the resource matrix, at least one accessible network element that can access the first network element and at least one visited network element that can be accessed by the first network element if the network element resource information sent by the first network element is not received within a preset time length; and adjust the traffic transmission direction of at least one accessible network element and at least one visited network element so that at least one accessible network element can access at least one visited network element.

[0177] Optionally, the network element resource information corresponding to the first network element includes multiple target network element resources corresponding to the first network element.

[0178] In some embodiments, the feature extraction module 820 is also used to determine the mapping data corresponding to multiple target network element resources corresponding to the first network element according to the resource mapping table to obtain the resource characteristics corresponding to the first network element; the resource mapping table contains each network element resource and the corresponding mapping value.

[0179] In an embodiment of the present application, a control device in a local area network generates a target feature matrix by extracting resource features of network element resource information of each network element, and generates a target adjacency matrix according to the communication information corresponding to each network element, and processes the target feature matrix and the target adjacency matrix through a graph convolutional neural network to obtain a resource matrix. Since the target feature matrix and the target adjacency matrix can accurately characterize the resource status of each network element in the local area network and the communication relationship between different network elements, the obtained resource matrix can accurately reflect the association relationship between different network elements in the local area network. Based on the resource matrix, the control device can manage the traffic of each network element in the local area network more accurately according to the association relationship between different network elements in the local area network, thereby improving the efficiency and accuracy of network element management and improving the intelligence level of network element management.

[0180] Fig. 9 FIG. 1 is a structural block diagram of an electronic device in an embodiment. Fig. 9 As shown, the electronic device 900 may include one or more of the following components: a processor 910, and a memory 920 coupled to the processor 910, wherein the memory 920 may store one or more computer programs, and the one or more computer programs may be configured to implement the methods described in the above embodiments when executed by one or more processors 910.

[0181] The processor 910 may include one or more processing cores. The processor 910 uses various interfaces and lines to connect various parts of the entire electronic device 900, and executes various functions and processes data of the electronic device 900 by running or executing instructions, programs, code sets or instruction sets stored in the memory 920, and calling data stored in the memory 920. Optionally, the processor 910 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 910 can integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 910, but may be implemented separately through a communication chip.

[0182] The memory 920 may include a random access memory (RAM) or a read-only memory (ROM). The memory 920 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 920 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc. The data storage area may also store data created by the electronic device 900 during use, etc.

[0183] It can be understood that the electronic device 900 may include more or fewer structural elements than those in the above structural block diagram, for example, including a power supply, input buttons, cameras, speakers, screens, RF (Radio Frequency) circuits, Wi-Fi (Wireless Fidelity) modules, Bluetooth modules, sensors, etc., and no limitation is made here.

[0184] An embodiment of the present application discloses a computer-readable storage medium storing a computer program, wherein the computer program implements the methods described in the above embodiments when executed by a processor.

[0185] The embodiments of the present application disclose a computer program product, including a computer program, and the computer program can be executed by a processor to implement the methods described in the above embodiments.

[0186] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM), etc.

[0187] As used herein, any reference to memory, storage, database, or other medium may include nonvolatile and / or volatile memory. Suitable nonvolatile memory may include ROM, programmable ROM (Programmable ROM, PROM), erasable PROM (Erasable PROM, EPROM), electrically erasable PROM (Electrically Erasable PROM, EEPROM) or flash memory. Volatile memory may include random access memory (random access memory, RAM), which is used as an external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (Static RAM, SRAM), dynamic RAM (Dynamic Random Access Memory, DRAM), synchronous DRAM (synchronous DRAM, SDRAM), double data rate SDRAM (Double Data Rate SDRAM, DDR SDRAM), enhanced SDRAM (Enhanced Synchronous DRAM, ESDRAM), synchronous link DRAM (Synchlink DRAM, SLDRAM), memory bus direct RAM (Rambus DRAM, RDRAM) and direct memory bus dynamic RAM (Direct Rambus DRAM, DRDRAM).

[0188] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. Those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required for the present application.

[0189] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the above-mentioned processes does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0190] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0191] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0192] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0193] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several requests for a computer device (which can be a personal computer, a server or a network device, etc., specifically a processor in a computer device) to perform some or all of the steps of the above-mentioned methods of various embodiments of the present application.

[0194] The above is a detailed introduction to a network element management method, device, electronic device and storage medium for a local area network disclosed in the embodiment of the present application. The principle and implementation method of the present application are described in detail using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those of ordinary skill in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A network element management method for a local area network, characterized in that: A control device applied to the local area network, wherein the local area network further comprises a plurality of network elements; the method comprises: Acquire network element resource information corresponding to the multiple network elements respectively, and communication information corresponding to the multiple network elements respectively; wherein the communication information corresponding to the first network element is used to indicate each other network element with which the first network element can communicate, and the first network element is any network element among the multiple network elements; Extracting characteristics of network element resource information corresponding to each of the network elements to obtain resource characteristics corresponding to each of the network elements; Generating a target feature matrix corresponding to the local area network according to resource features corresponding to each of the network elements; Generate a target adjacency matrix corresponding to the local area network according to the communication information respectively corresponding to the multiple network elements; Based on a graph convolutional neural network, generating a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix, wherein the resource matrix is ​​used to indicate the association relationship between the multiple network elements in the local area network; The traffic corresponding to each of the network elements is managed according to the resource matrix.

2. The method according to claim 1, characterized in that The communication information corresponding to the first network element includes communication information of the first network element accessing the second network element, and the communication information of the first network element accessing the second network element includes the number of first ports through which the first network element can access the second network element, the number of second ports of the second network element that can be accessed by the first network element, and the types of communication protocols supported by the first network element and the second network element; The second network element is any other network element with which the first network element can communicate; The generating a target adjacency matrix corresponding to the local area network according to the communication information respectively corresponding to the plurality of network elements comprises: Generate a matrix element corresponding to the first network element accessing the second network element according to the first port quantity, the second port quantity and the communication protocol type; According to any one network element in the local area network accessing a matrix element corresponding to another network element, a target adjacency matrix corresponding to the local area network is generated.

3. The method according to claim 1, characterized in that The graph convolutional neural network based method generates a resource matrix corresponding to the local area network according to the target feature matrix and the target adjacency matrix, including: Using the target feature matrix and the target adjacency matrix as the input matrix of the first layer of the graph convolutional neural network; Based on the graph convolution formula corresponding to the current layer, encode the input matrix of the current layer to obtain the output matrix of the current layer; Using the output matrix of the current layer and the target adjacency matrix as the input matrix of the next layer, continuing to execute the graph convolution formula corresponding to the current layer, encoding the input matrix of the current layer, and obtaining the output matrix of the current layer, until the output matrix of the last layer is obtained; The output matrix of the last layer is used as the resource matrix corresponding to the local area network.

4. The method according to claim 1, characterized in that: The training process of the graph convolutional neural network includes: Acquire a sample set, and construct a pre-trained graph convolutional neural network according to preset parameters; the sample set includes a plurality of sample feature matrices, respectively corresponding sample adjacency matrices, and respectively corresponding target resource matrices; Based on the pre-trained graph convolutional neural network, according to the first sample feature matrix and the corresponding sample adjacency matrix, a sample resource matrix corresponding to the first sample feature matrix is ​​generated; the first sample feature matrix is ​​any sample feature matrix of the sample set; Calculate the loss error corresponding to the pre-trained graph convolutional neural network according to the sample resource matrix and the corresponding target resource matrix corresponding to each of the sample feature matrices; If the loss error is greater than a preset loss value, adjusting the graph convolution formula corresponding to each layer in the pre-trained graph convolutional neural network to obtain a new pre-trained graph convolutional neural network; Repeat the steps of generating a sample resource matrix corresponding to the first sample feature matrix based on the pre-trained graph convolutional neural network according to the first sample feature matrix and the corresponding sample adjacency matrix until the loss error is less than or equal to the preset loss value, thereby obtaining a trained graph convolutional neural network.

5. The method according to claim 1, characterized in that The resource matrix includes network element vectors corresponding to the multiple network elements respectively, and the network element vector corresponding to the first network element is used to indicate the association relationship between the first network element and each network element in the local area network; The managing the traffic corresponding to each of the network elements according to the resource matrix includes: receiving an access request sent by the first network element, the access request including network element information corresponding to a third network element to be accessed by the first network element; the third network element is any network element in the local area network different from the first network element; Determine, according to the access request, a first network element vector corresponding to the first network element and a second network element vector corresponding to the third network element from the resource matrix; Calculating a vector distance between the first network element vector and the second network element vector to obtain a first distance; If the first distance is less than or equal to a preset distance threshold, the access request is forwarded to the third network element.

6. The method according to claim 1, characterized in that The managing the traffic corresponding to each of the network elements according to the resource matrix includes: If the network element resource information sent by the first network element is not received within a preset time length, determining at least one accessible network element that can access the first network element and at least one visited network element that can be accessed by the first network element according to the resource matrix; The traffic transmission direction between the at least one accessible network element and the at least one visited network element is adjusted so that the at least one accessible network element can access the at least one visited network element.

7. The method according to claim 1, characterized in that The network element resource information corresponding to the first network element includes multiple target network element resources corresponding to the first network element; The extracting the characteristics of the network element resource information corresponding to each of the network elements to obtain the resource characteristics corresponding to each of the network elements includes: Determine, according to the resource mapping table, mapping data corresponding to a plurality of target network element resources corresponding to the first network element, so as to obtain resource characteristics corresponding to the first network element; The resource mapping table contains various network element resources and corresponding mapping values.

8. A network element management device for a local area network, characterized in that: A control device applied to the local area network, wherein the local area network further comprises a plurality of network elements; the device comprises: An information acquisition module, used to acquire network element resource information corresponding to the multiple network elements respectively, and communication information corresponding to the multiple network elements respectively; wherein the communication information corresponding to the first network element is used to indicate each other network element with which the first network element can communicate, and the first network element is any network element among the multiple network elements; A feature extraction module, used to extract features of network element resource information corresponding to each network element, and obtain resource features corresponding to each network element; A matrix generation module, used to generate a target feature matrix corresponding to the local area network according to resource features corresponding to each of the network elements; The matrix generation module is further used to generate a target adjacency matrix corresponding to the local area network according to the communication information respectively corresponding to the multiple network elements; A graph convolution module, configured to generate a resource matrix corresponding to the local area network based on a graph convolutional neural network and according to the target feature matrix and the target adjacency matrix, wherein the resource matrix is ​​used to indicate an association relationship between the multiple network elements in the local area network; The network element management module is used to manage the traffic corresponding to each of the network elements according to the resource matrix.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor implements the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images