A network processing method, apparatus and device

Abstract

The application discloses a network processing method, device and equipment, and belongs to the technical field of communication. The network processing method comprises the following steps: acquiring a network topology structure of a target network, and determining graph structure data corresponding to each target node in the target network according to the network topology structure; performing equipment anomaly identification on network equipment corresponding to the target node according to a preset convolutional neural network and the graph structure data; in the case that the network equipment corresponding to the target node is abnormal, verifying the connectivity of a target link according to target graph structure data tree corresponding to the target network, and obtaining a link fault identification result; the target link is a link where global coordinate data corresponding to the target node is located. The network processing method realizes automatic inspection and digital monitoring of the network, does not require manual on-site inspection, improves the inspection efficiency, accuracy and data visualization degree, and saves human resources.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and more specifically to a network processing method, apparatus, and device. Background Technology

[0002] With the development of technologies such as the Internet of Things, big data, and artificial intelligence, 5G (5th Generation Mobile Communication Technology) networks are gradually becoming a crucial infrastructure for connecting the future. 5G networks, with their high-speed transmission, low latency, and massive connectivity, offer vast development opportunities for various industries. However, as the scale and complexity of networks increase, the operation and maintenance management of 5G networks faces unprecedented challenges.

[0003] 5G networks have extensive coverage, involving numerous base stations, devices, and users. Network failures in 5G networks are influenced by a variety of factors, including weather, equipment malfunctions, and signal interference. Traditional network maintenance methods primarily rely on manual inspections and analysis, which presents the following problems:

[0004] 1. Manual inspection usually requires operators to go to the site to check the status of each piece of equipment, which is not only tedious and time-consuming, but also requires a lot of human resources.

[0005] 2. Manual inspections are usually conducted periodically, making it impossible to obtain real-time equipment operating data and detect problems in a timely manner.

[0006] 3. Reports generated by manual analysis of manual inspections often lack detailed data support, making it difficult to conduct in-depth analysis and decision-making. Therefore, there is also the problem of incomplete inspection reports. Summary of the Invention

[0007] Therefore, the present invention provides a network processing method, apparatus, and device to solve the problem of network failures caused by various factors in the prior art.

[0008] To achieve the above objectives, a first aspect of the present invention provides a network processing method, the method comprising:

[0009] Obtain the network topology of the target network, and determine the graph structure data corresponding to each target node in the target network based on the network topology;

[0010] Based on a preset convolutional neural network and the graph structure data, the network device corresponding to the target node is identified as having device anomalies.

[0011] In the event of an anomaly in the network device corresponding to the target node, the connectivity of the target link is verified based on the target graph structure data tree corresponding to the target network to obtain the link fault identification result; the target link is the link where the global coordinate data corresponding to the target node is located.

[0012] Specifically, before obtaining the network topology of the target network and determining the graph structure data corresponding to each target node in the target network based on the network topology, the method further includes:

[0013] Based on the network topology and the line vector information between each target node, a graph structure data tree of the target network is created; the line vector information includes the start point, end point, length, bandwidth, and delay information of the line formed by at least two target nodes.

[0014] The 3D building information model corresponding to the target node is imported into the graph structure data tree to obtain the target graph structure data tree.

[0015] Specifically, determining the graph structure data corresponding to each target node in the target network based on the network topology includes:

[0016] Obtain the network operation data of the target node;

[0017] Key performance indicators are extracted from the network operation data; the key performance indicators include at least one of throughput, latency, and packet loss rate.

[0018] The trends and patterns of the key performance indicators are identified using a preset time series analysis model, and a visualization image is obtained based on the trends and patterns.

[0019] The visualized image is aligned with the image information of the network device corresponding to the target node to obtain graph structure data; wherein, the image information of the network device corresponding to the target node is captured by an artificial intelligence (AI) sensing device.

[0020] Specifically, the step of verifying the connectivity of the target link based on the target graph structure data tree corresponding to the target network to obtain the link fault identification result includes:

[0021] Obtain the target line vector information of the target link from the target graph structure data tree;

[0022] Based on the target line vector information, at least one sub-link in the target link is determined; the link start node of the target link is the global coordinate data corresponding to the abnormal target node;

[0023] Verify the connectivity of each sub-link; if at least one of the sub-links in the target link has abnormal connectivity, determine that the target link is an abnormal link.

[0024] Based on the abnormal link, determine the link fault identification result.

[0025] Specifically, determining the link fault identification result based on the abnormal link includes:

[0026] A diagnostic request is sent to the end node in the abnormal link; wherein the diagnostic request is used to request the diagnosis of the network status information of the end node, and the network status information includes at least one of the following: operating status, interface information, and traffic statistics;

[0027] Receive response data from the end node in response to the diagnostic request; wherein the response data includes at least one of the following: current status, connection information, and error information;

[0028] The response data is parsed to obtain diagnostic results;

[0029] Based on the diagnostic results, the link fault identification result is determined.

[0030] Specifically, after determining the link fault identification result based on the diagnostic result, the method further includes:

[0031] A test report is generated based on the diagnostic results.

[0032] Specifically, network processing methods also include:

[0033] The network topology of the target network is updated based on the target node corresponding to the abnormal network device and the link fault identification result.

[0034] Specifically, updating the network topology of the target network based on the target node corresponding to the abnormal network device and the link failure identification result includes:

[0035] Receive real-time image information of each network device in the target network, acquired through an AI sensing device; the real-time image information includes real-scene images of the network devices and the location of the network devices;

[0036] Align the real-scene image of the network device with the graph structure data of the target node, and map the real-time image information to the network topology of the target network to determine the mapping relationship of the real-time image information in the network topology;

[0037] Extract the visual features of the network device from the real-scene image;

[0038] The visual features are compared with the abnormal target nodes to obtain the comparison results;

[0039] Based on the comparison results, determine whether the abnormal target node exists in the real-world image;

[0040] If the abnormal target node exists in the real-world image, the network topology of the target network is updated according to the link fault identification result and the mapping relationship.

[0041] A second aspect of the present invention provides a network processing apparatus, the apparatus comprising:

[0042] The data acquisition module is used to acquire the network topology of the target network and determine the graph structure data corresponding to each target node in the target network based on the network topology.

[0043] An abnormal node identification module is used to identify abnormal devices in the network devices corresponding to the target node based on a preset convolutional neural network and the graph structure data.

[0044] The fault link determination module is used to verify the connectivity of the target link and obtain the link fault identification result based on the target graph structure data tree corresponding to the target network when the network device corresponding to the target node is abnormal; the target link is the link where the global coordinate data corresponding to the target node is located.

[0045] A third aspect of the present invention provides an electronic device, comprising:

[0046] One or more processors;

[0047] A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method according to the first aspect;

[0048] One or more I / O interfaces are connected between the processor and the memory and configured to enable information interaction between the processor and the memory.

[0049] The beneficial effects of the network processing method, apparatus, and device provided by the present invention include: using a preset convolutional neural network to identify anomalies in graph structure data of the topology to determine whether there are anomalies in the network, thereby determining the link failure status in the network when network devices are abnormal, realizing automated inspection and digital monitoring, eliminating the need for manual on-site visits, improving inspection efficiency, accuracy, and data visualization, and saving human resources. Attached Figure Description

[0050] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the following detailed description to explain the invention, but do not constitute a limitation thereof.

[0051] Figure 1 A flowchart illustrating a network processing method provided in an embodiment of the present invention;

[0052] Figure 2 This is a flowchart of a specific implementation method of step 100 in this embodiment of the invention;

[0053] Figure 3 This is a flowchart of a specific implementation method of step 101 in this embodiment of the invention;

[0054] Figure 4 This is a flowchart of a specific implementation method for step 103 in this embodiment of the invention;

[0055] Figure 5 This is a flowchart of a specific implementation method of step 1033 in this embodiment of the invention;

[0056] Figure 6 This is a flowchart of a specific implementation method for step 104 in this embodiment of the invention;

[0057] Figure 7 This is a schematic diagram of the structure of a network processing device provided in an embodiment of the present invention;

[0058] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0059] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0060] As used in this invention, the term "and / or" includes any and all combinations of one or more of the associated enumerated entries.

[0061] The terminology used in this invention is for describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms "a" and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0062] When the terms “comprising” and / or “made of” are used in this invention, the presence of the said feature, integral, step, operation, element and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or groups thereof is not excluded.

[0063] Unless otherwise specified, all terms used in this invention (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in common dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined by the invention.

[0064] Firstly, such as Figure 1 As shown, this embodiment of the invention provides a network processing method, which includes steps 101 to 103.

[0065] Step 101: Obtain the network topology of the target network, and determine the graph structure data corresponding to each target node in the target network based on the network topology.

[0066] Step 102: Based on the preset convolutional neural network and the graph structure data, perform device anomaly identification on the network device corresponding to the target node;

[0067] Step 103: In the event of an anomaly in the network device corresponding to the target node, verify the connectivity of the target link based on the target graph structure data tree corresponding to the target network to obtain the link fault identification result; the target link is the link where the global coordinate data corresponding to the target node is located.

[0068] In this embodiment, a pre-set convolutional neural network is used to identify anomalies in the graph structure data of the network topology to determine whether any anomalies exist in the network. This allows for the identification of link failures in the network when network devices exhibit anomalies, achieving automated inspection and digital monitoring. It enables comprehensive network analysis and rapid fault location without requiring on-site human intervention, improving inspection efficiency, accuracy, and data visualization, while saving human resources. Specifically, anomaly identification in the graph structure data refers to predicting the network devices corresponding to target nodes that may exhibit anomalies based on the network topology graph structure data.

[0069] In some embodiments, the graph structure data of each target node is acquired in real time.

[0070] The target network consists of several nodes and links connecting these nodes. Nodes correspond to network devices in the network (e.g., computers, switches, routers, etc.). The network topology is the layout and organization of the connections between the nodes in the target network. The graph structure data of each target node in the target network is used to describe the operating status of the network device corresponding to the target node. In some embodiments, the graph structure data includes basic information, status, and operating parameters of the network device.

[0071] The embodiments disclosed herein do not impose any special restrictions on the type of target network; it can be a 5G network, a 4G network, etc. As an example, when the target network is a 5G network, the network topology of the 5G network includes a core network 5GC (5G core) and a radio access side NG-RAN (Next Generation Radio Access Network).

[0072] In some embodiments, after step 101, the method further includes:

[0073] Identify and distinguish between core network nodes and access network nodes in the network topology of the target network;

[0074] Based on the core network nodes and the access network nodes, at least one node hierarchy is determined.

[0075] The core network nodes include the CU (Central Unit) responsible for the control plane and the DU (Distributed Unit) responsible for the user plane; the access network nodes include the AMF (Access and Mobility Management Function) unit and the UPF (User Plane Function) unit in the 5GC section. The pre-defined convolutional neural network architecture includes convolutional layers, pooling layers, and fully connected layers. In some embodiments, the training process of the pre-defined convolutional neural network is as follows:

[0076] Obtain a graph structure dataset for training, which includes data samples in normal and abnormal states. Each data sample includes node information and the relationship information between nodes. The graph structure dataset includes standard network topologies in normal states and abnormal network topologies in abnormal states. Each node in each network topology contains node information.

[0077] Data samples are generated from a graph-structured dataset, and each data sample is labeled. The dataset is divided into a training set, a validation set, and a test set. The label of a data sample indicates whether it is in a normal or abnormal state. In some embodiments, the labels are binary labels. As an example, a label of 1 indicates that the data sample is in a normal state, and a label of 0 indicates that the data sample is in an abnormal state. It is worth noting that the divided training, validation, and test sets should all contain data samples in both normal and abnormal states.

[0078] The network architecture to be trained includes convolutional layers, pooling layers, and fully connected layers, forming a convolutional neural network (CNN). In this embodiment, the number of convolutional and fully connected layers in the CNN to be trained is not limited; it can be one or more. CNNs are used for feature extraction and pattern recognition of graph-structured data.

[0079] Convert graph-structured data into input vectors that can be accepted by convolutional neural networks;

[0080] The node features of the node information in the data samples are mapped into vectors to adapt to the input format of the CNN, and the node information and edge information are mapped into appropriate vector representations to be input into the CNN.

[0081] The constructed CNN model is trained using a training set. During training, the CNN model identifies and learns whether the graph structure data in the data samples is in a normal or abnormal state. The weights and biases in the CNN are updated using the backpropagation algorithm based on the graph structure data in a normal or abnormal state. Furthermore, the training process of the pre-defined convolutional neural network may also include data preparation, model construction, training, and evaluation, and this application is not limited to these aspects.

[0082] The trained CNN model is validated using a validation set, and its accuracy is evaluated using a test set. If the model passes validation and its accuracy is greater than or equal to a preset accuracy, the trained CNN model is identified as the preset convolutional neural network, which can quickly identify abnormal target nodes in the network.

[0083] By inputting graph structure data obtained from the target node that is transmitting data in real time into a preset convolutional neural network for device anomaly identification, the identification result is obtained, thereby determining whether the network device corresponding to the node is abnormal. This achieves efficient identification of nodes corresponding to abnormal network devices in the network, improving the accuracy and speed of network processing.

[0084] In some embodiments, such as Figure 2 As shown, before step 101, there is also step 100, which includes the following steps 1001 and 1002.

[0085] Step 1001: Based on the network topology and the line vector information between each target node, create a graph structure data tree of the target network; the line vector information includes the start point, end point, length, bandwidth, and delay information of the line formed by at least two target nodes.

[0086] Step 1002: Import the 3D building information model corresponding to the target node into the graph structure data tree to obtain the target graph structure data tree.

[0087] In this embodiment, the graph structure data tree organizes graph structure data in a tree-like manner. Each node in the graph structure data tree can have zero or more child nodes, forming a hierarchical tree structure. The 3D building information model in this embodiment is created by scanning network devices with a laser scanner to obtain the 3D geometric data of the network devices, and then combining this data with a high-precision 3D model created using BIM (Building Information Modeling) software. The 3D building information model corresponding to the target node is imported into the graph structure data tree, thereby integrating the 3D BIM into the graph structure data tree, making the description of the graph structure data tree more comprehensive and accurate.

[0088] In some embodiments, step 1001 includes:

[0089] Step 10011: Obtain the network topology of the target network, each target node, and the line vector information between the target nodes; the line vector information includes the start point, end point, length, bandwidth, and delay information of the line formed by at least two target nodes.

[0090] Step 10012: Traverse each target node and the line vector information between each target node, create a corresponding node in the graph structure data tree for each target node, and assign a unique identifier to each node in the graph structure data tree to ensure that each target node corresponds to a unique node in the graph structure data tree;

[0091] Each node is identified and assigned a unique identifier, which is used to identify whether the node is a core network node or an access network node.

[0092] Step 10013: Using the line vector information between each target node, create edges representing the connection relationships between nodes in the graph structure data tree, and determine the node type (e.g., core network node, radio access side node), status information (e.g., operating status as core network node, radio access side node), and edge attributes of the edges between each node for each node. The edges include: edges between core network nodes (e.g., the edge between CU and DU), edges between core network nodes and access network nodes (e.g., the edge between CU and AMF), and edges between access network nodes (e.g., the edge between AMF and UPF). The edge attributes include bandwidth and latency information, which can more comprehensively describe the network connection quality in the graph structure data tree.

[0093] As an example, node status information includes normal status, warning status, fault status, etc.; bandwidth and latency information can be specifically indicated by data in the edge attributes, for example, bandwidth of 100Mbps and latency of 10ms, with the following formula parameters:

[0094] [\text{bandwidth}=\frac{\text{data volume}}{\text{transmission time}}]

[0095] [\text{delay}=\frac{\text{transmission time}}{\text{data volume}}].

[0096] Step 10014: Organize the nodes and edges in the graph structure data tree corresponding to each target node according to the hierarchical structure of the target network to form a graph structure data tree; in some embodiments, the hierarchical organization of the graph structure data tree includes: core network and radio access side. The hierarchical organization of this graph structure data tree can reflect the organizational relationship of target nodes at different levels in the target network.

[0097] In some embodiments, core network nodes (e.g., CU, DU) are organized at the root node (i.e., the first layer) in a graph structure data tree, and access network nodes (e.g., AMF, UPF) are organized at other nodes in the graph structure data tree besides the root node (i.e., other levels besides the first layer, such as the second layer and below).

[0098] Step 10015: Verify the graph structure data (i.e., node type, state information, edge attributes, etc.) in the formed graph structure data tree to ensure that the association between nodes and edges is accurate. The graph structure data tree that passes the verification is the constructed graph structure data tree. Store the graph structure data tree in the database so that subsequent fault analysis and correction operations can be performed based on the graph structure data tree. Also, create an index for the graph structure data tree to improve data retrieval efficiency and facilitate fast querying.

[0099] Through steps 10011 to 10015 above, a graph structure data tree is created for the target network, taking into full account the actual situation of the target network. This provides strong support for subsequent fault analysis and correction operations. At the same time, the graph structure data (i.e., node type, state information, edge attributes, etc.) is verified during the construction of the graph structure data tree to ensure the reliability of the data in the graph structure data tree. The graph structure data tree is stored in the database and an index is created, which can also improve retrieval efficiency.

[0100] In some embodiments, the three-dimensional building information model in step 1002 is created through the following steps:

[0101] Step 10021: Use a laser scanner to perform a comprehensive 3D scan of the network device corresponding to the target node to collect three-dimensional geometric data such as the size, shape, and position of the network device. Based on the device description record corresponding to the target node, obtain the material property data of the network device. For example, if the network device is a base station, the laser scanner captures the antenna shape, main unit volume, and placement coordinates of the base station. In other words, the three-dimensional geometric data of the base station includes (x, y, z) coordinates representing the placement position of the base station and (length, width, height) parameters representing the size and shape of the base station.

[0102] Step 10022: Import the collected solid geometry data into the BIM software, and create a 3D building information model of the network device in the BIM software based on the solid geometry data and material property data of the network device. For example, if the network device is a communication base station, the 3D building information model of the network device includes components such as antennas, supports, and main units, and each part has accurate dimensions and shape data.

[0103] Step 10023: Based on the line vector information between the target nodes corresponding to each network device, create a line model that is consistent with the actual network topology of the target network, thereby ensuring that the physical connection relationship between the line and the network device can be accurately represented. For example, in the case of a fiber optic line, the created line model can include information such as the path and length of the optical cable corresponding to the fiber optic line, corresponding to the network device connected to the fiber optic line.

[0104] Step 10024: Establish the coordinate system of the 3D building information model and accurately map the 3D building information model and line model of the network device into this coordinate system.

[0105] Step 10025: Based on the network topology of the target network, perform physical connection mapping on the 3D building information model and line model of the network devices to ensure that the 3D building information model can accurately reflect the connection layout relationship between the network devices in the actual target network (e.g., there is an actual physical connection relationship between the antenna and the host, and between the optical cable and the network device).

[0106] By constructing a 3D building information model and a line model that are consistent with the actual network topology of the target network, the connection layout relationship (physical connection relationship) between network devices in the target network is mapped in the coordinate system, so that the 3D building information model can realistically and reliably represent the network devices in the network topology.

[0107] In some embodiments, such as Figure 3As shown, step 101 involves obtaining the graph structure data of each target node in the target network, including steps 1011 and 1014.

[0108] Step 1011: Obtain the network operation data of the target node;

[0109] Step 1012: Extract key performance indicators from the network operation data; the key performance indicators include at least one of throughput, latency, and packet loss rate;

[0110] Step 1013: Use a preset time series analysis model to identify the trends and patterns of the key performance indicators, and obtain a visualization image based on the trends and patterns;

[0111] Step 1014: Align the visualized image with the image information of the network device corresponding to the target node to obtain graph structure data; wherein, the image information of the network device corresponding to the target node is captured by the AI ​​sensing device.

[0112] In the embodiments of this disclosure, prior to step 1012, the method further includes: data analysis and processing of network operation data, including cleaning and format conversion, i.e., cleaning the collected network operation data and converting it into a unified and standardized format. By performing data analysis and processing on the cleaned and format-converted network operation data, key performance indicators can be extracted more effectively. A preset time series analysis model is used to identify the trends and patterns of key performance indicators, thereby obtaining a visual image to provide intuitive graphical information, helping maintenance personnel quickly understand the network status of the target network. The visual image is aligned with the image information of the network devices corresponding to the target node captured by the AI ​​sensing device to obtain graph structure data. The AI ​​sensing device enables comprehensive monitoring of network devices.

[0113] This disclosure does not impose any special restrictions on the method of obtaining network operation data. It can be achieved by collecting real-time data through real-time monitoring of network devices to form a raw dataset of network operation data, which includes data traffic, signal strength, and connection status.

[0114] This disclosure does not impose any special limitations on the type of visualization image; it can be a line chart, bar chart, or heatmap, etc. As an example, when using a line chart to represent the trends of network throughput, latency, and packet loss rate, time series analysis can be performed using ARIMA (Autoregressive Integrated Moving Average) to obtain the analysis results. Based on the analysis results, a line chart is generated to clearly show the changing trends of the target network's network performance.

[0115] In some embodiments, such as Figure 4 As shown, step 103 includes the following steps 1031 and 1033.

[0116] Step 1031: Obtain the target line vector information of the target link from the target graph structure data tree;

[0117] Step 1032: Based on the target line vector information, determine at least one sub-link in the target link; the link start node of the target link is the global coordinate data corresponding to the abnormal target node;

[0118] Step 1033: Verify the connectivity of each sub-link. If the connectivity of at least one sub-link in the target link is abnormal, determine that the target link is an abnormal link.

[0119] Step 1034: Determine the link fault identification result based on the abnormal link.

[0120] In this embodiment, when the network device corresponding to the target node malfunctions, the link connected to the target node may also malfunction. Based on the global coordinate data of the target node, the node type and ID information of the link's starting node, and this node type and ID information, the target line vector information of the target node is queried from the target graph structure data tree. This target line vector information includes the start point, end point, length, bandwidth, and delay information of the sub-links corresponding to the target node. Test data packets are sent on each sub-link to check the connectivity of each sub-link, i.e., to check whether the communication between the start and end points of the sub-link is normal. If at least one sub-link in the target link exhibits abnormal connectivity, the target link is determined to be an abnormal link. For the abnormal link, a diagnostic request is sent to obtain detailed network status information to determine the link fault identification result. If no sub-links in the target link exhibit abnormal connectivity, the target link is determined to be a normal link.

[0121] By taking the target node as the starting point and verifying the connectivity of multiple target links corresponding to the target node based on the target line vector information, abnormal links can be successfully identified, which helps to quickly locate and resolve anomalies in the target network.

[0122] As an example, assuming the target network is a 5G network, if a network device corresponding to a target node in the target network malfunctions, the links connected to it may also malfunction. The node type and ID information of the link's starting node are obtained, where the node type is a base station and the ID is BS001. The target line vector information associated with base station BS001 is queried from the target graph structure data tree. Based on the target line vector information, at least one sub-link in the target link is identified. Test data packets are sent on the sub-links to check if the communication of the target link connected to base station BS001 is normal. If the connectivity of at least one sub-link in the target link is abnormal, the target link itself is also abnormal. If the target link is determined to be abnormal, a diagnostic request is sent to obtain detailed network status information, thereby confirming the link fault identification result.

[0123] In some embodiments, such as Figure 5 As shown, step 1034 includes the following steps 10341 and 10344.

[0124] Step 10341: Send a diagnostic request to the end node in the abnormal link; wherein the diagnostic request is used to request the diagnosis of the network status information of the end node, and the network status information includes at least one of the following: operating status, interface information, and traffic statistics;

[0125] Step 10342: Receive response data from the end node in accordance with the diagnostic request; wherein the response data includes at least one of the following: current status, connection information, and error information;

[0126] Step 10343: Analyze the response data to obtain the diagnostic results;

[0127] Step 10344: Determine the link fault identification result based on the diagnostic results.

[0128] In this embodiment, a request is sent to the network device corresponding to the target node based on SNMP (Simple Network Management Protocol) to obtain the network device's status information and measure the network's response time and path information. The end node of the target link is determined based on the link's starting node; this end node is the node to be diagnosed. SNMP authentication parameters are configured according to the end node's IP address, port number, and protocol type. A diagnostic request is then initiated to the end node using the configured SNMP to obtain the end node's network status information, which includes operating status, interface information, and traffic statistics.

[0129] The system listens for and receives response data from end nodes, parses the received response data in real time, and obtains diagnostic results. The response data includes at least one of the end node's current status, connection information, and error information. The diagnostic results are analyzed to check whether the target link has any connection timeouts, incorrect network configurations, abnormal or faulty device malfunctions, etc., thereby confirming the link fault identification result.

[0130] It is worth noting that the embodiments disclosed herein do not impose any special restrictions on the method of sending diagnostic requests to end nodes; it can be SNMP or other communication protocols.

[0131] In some embodiments, after step 10344, the method further includes:

[0132] Step 10345: Generate a test report based on the diagnostic results.

[0133] The detection report in this embodiment includes at least one of the location, type, and scope of impact of the abnormal link. By obtaining real-time network status information through network management protocols, analyzing and confirming abnormal links, and obtaining diagnostic results, not only can faults be effectively located, but a detection report can also be generated based on the anomaly to help find a repair solution.

[0134] In some embodiments, the network processing method further includes:

[0135] Step 104: Update the network topology of the target network based on the target node corresponding to the abnormal network device and the link fault identification result.

[0136] In the embodiments of this disclosure, fault analysis and location results are generated based on the location of abnormal network devices and link fault identification results. These results are then fed back into the graph structure data tree of the target network to correct the network topology, thereby updating the location information and correcting associations of abnormal network devices in the target network, thus improving the stability and maintainability of the target network.

[0137] By capturing real-time image information of network devices using AI sensing devices, and aligning this image information with visual images describing the trends and patterns of the network devices, graph structure data is obtained. This fusion of image information with target nodes in the network topology provides intuitive real-time information for verifying the location of network devices and correcting the network topology. Fault analysis of the graph structure data using a pre-set convolutional neural network makes the analysis more intelligent. The real-world verification (i.e., real-time image information) from the AI ​​sensing device provides real-time visual information for network processing. In the embodiments of this disclosure, the AI ​​sensing device is an AI visual sensing device.

[0138] In some embodiments, such as Figure 6 As shown, step 104 includes the following steps 1041 and 1046.

[0139] Step 1041: Receive real-time image information of each network device in the target network obtained through the AI ​​sensing device; the real-time image information includes real-scene images of the network devices and the location of the network devices;

[0140] Step 1042: Align the real-scene image of the network device with the graph structure data of the target node, and map the real-time image information to the network topology of the target network to determine the mapping relationship of the real-time image information in the network topology.

[0141] Step 1043: Extract the visual features of the network device from the real-scene image;

[0142] Step 1044: Compare the visual features with the abnormal target nodes to obtain the comparison results;

[0143] Step 1045: Based on the comparison results, determine whether the abnormal target node exists in the real-world image;

[0144] Step 1046: If the abnormal target node exists in the real-world image, update the network topology of the target network according to the link fault identification result and the mapping relationship.

[0145] The embodiments of this disclosure provide a real-time verification mechanism for location accuracy verification by using an AI sensing device to capture image information of network devices in real time. This real-time image information includes real-scene images of the network devices and their locations. The real-scene images are aligned with the graph structure data of the target nodes, and the image information is mapped to the network topology. Visual features of the network devices are extracted from the real-scene images, and these visual features are compared with abnormal target nodes. Based on the comparison results, it is determined whether there are network devices corresponding to abnormal target nodes in the real-scene images. If abnormal target nodes are found in the real-scene images, meaning the expected graph structure of each network device and its location in the real-scene images is inconsistent with the network devices corresponding to the abnormal target nodes, the data in the target network topology that does not match the real-scene images (i.e., the location information and association information related to the network devices) needs to be corrected. The location information of the network devices is updated, and their association relationships in the network topology are corrected. This helps ensure the accuracy and real-time performance of the network topology. Furthermore, using an AI sensing device to capture image information of network devices in real time ensures the timeliness of the verification result feedback.

[0146] The embodiments of this disclosure use a pre-set convolutional neural network to identify anomalies in the graph structure data of the topology to determine whether there are anomalies in the network. In the event of network device malfunctions, the link failure status in the network can be determined, realizing automated inspection, digital monitoring, and data processing. This not only improves the accuracy and efficiency of fault identification but also provides more intelligent, intuitive, and real-time support for network operation and maintenance. It can provide network operation and maintenance personnel with a comprehensive and efficient method for fault analysis and location, eliminating the need for manual on-site visits, improving inspection efficiency, accuracy, and data visualization, and saving human resources.

[0147] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but without changing the core design of the algorithm and process, are also within the scope of protection of this patent.

[0148] Secondly, such as Figure 7 As shown, an embodiment of the present invention provides a network processing device 700, the device comprising:

[0149] The data acquisition module 701 is used to acquire the network topology of the target network and determine the graph structure data corresponding to each target node in the target network based on the network topology.

[0150] An abnormal node identification module 702 is used to identify device anomalies in the network device corresponding to the target node based on a preset convolutional neural network and the graph structure data.

[0151] The fault link determination module 703 is used to verify the connectivity of the target link and obtain the link fault identification result based on the target graph structure data tree corresponding to the target network when the network device corresponding to the target node is abnormal; the target link is the link where the global coordinate data corresponding to the target node is located.

[0152] In some embodiments, the apparatus further includes: a device model building unit, configured to: before acquiring the network topology of the target network and determining the graph structure data corresponding to each target node in the target network based on the network topology, perform the following:

[0153] Based on the network topology and the line vector information between each target node, a graph structure data tree of the target network is created; the line vector information includes the start point, end point, length, bandwidth, and delay information of the line formed by at least two target nodes.

[0154] The 3D building information model corresponding to the target node is imported into the graph structure data tree to obtain the target graph structure data tree.

[0155] In some embodiments, the construction process of the 3D building information model in the device model construction module is as follows:

[0156] A laser scanner is used to perform a comprehensive 3D scan of the network device corresponding to the target node to collect three-dimensional geometric data such as the size, shape, and position of the network device. Based on the device's specifications, material property data is obtained. For example, if the network device is a base station, the laser scanner captures the antenna shape, main unit volume, and placement coordinates of the base station. In other words, the three-dimensional geometric data of the base station includes (x, y, z) coordinates representing its placement and (length, width, height) parameters representing its size and shape.

[0157] The collected solid geometry data is imported into BIM software, and a 3D building information model of the network device is created in the BIM software based on the solid geometry data and material property data of the network device. For example, if the network device is a communication base station, the 3D building information model of the network device includes components such as antennas, brackets, and main units, each with accurate dimensions and shape data.

[0158] Based on the line vector information between the target nodes corresponding to each network device, a line model is created that is consistent with the actual network topology of the target network, thereby ensuring that the physical connection relationship between the line and the network device can be accurately represented. For example, in the case of a fiber optic line, the created line model can include information such as the path and length of the optical cable corresponding to the fiber optic line, corresponding to the network device connected to the fiber optic line.

[0159] Establish a coordinate system for the 3D building information model, and accurately map the 3D building information model and line model of the network device into this coordinate system.

[0160] Based on the network topology of the target network, the physical connection relationships of the 3D building information model and line model of the network devices are mapped to ensure that the 3D building information model can accurately reflect the connection layout relationship between the network devices in the actual target network (e.g., there is an actual physical connection relationship between the antenna and the host, and between the optical cable and the network device).

[0161] By constructing a 3D building information model and a line model that are consistent with the actual network topology of the target network, the connection layout relationship (physical connection relationship) between network devices in the target network is mapped in the coordinate system, so that the 3D building information model can realistically and reliably represent the network devices in the network topology.

[0162] In some embodiments, the data acquisition module 701 includes a network operation data acquisition unit, a visualization image generation unit, and a graph structure data generation unit:

[0163] A network operation data acquisition unit is used to acquire the network operation data of the target node;

[0164] A visualization image generation unit is used to extract key performance indicators from the network operation data; the key performance indicators include at least one of throughput, latency, and packet loss rate; a preset time series analysis model is used to identify the trends and patterns of the key performance indicators, and a visualization image is obtained based on the trends and patterns;

[0165] The graph structure data generation unit is used to align the visualized image with the image information of the network device corresponding to the target node to obtain graph structure data; wherein, the image information of the network device corresponding to the target node is captured by the artificial intelligence (AI) perception device.

[0166] In some embodiments, the training process of the preset convolutional neural network in the abnormal node identification module 702 is as follows:

[0167] Obtain a graph structure dataset for training, which includes data samples in normal and abnormal states. Each data sample includes node information and the relationship information between nodes. The graph structure dataset includes standard network topologies in normal states and abnormal network topologies in abnormal states. Each node in each network topology contains node information.

[0168] Data samples are generated from a graph-structured dataset, and each data sample is labeled. The dataset is divided into a training set, a validation set, and a test set. The label of a data sample indicates whether it is in a normal or abnormal state. In some embodiments, the labels are binary labels. As an example, a label of 1 indicates that the data sample is in a normal state, and a label of 0 indicates that the data sample is in an abnormal state. It is worth noting that the divided training, validation, and test sets should all contain data samples in both normal and abnormal states.

[0169] The network architecture to be trained includes convolutional layers, pooling layers, and fully connected layers, forming a convolutional neural network (CNN). In this embodiment, the number of convolutional and fully connected layers in the CNN to be trained is not limited; it can be one or more. CNNs are used for feature extraction and pattern recognition of graph-structured data.

[0170] Convert graph-structured data into input vectors that can be accepted by convolutional neural networks;

[0171] The node features of the node information in the data samples are mapped into vectors to adapt to the input format of the CNN, and the node information and edge information are mapped into appropriate vector representations to be input into the CNN.

[0172] The constructed CNN model is trained using a training set. During training, the CNN model identifies and learns whether the graph structure data in the data samples is in a normal or abnormal state. The weights and biases in the CNN are updated using the backpropagation algorithm based on the graph structure data in a normal or abnormal state. Furthermore, the training process of the pre-defined convolutional neural network may also include data preparation, model construction, training, and evaluation, and this application is not limited to these aspects.

[0173] The trained CNN model is validated using a validation set, and its accuracy is evaluated using a test set. If the model passes validation and its accuracy is greater than or equal to a preset accuracy, the trained CNN model is identified as the preset convolutional neural network, which can quickly identify abnormal target nodes in the network.

[0174] By inputting graph structure data obtained from the target node that is transmitting data in real time into a preset convolutional neural network for device anomaly identification, the identification result is obtained, thereby determining whether the network device corresponding to the node is abnormal. This achieves efficient identification of the node corresponding to the abnormal network device in the network, improving the accuracy and speed of network fault identification.

[0175] In some embodiments, the fault link determination module 703 includes an acquisition unit, a processing unit, a verification unit, and an identification unit:

[0176] The acquisition unit is used to acquire the target line vector information of the target link from the target graph structure data tree;

[0177] The processing unit is configured to determine at least one sub-link in the target link based on the target line vector information; the link start node of the target link is the global coordinate data corresponding to the abnormal target node;

[0178] The verification unit is used to verify the connectivity of each of the sub-links, and if the connectivity of at least one of the sub-links in the target link is abnormal, the target link is determined to be an abnormal link.

[0179] The identification unit is used to determine the link fault identification result based on the abnormal link.

[0180] In some embodiments, the identification unit includes a transmitting subunit, a receiving subunit, a parsing subunit, and an identification subunit:

[0181] A sending subunit is configured to send a diagnostic request to the end node in the abnormal link; wherein the diagnostic request is used to request the diagnosis of the network status information of the end node, and the network status information includes at least one of the following: operating status, interface information, and traffic statistics;

[0182] A receiving subunit is configured to receive response data fed back by the end node in accordance with the diagnostic request; wherein the response data includes at least one of current status, connection information, and error information;

[0183] The parsing subunit is used to parse the response data to obtain diagnostic results;

[0184] The identification subunit is used to determine the link fault identification result based on the diagnostic results.

[0185] In some embodiments, the identification unit further includes a detection report generation subunit, configured to:

[0186] After determining the link fault identification result based on the diagnostic results, a detection report is generated based on the diagnostic results.

[0187] In some embodiments, the apparatus further includes a correction module for:

[0188] The network topology of the target network is updated based on the target node corresponding to the abnormal network device and the link fault identification result.

[0189] In some embodiments, the correction module includes an AI perception unit, an image alignment unit, a real-scene image feature extraction unit, a real-scene image feature comparison unit, and a network topology correction unit.

[0190] The AI ​​sensing unit is used to receive real-time image information of each network device in the target network obtained through the AI ​​sensing device; the real-time image information includes real-scene images of the network devices and the location of the network devices;

[0191] An image alignment unit is used to align the real-scene image of the network device with the graph structure data of the target node, and to map the real-time image information to the network topology of the target network, thereby determining the mapping relationship of the real-time image information in the network topology.

[0192] A real-scene image feature extraction unit is used to extract the visual features of the network device from the real-scene image;

[0193] A real-scene image feature comparison unit is used to compare the visual features with the abnormal target nodes to obtain a comparison result; and to determine whether the abnormal target nodes exist in the real-scene image based on the comparison result.

[0194] A network topology correction unit is used to update the network topology of the target network based on the link fault identification result and the mapping relationship when the abnormal target node exists in the real-world image.

[0195] The network topology correction unit includes a fault elimination subunit and a graph structure data correction subunit.

[0196] The fault elimination subunit is used to correct the abnormal target nodes in the network topology of the target network based on the abnormal target nodes and the link fault identification results; by correcting the network topology of the target network to eliminate faults in the network topology, the stability and efficient operation of the network can be ensured.

[0197] The graph structure data correction subunit is used to correct the location information and association information of the network devices corresponding to the abnormal target nodes in the network topology of the target network when there are abnormal target nodes in the real scene image.

[0198] Among them, the presence of abnormal target nodes in the real-world image indicates that the devices in the real-world image are inconsistent with the network devices corresponding to the expected graph structure data. It is necessary to correct the data in the target network topology that does not match the real-world image (i.e., the location information and association information related to the network devices), update the location information corresponding to the network devices, and correct the association relationship of the network devices, so as to ensure the accuracy and consistency of the network topology.

[0199] In this embodiment, the network processing device uses a pre-set convolutional neural network to identify anomalies in the graph structure data of the topology to determine whether there are any anomalies in the network. In the event of network device malfunctions, it can determine the link failure status in the network, realizing automated inspection, digital monitoring, and data processing. This not only improves the accuracy and efficiency of fault identification but also provides more intelligent, intuitive, and real-time support for network operation and maintenance. It can provide network operation and maintenance personnel with a comprehensive and efficient method for fault analysis and location, eliminating the need for on-site personnel, improving inspection efficiency, accuracy, and data visualization, and saving human resources.

[0200] The apparatus provided in the embodiments of the present invention has functions or includes modules that can be used to perform the methods described in the first aspect of the method embodiments above. Its specific implementation and technical effects can be referred to the description of the method embodiments above. For the sake of brevity, it will not be repeated here.

[0201] It should be noted that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0202] Reference Figure 8 This invention provides an electronic device comprising:

[0203] One or more processors 801;

[0204] The memory 802 stores one or more programs, which, when executed by one or more processors, enable the one or more processors to implement any of the network processing methods described above.

[0205] One or more I / O interfaces 803 are connected between the processor and memory and configured to enable information exchange between the processor and memory.

[0206] Among them, processor 801 is a device with data processing capabilities, including but not limited to central processing unit (CPU); memory 802 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (FLASH); I / O interface (read-write interface) 803 is connected between processor 801 and memory 802, and can realize information interaction between processor 801 and memory 802, including but not limited to data bus (Bus).

[0207] In some embodiments, the processor 801, memory 802, and I / O interface 803 are interconnected via a bus, and thus connected to other components of the computing device.

[0208] This embodiment also provides a computer-readable medium storing a computer program thereon. When the program is executed by a processor, it implements the network processing method provided in this embodiment. To avoid repetition, the specific steps of the network processing method will not be repeated here.

[0209] Those skilled in the art will understand that all or some of the steps, systems, or apparatuses in the methods, systems, and apparatuses described above can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0210] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0211] Those skilled in the art will understand that although some embodiments described herein include certain features that are included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this embodiment and form different embodiments.

[0212] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A network processing method, characterized in that, The method includes: Obtain the network topology of the target network, and determine the graph structure data corresponding to each target node in the target network based on the network topology; Based on a preset convolutional neural network and the graph structure data, the network device corresponding to the target node is identified as having device anomalies. In the event of an anomaly in the network device corresponding to the target node, the connectivity of the target link is verified based on the target graph structure data tree corresponding to the target network to obtain the link fault identification result; the target link is the link where the global coordinate data corresponding to the target node is located. The step of determining the graph structure data corresponding to each target node in the target network based on the network topology includes: Obtain the network operation data of the target node; Key performance indicators are extracted from the network operation data; the key performance indicators include at least one of throughput, latency, and packet loss rate. The trends and patterns of the key performance indicators are identified using a preset time series analysis model, and a visualization image is obtained based on the trends and patterns. The visualized image is aligned with the image information of the network device corresponding to the target node to obtain graph structure data; wherein, the image information of the network device corresponding to the target node is captured by an artificial intelligence (AI) sensing device.

2. The network processing method according to claim 1, characterized in that, Before obtaining the network topology of the target network and determining the graph structure data corresponding to each target node in the target network based on the network topology, the method further includes: Based on the network topology and the line vector information between each target node, a graph structure data tree of the target network is created; the line vector information includes the start point, end point, length, bandwidth, and delay information of the line formed by at least two target nodes. The 3D building information model corresponding to the target node is imported into the graph structure data tree to obtain the target graph structure data tree.

3. The network processing method according to claim 1, characterized in that, The step of verifying the connectivity of the target link based on the target graph structure data tree corresponding to the target network to obtain the link fault identification result includes: Obtain the target line vector information of the target link from the target graph structure data tree; Based on the target line vector information, at least one sub-link in the target link is determined; the link start node of the target link is the global coordinate data corresponding to the abnormal target node; Verify the connectivity of each sub-link; if at least one of the sub-links in the target link has abnormal connectivity, determine that the target link is an abnormal link. Based on the abnormal link, determine the link fault identification result.

4. The network processing method according to claim 3, characterized in that, The step of determining the link fault identification result based on the abnormal link includes: A diagnostic request is sent to the end node in the abnormal link; wherein the diagnostic request is used to request the diagnosis of the network status information of the end node, and the network status information includes at least one of the following: operating status, interface information, and traffic statistics; Receive response data from the end node in response to the diagnostic request; wherein the response data includes at least one of the following: current status, connection information, and error information; The response data is parsed to obtain diagnostic results; Based on the diagnostic results, the link fault identification result is determined.

5. The network processing method according to claim 4, characterized in that, After determining the link fault identification result based on the diagnostic result, the method further includes: A test report is generated based on the diagnostic results.

6. The network processing method according to claim 1, characterized in that, Also includes: The network topology of the target network is updated based on the target node corresponding to the abnormal network device and the link fault identification result.

7. The network processing method according to claim 6, characterized in that, The step of updating the network topology of the target network based on the target node corresponding to the abnormal network device and the link failure identification result includes: Receive real-time image information of each network device in the target network, acquired through an AI sensing device; the real-time image information includes real-scene images of the network devices and the location of the network devices; Align the real-world image of the network device with the graph structure data of the target node, and map the real-time image information to the network topology of the target network to determine the mapping relationship of the real-time image information in the network topology. Extract the visual features of the network device from the real-scene image; The visual features are compared with the abnormal target nodes to obtain the comparison results; Based on the comparison results, determine whether the abnormal target node exists in the real-world image; If the abnormal target node exists in the real-world image, the network topology of the target network is updated according to the link fault identification result and the mapping relationship.

8. A network processing device, characterized in that, The device includes: The data acquisition module is used to acquire the network topology of the target network and determine the graph structure data corresponding to each target node in the target network based on the network topology. An abnormal node identification module is used to identify abnormal devices in the network devices corresponding to the target node based on a preset convolutional neural network and the graph structure data. The fault link determination module is used to verify the connectivity of the target link and obtain the link fault identification result based on the target graph structure data tree corresponding to the target network when the network device corresponding to the target node is abnormal; the target link is the link where the global coordinate data corresponding to the target node is located. The step of determining the graph structure data corresponding to each target node in the target network based on the network topology includes: Obtain the network operation data of the target node; Key performance indicators are extracted from the network operation data; the key performance indicators include at least one of throughput, latency, and packet loss rate. The trends and patterns of the key performance indicators are identified using a preset time series analysis model, and a visualization image is obtained based on the trends and patterns. The visualized image is aligned with the image information of the network device corresponding to the target node to obtain graph structure data; wherein, the image information of the network device corresponding to the target node is captured by an artificial intelligence (AI) sensing device.

9. An electronic device, characterized in that, include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method according to any one of claims 1-7; One or more I / O interfaces are connected between the processor and the memory and configured to enable information interaction between the processor and the memory.

Images