Route flapping positioning method and apparatus, non-volatile readable storage medium, and electronic device

Abstract

Embodiments of the present application provide a route flapping positioning method and apparatus, a non-volatile readable storage medium, and an electronic device. The method comprises: acquiring dynamic information of a routing network when route flapping occurs in the routing network, the dynamic information of the routing network being used for representing routing line transformation among a plurality of autonomous domains in the routing network within a preset period in which the route flapping occurs; converting the dynamic information of the routing network into a matching first data matrix; acquiring a second data matrix matching static information of the routing network, the static information of the routing network being used for representing a topological relationship among the plurality of autonomous domains; and using a preset neural network model to perform data fitting on the first data matrix and the second data matrix, so as to obtain a prediction result corresponding to each autonomous domain, and according to the prediction result corresponding to each autonomous domain, determining an abnormal autonomous domain from among the plurality of autonomous domains, the prediction result corresponding to each autonomous domain being used for representing the probability of an abnormality occurring in each autonomous domain.

Description

Method and device for locating routing oscillation, non-volatile readable storage medium, and electronic device

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to a Chinese patent application filed with the Patent Office of China on June 13, 2023, with application number 2023106960600, entitled “Method and device for locating routing oscillations, storage medium and electronic device,” the entire contents of which are incorporated herein by reference. Technical Field

[0003] Embodiments of the present application relate to the field of computers, and in particular, to a method and device for locating routing oscillations, a non-volatile readable storage medium, and an electronic device. Background Art

[0004] Routing is a fundamental concept in data communication networks. Routing is the process of transferring information from a source address to a destination address across interconnected networks. Routing occurs at Layer 3 (the network layer) in the Open Systems Interconnection Reference Model (OSI). Devices that perform routing and forwarding functions are generally referred to as routers.

[0005] The Border Gateway Protocol (BGP) is a protocol used to implement routing information exchange between autonomous domains. However, during the routing information exchange process, due to the different routing exchange protocols and configurations of different autonomous domains, routing policy conflicts may occur, resulting in routing oscillations in the routing network. However, routing networks are generally complex, and it is difficult to directly determine the abnormal autonomous domain based on the routing policies of each autonomous domain.

[0006] It can be seen that the routing oscillation location method in the related art has the problem of low efficiency in locating abnormal autonomous domains due to the complex topological relationship in the routing network.

[0007] Summary of the Invention

[0008] Embodiments of the present application provide a method and apparatus for locating routing oscillations, a non-volatile readable storage medium, and an electronic device to at least address the problem of low efficiency in locating abnormal autonomous domains due to the complex topological relationships in the routing network in the routing oscillation locating methods of the related art.

[0009] According to one embodiment of the present application, a method for locating routing oscillation is provided, comprising: in the event of routing oscillation in a routing network, obtaining dynamic information of the routing network, wherein the dynamic information of the routing network is used to indicate routing line changes between multiple autonomous domains in the routing network within a preset period in which routing oscillation occurs; converting the dynamic information of the routing network into a first data matrix that matches the dynamic information of the routing network; obtaining a second data matrix that matches the static information of the routing network, wherein the static information of the routing network is used to indicate a topological relationship between multiple autonomous domains; performing data fitting on the first data matrix and the second data matrix using a preset neural network model to obtain a prediction result corresponding to each of the multiple autonomous domains, and determining an abnormal autonomous domain from the multiple autonomous domains based on the prediction result corresponding to each autonomous domain, wherein the prediction result corresponding to each autonomous domain is used to indicate the probability of an abnormality occurring in each autonomous domain.

[0010] According to another embodiment of the present application, a positioning device for routing oscillation is provided, comprising: a first acquisition unit, configured to acquire dynamic information of the routing network when routing oscillation occurs in the routing network, wherein the dynamic information of the routing network is configured to represent the routing line transformation between multiple autonomous domains in the routing network within a preset period when routing oscillation occurs; a first conversion unit, configured to convert the dynamic information of the routing network into a first data matrix that matches the dynamic information of the routing network; a second acquisition unit, configured to acquire a second data matrix that matches the static information of the routing network, wherein the static information of the routing network is configured to represent the topological relationship between multiple autonomous domains; a fitting unit, configured to perform data fitting on the first data matrix and the second data matrix using a preset neural network model to obtain a prediction result corresponding to each of the multiple autonomous domains, wherein the prediction result corresponding to each autonomous domain is set to represent the probability of an abnormality occurring in each autonomous domain; and a first determination unit, configured to determine an abnormal autonomous domain from the multiple autonomous domains based on the prediction result corresponding to each autonomous domain.

[0011] According to another embodiment of the present application, a non-volatile readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when running.

[0012] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0013] Through the embodiments of the present application, a method of combining the dynamic information and static information of the routing network to locate abnormal autonomous domains is adopted. When routing oscillation occurs in the routing network, the dynamic information of the routing network is obtained, and the dynamic information indicates the routing line transformation between multiple autonomous domains in the routing network within a preset period of routing oscillation; so that when routing oscillation occurs in the routing network, the routing line transformation information between autonomous domains can be periodically obtained; the dynamic information of the routing network is converted into a first data matrix that matches the dynamic information of the routing network, and a second data matrix that matches the static information of the routing network is obtained; the routing line transformation between autonomous domains of the routing network and the topological relationship between autonomous domains are respectively converted into data matrices to facilitate the subsequent determination of abnormal autonomous domains; a preset neural network model is used to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each autonomous domain, and the abnormal autonomous domain is determined based on the prediction result corresponding to each autonomous domain, wherein the prediction result is used to represent the probability of an abnormality in each autonomous domain. This approach can reduce the impact of complex topological relationships on the efficiency of locating abnormal autonomous domains in routing networks when route oscillation occurs, achieving the technical effect of improving the efficiency of locating abnormal autonomous domains. This solves the problem of low efficiency in locating abnormal autonomous domains caused by complex topological relationships in routing networks in existing routing oscillation locating methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] FIG1 is a schematic diagram of the architecture of an optional positioning system for routing oscillation provided in an embodiment of the present application;

[0015] FIG2 is a schematic structural diagram of an optional electronic device provided in an embodiment of the present application;

[0016] FIG3 is a flow chart of an optional method for locating route oscillation according to an embodiment of the present application;

[0017] FIG4 is a schematic diagram of an optional boundary vector map according to an embodiment of the present application;

[0018] FIG5 is a schematic diagram of an optional adjacency matrix conversion according to an embodiment of the present application;

[0019] FIG6 is a schematic diagram of an optional data matrix processing according to an embodiment of the present application;

[0020] FIG7 is a structural block diagram of an optional positioning device for routing oscillation according to an embodiment of the present application;

[0021] FIG8 is a structural block diagram of a computer system of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] The embodiments of the present application provide a method and apparatus for locating routing oscillations, a non-volatile readable storage medium, and an electronic device, which can quickly and accurately locate abnormal autonomous domains. The following describes exemplary applications of the electronic devices provided by the embodiments of the present application. The electronic devices provided by the embodiments of the present application can be implemented as various types of terminal devices, such as routing devices, laptop computers, tablet computers, desktop computers, set-top boxes, mobile devices (e.g., mobile phones, portable music players, personal digital assistants, dedicated messaging devices, portable gaming devices), and can also be implemented as servers or other types of processing devices. The following describes exemplary applications when the device is implemented as a routing device.

[0025] According to one aspect of an embodiment of the present application, a method for positioning a routing oscillation is provided. Optionally, in this embodiment, the above-mentioned method for positioning a routing oscillation can be applied to a positioning system for routing oscillation as shown in FIG1 . As shown in FIG1 , the positioning system 100 for routing oscillation may include: a routing network 101, a network 102 and a processing device 103. In order to realize a positioning application supporting routing oscillation, the routing network 101 includes a plurality of autonomous domains (autonomous domain 1, autonomous domain 2, and autonomous domain 3 are shown as examples), and the autonomous domains can exchange routing information based on a preset border gateway protocol. The routing network 101 can be connected to the processing device 103 through the network 102, and the network 102 may include but is not limited to at least one of the following: a wired network, a wireless network. The above-mentioned wired network may include but is not limited to at least one of the following: a wide area network, a metropolitan area network, and a local area network. The above-mentioned wireless network may include but is not limited to at least one of the following: WIFI (Wireless Fidelity), Bluetooth.

[0026] The processing device 103 can be configured to obtain static information of the routing network and, in the event of routing oscillation in the routing network 101, obtain dynamic information of the routing network 101. The dynamic information of the routing network 101 can be obtained based on configuration information, and the dynamic information of the routing network 101 can be obtained from the border routers of each autonomous domain in the routing network 101 via the network 102. The processing device 103 can convert the dynamic information of the routing network 101 into a matching first data matrix, convert the static information of the routing network 101 into a matching second data matrix, and use a preset neural network model to perform data fitting on the first data matrix and the second data matrix to obtain prediction results corresponding to each autonomous domain of the routing network 101, thereby determining an abnormal autonomous domain (i.e., an autonomous domain that causes routing oscillation and is in an abnormal state). The processing device 103 can send control instructions to the associated autonomous domains in the routing network 101 via the network 102 based on the determined abnormal autonomous domain to suppress network oscillation.

[0027] The network devices in the routing network 101 (for example, border routers or other devices in a certain autonomous domain) and the processing device 103 can be independent physical devices (for example, physical servers), or they can be device clusters (for example, server clusters) or distributed systems composed of multiple physical devices. They can also be cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0028] Optionally, in this embodiment, the above-mentioned method for locating routing oscillations can be performed by an electronic device as shown in Figure 2. As shown in Figure 2, the electronic device 200 can be a physical network device or processing device 103 of the autonomous domain in the above-mentioned routing network 101. The electronic device 200 includes: at least one processor 201, at least one network interface 202, a bus system 203 and a memory 204. The various components in the electronic device 200 are coupled together through the bus system 203. It can be understood that the bus system 203 is configured to achieve connection and communication between these components. In addition to the data bus, the bus system 203 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, all buses are labeled as bus system 203 in Figure 2.

[0029] The processor 201 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0030] The memory 204 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical drives, etc. The memory 204 may optionally include one or more storage devices that are physically remote from the processor 201.

[0031] The memory 204 includes volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 204 described in the embodiments of the present application is intended to include any suitable type of memory.

[0032] In some embodiments, the memory 204 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplified below.

[0033] Operating system 2041, including system programs configured to handle various basic system services and perform hardware-related tasks, such as the framework layer, core library layer, and driver layer, configured to implement various basic services and handle hardware-based tasks;

[0034] The network communication module 2402 is configured to reach other computing devices via one or more (wired or wireless) network interfaces 202. Exemplary network interfaces 202 include Bluetooth, Wi-Fi, and USB.

[0035] In some embodiments, the apparatus provided in the embodiments of the present application can be implemented using software. FIG2 shows a routing oscillation location apparatus 2043 stored in memory 204. This apparatus can be software in the form of a program or plug-in, and includes the following software units: a first acquisition unit, a first conversion unit, a second acquisition unit, a fitting unit, and a first determination unit. These units are logical and can be arbitrarily combined or separated depending on the functions implemented. The functions of each unit will be described below.

[0036] When the Border Gateway Protocol is applied in a complex network environment, the presence of a large number of autonomous domains in the routing network causes frequent routing line changes. Frequent routing line switching consumes a large amount of bandwidth and processor resources, resulting in routing oscillation, which affects the normal operation of the routing network.

[0037] In the related art, a static method is usually adopted to analyze the routing strategies of each known autonomous domain under the routing network, determine the abnormal autonomous domain and change the routing strategy of the abnormal autonomous domain to suppress routing oscillation. However, when the routing network architecture is relatively complex, there are a large number of autonomous domains in the routing network, resulting in the analysis of the routing strategy of each autonomous domain consuming a lot of processor resources and time, and the efficiency of locating the abnormal autonomous domain is low; and because the routing selection of the border gateway protocol is a complex process, the routing strategy of the autonomous domain is a distributed design, and it is difficult to simply use static information to determine routing problems and suppress routing oscillation.

[0038] In this regard, a dynamic method can be used to suppress abnormal autonomous domains when routing oscillations occur. That is, the border routing protocols of each autonomous domain in the routing network use a penalty value to measure the stability of the autonomous domain. The higher the penalty value, the more unstable the autonomous domain. When the penalty value exceeds the suppression threshold, the autonomous domain is regarded as an abnormal autonomous domain and is suppressed in the current routing network. However, when the number of routing oscillations in a complex routing network is large, the dynamic method is easily interfered by a large amount of routing change information, which reduces the performance of the detection method. In addition, since the dynamic method records a large amount of routing information, the bandwidth resource consumption of the routing device is greatly increased.

[0039] Through the embodiments of the present application, a method of combining dynamic information and static information of the routing network to locate abnormal autonomous domains is adopted, and the routing line transformation between multiple autonomous domains in the routing network within the routing oscillation period is set as dynamic information, and the topological structure between multiple autonomous domains in the routing network is set as static information; and the dynamic information and the static information are converted into a first data matrix and a second data matrix respectively, thereby greatly improving the information processing efficiency of the routing device and saving storage resources; using a preset neural network model to fit the first data matrix and the second data matrix to obtain a prediction result corresponding to each autonomous domain, and the prediction result is used to represent the probability of an abnormality in the autonomous domain; thereby, the real-time status of each autonomous domain can be obtained, and the abnormal autonomous domain can be accurately located to quickly stabilize the routing network.

[0040] Optionally, the above-mentioned method for locating route oscillation may be executed solely by processing device 103, or jointly by processing device 103 and network devices in routing network 101, or by other processing devices other than processing device 103 and routing network 101. As an optional implementation, taking the method for locating route oscillation in this embodiment executed by processing device 103 as an example, as shown in FIG3 , the process of the above-mentioned method for locating route oscillation may include the following steps:

[0041] In step S301, dynamic information of the routing network is obtained.

[0042] In the case of routing network oscillation, dynamic information of the routing network is obtained, wherein the dynamic information of the routing network is used to indicate routing line changes between multiple autonomous domains in the routing network within a preset period of routing oscillation.

[0043] In some examples, the unit of the routing network may be one or more common network groups, that is, a separate manageable network unit, such as a local area network in a specific area. The routing network may include multiple autonomous domains.

[0044] Route flapping is used to indicate that a routing line between autonomous domains in a routing network repeatedly disappears or reappears, resulting in an abnormal routing network. The Border Gateway Protocol (BGP) can be used to read the routing information exchange protocol of autonomous domains in the routing network, that is, to obtain the routing lines (communication paths) used to transmit routing information between autonomous domains. The configuration of the Border Gateway Protocol and the routing selection of autonomous domains is related to the topology of the entire routing network and is also related to human factors. The Border Gateway Protocol has the advantages of flexibility and configurability. For example, the flexibility of the Border Gateway Protocol can allow communication between autonomous domains, that is, autonomous domains can share routing information; the configurability of the Border Gateway Protocol can allow the configuration user of the autonomous domain to manually select the routing strategy.

[0045] An autonomous domain is also called an autonomous system (AS). In the Internet, an autonomous domain is a small unit that determines the various routing protocols that should be adopted in the current routing network. An autonomous domain can also be called a routing domain.

[0046] In some embodiments, the dynamic information of the routing network may be the number of changes in routing lines between multiple autonomous domains when the source autonomous domain transmits routing information to the target autonomous domain within a preset period of routing oscillation.

[0047] In some embodiments, the dynamic information of the routing network may be the frequency of routing line changes when the routing information is transmitted from the starting autonomous domain to the target autonomous domain within a preset period of sending routing oscillations, that is, the ratio of the number of routing line changes to the preset period length.

[0048] It should be noted that the processing device can determine whether routing oscillation occurs in each period based on the set time window. The preset period can be the period in which routing oscillation occurs. The length of each period can be set by the user according to needs, or it can be the default setting length of the routing network. This is not limited in this embodiment.

[0049] Through the embodiments of the present application, when a routing network oscillates, dynamic information of the routing network can be obtained according to the period to which the routing oscillation occurs, the period length can be customized, and dynamic information of the routing network can be obtained efficiently.

[0050] In step S302 , the dynamic information of the routing network is converted into a first data matrix that matches the dynamic information of the routing network.

[0051] In some examples, a first data matrix that matches the dynamic information of the routing network can be constructed based on the number of autonomous domains in the routing network; and the element value of each element in the first data matrix is ​​set based on the number of routing line changes between multiple autonomous domains within a preset period of routing oscillation.

[0052] In some embodiments, if there are five autonomous domains in the routing network, a first data matrix with the same number of rows and columns can be constructed, that is, the number of rows and columns of the first data matrix is ​​5. The starting autonomous domain when the starting autonomous domain transmits routing information to the target autonomous domain within a preset period of routing oscillation is used as the horizontal axis of the first data matrix, the target autonomous domain is used as the vertical axis of the first data matrix, and the number of routing line changes from the starting autonomous domain to the target autonomous domain is used as the numerical value of the corresponding element of the first data matrix.

[0053] Through the embodiments of the present application, the dynamic information of the routing network can be visualized, that is, the dynamic information of the routing network can be converted into the form of a data matrix to facilitate subsequent data analysis.

[0054] In step S303, a second data matrix matching the static information of the routing network is obtained.

[0055] In some examples, static information of a routing network is used to represent the topological relationship between multiple autonomous domains. For example, the static information of a routing network may be the number of autonomous domains in the routing network and the communication paths between autonomous domains.

[0056] In some embodiments, the second data matrix matching the static information of the routing network can represent the communication path between autonomous domains, where the directed connection relationship between the autonomous domains can represent the border gateway protocol between the autonomous domains, that is, the communication path between the autonomous domains.

[0057] It should be noted that the static information of the routing network is preset information in the routing network, and the static information of the routing network can be obtained by reading the border gateway protocol of each autonomous domain in the routing network.

[0058] Through the embodiments of the present application, the static information of the routing network can be visualized, that is, the static information of the routing network can be converted into the form of a data matrix to facilitate subsequent data analysis.

[0059] In step S304, a preset neural network model is used to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each autonomous domain in the plurality of autonomous domains.

[0060] In some examples, the preset neural network model can be a neural network model such as a convolutional neural network model or a residual neural network model.

[0061] For routing networks with simple border gateway structures, a shallow convolutional neural network can be used to process the corresponding first data matrix and second data matrix. For routing networks with complex border network structures, a residual neural network can be used to process the corresponding first data matrix and second data matrix. Here, a gateway complexity threshold can be set to represent the complexity of the border network structure of the routing network. For example, a routing network with more than 5 autonomous domains is set as having a complex border gateway structure, and a routing network with less than or equal to 5 autonomous domains is set as having a simple border gateway structure. For example, the shallow convolutional neural network can be LeNet-5, and the residual network model can be ResNet.

[0062] In some embodiments, a convolutional neural network model is used to perform convolution fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each autonomous domain. The value range of the prediction result is 01, which represents the probability of an abnormality in the autonomous domain.

[0063] Optionally, a residual neural network model can be used to perform regression fitting on the first data matrix and the second data matrix, and the number of output nodes of the fully connected layer of the residual neural network model is set to the number of autonomous domains in the routing network to obtain the corresponding prediction results for each autonomous domain. The value range of the prediction results is 01 to represent the probability of an abnormality in the autonomous domain.

[0064] Through the embodiments of the present application, different preset neural network models can be selected to process the data matrix according to the border gateway structure of the routing network, thereby improving data processing efficiency; the abnormal probability of each autonomous domain is expressed in numerical form, making the prediction results visual.

[0065] In step S305, an abnormal autonomous domain is determined from the multiple autonomous domains according to the prediction result corresponding to each autonomous domain.

[0066] In some embodiments, the prediction result corresponding to each autonomous domain is used to represent the probability of an abnormality occurring in each autonomous domain. The abnormal autonomous domain can be used to represent an abnormal autonomous domain in the routing network when network oscillation occurs. Here, the abnormal autonomous domain cannot normally transmit routing information to other autonomous domains.

[0067] The priority of the routing policy corresponding to each autonomous domain in the routing network can be adjusted according to the numerical value of the prediction result corresponding to each autonomous domain in the routing network. For example, the prediction results of each autonomous domain in the routing network are sorted in reverse order to obtain an autonomous domain priority list; the priority of the routing policy corresponding to the autonomous domain in the front of the autonomous domain priority list is higher than the priority of the routing policy corresponding to the autonomous domain in the back.

[0068] Through the embodiments of the present application, the numerical values ​​of the corresponding prediction results of each autonomous domain are sorted to determine the abnormal autonomous domain and the priority of each autonomous domain, which can quickly and accurately locate the abnormal autonomous domain and obtain the situation of each autonomous domain.

[0069] Through the above steps of the embodiment of the present application, when routing oscillation occurs in the routing network, dynamic information of the routing network is obtained, wherein the dynamic information of the routing network is used to indicate the routing line transformation between multiple autonomous domains in the routing network within a preset period when routing oscillation occurs; the dynamic information of the routing network is converted into a first data matrix that matches the dynamic information of the routing network; a second data matrix that matches the static information of the routing network is obtained, wherein the static information of the routing network is used to indicate the topological relationship between multiple autonomous domains; a preset neural network model is used to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each of the multiple autonomous domains, and based on the prediction result corresponding to each autonomous domain, an abnormal autonomous domain is determined from the multiple autonomous domains, wherein the prediction result corresponding to each autonomous domain is used to indicate the probability of an abnormality in each autonomous domain, thereby solving the problem of low efficiency of abnormal autonomous domain positioning due to the complex topological relationship in the routing network in the routing oscillation positioning method in the related art, thereby improving the efficiency of abnormal autonomous domain positioning.

[0070] In an exemplary embodiment, converting the dynamic information of the routing network into a first data matrix that matches the dynamic information of the routing network includes:

[0071] S11, determining the number of line changes between each autonomous domain and an adjacent domain of each autonomous domain in the plurality of autonomous domains based on dynamic information of the routing network;

[0072] S12 , generating a first data matrix that matches the dynamic information of the routing network according to the number of line changes between each autonomous domain and its adjacent domains.

[0073] In some examples, dynamic information about a routing network is used to indicate routing line changes between multiple autonomous domains in the routing network within a preset period of routing oscillation. Here, based on the routing line change information between the multiple autonomous domains, the number of routing line changes between each autonomous domain and its adjacent autonomous domain (neighboring domain) in the routing network is determined. Neighboring domains of an autonomous domain are used to indicate autonomous domains that have a direct communication path with the current autonomous domain. That is, autonomous domains can directly transmit routing information to their adjacent domains without forwarding by other autonomous domains.

[0074] In some embodiments, a first data matrix with the same number of rows and columns is generated based on the number of autonomous domains in the routing network; the number of line changes between each autonomous domain and its adjacent autonomous domain is used as the value of the corresponding element in the first data matrix to generate a first data matrix that matches the dynamic information of the routing network.

[0075] Through the embodiments of the present application, a first data matrix matching the dynamic information of the routing network can be generated according to the number of line changes between adjacent autonomous domains in the routing network, thereby visualizing the line changes between autonomous domains to facilitate subsequent data analysis and processing.

[0076] In an exemplary embodiment, generating a first data matrix that matches dynamic information of a routing network based on the number of line changes between each autonomous domain and its neighboring domains includes:

[0077] S21, constructing a first initial data matrix having the same number of rows and columns based on the number of autonomous domains in the routing network, wherein the number of rows and columns of the first initial data matrix is ​​the same as the number of autonomous domains in the routing network;

[0078] S22 , setting the element value of each element in the first initial data matrix according to the number of line changes between each autonomous domain and its adjacent domain, to obtain a first data matrix.

[0079] For example, referring to Figure 4, which is a boundary vector diagram provided in an embodiment of the present application, the diagram includes five autonomous domains, namely AS1, AS2, AS3, AS4, and AS5; the directed connections between the autonomous domains represent the communication paths between the autonomous domains. A first initial data matrix with 5 rows and columns can be constructed. Based on dynamic information in the routing network, the number of line changes between each autonomous domain and its adjacent domains is obtained and set as the element value of each element in the first initial data matrix to obtain a first data matrix.

[0080] Here, the first data matrix can be used as the third layer data matrix (channel 3) of the multi-layer data (feature) matrix in the preset neural network model.

[0081] Please refer to Table 1, which is the first data matrix table provided in an embodiment of the present application, wherein the row autonomous domain (the autonomous domain corresponding to the row) represents the starting autonomous domain for transmitting routing information, the column autonomous domain (the autonomous domain corresponding to the column) represents the target autonomous domain for transmitting routing information, and the element value represents the number of line changes between each autonomous domain and the adjacent domain of each autonomous domain.

[0082] Table 1

[0083] Through the embodiments of the present application, a first data matrix is ​​constructed based on the number of autonomous domains in the routing network, and the number of line changes between each autonomous domain and the adjacent domains of each autonomous domain is used as the value of the first data matrix; the number of line changes of autonomous domains in the routing network is visualized, and at the same time, the data matrix form facilitates subsequent data analysis and processing.

[0084] In an exemplary embodiment, the element value of each element in the first initial data matrix is ​​set according to the number of line changes between each autonomous domain and its adjacent domain to obtain the first data matrix, including:

[0085] S31, setting the element value of each element with different row and column numbers in the first initial data matrix to the line conversion frequency between the autonomous domain corresponding to the row and the autonomous domain corresponding to the column, and setting the element value of each element with the same row and column number in the first initial data matrix to 0, to obtain the first data matrix.

[0086] In some examples, the frequency of line changes between the autonomous domain corresponding to the row and the autonomous domain corresponding to the column of an element may be a proportional relationship between the number of line changes between the autonomous domain corresponding to the row and the autonomous domain corresponding to the column of the element and the duration of a preset period. Here, the duration of the preset period may be set based on dynamic or static information of the routing network, or the duration between two time nodes may be specified as the duration of the preset period.

[0087] Through the embodiments of the present application, multiple methods for determining the numerical values ​​of elements in the first data matrix are provided to flexibly generate a first data matrix that matches the dynamic information of the routing network, thereby meeting various routing oscillation determination requirements.

[0088] In an exemplary embodiment, obtaining a second data matrix matching static information of a routing network includes:

[0089] S41, obtaining a border vector map corresponding to the routing network, wherein the border vector map is a directed graph constructed according to a border gateway protocol between multiple autonomous domains and used to represent a topological structure between the multiple autonomous domains, and the static information of the routing network includes the border vector map;

[0090] S42: Convert the boundary vector map into an adjacency matrix matching the boundary vector map to obtain a first sub-data matrix, wherein the second data matrix includes the first sub-data matrix.

[0091] The second data matrix can be pre-generated and stored, or generated after determining that routing oscillation has occurred. In some examples, the second data matrix can include an adjacency matrix converted from a boundary vector map corresponding to the routing network, i.e., a first sub-data matrix. Here, the static information of the routing network includes a boundary vector map, which is a directed graph constructed according to a border gateway protocol between multiple autonomous domains and used to represent the topological structure between multiple autonomous domains.

[0092] The Border Gateway Protocol (BGP) is a protocol used to exchange routing information between autonomous domains. It enables communication between autonomous domains to disseminate routing information. The BGP is an exterior gateway protocol that allows one autonomous domain to communicate with another. The BGP allows an autonomous domain to advertise the reachability of its internal networks (i.e., communication paths) or routing information for other networks reachable through it to other autonomous domains. Furthermore, autonomous domains can retrieve this information from other autonomous domains.

[0093] In some examples, a boundary vector diagram is used to represent a directed graph of the topological structure between multiple autonomous domains. Refer to Figure 4, which is a boundary vector diagram provided in an embodiment of the present application; it includes 5 autonomous domains, namely AS1, AS2, AS3, AS4, and AS5; the directed connections between the autonomous domains represent the communication paths between the autonomous domains.

[0094] In some examples, a graph embedding process is performed on the boundary vector map in the static information of the routing network, and the graph is encoded based on the topological structure and node-to-node relationships of the graph to obtain a first sub-data matrix that matches the static information of the routing network. Optionally, for a scenario in which a second data matrix is ​​generated after determining that routing oscillation has occurred, a boundary vector map corresponding to the routing network can be obtained, and the boundary vector map can be converted into an adjacency matrix that matches the boundary vector map to obtain a first sub-data matrix. Here, the second data matrix includes at least one sub-data matrix, and the first sub-data matrix belongs to the second data matrix.

[0095] In some embodiments, the adjacency matrix is ​​a matrix representing the adjacent relationship between the nodes of the boundary vector map. Referring to Figure 5, Figure 5 is a schematic diagram of the adjacency matrix conversion provided by an embodiment of the present application, including a boundary vector map and an adjacency matrix. As shown in Figure 5, the boundary vector map includes nodes V1, V2, V3, V4, and the directed connection relationship of each node; the row coordinates and column coordinates of the adjacency matrix are the same and are all nodes in the boundary vector map. When there is a directed connection relationship between two nodes, the corresponding element value in the adjacency matrix is ​​1; when there is no directed connection relationship between two nodes, the corresponding element value in the adjacency matrix is ​​0.

[0096] Through the embodiments of the present application, a boundary vector map corresponding to the routing network can be constructed through the border gateway protocols of multiple autonomous domains in the routing network, and a matching adjacency matrix can be obtained as the first sub-data matrix based on the boundary vector map conversion to visualize the relationship between the autonomous domains in the routing network.

[0097] In an exemplary embodiment, before obtaining the boundary vector map corresponding to the routing network, the method further includes:

[0098] S51, determining a communication path between the plurality of autonomous domains based on a border gateway protocol corresponding to the plurality of autonomous domains in the routing network;

[0099] S52 , mapping each of the multiple autonomous domains to a node in a directed graph, and adding a directed edge between two nodes having a communication path between the corresponding autonomous domains to obtain a boundary vector graph.

[0100] In some examples, the Border Gateway Protocol allows an autonomous domain to notify other autonomous domains of the reachability information of its internal network, and can determine the communication path between each autonomous domain based on the Border Gateway Protocol of each autonomous domain in the routing network. The communication path can be unidirectional or bidirectional.

[0101] In some embodiments, each autonomous domain in the routing network is mapped to each node in a directed graph, and directed edges are added between the nodes corresponding to the autonomous domains based on the communication paths between each autonomous domain and its adjacent autonomous domains and the transmission direction of the communication paths to obtain a boundary vector diagram; the boundary vector diagram can be referred to Figure 4 and will not be repeated here.

[0102] Through the embodiments of the present application, a border vector diagram can be constructed based on the autonomous domains in the routing network and the border gateway protocols of the autonomous domains to visualize the communication relationships between the autonomous domains in the routing network.

[0103] In an exemplary embodiment, converting the boundary vector map into an adjacency matrix matching the boundary vector map to obtain a first sub-data matrix includes:

[0104] S61, constructing an initial adjacency matrix with the same number of rows and columns based on the number of nodes in the boundary vector map, wherein the number of rows and columns of the initial adjacency matrix is ​​the number of nodes in the boundary vector map;

[0105] S62, setting the element value of each element in the initial adjacency matrix according to the connectivity relationship between the node corresponding to the row where each element is located and the node corresponding to the column where each element is located, to obtain a first sub-data matrix.

[0106] In some examples, the number of nodes in the boundary vector map is equal to the number of autonomous domains in the routing network. In some embodiments, the boundary vector map can be pre-generated based on preset static information of the routing network, or generated in real time after adjusting communication paths between autonomous domains in the routing network based on dynamic information of the routing network. It should be noted that the first sub-data matrix can change element values ​​in real time as the boundary vector map changes.

[0107] An initial adjacency matrix is ​​constructed based on the boundary vector map. According to the connectivity relationship between the nodes corresponding to the rows and columns of each element in the initial adjacency matrix, the element values ​​of each element in the initial adjacency matrix are set to obtain a first sub-data matrix. Please refer to the above embodiment and will not repeat them here.

[0108] Through the embodiments of the present application, an initial adjacency matrix can be constructed based on the boundary vector diagram of the routing network, and element values ​​can be set according to the connectivity relationship between nodes to obtain a first sub-data matrix representing the connectivity relationship of autonomous domains in the routing network.

[0109] In an exemplary embodiment, the element value of each element in the initial adjacency matrix is ​​set according to the connectivity relationship between the node corresponding to the row where each element is located and the node corresponding to the column where each element is located, to obtain a first sub-data matrix, including:

[0110] S71, setting the element value of each element with the same row and column number in the initial adjacency matrix, and the element value of each element in which there is an edge in the direction from the node corresponding to the row to the node corresponding to the column to the first value, and setting the element value of each element in the initial adjacency matrix in which there is no edge in the direction from the node corresponding to the row to the node corresponding to the column to the second value, to obtain a first sub-data matrix, wherein the first value and the second value are different values.

[0111] In some embodiments, referring to Figure 4, there are 5 autonomous domains in the routing network shown in Figure 4, then the number of boundary vector diagram nodes corresponding to the routing network is 5, and a first sub-data matrix with the same number of rows and columns is constructed, and the number of rows and columns of the first sub-data matrix is ​​both 5, and the 5 nodes are respectively used as the row axis and column axis of the first sub-data matrix.

[0112] Referring to Table 2, Table 2 is a first sub-data matrix provided by an embodiment of the present application, including nodes AS1, AS2, AS3, AS4, and AS5. The corresponding element values ​​in the table are used to indicate whether there is an edge in the direction from the node corresponding to the row of the element to the node corresponding to the column of the element. The element values ​​of each element with the same row and column number, and the element values ​​of each element in the direction from the node corresponding to the row of the element to the node corresponding to the column of the element, are set to the first value, i.e., 1; the element values ​​of each element in the direction from the node corresponding to the row of the element to the node corresponding to the column of the element, where no edge exists, are set to the second value, i.e., 0. In this way, the topological relationship of the boundary vector diagram corresponding to Figure 4 can be converted into a 5*5 matrix form.

[0113] Here, the first sub-data matrix can be used as the first layer data matrix (channel 1) of the multi-layer data (feature) matrix in the preset neural network model.

[0114] Table 2

[0115] Through the embodiments of the present application, an initial adjacency matrix can be constructed based on the boundary vector diagram of the routing network, and the first value and the second value can be set according to the connectivity relationship between the nodes to respectively represent different connection conditions between the nodes.

[0116] In an exemplary embodiment, obtaining a second data matrix matching static information of a routing network further includes:

[0117] S81, constructing a second initial data matrix with the same number of rows and columns based on the number of nodes in the boundary vector map, wherein the number of rows and columns of the second initial data matrix is ​​the same as the number of nodes in the boundary vector map;

[0118] S82, setting the element value of each element in the second initial data matrix according to the weight value of the edge between the node corresponding to the row where each element is located and the node corresponding to the column where the element is located, to obtain a second sub-data matrix, wherein the second data matrix also includes a second sub-data matrix.

[0119] In some examples, the weight value is related to the static information of the routing network. The weight value of the edge between the nodes corresponding to the autonomous domain can be set according to the number of communication paths between each autonomous domain and the domains adjacent to each autonomous domain in the routing network. For example, the weight value of the element with one communication path between the autonomous domains corresponding to two nodes is set to 0.5, and the weight value of the element with two communication paths is set to 1.

[0120] In some examples, the weight value of the corresponding element may also be set according to the quality of the communication path between autonomous domains in the routing network, that is, the load of the communication path is proportional to the weight value of the corresponding element.

[0121] Through the embodiments of the present application, the weight values ​​of the edges between nodes can be customized according to the actual conditions of each autonomous domain in the routing network, so that the obtained second sub-data matrix can meet the predetermined requirements.

[0122] In an exemplary embodiment, the element value of each element in the second initial data matrix is ​​set according to the weight value of the edge between the node corresponding to the row where each element is located and the node corresponding to the column where each element is located in the second initial data matrix, to obtain the second sub-data matrix, including:

[0123] S91, set the element value of each element with different row and column numbers in the second initial data matrix to the weight value of the edge in the direction from the node corresponding to the row to the node corresponding to the column, and set the element value of each element with the same row and column number in the second initial data matrix to the number of connections between the node corresponding to the row and other nodes in the boundary vector diagram except the node corresponding to the row, to obtain the second sub-data matrix.

[0124] In some embodiments, referring to Figure 4, the routing network shown in Figure 4 has 5 autonomous domains, then the number of boundary vector diagram nodes corresponding to the routing network is 5, and a second initial data matrix with the same number of rows and columns is constructed, and the number of rows and columns of the first sub-data matrix is ​​both 5, and the 5 nodes are respectively used as the row axis and column axis of the first sub-data matrix.

[0125] Refer to Table 3, which is a schematic diagram of the second sub-data matrix provided in an embodiment of the present application, including nodes AS1, AS2, AS3, AS4, and AS5. The element values ​​of each element with the same row and column number are set to the number of connections between the node corresponding to the row and the node other than the node corresponding to the row in the boundary vector diagram (as shown in Figure 4), and the element values ​​of each element with different row and column numbers are set to the weight value of the edge in the direction from the node corresponding to the row to the node corresponding to the column, thereby obtaining the second sub-data matrix.

[0126] Here, the second sub-data matrix can be used as the second layer data matrix (channel 2) of the multi-layer data (feature) matrix in the preset neural network model.

[0127] Table 3

[0128] It should be noted that the first data matrix and the second data matrix have the same number of rows and columns and data format, and only the element values ​​in the matrices are different. The second data matrix includes a first sub-data matrix and a second sub-data matrix.

[0129] Here, the first data matrix matches the dynamic information of the routing network and can represent the changes in routing lines between multiple autonomous domains within a preset period of routing oscillation; the second data matrix matches the static information of the routing network and is used to represent the topological relationship between multiple autonomous domains, wherein the first sub-data matrix is ​​used to indicate whether there is a communication path between the autonomous domains in the routing network, and the second sub-data matrix is ​​used to indicate the number of communication paths between each autonomous domain in the routing network and the adjacent domains of each autonomous domain, as well as the weight of each communication path.

[0130] Through the embodiments of the present application, the numerical values ​​of elements with the same number of rows and columns can be set according to the number of connections from each node to other nodes, and the numerical values ​​of elements with different numbers of rows and columns can be set according to the weight value of the edge in the direction from the node corresponding to the row of the element to the node corresponding to the column of the element, thereby making the obtained second sub-data matrix meet the predetermined weight and number of node connections.

[0131] In an exemplary embodiment, obtaining a second data matrix matching static information of a routing network includes:

[0132] S101: Read a second data matrix that matches static information of a routing network according to a preset storage path, wherein the second data matrix is ​​pre-stored in the preset storage path.

[0133] In some embodiments, existing routing network static information can be read from a preset storage path to obtain a boundary vector map corresponding to the routing network. The boundary vector map is converted into a corresponding adjacency matrix to pre-generate a second data matrix. The generated second data matrix is ​​then stored in the preset storage path for subsequent access. The second data matrix includes a first sub-data matrix and a second sub-data matrix, and the generation method is not further described here.

[0134] In some embodiments, according to a preset period of routing oscillation, a second data matrix matching the static information of the routing network is periodically read from a preset storage path, so as to change the value of the second data matrix in real time according to changes in the static information of the routing network.

[0135] It should be noted that the static information of the routing network will be passively updated according to the dynamic information of the routing network. For example, when routing oscillation occurs in the routing network, the abnormal autonomous domain and the communication paths related to the abnormal autonomous domain will be temporarily blocked. Here, the blocking duration can be the same as the preset period of routing oscillation, or it can be set by the routing network.

[0136] Through the embodiments of the present application, the second data matrix can be read according to a preset storage path, and the current static matrix can also be generated or updated in real time according to the static information of the routing network, thereby improving the generation and update efficiency of the second data matrix.

[0137] In an exemplary embodiment, when routing flapping occurs in a routing network, obtaining dynamic information of the routing network includes:

[0138] S111 , when route oscillation occurs in the routing network, dynamic information of each autonomous domain is obtained from a border server of each autonomous domain in a plurality of autonomous domains to obtain dynamic information of the routing network.

[0139] The dynamic information of each autonomous domain is used to indicate the routing line change between each autonomous domain and the adjacent domains of each autonomous domain in the multiple autonomous domains within a preset period when routing oscillation occurs.

[0140] In some embodiments, when routing oscillation occurs in the routing network, the routing line transformation information of each autonomous domain within the current routing oscillation period is read to obtain the dynamic information of the current routing oscillation period of each autonomous domain, and the dynamic information of the current routing oscillation period of all autonomous domains is used as the current dynamic information of the routing network.

[0141] In some embodiments, a detection period is pre-set, and the boundary server of each autonomous domain in the routing network is periodically read to obtain dynamic information of each autonomous domain. Here, the predetermined detection period is related to the static information of the routing network. For example, the detection period interval is positively correlated with the number of autonomous domains in the routing network, that is, the more autonomous domains there are in the routing network, the longer the interval of the pre-set detection period.

[0142] Through the embodiments of the present application, in the event of routing oscillation, the dynamic information of each autonomous domain can be automatically read to store the dynamic information of the autonomous domain in real time.

[0143] In an exemplary embodiment, when routing flapping occurs in a routing network, obtaining dynamic information of the routing network includes:

[0144] S121 , when routing oscillation occurs in the routing network, extract dynamic information of each autonomous domain from pre-stored routing line change information of each autonomous domain within multiple cycles to obtain dynamic information of the routing network.

[0145] The dynamic information of each autonomous domain is used to indicate the routing line change between each autonomous domain and the adjacent domains of each autonomous domain in the multiple autonomous domains within a preset period when routing oscillation occurs.

[0146] In some embodiments, when routing oscillation occurs in the routing network, all pre-saved routing line transformation information of each autonomous domain in the current cycle and the previous cycle is read to obtain all dynamic information of each autonomous domain, and all routing information of all autonomous domains is used as all dynamic information of the routing network.

[0147] Through the embodiments of the present application, when routing oscillation is sent, dynamic information of the routing network in the current routing oscillation period or dynamic information of all routing networks can be read as needed to meet various routing network detection requirements.

[0148] In an exemplary embodiment, a preset neural network model is used to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each of the plurality of autonomous domains, including:

[0149] S131, input the first data matrix and the second data matrix into the preset neural network model as multi-layer feature matrices, and obtain the target prediction result regressed by the preset neural network model, wherein the target prediction result includes the prediction result corresponding to each autonomous domain, and the first data matrix and the second data matrix correspond to different layers of the multi-layer feature matrix.

[0150] The prediction result corresponding to the autonomous domain is used to represent the probability of anomaly in each autonomous domain. The value range of the prediction result is 01. The larger the value of the prediction result, the greater the probability of anomaly in the corresponding autonomous domain.

[0151] In some embodiments, the matrix structures of the first data matrix and the second data matrix are the same, both are the number of autonomous domains in the routing network, and only the stored routing information is different.

[0152] For example, the first data matrix matches the dynamic information of the routing network and can represent the changes in routing lines between multiple autonomous domains within a preset period of routing oscillation; the second data matrix matches the static information of the routing network and is used to represent the topological relationship between multiple autonomous domains, wherein the first sub-data matrix is ​​used to indicate whether there is a communication path between the autonomous domains in the routing network, and the second sub-data matrix is ​​used to indicate the number of communication paths between each autonomous domain in the routing network and the adjacent domains of each autonomous domain, as well as the weight of each communication path.

[0153] Refer to Figure 6, which is a schematic diagram of data matrix processing provided by an embodiment of the present application, including a first data matrix, a first sub-data matrix, a second sub-data matrix, a preset neural network model and an abnormal autonomous domain (target prediction result).

[0154] The preset neural network model can be a convolutional neural network or a residual neural network, etc., including an input layer, a hidden layer, and an output layer. Here, the input layer is used to receive the first data matrix and the second data matrix (including the first sub-data matrix and the second sub-data matrix), the hidden layer can be a convolutional layer, a pooling layer, etc. for performing convolution, fitting, regression and other processing, and the output layer can be a fully connected layer for outputting the prediction results of each autonomous domain.

[0155] In some embodiments, the first data matrix and the second data matrix are integrated into a three-dimensional data matrix. The data structure of the three-dimensional data matrix is ​​similar to the data structure of the convolutional neural network and has a topological structure. Therefore, feature extraction is performed through a pre-trained convolutional neural network to obtain the prediction results for each autonomous domain.

[0156] In some embodiments, the first data matrix and the second data matrix can also be fitted through a residual neural network. On the fully connected layer of the residual neural network, the number of output nodes is set to the number of autonomous domains to obtain the prediction results for each autonomous domain.

[0157] It should be noted that the multi-layer feature matrix may include data matrices for multiple channels, and may include a first data matrix and a second data matrix. The first data matrix may include at least one sub-data matrix, and the second data matrix may include at least one sub-data matrix. Each sub-data matrix may have the same number of rows and columns as the number of autonomous domains in the routing network. Each sub-data matrix may correspond to the data matrix of one channel in the multi-layer feature matrix.

[0158] In some embodiments, the topological relationship of the routing network, that is, the edge information (communication path) between autonomous domains, is used as the first layer data matrix (channel 1) of the multi-layer data matrix, and the static information of other routing networks is used as the second layer data matrix (channel 2) of the multi-layer data matrix. Here, the static information of other routing networks can be the priority of the autonomous domain routing strategy, the number of communication paths of the autonomous domain, etc.; the frequency of change of routing lines in the routing network within a certain period of time (for example, the preset period of routing oscillation) is used as the third layer data matrix (channel 3) of the multi-layer data matrix. The storage information of the third layer data matrix also includes the frequency of change of routing lines when the autonomous domain in the routing network sends and receives routing information from adjacent domains; the three layers of data matrices are collectively used as the feature data of the preset neural network.

[0159] For example, the first sub-data matrix can be used as the data matrix of channel 1, the second sub-data matrix can be used as the data matrix of channel 2, and the first data matrix can be used as the data matrix of channel 3, thereby obtaining a three-dimensional data matrix.

[0160] Since the form of the multi-layer data matrix is ​​similar to the data features of the convolutional neural network and has a certain topological structure, convolution can be used to extract features from the data matrix. For complex network structures, we can use residual neural networks for fitting. On the fully connected layer, the number of output nodes is set to the number of domains. Based on the output, the domains with a greater possibility of problems are determined, and their suppression is increased to achieve the purpose of quickly stabilizing the network.

[0161] Through the embodiments of the present application, data matrix processing is performed through a pre-trained neural network model, which can efficiently integrate the static information and dynamic information of the routing network corresponding to the data matrix to accurately output the prediction results of each autonomous domain.

[0162] In an exemplary embodiment, after determining an abnormal autonomous domain from the plurality of autonomous domains based on the prediction results corresponding to each autonomous domain, the method further includes:

[0163] S141, generating labels corresponding to the first data matrix and the second data matrix according to autonomous domains in which abnormalities occur and autonomous domains in which no abnormalities occur among the plurality of autonomous domains;

[0164] S142, storing the first data matrix and the second data matrix as a new training sample of the preset neural network model, wherein the sample label of the new training sample is a label corresponding to the first data matrix and the second data matrix, and the new training sample is used to perform model training on the preset neural network model to update the model parameters of the preset neural network model.

[0165] In some embodiments, a training loss value is determined based on the labels corresponding to the first data matrix of the training sample set, the labels corresponding to the second data matrix, and the true abnormal autonomous domain labels, and the current neural network model is iteratively trained according to the training loss value and the training sample set until the training loss value meets the training end condition, and the model at the end of training is determined as the updated preset neural network model;

[0166] Through the embodiments of the present application, a training database for a neural network is constructed, and static and dynamic information of the routing network is recorded when the network oscillates to be used as subsequent training data, thereby quickly locating abnormal autonomous domains.

[0167] In an exemplary embodiment, determining an abnormal autonomous domain from a plurality of autonomous domains based on the prediction results corresponding to each autonomous domain includes:

[0168] S151 , based on the prediction results corresponding to each autonomous domain, select an autonomous domain whose probability of abnormality is greater than or equal to a preset probability threshold from multiple autonomous domains to obtain an abnormal autonomous domain.

[0169] The prediction result corresponding to the autonomous domain represents the probability of anomaly occurring in the autonomous domain. The value range of the prediction result is 01. The larger the value of the prediction result, the greater the probability of anomaly occurring in the corresponding autonomous domain. The preset probability threshold can be related to the static information of the routing network.

[0170] For example, a preset probability threshold is set according to the number of autonomous domains in the routing network, wherein the preset probability threshold is positively correlated with the number of autonomous domains in the routing network.

[0171] It should be noted that the preset probability threshold may also be related to static information of other routing networks, such as the maximum load of the communication path between autonomous domains.

[0172] Through the embodiments of the present application, it is possible to automatically locate abnormal autonomous domains based on the prediction results of each autonomous domain and a preset probability threshold.

[0173] In an exemplary embodiment, after determining an abnormal autonomous domain from the plurality of autonomous domains based on the prediction results corresponding to each autonomous domain, the method further includes:

[0174] S161: Send policy adjustment instruction information to border routers of adjacent domains of an abnormal autonomous domain among multiple autonomous domains, wherein the policy adjustment instruction information is used to instruct to lower the priority of the routing policy associated with the abnormal autonomous domain.

[0175] The adjacent domain of an autonomous domain is used to represent the autonomous domain that has a direct communication path with the current autonomous domain. That is, the autonomous domain and the adjacent domain of the autonomous domain can directly transmit routing information through the border router without the need for forwarding by other autonomous domains.

[0176] Policy adjustment indication information is sent to the border routers of the adjacent domains of the abnormal autonomous domain, so that the adjacent domains respond to the policy adjustment indication information and change the current routing policy. For example, the communication path of other normal autonomous domains is selected for forwarding, and the priority of the routing policy of the abnormal autonomous domain is adjusted. According to the output results of the routing network, the abnormal autonomous domain and the autonomous domain with a higher possibility of abnormality (the autonomous domain in which the probability of abnormality among multiple autonomous domains is greater than or equal to the preset probability threshold) are inspected and repaired, thereby solving the routing oscillation of the routing network.

[0177] In some embodiments, based on the border gateway protocol of the normal autonomous domain, the border network protocol of the abnormal autonomous domain is adjusted or the maximum load of the abnormal autonomous domain is reduced so that the abnormal autonomous domain is restored to a normal autonomous domain.

[0178] Through the embodiments of the present application, after locating an abnormal autonomous domain in a routing network, the routing oscillation in the routing network is automatically suppressed by adjusting the priority of the routing policy. At the same time, the abnormal autonomous domain is adapted and adjusted according to the border gateway protocol of the normal autonomous domain to alleviate the impact of the abnormal autonomous domain. The routing oscillation locating method of the embodiment of the present application is more accurate than traditional static methods and does not require a large amount of testing compared to dynamic methods. It improves efficiency while reducing costs and can suppress network oscillation faster and better.

[0179] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all optional embodiments, and the actions and modules involved are not necessarily required by this application.

[0180] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a non-volatile readable storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the method of each embodiment of the present application.

[0181] According to another aspect of the embodiments of the present application, a device for locating route oscillation is also provided, which is used to implement the method for locating route oscillation provided in the above embodiments. Details that have been described will not be repeated here. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are optionally implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0182] FIG7 is a structural block diagram of an optional positioning device for routing oscillation according to an embodiment of the present application. As shown in FIG7 , the device includes:

[0183] The first acquiring unit 701 is configured to acquire dynamic information of the routing network when routing flapping occurs in the routing network, wherein the dynamic information of the routing network is configured to indicate routing line changes between multiple autonomous domains in the routing network within a preset period of routing flapping;

[0184] A first conversion unit 702 is configured to convert the dynamic information of the routing network into a first data matrix that matches the dynamic information of the routing network;

[0185] A second acquiring unit 703 is configured to acquire a second data matrix that matches static information of the routing network, wherein the static information of the routing network is configured to represent a topological relationship between multiple autonomous domains;

[0186] a fitting unit 704 configured to perform data fitting on the first data matrix and the second data matrix using a preset neural network model to obtain a prediction result corresponding to each of the plurality of autonomous domains, wherein the prediction result corresponding to each autonomous domain is configured to represent a probability of an abnormality occurring in each autonomous domain;

[0187] The first determining unit 705 is configured to determine an abnormal autonomous domain from the multiple autonomous domains according to the prediction result corresponding to each autonomous domain.

[0188] It should be noted that the first acquisition unit 701 in this embodiment can be set to execute the above step S301, the first conversion unit 702 in this embodiment can be set to execute the above step S302, the second acquisition unit 703 in this embodiment can be set to execute the above step S303, the fitting unit 704 in this embodiment can be set to execute the above step S304, and the first determination unit 705 in this embodiment can be set to execute the above step S305.

[0189] Through the embodiments provided by the present application, when routing oscillation occurs in a routing network, dynamic information of the routing network is obtained, wherein the dynamic information of the routing network is set to represent the routing line transformation between multiple autonomous domains in the routing network within a preset period when routing oscillation occurs; the dynamic information of the routing network is converted into a first data matrix that matches the dynamic information of the routing network; a second data matrix that matches the static information of the routing network is obtained, wherein the static information of the routing network is set to represent the topological relationship between multiple autonomous domains; a preset neural network model is used to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each of the multiple autonomous domains, and based on the prediction result corresponding to each autonomous domain, an abnormal autonomous domain is determined from the multiple autonomous domains, wherein the prediction result corresponding to each autonomous domain is set to represent the probability of an abnormality occurring in each autonomous domain. This solves the problem of low efficiency of abnormal autonomous domain positioning due to the complex topological relationship in the routing network in the routing oscillation positioning method in the related art, thereby improving the efficiency of abnormal autonomous domain positioning.

[0190] In an exemplary embodiment, the first conversion unit includes:

[0191] a determination module configured to determine the number of line changes between each autonomous domain and an adjacent domain of each autonomous domain in the plurality of autonomous domains based on dynamic information of the routing network;

[0192] The generating module is configured to generate a first data matrix matching the dynamic information of the routing network according to the number of line changes between each autonomous domain and its adjacent domains.

[0193] In an exemplary embodiment, the generation module includes:

[0194] A first construction submodule is configured to construct a first initial data matrix having the same number of rows and columns based on the number of autonomous domains in the routing network, wherein the number of rows and columns of the first initial data matrix is ​​the number of autonomous domains in the routing network;

[0195] The first setting submodule is configured to set the element value of each element in the first initial data matrix according to the number of line changes between each autonomous domain and its adjacent domain, so as to obtain the first data matrix.

[0196] In an exemplary embodiment, the first setting submodule includes:

[0197] The first setting subunit is configured to set the element value of each element in the first initial data matrix with different numbers of rows and columns to the line conversion frequency between the autonomous domain corresponding to the row and the autonomous domain corresponding to the column, and to set the element value of each element in the first initial data matrix with the same number of rows and columns to 0, thereby obtaining the first data matrix.

[0198] In an exemplary embodiment, the second acquiring unit includes:

[0199] a first acquisition module configured to acquire a border vector map corresponding to the routing network, wherein the border vector map is a directed graph constructed according to a border gateway protocol between multiple autonomous domains and configured to represent a topology structure between the multiple autonomous domains, and the static information of the routing network includes the border vector map;

[0200] The conversion module is configured to convert the boundary vector map into an adjacency matrix matching the boundary vector map to obtain a first sub-data matrix, wherein the second data matrix includes the first sub-data matrix.

[0201] In an exemplary embodiment, the apparatus further comprises:

[0202] a second determining unit configured to determine, before obtaining a border vector map corresponding to the routing network, a communication path between the plurality of autonomous domains based on a border gateway protocol corresponding to the plurality of autonomous domains in the routing network;

[0203] The execution unit is configured to map each of the plurality of autonomous domains into a node in a directed graph, and add a directed edge between two nodes having a communication path between the corresponding autonomous domains to obtain a boundary vector graph.

[0204] In one exemplary embodiment, the conversion module includes:

[0205] The second construction submodule is configured to construct an initial adjacency matrix with the same number of rows and columns based on the number of nodes in the boundary vector map, wherein the number of rows and columns of the initial adjacency matrix is ​​the number of nodes in the boundary vector map;

[0206] The second setting submodule is configured to set the element value of each element in the initial adjacency matrix according to the connectivity relationship between the node corresponding to the row where each element is located and the node corresponding to the column where the element is located, so as to obtain the first sub-data matrix.

[0207] In an exemplary embodiment, the second setting submodule includes:

[0208] The second setting sub-unit is configured to set the element values ​​of each element in the initial adjacency matrix with the same row and column number, and the element value of each element in which there is an edge from the node corresponding to the row to the node corresponding to the column to the first value, and to set the element value of each element in the initial adjacency matrix in which there is no edge from the node corresponding to the row to the node corresponding to the column to the second value, to obtain a first sub-data matrix, wherein the first value and the second value are different values.

[0209] In an exemplary embodiment, the second acquiring unit further includes:

[0210] A construction module is configured to construct a second initial data matrix with the same number of rows and columns based on the number of nodes in the boundary vector map, wherein the number of rows and columns of the second initial data matrix is ​​the number of nodes in the boundary vector map;

[0211] The setting module is configured to set the element value of each element in the second initial data matrix according to the weight value of the edge between the node corresponding to the row where each element is located and the node corresponding to the column where the element is located, so as to obtain the second sub-data matrix, wherein the second data matrix also includes the second sub-data matrix.

[0212] In one exemplary embodiment, the setup module includes:

[0213] The third setting submodule is configured to set the element value of each element in the second initial data matrix with different numbers of rows and columns to the weight value of the edge in the direction from the node corresponding to the row to the node corresponding to the column, and set the element value of each element in the second initial data matrix with the same number of rows and columns to the number of connections between the node corresponding to the row and other nodes in the boundary vector diagram except the node corresponding to the row, to obtain a second sub-data matrix.

[0214] In an exemplary embodiment, the second acquiring unit includes:

[0215] The reading module is configured to read a second data matrix matching the static information of the routing network according to a preset storage path, wherein the second data matrix is ​​pre-stored in the preset storage path.

[0216] In an exemplary embodiment, the first acquiring unit includes:

[0217] The second acquisition module is configured to obtain dynamic information of each autonomous domain from the boundary server of each autonomous domain in the multiple autonomous domains when routing oscillation occurs in the routing network, thereby obtaining dynamic information of the routing network, wherein the dynamic information of each autonomous domain is configured to indicate the routing line changes between each autonomous domain and the adjacent domains of each autonomous domain in the multiple autonomous domains within a preset period when routing oscillation occurs.

[0218] In an exemplary embodiment, the first acquiring unit includes:

[0219] The third acquisition module is configured to extract dynamic information of each autonomous domain from pre-saved routing line transformation information of each autonomous domain within multiple cycles when routing oscillation occurs in the routing network, thereby obtaining dynamic information of the routing network, wherein the dynamic information of each autonomous domain is configured to represent routing line transformations between each autonomous domain and adjacent domains of each autonomous domain in the multiple autonomous domains within a preset cycle when routing oscillation occurs.

[0220] In an exemplary embodiment, the fitting unit includes:

[0221] The input module is configured to input the first data matrix and the second data matrix as multi-layer feature matrices into a preset neural network model to obtain a target prediction result regressed by the preset neural network model, wherein the target prediction result includes a prediction result corresponding to each autonomous domain, and the first data matrix and the second data matrix correspond to different layers of the multi-layer feature matrix.

[0222] In an exemplary embodiment, the apparatus further comprises:

[0223] a generating unit configured to, after determining an abnormal autonomous domain from a plurality of autonomous domains based on a prediction result corresponding to each autonomous domain, generate labels corresponding to the first data matrix and the second data matrix based on autonomous domains in which the abnormality occurs and autonomous domains in which the abnormality does not occur among the plurality of autonomous domains;

[0224] The storage unit is configured to store the first data matrix and the second data matrix as a new training sample of the preset neural network model, wherein the sample label of the new training sample is a label corresponding to the first data matrix and the second data matrix, and the new training sample is configured to perform model training on the preset neural network model to update the model parameters of the preset neural network model.

[0225] In an exemplary embodiment, the first determining unit includes:

[0226] The selection module is configured to select an autonomous domain whose probability of abnormality is greater than or equal to a preset probability threshold from multiple autonomous domains based on the prediction results corresponding to each autonomous domain, to obtain an abnormal autonomous domain.

[0227] In an exemplary embodiment, the apparatus further comprises:

[0228] The sending unit is configured to send policy adjustment indication information to the border routers of the adjacent domains of the abnormal autonomous domain in the multiple autonomous domains after determining the abnormal autonomous domain from the multiple autonomous domains based on the prediction results corresponding to each autonomous domain, wherein the policy adjustment indication information is configured to indicate to lower the priority of the routing policy associated with the abnormal autonomous domain.

[0229] It should be noted that the above modules can be implemented through software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0230] According to another aspect of the embodiments of the present application, a non-volatile readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when running.

[0231] In an exemplary embodiment, the non-volatile readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0232] According to one aspect of the present application, a computer program product is provided, comprising a computer program / instruction containing program code configured to execute the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication portion 809 and / or installed from a removable medium 811. When the computer program is executed by the central processing unit 801, the various functions provided in the embodiments of the present application are performed. The serial numbers of the embodiments of the present application are for descriptive purposes only and do not represent the merits of the embodiments.

[0233] Fig. 8 schematically shows a computer system structure block diagram of the electronic device configured to implement an embodiment of the present application. As shown in Figure 8, computer system 800 includes a central processing unit 801 (Central Processing Unit, CPU), which can perform various appropriate actions and processes according to the program stored in a read-only memory 802 (Read-Only Memory, ROM) or the program loaded from a storage portion 808 into a random access memory 803 (Random Access Memory, RAM). In the random access memory 803, various programs and data required for system operation are also stored. The central processing unit 801, the read-only memory 802 and the random access memory 803 are connected to each other via a bus 804. An input / output interface 805 (Input / Output interface, i.e., an I / O interface) is also connected to the bus 804.

[0234] The following components are connected to the input / output interface 805: an input section 806 including a keyboard, a mouse, and the like; an output section 807 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 808 including a hard disk; and a communication section 809 including a network interface card such as a local area network card or a modem. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the input / output interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 810 as needed, so that a computer program read therefrom can be installed into the storage section 808 as needed.

[0235] In particular, according to an embodiment of the present application, the processes described in the various method flow charts can be implemented as computer software programs. For example, an embodiment of the present application includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code configured to execute the methods shown in the flow charts. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 809 and / or installed from a removable medium 811. When the computer program is executed by the central processing unit 801, the various functions defined in the system of the present application are executed.

[0236] It should be noted that the computer system 800 of the electronic device shown in FIG8 is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0237] According to another aspect of the embodiments of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any of the above method embodiments.

[0238] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the input / output resource pool, and the input / output device is connected to the input / output resource pool.

[0239] For optional examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementations, and this embodiment will not be described in detail here.

[0240] Obviously, those skilled in the art should understand that the various modules or steps of the above-mentioned embodiments of the present application can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented using program codes executable by the computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0241] The above are merely optional embodiments of the present application and are not intended to limit the embodiments of the present application. For those skilled in the art, the embodiments of the present application may be modified and varied in various ways. Any modifications, equivalent replacements, improvements, etc. made within the principles of the embodiments of the present application shall be included in the scope of protection of the embodiments of the present application.

Claims

1. A method for locating route oscillation, characterized in that: include: In the case where routing oscillation occurs in the routing network, dynamic information of the routing network is obtained, wherein the dynamic information of the routing network is used to indicate the routing line changes between multiple autonomous domains in the routing network within a preset period when the routing oscillation occurs; converting the dynamic information of the routing network into a first data matrix matching the dynamic information of the routing network; Acquire a second data matrix matching the static information of the routing network, wherein the static information of the routing network is used to represent the topological relationship between the multiple autonomous domains; Use a preset neural network model to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each of the multiple autonomous domains, and determine an abnormal autonomous domain from the multiple autonomous domains based on the prediction result corresponding to each autonomous domain, wherein the prediction result corresponding to each autonomous domain is used to represent the probability of an abnormality occurring in each autonomous domain.

2. The method according to claim 1, characterized in that The converting the dynamic information of the routing network into a first data matrix matching the dynamic information of the routing network comprises: Determining the number of line changes between each autonomous domain and an adjacent domain of each autonomous domain in the plurality of autonomous domains according to the dynamic information of the routing network; The first data matrix matching the dynamic information of the routing network is generated according to the number of line changes between each autonomous domain and its adjacent domain.

3. The method according to claim 2, characterized in that The step of generating the first data matrix matching the dynamic information of the routing network according to the number of line changes between each autonomous domain and an adjacent domain of each autonomous domain comprises: Based on the number of autonomous domains in the routing network, construct a first initial data matrix with the same number of rows and columns, wherein the number of rows and columns of the first initial data matrix is ​​the number of autonomous domains in the routing network; According to the number of line changes between each autonomous domain and its adjacent domain, the element value of each element in the first initial data matrix is ​​set to obtain the first data matrix.

4. The method according to claim 3, characterized in that The step of setting the element value of each element in the first initial data matrix according to the number of line changes between each autonomous domain and an adjacent domain of each autonomous domain to obtain the first data matrix includes: The element value of each element with different row and column numbers in the first initial data matrix is ​​set to the line conversion frequency between the autonomous domain corresponding to the row and the autonomous domain corresponding to the column, and the element value of each element with the same row and column number in the first initial data matrix is ​​set to 0, to obtain the first data matrix.

5. The method according to claim 1, characterized in that The acquiring of a second data matrix matching the static information of the routing network comprises: Acquire a border vector map corresponding to the routing network, wherein the border vector map is a directed graph constructed according to a border gateway protocol between the multiple autonomous domains and used to represent the topological structure between the multiple autonomous domains, and the static information of the routing network includes the border vector map; The boundary vector map is converted into an adjacency matrix matching the boundary vector map to obtain a first sub-data matrix, wherein the second data matrix includes the first sub-data matrix.

6. The method according to claim 5, characterized in that Before obtaining the boundary vector graph corresponding to the routing network, the method further includes: Determining, based on border gateway protocols corresponding to the plurality of autonomous domains in the routing network, communication paths between the plurality of autonomous domains; Each of the plurality of autonomous domains is mapped to a node in a directed graph, and a directed edge is added between two nodes having a communication path between the corresponding autonomous domains to obtain the boundary vector graph.

7. The method according to claim 5, characterized in that The step of converting the boundary vector map into an adjacency matrix matching the boundary vector map to obtain a first sub-data matrix includes: Based on the number of nodes in the boundary vector map, construct an initial adjacency matrix with the same number of rows and columns, wherein the number of rows and columns of the initial adjacency matrix is ​​the number of nodes in the boundary vector map; According to the connectivity relationship between the nodes corresponding to the rows and the columns of the elements in the initial adjacency matrix, the element value of each element in the initial adjacency matrix is ​​set to obtain the first sub-data matrix.

8. The method according to claim 7, characterized in that The step of setting the element value of each element in the initial adjacency matrix according to the connectivity relationship between the node corresponding to the row where each element in the initial adjacency matrix is ​​located and the node corresponding to the column where each element is located, to obtain the first sub-data matrix, comprises: The element values ​​of each element with the same row and column number in the initial adjacency matrix and the element value of each element in which there is an edge from the node corresponding to the row to the node corresponding to the column are set to a first value, and the element value of each element in the initial adjacency matrix in which there is no edge from the node corresponding to the row to the node corresponding to the column is set to a second value, to obtain the first sub-data matrix, wherein the first value and the second value are different values.

9. The method according to claim 5, characterized in that The acquiring of a second data matrix matching the static information of the routing network further includes: Based on the number of nodes in the boundary vector graph, construct a second initial data matrix with the same number of rows and columns, wherein the number of rows and columns of the second initial data matrix is ​​the number of nodes in the boundary vector graph; According to the weight value of the edge between the node corresponding to the row where each element in the second initial data matrix is ​​located and the node corresponding to the column where each element is located, the element value of each element in the second initial data matrix is ​​set to obtain a second sub-data matrix, wherein the second data matrix also includes the second sub-data matrix.

10. The method according to claim 9, characterized in that The step of setting the element value of each element in the second initial data matrix according to the weight value of the edge between the node corresponding to the row where each element in the second initial data matrix is ​​located and the node corresponding to the column where each element is located, to obtain the second sub-data matrix, comprises: The element value of each element with different row and column numbers in the second initial data matrix is ​​set to the weight value of the edge in the direction from the node corresponding to the row to the node corresponding to the column, and the element value of each element with the same row and column number in the second initial data matrix is ​​set to the number of connections between the node corresponding to the row and other nodes in the boundary vector diagram except the node corresponding to the row, to obtain the second sub-data matrix.

11. The method according to claim 1, characterized in that: The acquiring of a second data matrix matching the static information of the routing network comprises: The second data matrix matching the static information of the routing network is read according to a preset storage path, wherein the second data matrix is ​​pre-stored in the preset storage path.

12. The method according to claim 1, characterized in that When routing oscillation occurs in the routing network, obtaining dynamic information of the routing network includes: In the case where the routing oscillation occurs in the routing network, the dynamic information of each autonomous domain is obtained from the boundary server of each autonomous domain in the multiple autonomous domains to obtain the dynamic information of the routing network, wherein the dynamic information of each autonomous domain is used to indicate the change of routing lines between each autonomous domain and the adjacent domains of each autonomous domain in the multiple autonomous domains within the preset period when the routing oscillation occurs.

13. The method according to claim 1, characterized in that When routing oscillation occurs in the routing network, obtaining dynamic information of the routing network includes: In the case where the routing oscillation occurs in the routing network, the dynamic information of each autonomous domain is extracted from the pre-saved routing line transformation information of each autonomous domain within multiple cycles to obtain the dynamic information of the routing network, wherein the dynamic information of each autonomous domain is used to indicate the routing line transformation between each autonomous domain and the adjacent domains of each autonomous domain among the multiple autonomous domains within the preset cycle when the routing oscillation occurs.

14. The method according to claim 1, characterized in that The using a preset neural network model to perform data fitting on the first data matrix and the second data matrix to obtain a prediction result corresponding to each autonomous domain in the plurality of autonomous domains includes: The first data matrix and the second data matrix are input into the preset neural network model as multi-layer feature matrices to obtain a target prediction result regressed by the preset neural network model, wherein the target prediction result includes a prediction result corresponding to each autonomous domain, and the first data matrix and the second data matrix correspond to different layers of the multi-layer feature matrix.

15. The method according to claim 1, characterized in that After determining an abnormal autonomous domain from the multiple autonomous domains according to the prediction results corresponding to each autonomous domain, the method further includes: Generate labels corresponding to the first data matrix and the second data matrix according to autonomous domains in which abnormalities occur and autonomous domains in which abnormalities do not occur among the multiple autonomous domains; The first data matrix and the second data matrix are stored as a newly added training sample of the preset neural network model, wherein the sample label of the newly added training sample is the label corresponding to the first data matrix and the second data matrix, and the newly added training sample is used to perform model training on the preset neural network model to update the model parameters of the preset neural network model.

16. The method according to claim 1, characterized in that The step of determining an abnormal autonomous domain from the plurality of autonomous domains according to the prediction result corresponding to each autonomous domain comprises: According to the prediction results corresponding to each autonomous domain, an autonomous domain with a probability of abnormality greater than or equal to a preset probability threshold is selected from the multiple autonomous domains to obtain the abnormal autonomous domain.

17. The method according to any one of claims 1 to 16, characterized in that After determining an abnormal autonomous domain from the multiple autonomous domains according to the prediction results corresponding to each autonomous domain, the method further includes: Sending policy adjustment indication information to a border router of an adjacent domain of the abnormal autonomous domain among the multiple autonomous domains, wherein the policy adjustment indication information is used to indicate to lower the priority of the routing policy associated with the abnormal autonomous domain.

18. A positioning device for routing oscillation, characterized in that: include: A first acquisition unit is configured to acquire dynamic information of the routing network when routing oscillation occurs in the routing network, wherein the dynamic information of the routing network is configured to indicate that routing lines between multiple autonomous domains in the routing network are changed within a preset period when the routing oscillation occurs; A first conversion unit, configured to convert the dynamic information of the routing network into a first data matrix matching the dynamic information of the routing network; A second acquisition unit is configured to acquire a second data matrix matching the static information of the routing network, wherein the static information of the routing network is configured to represent a topological relationship between the multiple autonomous domains; a fitting unit, configured to perform data fitting on the first data matrix and the second data matrix using a preset neural network model to obtain a prediction result corresponding to each of the plurality of autonomous domains, wherein the prediction result corresponding to each autonomous domain is configured to represent a probability of an abnormality occurring in each autonomous domain; The first determining unit is configured to determine an abnormal autonomous domain from the multiple autonomous domains according to a prediction result corresponding to each autonomous domain.

19. A non-volatile readable storage medium, characterized in that: The non-volatile readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 17 when executed by a processor.

20. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 17 are implemented.

Images