Route search method and device, electronic equipment and readable storage medium

Abstract

Embodiments of the present application provide a routing search method and device, electronic equipment and computer readable storage medium, and relate to the technical field of network transmission. The method comprises: constructing a plurality of first network routing paths based on a search starting point network element and a search end point network element; scoring each first network routing path according to a trained routing search model to obtain a target search value corresponding to each first network routing path; selecting a first network routing path with a target search value greater than a preset threshold as a second network routing path to complete a corresponding network transmission service based on the second network routing path. The routing search model is trained by a standard routing path and a candidate routing path corresponding to the standard routing path, the second network routing path is obtained by screening the first network routing path, and the utilization rate of the routing network is improved.

Description

Technical Field

[0001] This application relates to the field of network transmission technology, and more specifically, to a route search method, apparatus, electronic device, and computer-readable storage medium. Background Technology

[0002] With the rapid development of the Internet, IP (Internet Protocol) networks have gradually become the standard for modern networks. When building a network using the IP protocol, routing devices are needed to interconnect various IP subnets, and routing mechanisms are used between IP subnets to form a hierarchical Internet through IP gateways. Multiprotocol label switching is a new generation of high-speed IP backbone network switching standard, which can effectively improve network speed and scalability, and increase routing forwarding rate.

[0003] Multiprotocol Label Switching (MPLS) employs a control-driven model to initialize the allocation and distribution of label bundles, establishing an LSP (Label Switched Path) by connecting several label switching points. In existing technologies, LSPs, or network routing paths, are often manually configured based on experience. This allows for network traffic splitting based on service scenarios or prioritizing routing paths according to network point devices to complete relevant services. However, with the continuous expansion and addition of services, this method is unsuitable for networks with unstable loads and significant topology changes, resulting in low routing and forwarding efficiency. Summary of the Invention

[0004] This application provides a route search method, apparatus, electronic device, and computer-readable storage medium, which can solve the problem of low routing forwarding efficiency. The technical solution is as follows:

[0005] According to one aspect of the embodiments of this application, a route search method is provided, the method comprising:

[0006] Multiple first network routing paths are constructed based on the search starting network element and the search ending network element;

[0007] Each first network routing path is scored based on the trained routing search model to obtain the target search value corresponding to each first network routing path; the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the candidate routing path and the corresponding standard routing path have the same start and end network elements;

[0008] The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0009] In one possible implementation, the above route search model is trained as follows:

[0010] Standard routing paths are obtained by filtering from historical routing data; the standard routing path includes at least one of the primary routing path and the backup routing path.

[0011] Determine multiple candidate routing paths corresponding to each standard routing path;

[0012] The initial model is updated at least once based on the standard route path and candidate route paths until the training termination condition is met. The initial model that meets the training termination condition is then used as the route search model.

[0013] The update operations include:

[0014] Input the routing feature data of the standard routing path and the candidate routing path into the initial model to obtain the search value corresponding to each routing feature data;

[0015] Determine the first difference between the search value of the standard route path and the target threshold, and the second difference between the search value of the candidate route path and the target threshold;

[0016] The parameters of the initial model are updated based on the first and second differences.

[0017] In one possible implementation, the determination of multiple candidate routing paths corresponding to each standard routing path includes:

[0018] Determine the starting and ending network elements for each standard routing path;

[0019] For each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element;

[0020] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, the candidate routing path is determined based on the connection path between the starting network element and the ending network element in the third network level; wherein, the third network level is greater than or equal to the first network level.

[0021] In yet another possible implementation, the above method also includes:

[0022] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, the starting network element or the ending network element is connected upward to the fourth network level. The candidate routing path is determined based on the connection path between the starting network element and the ending network element in the fourth network level. The fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level between the first network level and the second network level.

[0023] In yet another possible implementation, the above method also includes:

[0024] When the core layer network elements connected to the starting network element and the ending network element are different, determine at least two first network elements connected to the starting network element and at least two second network elements connected to the ending network element in the core layer.

[0025] Determine the core layer connectivity path for each network element matching pair; wherein, the network element matching pair is determined by pairwise matching of any first network element and any second network element;

[0026] Based on the connection path of each core layer, construct the corresponding candidate routing path.

[0027] In another possible implementation, the above routing feature data is extracted based on the following method:

[0028] Determine the network data, port data, and topology connection edge data included in each routing path; routing paths include standard routing paths and candidate routing paths;

[0029] Convert network data into network element feature vectors, port data into port feature vectors, and topology connection edge data into edge feature vectors;

[0030] The mean of the feature vectors of each network element is used as the corresponding network element vector, the mean of the feature vectors of each port is used as the corresponding port connection, and the mean of the feature vectors of each edge is used as the edge vector.

[0031] By concatenating the network element vector, port vector, and edge vector, routing feature data is obtained.

[0032] In another possible implementation, the above-mentioned conversion of network data into network element feature vectors includes:

[0033] The network elements in the network data are filtered to obtain the target network elements, and the target network element feature data corresponding to each target network element is determined.

[0034] For each target network element, identify the adjacent network elements that are adjacent to and connected to the target network element, and obtain the adjacent network element data and the weight of each adjacent network element data; perform a weighted summation of the adjacent network element data to obtain the adjacent network element feature data corresponding to the target network element.

[0035] The sum of the target network element feature data and the feature data of the adjacent network elements of each target network element is used as the feature vector of the target network element.

[0036] The network element feature vector corresponding to the network data is determined based on the feature vector of each target network element.

[0037] According to another aspect of the embodiments of this application, a route search device is provided, the device comprising:

[0038] The module is used to construct multiple first network routing paths based on the search starting network element and the search ending network element;

[0039] The scoring module is used to score each first network routing path according to the trained routing search model, and obtain the target search value corresponding to each first network routing path. The target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data. The routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path. The candidate routing path and the corresponding standard routing path have the same start and end network elements.

[0040] The selection module is used to select a first network routing path whose target search value is greater than a preset threshold as a second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0041] In one possible implementation, the above-described apparatus includes a training module for:

[0042] Standard routing paths are obtained by filtering from historical routing data; the standard routing path includes at least one of the primary routing path and the backup routing path.

[0043] Determine multiple candidate routing paths corresponding to each standard routing path;

[0044] The initial model is updated at least once based on the standard route path and candidate route paths until the training termination condition is met. The initial model that meets the training termination condition is then used as the route search model.

[0045] The update operations include:

[0046] Input the routing feature data of the standard routing path and the candidate routing path into the initial model to obtain the search value corresponding to each routing feature data;

[0047] Determine the first difference between the search value of the standard route path and the target threshold, and the second difference between the search value of the candidate route path and the target threshold;

[0048] The parameters of the initial model are updated based on the first and second differences.

[0049] In one possible implementation, the training module described above, when determining multiple candidate routing paths corresponding to each standard routing path, is used for:

[0050] Determine the starting and ending network elements for each standard routing path;

[0051] For each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element;

[0052] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, the candidate routing path is determined based on the connection path between the starting network element and the ending network element in the third network level; wherein, the third network level is greater than or equal to the first network level.

[0053] In yet another possible implementation, the training module described above is also used for:

[0054] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, the starting network element or the ending network element is connected upward to the fourth network level. The candidate routing path is determined based on the connection path between the starting network element and the ending network element in the fourth network level. The fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level between the first network level and the second network level.

[0055] In yet another possible implementation, the training module described above is also used for:

[0056] When the core layer network elements connected to the starting network element and the ending network element are different, determine at least two first network elements connected to the starting network element and at least two second network elements connected to the ending network element in the core layer.

[0057] Determine the core layer connectivity path for each network element matching pair; wherein, the network element matching pair is determined by pairwise matching of any first network element and any second network element;

[0058] Based on the connection path of each core layer, construct the corresponding candidate routing path.

[0059] In another possible implementation, the training module described above is also used for:

[0060] Determine the network data, port data, and topology connection edge data included in each routing path; routing paths include standard routing paths and candidate routing paths;

[0061] Convert network data into network element feature vectors, port data into port feature vectors, and topology connection edge data into edge feature vectors;

[0062] The mean of the feature vectors of each network element is used as the corresponding network element vector, the mean of the feature vectors of each port is used as the corresponding port connection, and the mean of the feature vectors of each edge is used as the edge vector.

[0063] By concatenating the network element vector, port vector, and edge vector, routing feature data is obtained.

[0064] In another possible implementation, the training module described above, when converting network data into network element feature vectors, is used for:

[0065] The network elements in the network data are filtered to obtain the target network elements, and the target network element feature data corresponding to each target network element is determined.

[0066] For each target network element, identify the adjacent network elements that are adjacent to and connected to the target network element, and obtain the adjacent network element data and the weight of each adjacent network element data; perform a weighted summation of the adjacent network element data to obtain the adjacent network element feature data corresponding to the target network element.

[0067] The sum of the target network element feature data and the feature data of the adjacent network elements of each target network element is used as the feature vector of the target network element.

[0068] The network element feature vector corresponding to the network data is determined based on the feature vector of each target network element.

[0069] According to another aspect of the present application, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method shown in the first aspect of the present application.

[0070] According to another aspect of the embodiments of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method shown in the first aspect of the embodiments of this application.

[0071] According to one aspect of the present application, a computer program product is provided, which includes a computer program that, when executed by a processor, implements the steps of the method shown in the first aspect of the present application.

[0072] The beneficial effects of the technical solutions provided in this application are:

[0073] This application embodiment constructs multiple first network routing paths by searching starting and ending network elements. Then, it scores each first network routing path according to a trained routing search model to obtain the target search data value corresponding to each first network routing path. The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, thereby completing the corresponding network transmission service based on the second network routing path. This application embodiment trains the routing search model based on standard routing paths and their corresponding candidate routing paths. Since the target search value can be used to characterize the similarity between the first network routing path and the standard routing path in historical routing data, the routing search model can perform data fitting on historical routing data and filter each first network routing path. Based on the selected second network routing path, the utilization rate of the routing network can be effectively improved. Unlike the manual configuration of network routing paths in the prior art, this application is applicable to high-volume IP networks, enhancing the routing and forwarding efficiency in IP networks. Attached Figure Description

[0074] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0075] Figure 1 This is a schematic diagram illustrating an application scenario of a route search method provided in an embodiment of this application.

[0076] Figure 2 A flowchart illustrating a route search method provided in an embodiment of this application;

[0077] Figure 3 A schematic diagram illustrating the model training process in a route search method provided in this application embodiment;

[0078] Figure 4 A schematic diagram illustrating the process of constructing candidate route paths in a route search method provided in an embodiment of this application;

[0079] Figure 5 This is a schematic diagram illustrating another process for constructing candidate route paths in a route search method provided in an embodiment of this application;

[0080] Figure 6 This is a schematic diagram illustrating another step in constructing candidate routing paths in a routing search method provided in an embodiment of this application.

[0081] Figure 7 A flowchart illustrating an example of a route search method provided in an embodiment of this application;

[0082] Figure 8 This is a schematic diagram of the structure of a route search device provided in an embodiment of this application;

[0083] Figure 9 This is a schematic diagram of the structure of a routing search electronic device provided in an embodiment of this application. Detailed Implementation

[0084] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0085] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.”

[0086] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0087] Multi-Protocol Label Switching (MPLS) resides between the link layer and network layer in the TCP / IP protocol stack. It replaces IP forwarding with label switching, separating path selection and data forwarding. This allows it to provide connectivity services to the IP layer and obtain services from the link layer. In an MPLS network, the path that an IP packet traverses through is defined by a label and is called an LSP (Label Switched Path). An LSP is a unidirectional path, following the direction of the data flow.

[0088] Because MPLS technology isolates the label distribution mechanism from the data flow, its implementation does not depend on a specific data link layer protocol and can support various physical and link layer technologies (Ethernet, Frame Relay, optical transmission, etc.). MPLS uses a control-driven model to initialize the allocation and distribution of label bundles to establish an LSP. An LSP is established by connecting several label switching points. An LSP is unidirectional, and full-duplex services require two LSPs.

[0089] For telecommunications operators, MPLS has applications in packet transport networks such as traffic engineering, service level (SLR) architecture, and virtual private networks (VPNs). For example, traditional routing methods based on the shortest path principle can lead to congestion on some links while other links remain idle and underutilized. MPLS traffic engineering can utilize methods such as path selection, load balancing, path backup, fault recovery, path prioritization, and collision detection to achieve efficient use of network resources. Regarding service level (SLR), some services require higher and faster response times, while services like video are less sensitive to latency. LSRs have a prioritization mechanism that can prioritize the transmission of LSPs.

[0090] LSPs can be constructed in two ways: static LSPs and dynamic LSPs. The inventors discovered that manual configuration of static routes is ill-suited to complex and large network structures; while dynamic routes mostly focus on specific aspects and cannot meet all service requirements. For example, the greedy search algorithm Dijkstra's algorithm based on shortest distance does not consider channel bandwidth, potentially leading to network congestion. The heuristic search algorithm A* improves search efficiency, but it is more suitable for static networks and generates locally optimal solutions.

[0091] The inventors also discovered that various routing algorithms based on supervised learning and reinforcement learning have emerged in recent years. These include routing algorithms combining deep neural networks, deep belief networks, and recurrent neural networks. While most of these methods have been validated in simulation environments, they suffer from difficulties in deployment, handling unexpected network events, and high model training costs.

[0092] The routing search method, apparatus, electronic device, and computer-readable storage medium provided in this application are intended to solve the above-mentioned technical problems of the prior art.

[0093] The technical solutions of this application and their effects are described below through several exemplary embodiments. It should be noted that the following embodiments can be referenced, borrowed from, or combined with each other. Identical terms, similar features, and similar implementation steps in different embodiments will not be repeated.

[0094] like Figure 1 As shown, the route search method of this application can be applied to... Figure 1 In the scenario shown, specifically, server 101 obtains the search starting network element and the search ending network element from terminal 102, and constructs multiple first network routing paths based on the search starting network element and the search ending network element; it scores each first network routing path according to the trained routing search model to obtain the target search value corresponding to each first network routing path; wherein, the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; server 101 selects the first network routing path with a target search value greater than a preset threshold as the second network routing path, and sends the second network routing path to the terminal, and the terminal completes the corresponding network transmission service based on the second network routing path.

[0095] Figure 1 In the scenario shown, the above route search method can be performed on the server; in other scenarios, it can also be performed on the terminal.

[0096] Those skilled in the art will understand that the “terminal” used herein can be a mobile phone, tablet computer, PDA (Personal Digital Assistant), MID (Mobile Internet Device), etc.; and the “server” can be implemented using a standalone server or a server cluster composed of multiple servers.

[0097] This application provides a route search method, such as... Figure 2 As shown, this method, which can be applied to servers or terminals performing route searches, includes:

[0098] S201, constructs multiple first network routing paths based on the search starting network element and the search ending network element.

[0099] The first network routing path mentioned above is a connectable path from the starting network element to the ending network element.

[0100] Specifically, the server or terminal used for route search can perform path search based on the starting network element and the ending network element to construct multiple first network routing paths.

[0101] S202, score each first network routing path according to the trained routing search model to obtain the target search value corresponding to each first network routing path.

[0102] The target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the candidate routing path and the corresponding standard routing path have the same start and end network elements.

[0103] Furthermore, the aforementioned standard routing path can be a static routing path configured based on historical manual rules. A standard routing path can represent existing, universal routing selection rules. One standard routing path can correspond to multiple candidate routing paths.

[0104] Specifically, the server or terminal used for route search can perform path search based on the starting and ending network elements of the standard route path to construct candidate route paths corresponding to each standard route path. The process of constructing these candidate route paths can be the same as the process of constructing the first network route path, and will be described in detail below.

[0105] S203, select the first network routing path whose target search value is greater than a preset threshold as the second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0106] In some implementations, the first network routing path with the largest target search value can be selected as the second network routing path, and the second network routing path can be used as the optimal path to complete the network transmission service based on the optimal path.

[0107] In other implementations, the target search values ​​can be sorted from high to low, and the first network routing path with the top two target search values ​​can be selected as the second network routing path. The two second network routing paths can be used as primary and backup routing paths to complete network transmission services based on the primary and backup routing paths.

[0108] This application embodiment constructs multiple first network routing paths by searching starting and ending network elements. Then, it scores each first network routing path according to a trained routing search model to obtain the target search data value corresponding to each first network routing path. The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, thereby completing the corresponding network transmission service based on the second network routing path. This application embodiment trains the routing search model based on standard routing paths and their corresponding candidate routing paths. Since the target search value can be used to characterize the similarity between the first network routing path and the standard routing path in historical routing data, the routing search model can perform data fitting on historical routing data and filter each first network routing path. Based on the selected second network routing path, the utilization rate of the routing network can be effectively improved. Unlike the manual configuration of network routing paths in the prior art, this application is applicable to high-volume IP networks, enhancing the routing and forwarding efficiency in IP networks.

[0109] This application provides one possible implementation method, such as... Figure 3 As shown, the above route search model is trained in the following way:

[0110] S301, selects the standard routing path from historical routing data.

[0111] The standard routing path includes at least one of the primary routing path and the backup routing path.

[0112] S302, determine multiple candidate routing paths corresponding to each standard routing path;

[0113] S303: Based on the standard routing path and candidate routing paths, perform at least one update operation on the initial model until the training termination condition is met, and use the initial model that meets the training termination condition as the routing search model.

[0114] The update operations include:

[0115] Input the routing feature data of the standard routing path and the candidate routing path into the initial model to obtain the search value corresponding to each routing feature data;

[0116] Determine the first difference between the search value of the standard route path and the target threshold, and the second difference between the search value of the candidate route path and the target threshold;

[0117] The parameters of the initial model are updated based on the first and second differences.

[0118] The training termination condition may include a first difference of 0 and a second difference greater than a second threshold; or the training termination condition may also include the number of the above update operations reaching a preset third threshold.

[0119] Specifically, the search value can be calculated based on the following loss function (3):

[0120]

[0121]

[0122]

[0123] Where, N t N represents the number of network elements that are identical in the current routing path and the standard routing path. a N represents the number of network elements in the current routing path. s This represents the number of network elements in the standard routing path. The current route can include standard routing paths and candidate routing paths. When the current route is a standard routing path, the search value for the standard routing path can be score1; when the current route is a candidate routing path, the search value for the candidate routing path is score2.

[0124] Specifically, training termination conditions may include: setting the target threshold to 1, the first difference (|score1-1|) being 0, and the second difference (|score2-1|) being greater than 0.1.

[0125] In this embodiment, the initial model is trained using routing feature data of standard routing paths and candidate routing paths. Then, the search values ​​of the standard routing paths and candidate routing paths are determined respectively. When the current model meets the training termination condition based on the search values ​​of the standard routing paths and candidate routing paths, it indicates that the feature similarity between the current standard routing path and the standard routing path in the historical network data is high, but the feature similarity between the candidate routing path and the standard routing path in the historical network data is low. At this point, the model converges, and the routing search model is obtained. This allows the routing search model to effectively fit the relationship between the features of the routing paths in the historical network data and the search values, thereby enhancing the model's representational ability.

[0126] This application provides one possible implementation method, such as... Figure 4 As shown, the above determination of multiple candidate routing paths corresponding to each standard routing path includes:

[0127] S401 determines the starting and ending network elements for each standard routing path.

[0128] S402, for each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element.

[0129] The current transmission network can be a packet transport network, and its network architecture can include three network layers, namely, from bottom to top: access layer, aggregation layer and core layer, with the core layer being the highest network layer and the access layer being the lowest network layer.

[0130] S403, when the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, then the candidate routing path is determined based on the connection path between the starting network element and the ending network element in the third network level; wherein, the third network level is greater than or equal to the first network level.

[0131] In this embodiment, when the starting network element and the ending network element can be connected to the same core layer network element, and the starting network element and the ending network element are at the same network level, the connection path between the starting network element and the ending network element in the first network level is determined; when there is a connection path between the starting network element and the ending network element based on the first network level, a candidate routing path is determined based on the aforementioned connection path; when there is no connection path between the starting network element and the ending network element based on the first network level, the search proceeds upwards level by level until it is determined that there is a connection path between the starting network element and the ending network element in the third network level, and a candidate routing path is determined based on the aforementioned searched connection path.

[0132] This application provides one possible implementation method, such as... Figure 5 As shown, the above method also includes:

[0133] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, the starting network element or the ending network element is connected upward to the fourth network level. The candidate routing path is determined based on the connection path between the starting network element and the ending network element in the fourth network level. The fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level between the first network level and the second network level.

[0134] In this embodiment, when the starting network element and the ending network element can be connected to the same core layer network element, but the starting network element and the ending network element are not at the same network layer, the lower layer of the starting network element and the ending network element can be searched upwards to the fifth network layer so that the starting network element and the ending network element are at the same network layer. Then, if there is a connection path between the starting network element and the ending network element based on the fifth network layer, a candidate routing path is determined based on the above connection path. If there is no connection path between the starting network element and the ending network element based on the fifth network layer, the search is performed level by level upwards until it is determined that there is a connection path between the starting network element and the ending network element in the fourth network layer, and a candidate routing path is determined based on the above-searched connection path.

[0135] This application provides one possible implementation method, such as... Figure 6 As shown, the above method also includes:

[0136] S501, when the core layer network elements connected to the starting network element and the ending network element are not the same, determine at least two first network elements connected to the starting network element and at least two second network elements connected to the ending network element in the core layer.

[0137] S502, determine the core layer connectivity path for each network element matching pair.

[0138] Among them, the network element matching pair is determined by matching any first network element with any second network element pairwise.

[0139] Specifically, any first network element and any second network element can be paired to form a network element pair, and the core layer connectivity path of each network element pair can be determined.

[0140] S503 constructs corresponding candidate routing paths based on the connectivity paths of each core layer.

[0141] In some implementations, when there are too many candidate routing paths obtained based on the above, i.e. the network search space is too large, the candidate routing paths can be reduced based on the single-layer link length and search depth of the candidate routing paths to ensure the reference value and data validity of the candidate routing paths.

[0142] In this embodiment, the process of constructing candidate routing paths based on starting network elements and ending network elements is also applicable to constructing first network routing paths based on searching starting network elements and searching ending network elements. Therefore, the specific steps for constructing the first network routing path will not be described in detail in this embodiment.

[0143] This application provides a possible implementation method, in which the above-mentioned routing feature data is extracted based on the following method:

[0144] S601 determines the network data, port data, and topology connection edge data included in each routing path.

[0145] The routing path includes standard routing paths and candidate routing paths.

[0146] Specifically, the routing path can be described based on the following array:

[0147] (A,a,e1,B,b,e2,C,c,e3…,Z,z,e n (4)

[0148] Where A, B, ..., Z are network elements, a, b, ..., z are ports, and e1, e2, ..., e n These are topological connecting edges.

[0149] S602 converts network data into network element feature vectors, port data into port feature vectors, and topology connection edge data into edge feature vectors.

[0150] S603 uses the mean of the feature vectors of each network element as the corresponding network element vector, the mean of the feature vectors of each port as the corresponding port connection, and the mean of the feature vectors of each edge as the edge vector.

[0151] Specifically, considering that the length of routes varies and the number of network elements and ports changes, an average value can be used to unify the feature dimensions.

[0152] S604 concatenates the network element vector, port vector, and edge vector to obtain routing feature data.

[0153] This application embodiment performs feature transformation on the network element data, port data, and topology connection edge data included in the routing path to obtain corresponding network element feature vectors, port feature vectors, and edge feature vectors. Then, the average value of the above feature vectors is taken, with the same feature length for each route. Next, the averaged network element vectors, port vectors, and edge vectors are concatenated to obtain routing feature data. This can fully extract the feature data of each routing path, laying a good foundation for subsequent model training.

[0154] This application provides a possible implementation method for converting network data into network element feature vectors, including:

[0155] S701, filter the network elements in the network data to obtain the target network elements, and determine the target network element feature data corresponding to each target network element.

[0156] Network data can include whether the network is blocked, service level, and signal type.

[0157] Specifically, invalid network elements in the network data can be removed to filter out the target network elements.

[0158] S702, for each target network element, determine the adjacent network elements that are adjacent to and connected to the target network element, and obtain the adjacent network element data and the weight of each adjacent network element data; perform weighted summation on the adjacent network element data to obtain the adjacent network element feature data corresponding to the target network element.

[0159] Meanwhile, after extracting the target network element feature data, neighbor sampling can be performed on the target network element to determine the adjacent network elements that are adjacent to and connected to the target network element.

[0160] S703, the sum of the target network element feature data and the feature data of adjacent network elements for each target network element is used as the feature vector of the target network element.

[0161] S704, determine the network element feature vector corresponding to the network element data based on the feature vector of each target network element.

[0162] In this embodiment, by filtering network elements in the network data, invalid network element data can be removed. At the same time, neighboring network elements of the current target network element can be sampled, and the feature vector of the target network element can be determined based on the feature data of neighboring network elements and the feature data of the target network element. This realizes the embedding of the features of neighboring network elements into the feature data of the current target network element, realizes the aggregation of network element feature data, and makes the obtained network element feature vector more accurately represent the network element features, effectively improving the model training effect.

[0163] To better understand the above route search method, the following will combine... Figure 7 A detailed example of a route search method according to this application is provided, which includes the following steps:

[0164] S801, determine the first network layer of the starting network element and the second network layer of the ending network element.

[0165] The current transmission network can be a packet transport network, and its network architecture can include three network layers, namely, from bottom to top: access layer, aggregation layer and core layer, with the core layer being the highest network layer and the access layer being the lowest network layer.

[0166] S802, when the core layer network elements connected to the search originating network element and the search destination network element are the same, and the first network level is equal to the second network level, then the first network routing path is determined based on the connection path between the search originating network element and the search destination network element at the third network level. Wherein, the third network level is greater than or equal to the first network level.

[0167] When the core layer network elements connected to the starting and ending network elements are the same, but the first network level is not equal to the second network level, the starting or ending network element is connected upwards to the fourth network level. The first network routing path is determined based on the connection path between the starting and ending network elements in the fourth network level. The fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level between the first and second network levels.

[0168] When the core layer network elements connected to the starting network element and the ending network element are different, determine at least two first network elements connected to the starting network element and at least two second network elements connected to the ending network element in the core layer; determine the core layer connection path of each network element matching pair, and construct the corresponding first network routing path based on each core layer connection path.

[0169] S803: Based on the trained routing search model, score each first network routing path to obtain the target search value corresponding to each first network routing path.

[0170] The target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the candidate routing path and the corresponding standard routing path have the same start and end network elements.

[0171] The above route search model was trained in the following way:

[0172] Standard routing paths are obtained by filtering historical routing data. These standard routing paths include at least one of a primary routing path and a backup routing path. Multiple candidate routing paths are determined for each standard routing path. The initial model is updated at least once based on the standard and candidate routing paths until the training termination condition is met. The initial model that meets the training termination condition is then used as the routing search model.

[0173] The update operations include:

[0174] The routing feature data of the standard routing path and the candidate routing path are input into the initial model to obtain the search value corresponding to each routing feature data; the first difference between the search value of the standard routing path and the target threshold, and the second difference between the search value of the candidate routing path and the target threshold are determined; the parameters of the initial model are updated based on the first difference and the second difference.

[0175] The training termination condition may include a first difference of 0 and a second difference greater than a second threshold; or the training termination condition may also include the number of the above update operations reaching a preset third threshold.

[0176] S804, select the first network routing path whose target search value is greater than a preset threshold as the second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0177] This application embodiment constructs multiple first network routing paths by searching starting and ending network elements. Then, it scores each first network routing path according to a trained routing search model to obtain the target search data value corresponding to each first network routing path. The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, thereby completing the corresponding network transmission service based on the second network routing path. This application embodiment trains the routing search model based on standard routing paths and their corresponding candidate routing paths. Since the target search value can be used to characterize the similarity between the first network routing path and the standard routing path in historical routing data, the routing search model can perform data fitting on historical routing data and filter each first network routing path. Based on the selected second network routing path, the utilization rate of the routing network can be effectively improved. Unlike the manual configuration of network routing paths in the prior art, this application is applicable to high-volume IP networks, enhancing the routing and forwarding efficiency in IP networks.

[0178] This application provides a route search device, such as... Figure 8 As shown, the route search device 80 may include: a construction module 801, a scoring module 802, and a selection module 803;

[0179] Among them, the construction module 801 is used to construct multiple first network routing paths based on the search starting network element and the search ending network element;

[0180] The scoring module 802 is used to score each first network routing path according to the trained routing search model to obtain the target search value corresponding to each first network routing path; wherein, the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the candidate routing path and the corresponding standard routing path have the same start and end network elements;

[0181] The selection module 803 is used to select a first network routing path whose target search value is greater than a preset threshold as a second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0182] This application provides a possible implementation, wherein the apparatus includes a training module for:

[0183] Standard routing paths are obtained by filtering from historical routing data; the standard routing path includes at least one of the primary routing path and the backup routing path.

[0184] Determine multiple candidate routing paths corresponding to each standard routing path;

[0185] The initial model is updated at least once based on the standard route path and candidate route paths until the training termination condition is met. The initial model that meets the training termination condition is then used as the route search model.

[0186] The update operations include:

[0187] Input the routing feature data of the standard routing path and the candidate routing path into the initial model to obtain the search value corresponding to each routing feature data;

[0188] Determine the first difference between the search value of the standard route path and the target threshold, and the second difference between the search value of the candidate route path and the target threshold;

[0189] The parameters of the initial model are updated based on the first and second differences.

[0190] This application embodiment provides a possible implementation method in which the above-mentioned training module, when determining multiple candidate routing paths corresponding to each standard routing path, is used for:

[0191] Determine the starting and ending network elements for each standard routing path;

[0192] For each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element;

[0193] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, the candidate routing path is determined based on the connection path between the starting network element and the ending network element in the third network level; wherein, the third network level is greater than or equal to the first network level.

[0194] This application provides a possible implementation method in which the above-mentioned training module is further used for:

[0195] When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, the starting network element or the ending network element is connected upward to the fourth network level. The candidate routing path is determined based on the connection path between the starting network element and the ending network element in the fourth network level. The fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level between the first network level and the second network level.

[0196] This application provides a possible implementation method in which the above-mentioned training module is further used for:

[0197] When the core layer network elements connected to the starting network element and the ending network element are different, determine at least two first network elements connected to the starting network element and at least two second network elements connected to the ending network element in the core layer.

[0198] Determine the core layer connectivity path for each network element matching pair; wherein, the network element matching pair is determined by pairwise matching of any first network element and any second network element;

[0199] Based on the connection path of each core layer, construct the corresponding candidate routing path.

[0200] This application provides a possible implementation method in which the above-mentioned training module is further used for:

[0201] Determine the network data, port data, and topology connection edge data included in each routing path; routing paths include standard routing paths and candidate routing paths;

[0202] Convert network data into network element feature vectors, port data into port feature vectors, and topology connection edge data into edge feature vectors;

[0203] The mean of the feature vectors of each network element is used as the corresponding network element vector, the mean of the feature vectors of each port is used as the corresponding port connection, and the mean of the feature vectors of each edge is used as the edge vector.

[0204] By concatenating the network element vector, port vector, and edge vector, routing feature data is obtained.

[0205] This application provides a possible implementation method in which the training module, when converting network data into network element feature vectors, is used for:

[0206] The network elements in the network data are filtered to obtain the target network elements, and the target network element feature data corresponding to each target network element is determined.

[0207] For each target network element, identify the adjacent network elements that are adjacent to and connected to the target network element, and obtain the adjacent network element data and the weight of each adjacent network element data; perform a weighted summation of the adjacent network element data to obtain the adjacent network element feature data corresponding to the target network element.

[0208] The sum of the target network element feature data and the feature data of the adjacent network elements of each target network element is used as the feature vector of the target network element.

[0209] The network element feature vector corresponding to the network data is determined based on the feature vector of each target network element.

[0210] The apparatus in this application embodiment can execute the method provided in this application embodiment, and the implementation principle is similar. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the apparatus, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0211] This application embodiment constructs multiple first network routing paths by searching starting and ending network elements. Then, it scores each first network routing path according to a trained routing search model to obtain the target search data value corresponding to each first network routing path. The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, thereby completing the corresponding network transmission service based on the second network routing path. This application embodiment trains the routing search model based on standard routing paths and their corresponding candidate routing paths. Since the target search value can be used to characterize the similarity between the first network routing path and the standard routing path in historical routing data, the routing search model can perform data fitting on historical routing data and filter each first network routing path. Based on the selected second network routing path, the utilization rate of the routing network can be effectively improved. Unlike the manual configuration of network routing paths in the prior art, this application is applicable to high-volume IP networks, enhancing the routing and forwarding efficiency in IP networks.

[0212] This application provides an electronic device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of a route search method. Compared with related technologies, this application can achieve the following: Multiple first network routing paths are constructed by searching starting network elements and ending network elements; then, each first network routing path is scored according to a trained route search model to obtain a target search data value corresponding to each first network routing path; and the first network routing path with a target search value greater than a preset threshold is selected as a second network routing path to complete the corresponding network transmission service based on the second network routing path. This application trains a route search model based on standard routing paths and candidate routing paths corresponding to the standard routing paths. Since the target search value can be used to characterize the similarity between the first network routing path and the standard routing path in historical routing data, the route search model can perform data fitting on historical routing data and filter each first network routing path. Based on the selected second network routing path, the utilization rate of the routing network can be effectively improved. Unlike the manual configuration of network routing paths in the prior art, this application can be applied to IP networks with high traffic volume, enhancing the routing forwarding efficiency in IP networks.

[0213] In one alternative embodiment, an electronic device is provided, such as Figure 9 As shown, Figure 9The illustrated electronic device 90 includes a processor 901 and a memory 903. The processor 901 and the memory 903 are connected, for example, via a bus 902. Optionally, the electronic device 90 may further include a transceiver 904, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 904 is not limited to one type, and the structure of the electronic device 90 does not constitute a limitation on the embodiments of this application.

[0214] Processor 901 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 901 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0215] Bus 902 may include a pathway for transmitting information between the aforementioned components. Bus 902 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 902 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0216] The memory 903 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of carrying or storing computer programs and capable of being read by a computer, without limitation herein.

[0217] The memory 903 stores computer programs that execute embodiments of this application, and its execution is controlled by the processor 901. The processor 901 executes the computer programs stored in the memory 903 to implement the steps shown in the foregoing method embodiments.

[0218] Electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, and tablets, as well as fixed terminals such as digital TVs and desktop computers.

[0219] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement the steps and corresponding content of the aforementioned method embodiments.

[0220] This application provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the following actions:

[0221] Multiple first network routing paths are constructed based on the search starting network element and the search ending network element;

[0222] Each first network routing path is scored based on the trained routing search model to obtain the target search value corresponding to each first network routing path; the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the candidate routing path and the corresponding standard routing path have the same start and end network elements;

[0223] The first network routing path with a target search value greater than a preset threshold is selected as the second network routing path, so as to complete the corresponding network transmission service based on the second network routing path.

[0224] The terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0225] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.

[0226] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.

Claims

1. A route search method, characterized in that, include: Multiple first network routing paths are constructed based on the search starting network element and the search ending network element; Each first network routing path is scored according to the trained routing search model to obtain a target search value corresponding to each first network routing path; wherein, the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the standard routing path is a static routing path configured based on historical manual rules, including at least one of the primary routing path and the backup routing path, and the candidate routing path has the same start and end network elements as the corresponding standard routing path; A first network routing path with a target search value greater than a preset threshold is selected as a second network routing path, so as to complete the corresponding network transmission service based on the second network routing path; Determine multiple candidate routing paths corresponding to each standard routing path, including: Determine the starting and ending network elements for each of the aforementioned standard routing paths; For each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element; When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, then the candidate routing path is determined based on the connection path between the starting network element and the ending network element at the third network level; wherein, the third network level is greater than or equal to the first network level; When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, then the starting network element or the ending network element is connected upwards to the fourth network level, and a candidate routing path is determined based on the connection path of the starting network element and the ending network element in the fourth network level; wherein, the fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level among the first network level and the second network level; When the core layer network elements connected to the starting network element and the ending network element are different, a network element matching pair is constructed in the core layer, and the candidate routing path is determined based on the core layer connection path of the network element matching pair. The network element matching pair includes a first network element connected to the starting network element and a second network element connected to the ending network element.

2. The method according to claim 1, characterized in that, The routing search model was trained in the following manner: Standard routing paths are obtained by filtering through historical routing data; Determine multiple candidate routing paths corresponding to each standard routing path; The initial model is updated at least once based on the standard routing path and candidate routing path until the training termination condition is met. The initial model that meets the training termination condition is then used as the routing search model. The update operation includes: The routing feature data of the standard routing path and the candidate routing path are input into the initial model to obtain the search value corresponding to each routing feature data. Determine a first difference between the search value of the standard routing path and the target threshold, and a second difference between the search value of the candidate routing path and the target threshold; The parameters of the initial model are updated based on the first difference and the second difference.

3. The method according to claim 1, characterized in that, The step of constructing network element matching pairs in the core layer and determining the candidate routing path based on the core layer connectivity path of the network element matching pairs includes: Identify at least two first network elements connected to the starting network element in the core layer, and at least two second network elements connected to the ending network element; Determine the core layer connectivity path for each network element matching pair; wherein, the network element matching pair is determined by pairwise matching of any first network element and any second network element; Based on each of the core layer connectivity paths, a corresponding candidate routing path is constructed.

4. The method according to claim 2, characterized in that, The routing feature data was extracted based on the following method: Determine the network data, port data, and topology connection edge data included in each routing path; the routing path includes standard routing paths and candidate routing paths; The network data is converted into network element feature vectors, the port data is converted into port feature vectors, and the topology connection edge data is converted into edge feature vectors. The mean value of each network element feature vector is used as the corresponding network element vector, the mean value of each port feature vector is used as the corresponding port connection, and the mean value of each edge feature vector is used as the edge vector. The routing feature data is obtained by concatenating the network element vector, port vector, and edge vector.

5. The method according to claim 4, characterized in that, The step of converting the network data into network element feature vectors includes: The network elements in the network data are filtered to obtain the target network elements, and the target network element feature data corresponding to each target network element is determined. For each target network element, determine the adjacent network elements that are adjacent to and connected to the target network element, and obtain the adjacent network element data and the weight of each adjacent network element data; perform a weighted summation of the adjacent network element data to obtain the adjacent network element feature data corresponding to the target network element. The sum of the target network element feature data and the feature data of adjacent network elements for each target network element is used as the feature vector of the target network element. The network element feature vector corresponding to the network data is determined based on the feature vector of each target network element.

6. A route search device, characterized in that, include: The module is used to construct multiple first network routing paths based on the search starting network element and the search ending network element; The scoring module is used to score each first network routing path according to the trained routing search model to obtain the target search value corresponding to each first network routing path; wherein, the target search value is used to characterize the similarity between the first network routing path and the standard routing path in the historical routing data; the routing search model is trained based on the standard routing path and the candidate routing path corresponding to the standard routing path; the standard routing path is a static routing path configured based on historical manual rules, including at least one of the main routing path and the backup routing path, and the candidate routing path has the same start and end network elements as the corresponding standard routing path; The selection module is used to select a first network routing path whose target search value is greater than a preset threshold as a second network routing path, so as to complete the corresponding network transmission service based on the second network routing path. Determine multiple candidate routing paths corresponding to each standard routing path, including: Determine the starting and ending network elements for each of the aforementioned standard routing paths; For each standard routing path, determine the first network layer of the starting network element and the second network layer of the ending network element; When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is equal to the second network level, then the candidate routing path is determined based on the connection path between the starting network element and the ending network element at the third network level; wherein, the third network level is greater than or equal to the first network level; When the core layer network elements connected to the starting network element and the ending network element are the same, and the first network level is not equal to the second network level, then the starting network element or the ending network element is connected upwards to the fourth network level, and a candidate routing path is determined based on the connection path of the starting network element and the ending network element in the fourth network level; wherein, the fourth network level is greater than or equal to the fifth network level, and the fifth network level is the largest network level among the first network level and the second network level; When the core layer network elements connected to the starting network element and the ending network element are different, a network element matching pair is constructed in the core layer, and the candidate routing path is determined based on the core layer connection path of the network element matching pair. The network element matching pair includes a first network element connected to the starting network element and a second network element connected to the ending network element.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images