A method and device for planning an adaptive protection path in a metropolitan transport network

Abstract

The application provides a kind of metropolitan area transport network adaptive protection path planning method and device.The method is executed on metropolitan area transport network, and metropolitan area transport network is composed of multiple nodes.Network topology information of metropolitan area transport network is stored as graph data, and the traffic load of each service flow in the next time period is predicted based on the pre-trained traffic prediction model and the obtained historical traffic information, and the residual capacity of each link in the next time period in the metropolitan area transport network, the available time slot type and the corresponding number are estimated.The traffic prediction model uses graph convolution network to perform multi-layer feature extraction on the structure of network topology graph, constructs load balancing constraint, minimization blocking rate constraint and minimization resource consumption constraint, and generates protection path and protection time slot allocation scheme for each service flow according to the residual capacity of each link in the next time period, the available time slot type and the corresponding number, to realize more efficient fault recovery and protection path selection.

Description

Technical Field

[0001] This invention relates to the field of protection path planning methods, and in particular to an adaptive protection path planning method and apparatus for metropolitan area transmission networks. Background Technology

[0002] Metropolitan area network (MAN) transport networks are networks that cover urban areas, provide a wide range of services, and support multiple communication protocols. In the current architecture of MAN transport networks, ensuring the stability and continuity of data transmission is crucial for achieving seamless communication services. With the widespread adoption of 5G mobile communication technology and the rapid development of future technologies, the data traffic carried by MAN transport networks is increasing, and traditional static protection mechanisms are no longer sufficient to adapt to the ever-increasing network load and dynamically changing network topology.

[0003] Existing technologies have significant shortcomings in handling unexpected network failures, dynamic load balancing, and resource utilization. Currently, network protection is typically achieved using conventional proactive protection methods. These methods aim to prepare for failures by sending the same data along multiple paths so that network reconfiguration is unnecessary when a failure occurs. While this provides a basic safety net for network failures, the approach cannot dynamically adapt to real-time changes in network traffic and can easily lead to excessive resource reservation or resource shortages at critical moments. Reactive methods, while saving resources after a failure, suffer from long response times from failure occurrence to recovery, impacting network responsiveness. Traditional optimization algorithms, due to their large solution space, struggle to find optimal solutions or have excessively long solution times, making them unsuitable for real-time applications. These methods fail to adequately address resource conflicts under shared protection mechanisms and signal processing delays that may occur during large-scale failures. They cannot establish a network protection mechanism with high resource utilization, load balancing, and stable and continuous transmission, nor can they achieve more efficient resource utilization and protection path selection. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide an adaptive protection path planning method and apparatus for metropolitan area transport networks to eliminate or improve one or more defects existing in the prior art. The present invention uses a graph convolutional network (GCN) to predict the traffic of the metropolitan area transport network for the next time period and generate protection paths for each service flow, thereby achieving protection path load balancing, minimizing the blocking rate and minimizing resource consumption. This solves the problems of resource waste, fault recovery and low efficiency of protection path selection under unstable network conditions in the prior art, thereby achieving more efficient resource utilization, fault recovery and protection path selection.

[0005] One aspect of the present invention provides an adaptive protection path planning method for a metropolitan area transport network, the method being executed on a metropolitan area transport network consisting of multiple nodes, the method comprising the following steps:

[0006] The network topology information of the metropolitan area transport network is obtained and stored as graph data. Historical traffic information of multiple service flows is obtained. The historical traffic information is used to record the timestamp, number of bytes transmitted, and Internet Protocol address of the service flows in each time period.

[0007] Based on the pre-trained traffic prediction model, the traffic load of each service flow in the next time period is predicted according to the historical traffic information of each service flow, and the remaining capacity, available time slot types and corresponding quantities of each link in the metropolitan area transmission network in the next time period are estimated according to the traffic load. The traffic prediction model uses a graph convolutional network to perform multi-layer feature extraction on the structure of the network topology graph, wherein the features of each layer are obtained by aggregating the feature information of each node and its neighbors.

[0008] Construct load balancing constraints, minimum blocking rate constraints, and minimum resource consumption constraints, and generate protection paths and protection time slot allocation schemes for each service flow based on the remaining capacity of each link, the types and corresponding quantities of available time slots in the next time period.

[0009] In some embodiments, the pre-training method for the traffic prediction model includes:

[0010] Obtain a training sample set of network topology graphs, wherein the training sample set includes sample feature nodes of the network topology;

[0011] Obtain an initial graph convolutional network, which takes the feature nodes of layer l as input and the final node features after iterative updates of the feature nodes of layer l as output;

[0012] Obtain a training sample set for the prediction layer, which includes the final node features of the network topology and sample traffic data of multiple service flows;

[0013] The initial traffic prediction model is trained using the training sample set of the prediction layer. The initial traffic prediction model takes the final node features of the network topology from time tp to time t of each service flow and the sample traffic data as input, and takes the sample traffic data at time t+1 as output. The parameters of the initial traffic prediction model are updated based on the mean square error loss function to obtain the traffic prediction model.

[0014] In some embodiments, the initial graph convolutional network is randomly initialized.

[0015] In some embodiments, the initial graph convolutional network includes a set number of feature layers and a prediction layer. The prediction layer predicts the traffic information of each node in the next time period based on the node features output by the last feature layer and the sample traffic information of the current time period. The expression is:

[0016]

[0017] Among them, W (pred) b represents the weights of the prediction layer. (pred) H represents the bias of the prediction layer, L represents the number of layers in the feature layer, and H represents the bias of the prediction layer. (L) F represents the node features of the Lth feature layer in the time period t. t This represents the traffic information during time period t, and ReLU represents the non-linear activation function.

[0018] In some embodiments, the features of each layer are obtained by aggregating the feature information of each node and its neighbors. The expression for aggregating the feature information of each node and its neighbors is as follows:

[0019]

[0020]

[0021] Where A represents the adjacency matrix of the metropolitan area transmission network topology, and I represents the identity matrix. This represents the adjacency matrix with added self-loops. yes The degree matrix, W represents the node feature of node v in the l-th feature layer. (l) Let σ represent the learnable weight matrix of the l-th feature layer, and σ represent the activation function.

[0022] In some embodiments, the expression for the load balancing constraint is:

[0023]

[0024]

[0025]

[0026] Where E represents the set of links in the metropolitan area transmission network, e represents a link in the set of links, and U e This indicates the traffic load of link e in the metropolitan area transport network. This represents the average traffic load of the links in the metropolitan area transport network. Indicates the business flow f for the next time period i Traffic load, This indicates the available capacity of link e in the next time period.

[0027] In some embodiments, the expression for minimizing the blocking rate constraint is:

[0028]

[0029]

[0030] Where B represents the proportion of blocked requests out of all requests, I block (f i Define the i-th business flow f i Is it blocked? block (f i A value of 1 indicates that the business flow f is equal to 1. i Unable to process due to insufficient resources, I block (f i A value of 0 indicates that the business flow f is equal to 0. i Because of sufficient resources, N can be successfully processed. f This indicates the total number of business flows.

[0031] In some embodiments, the expression for minimizing resource consumption constraints is:

[0032]

[0033] in, Indicates the service flow f i Protection path length, Indicates the service flow f i The number of time slots required on the protection path, N f w represents the total number of business flows. L w represents the weighting coefficient for the length of the protection path. S The weighting coefficient represents the number of time slots required on the protection path.

[0034] In some embodiments, the method further includes:

[0035] The load balancing constraint, the minimum blocking rate constraint, and the minimum resource consumption constraint are combined to construct a comprehensive constraint, the expression of which is:

[0036]

[0037] in, Indicates the service flow f i The number of time slots required on the protection path, Indicates the service flow f i In the protection path of the next time period, σ U This represents the load balancing constraint, where B represents the proportion of blocked requests among all requests, and w L w represents the weighting coefficient for the length of the protection path. SA weighting coefficient representing the number of time slots required on the protection path. Indicates the service flow f i The protection path length, α, β and γ are weighting coefficients that balance the load balancing constraint, the blocking rate constraint and the resource consumption constraint.

[0038] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the steps of the above-described method.

[0039] The beneficial effects of the present invention are at least as follows:

[0040] This invention provides an adaptive protection path planning method and apparatus for a metropolitan area transport network. The method is executed on a metropolitan area transport network composed of multiple nodes. The network topology information of the metropolitan area transport network is stored as graph data. Based on a pre-trained traffic prediction model and acquired historical traffic information, the traffic load of each service flow in the next time period is predicted. The remaining capacity, available time slot types, and corresponding quantities of each link in the metropolitan area transport network in the next time period are estimated. The traffic prediction model uses a graph convolutional network to extract multi-layer features from the network topology graph, constructing load balancing constraints, minimum blocking rate constraints, and minimum resource consumption constraints. Based on the remaining capacity, available time slot types, and corresponding quantities of each link in the next time period, a protection path and protection time slot allocation scheme are generated for each service flow, achieving more efficient fault recovery and protection path selection, and avoiding problems such as resource waste and low efficiency in fault recovery and protection path selection.

[0041] Furthermore, this invention employs the Node2Vec model to learn the low-dimensional feature vectors of each node in the network. Through these low-dimensional feature vectors, the domain structure of the nodes and the high-order similarity between nodes can be obtained, thereby acquiring more in-depth network structure information and simplifying the model and subsequent calculations.

[0042] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the description, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures specifically pointed out in the description and drawings.

[0043] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0044] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:

[0045] Figure 1 This is a flowchart illustrating the adaptive protection path planning method for metropolitan area transmission networks according to an embodiment of the present invention.

[0046] Figure 2 This is a pre-training method for the traffic prediction model according to an embodiment of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0048] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0049] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0050] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0051] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0052] In existing technologies, conventional active protection methods send the same data along multiple paths so that the network does not need to be reconfigured when a fault occurs. However, these methods cannot dynamically adapt to real-time changes in network traffic and are prone to over-reservation of resources or insufficient resources at critical moments. Reactive methods can save resources after a fault occurs, but the response time from fault occurrence to recovery is long, affecting the network's response timeliness. These methods cannot achieve efficient resource utilization, fault recovery, and protection path selection. This invention proposes an adaptive protection path planning method for metropolitan area transport networks. It uses a graph convolutional network to predict the traffic of the metropolitan area transport network for the next time period and generates protection paths for each service flow. This achieves protection path load balancing, minimizes the blocking rate, and minimizes resource consumption, solving the problems of resource waste, fault recovery, and low efficiency in protection path selection under unstable network conditions in existing technologies.

[0053] Figure 1 This is a flowchart illustrating the adaptive protection path planning method for metropolitan area transmission networks according to an embodiment of the present invention. Figure 2 This invention provides a pre-training method for a traffic prediction model according to an embodiment of the present invention. Specifically, this application provides an adaptive protection path planning method for a metropolitan area transport network. This method is executed on a metropolitan area transport network, which consists of multiple nodes. The method includes the following steps S101 to S103:

[0054] Step S101: Obtain the network topology information of the metropolitan area transport network and store it as graph data. Obtain the historical traffic information of multiple service flows. The historical traffic information is used to record the timestamp, number of bytes transmitted, and Internet Protocol address of the service flows in each time period.

[0055] Step S102: Based on the pre-trained traffic prediction model, predict the traffic load of each service flow in the next time period according to the historical traffic information of each service flow, and estimate the remaining capacity, available time slot types and corresponding quantities of each link in the metropolitan area transmission network in the next time period according to the traffic load; the traffic prediction model uses a graph convolutional network to extract multi-layer features of the network topology graph, wherein the features of each layer are obtained by aggregating the feature information of each node and its neighbors.

[0056] Step S103: Construct load balancing constraints, minimum blocking rate constraints, and minimum resource consumption constraints, and generate protection paths and protection time slot allocation schemes for each service flow based on the remaining capacity of each link, the types of available time slots, and the corresponding quantities for the next time period.

[0057] In step S101, the present invention is executed on a metropolitan area transmission network. The service flow is the data, signals, and information that pass through various nodes and links in the network topology graph during transmission. The network topology information is stored as graph data, which can obtain the node correlation in the network topology and provide more accurate data information for subsequent traffic prediction. Graph data is a complex data structure composed of a set of nodes and a set of links. The storage structure of graph data includes an adjacency matrix, an adjacency matrix, a cross-linked list, an adjacency multi-list, and an edge set array. In the present invention, the node correlation in the network topology is represented by an adjacency matrix. When the adjacency matrix is ​​equal to 1, it indicates that there is a direct link connection between two nodes.

[0058] Historical traffic information is information contained in traffic data within a certain time period, recording the timestamp, number of bytes transmitted, and Internet protocol address of the business flow in each time period.

[0059] In step S102, the pre-training method for the traffic prediction model includes the following steps S201 to S204:

[0060] Step S201: Obtain a network topology training sample set, which includes sample feature nodes of the network topology.

[0061] Step S202: Obtain the initial graph convolutional network. The initial graph convolutional network takes l layers of feature nodes as input and the final node features after iterative updates of l layers of feature nodes as output.

[0062] Step S203: Obtain the prediction layer training sample set, which includes the final node features of the network topology and sample traffic data of multiple service flows.

[0063] Step S204: The initial traffic prediction model is trained using the prediction layer training sample set. The initial traffic prediction model takes the final node characteristics and sample traffic data of the network topology from time tp to time t of each service flow as input and the sample traffic data at time t+1 as output. The parameters of the initial traffic prediction model are updated based on the mean square error loss function to obtain the traffic prediction model.

[0064] Specifically, the network topology information stored as graph data includes node and link information. The traffic prediction model primarily focuses on the topological relationships between nodes and links in the network topology view. In some embodiments, the initial graph convolutional network is randomly initialized. Further, a Node2Vec model is used to learn the low-dimensional feature vectors of each node in the network. These low-dimensional feature vectors reveal the node's neighborhood structure and high-order similarity between nodes. High-order similarity provides more in-depth network structure information, thereby simplifying the model and subsequent computations and reducing storage space. The graph convolutional network is used to update the features in the network topology graph multiple times, and the features of the node itself and its neighbors are weighted and averaged to obtain the final features.

[0065] The initial graph convolutional network updates its parameters based on the mean squared error loss function, L. MSE The expression is:

[0066]

[0067] Where N represents the number of samples in the pre-trained traffic prediction model; y i This represents the actual flow value of the i-th sample; This represents the traffic value of the graph convolutional network traffic prediction model.

[0068] In some embodiments, the features of each layer are obtained by aggregating the feature information of each node and its neighbors. The expression for aggregating the feature information of each node and its neighbors is as follows:

[0069]

[0070]

[0071] Where A represents the adjacency matrix of the metropolitan area transport network topology, and I represents the identity matrix. This represents the adjacency matrix with added self-loops. yes The degree matrix, W represents the node features of node v in the l-th feature layer. (l) Let σ represent the learnable weight matrix of the l-th feature layer, and σ represent the activation function.

[0072] In some embodiments, the initial graph convolutional network includes a predetermined number of feature layers and a prediction layer. The prediction layer predicts the traffic information of each node in the next time period based on the node features output by the last feature layer and the sample traffic information in the current time period. The expression is:

[0073]

[0074] Among them, W (pred) b represents the weights of the prediction layer. (pred) H represents the bias of the prediction layer, L represents the number of feature layers, and H represents the bias of the prediction layer. (L) F represents the node features of the Lth feature layer in time period t. t This represents the traffic information during time period t, and ReLU represents the non-linear activation function.

[0075] In step S103, the selection of the protection path in this invention requires load balancing, minimizing the blocking rate, and minimizing resource consumption. The load balancing constraint requires that the links in the network topology diagram of the metropolitan area transmission network evenly distribute the transmission traffic and load of the service flow in the next time period. Minimizing the blocking rate requires that the service flow has sufficient transmission resources for transmission. Minimizing resource consumption requires that the service flow improves resource utilization under the conditions of load balancing and sufficient transmission resources. When a network failure occurs, it can avoid the problems of load imbalance, unstable service flow transmission, and resource waste caused by traditional protection methods.

[0076] In some embodiments, the expression for the load balancing constraint is:

[0077]

[0078]

[0079]

[0080] Where E represents the set of links in the metropolitan area transport network, e represents a link in the set of links, and U e This indicates the traffic load on link e in the metropolitan area network. This represents the average traffic load of links in a metropolitan area network. Indicates the business flow f for the next time period i Traffic load, This indicates the available capacity of link e in the next time period.

[0081] In some embodiments, the expression for minimizing the blocking rate constraint is:

[0082]

[0083]

[0084] Where B represents the proportion of blocked requests out of all requests, I block (f i Define the i-th business flow f i Is it blocked? block (f i A value of 1 indicates that the business flow f is equal to 1. iUnable to process due to insufficient resources, I block (f i A value of 0 indicates that the business flow f is equal to 0. i Because of sufficient resources, N can be successfully processed. f This indicates the total number of business flows.

[0085] In some embodiments, the expression for minimizing resource consumption constraints is:

[0086]

[0087] in, Represents the business flow f i Protection path length, Represents the business flow f i The number of time slots required on the protection path, N f w represents the total number of business flows. L The weighting coefficient w represents the length of the protected path. S The weighting coefficient represents the number of time slots required on the protection path.

[0088] In some embodiments, the method further includes:

[0089] By combining the load balancing constraint, the minimum blocking rate constraint, and the minimum resource consumption constraint, a comprehensive constraint is constructed. The expression for the comprehensive constraint is:

[0090]

[0091] in, Represents the business flow f i The number of time slots required on the protection path, Represents the business flow f i In the protection path of the next time period, σ U This represents the load balancing constraint, where B represents the proportion of blocked requests out of all requests, and w L The weighting coefficient w represents the length of the protected path. S A weighting coefficient representing the number of time slots required on the protection path. Represents the business flow f i The protection path length is given by α, β, and γ, which are the weighting coefficients for load balancing constraints, blocking rate constraints, and resource consumption constraints.

[0092] Furthermore, customer flow f i The expression for the time slot allocation scheme is:

[0093]

[0094] in, Indicates the business flow f for the next time period i Traffic load, Represents the business flow f i The number of time slots required on the protection path.

[0095] Specifically, S DP This invention is a hybrid time slot allocation method based on dynamic programming. It divides the time slot allocation problem into multiple stages according to the number of service flows in the network. Each stage independently handles the time slot allocation problem for one service flow. State variables are defined in each stage to represent the available time slot resources in the current stage, and a time slot allocation decision is made based on the current stage's state. A state transition equation is determined based on the current state and the decision, and an index function is defined to evaluate the decision results of each stage. By establishing the basic equations of dynamic programming, the optimal decision for each stage is obtained, ultimately leading to the optimal time slot allocation scheme for the entire process.

[0096] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0097] The present invention will now be described with reference to a specific embodiment:

[0098] 1. Parameter Description:

[0099] In this invention, the network topology is represented as a graph G = (V, E), where V is the set of nodes and E is the set of links. The capacity of each link e ∈ E is B. e The available capacity of link e in the next time period t+1 is denoted as The connection relationship between nodes is represented by the adjacency relation A. ij =1 indicates that there is a direct link between nodes i and j. Service flow f i The protection paths for time periods t and t+1 are respectively represented as follows: and Its specific form can be represented by a sequence of nodes, i.e. (v s ,...,v u ), v s ,v u ∈V. v s and v u These represent the starting node, intermediate nodes, and destination node of the protection path, respectively. For example... The path of the first service flow is from the starting node v1 to the destination node v5, passing through intermediate nodes v2 and v4. The time slot type on each link is represented as k∈(1,2,...,J), and the number of each type of time slot is represented as N. k Customer flow f i The time slot allocation scheme is expressed as

[0100]

[0101] in and Let f represent the i-th customer flow respectively. i The time slots used and their corresponding number. Definition f i The number of timeslots of type k used. For customer flow f i The allocated protection time slot bandwidth is denoted as Not less than the predicted bandwidth of the customer flow:

[0102]

[0103] Define the business traffic set F = {f i |i=1,2,…,N f}, f i Let f represent the i-th business flow. i Historical traffic information during the time period from tp to t is recorded as follows: The traffic prediction mechanism will be based on F t The traffic bandwidth of the service flow predicted by the learned topology information in time period t+1 is denoted as...

[0104] When constructing a traffic prediction model based on Graph Convolutional Networks (GCNs), the main focus is on how to utilize the graph's topology and the relationships between nodes. To this end, the Node2Vec method is first employed to learn the low-dimensional feature vectors of each node in the network. These vectors capture the neighborhood structure of the nodes and the high-order similarity between them. Next, for any node v in the graph, within the GCN framework, the feature update of node v at layer l+1 is performed. This is achieved by a weighted average of the features of its neighboring nodes:

[0105]

[0106] in, (A is the adjacency matrix of the metropolitan area transport network topology, and I is the identity matrix used to add self-loops). yes The degree matrix, H (l) W is the feature representation of node v in the l-th feature layer. (l) σ is the learnable weight matrix of the l-th feature layer, and σ is the activation function.

[0107] After L layers of graph convolution, the final feature representation of node v is obtained. To predict traffic in the next time period, a prediction layer is used. This layer combines node characteristics and historical traffic information, and the goal of the prediction is to estimate the traffic load of each service flow in the next time period. And according to Estimate the remaining capacity of the link. And available time slots.

[0108]

[0109] Among them, H (L) This is the output of the last layer of the graph convolutional network, representing the node features during time interval t. These features are correlated with the current traffic information F. t The union is represented as [H] (L) ,F t ], as input to the prediction layer. W (pred) and b (pred) These are the weights and biases of the prediction layer, respectively. The ReLU activation function ensures that the predicted flux is non-negative.

[0110] 2. Objective function:

[0111] To more comprehensively evaluate the overall efficiency of the network, it is necessary to consider metrics such as load balancing, congestion rate, bandwidth utilization, and time slot utilization. We designed a comprehensive objective function that weighs multiple important dimensions to optimize routing planning and time slot allocation decisions.

[0112] 2.1 Load balancing objectives:

[0113] The traffic load of link e during the time period t+1 is obtained through To quantify. Among them... For the predicted bandwidth of the service flow, Given the available capacity of the link, the load balancing objective is:

[0114]

[0115] Where E is the set of links in the metropolitan area transport network. This represents the average traffic load of link e in the metropolitan area transmission network.

[0116] 2.2 Blocking Rate Target:

[0117] Define indicator function I block (f i ) represents the business flow f i Is it blocked?

[0118]

[0119] The blocking rate B can be defined as the proportion of requests that are blocked out of all requests:

[0120]

[0121] Where N ff represents the total number of business flows. i For a single business flow, the objective function for the blocking rate is:

[0122]

[0123] 2.3 Resource Consumption Target:

[0124] The goal of the resource consumption section is to minimize the protection path length and time slot resource consumption in the network. Define the service flow f. i The protection path length is definition This represents the number of time slots required along the path. The objective function in this part focuses on the use of these two resources to ensure efficient utilization of network resources while meeting transmission requirements.

[0125]

[0126] Among them, w L and w S These are the weighting coefficients for the protection path length and the weighting coefficients for the number of time slot resources, respectively.

[0127] To improve the aforementioned network performance, we propose a comprehensive objective function that combines network load balancing, traffic flow blocking probability, and the protection path length and time slot resource consumption of traffic flows:

[0128]

[0129] The overall objective function balances these different performance indicators through weighting coefficients α, β, and γ, thereby achieving a globally optimal protection path planning scheme.

[0130] 3. Adaptive protection path planning algorithm:

[0131] The adaptive protection routing planning algorithm based on graph convolutional networks proposed in this invention consists of two sub-algorithms: a traffic prediction algorithm based on graph convolutional networks and a protection path planning algorithm. The first sub-algorithm uses graph convolutional networks to deeply learn network topology and historical traffic data to predict future service flow bandwidth requirements. This traffic prediction provides key input parameters for protection path planning. The second sub-algorithm utilizes these prediction results, combined with the current network state, to calculate the optimal protection path and time slot allocation, thereby formulating a protection routing strategy aimed at improving network robustness and resource utilization efficiency. The entire algorithm framework strives to maximize network performance while ensuring service continuity and quality of service.

[0132] 3.1 Business Traffic Prediction Algorithm:

[0133] This algorithm proposes a graph convolutional network (GCN)-based method to predict the bandwidth requirements of various service flows in a network in the next time period. Through historical data analysis, this method learns complex patterns in network traffic and predicts future bandwidth demands accordingly, thus providing decision support for protection path planning. The model utilizes graph-structured data to capture the node relationships in the network topology, providing accurate bandwidth predictions for each service flow.

[0134] In constructing a traffic prediction model based on Graph Convolutional Networks (GCNs), a feature vector containing static attributes is first built for each node in the graph, primarily including static attributes such as node type or location. Utilizing the graph's topology, the model iteratively updates node features through a series of graph convolutional layers. Each layer's update is based on the aggregation of node and neighbor information and the application of non-linear activation functions, which helps capture the structural relationships between nodes in the graph. The main function of the GCN layers is to extract graph structural features, while historical traffic information is integrated in the model's subsequent prediction stages for the final traffic prediction.

[0135] 3.2 Protection Path Planning Algorithm:

[0136] This algorithm utilizes the results of Algorithm 1 and also employs the idea of ​​graph neural networks to combine information such as network topology features, bandwidth capacity, and predicted service traffic to construct an active protection routing algorithm.

[0137] First, the algorithm calculates the current and predicted bandwidth requirements of each service flow to determine the remaining capacity of the network links. Next, it assesses the network load using a load balancing function and calculates the blocking probability for each service flow. The algorithm further calculates the length of the protection path and, based on this information and the network load and capacity status, generates a protection route for each service flow. Furthermore, the algorithm handles time slot resource allocation, employing a flexible time slot allocation strategy to adapt to the protection requirements of each service flow. Finally, the algorithm outputs the protection routes and time slot allocation results for implementation in the network, ensuring data transmission reliability and network robustness.

[0138] In summary, this invention provides an adaptive protection path planning method and apparatus for a metropolitan area transport network. The method is executed on a metropolitan area transport network composed of multiple nodes. The network topology information of the metropolitan area transport network is stored as graph data. Based on a pre-trained traffic prediction model and acquired historical traffic information, the traffic load of each service flow in the next time period is predicted. The remaining capacity, available time slot types, and corresponding quantities of each link in the metropolitan area transport network in the next time period are estimated. The traffic prediction model uses a graph convolutional network to extract multi-layer features from the network topology graph, constructing load balancing constraints, minimum blocking rate constraints, and minimum resource consumption constraints. Based on the remaining capacity, available time slot types, and corresponding quantities of each link in the next time period, a protection path and protection time slot allocation scheme are generated for each service flow, achieving more efficient fault recovery and protection path selection, and avoiding problems such as resource waste and low efficiency in fault recovery and protection path selection.

[0139] Furthermore, this invention employs the Node2Vec model to learn the low-dimensional feature vectors of each node in the network. Through these low-dimensional feature vectors, the domain structure of the nodes and the high-order similarity between nodes can be obtained, thereby acquiring more in-depth network structure information and simplifying the model and subsequent calculations.

[0140] This invention also provides a computer device, which may include a processor and a memory, wherein the processor and the memory may be connected via a bus or other means.

[0141] The processor can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0142] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the button blocking method of the vehicle display device in this embodiment of the invention. The processor executes various functional applications and data processing by running the non-transitory software programs, instructions, and modules stored in the memory.

[0143] The memory may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory may include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0144] The one or more modules are stored in the memory, and when executed by the processor, they perform the method described in this embodiment.

[0145] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned edge computing server deployment method. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.

[0146] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0147] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0148] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0149] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for adaptive protection path planning in a metropolitan area transmission network, characterized in that, The method is executed on a metropolitan area transport network, which consists of multiple nodes, and the method includes the following steps: The network topology information of the metropolitan area transport network is obtained and stored as graph data. Historical traffic information of multiple service flows is obtained. The historical traffic information is used to record the timestamp, number of bytes transmitted, and Internet Protocol address of the service flows in each time period. Based on the pre-trained traffic prediction model, the traffic load of each service flow in the next time period is predicted according to the historical traffic information of each service flow, and the remaining capacity, available time slot types and corresponding quantities of each link in the metropolitan area transmission network in the next time period are estimated according to the traffic load. The traffic prediction model uses a graph convolutional network to perform multi-layer feature extraction on the structure of the network topology graph, wherein the features of each layer are obtained by aggregating the feature information of each node and its neighbors. Construct load balancing constraints, minimum blocking rate constraints, and minimum resource consumption constraints, and generate protection paths and protection time slot allocation schemes for each service flow based on the remaining capacity of each link, the types and corresponding quantities of available time slots in the next time period.

2. The adaptive protection path planning method for metropolitan area transmission networks according to claim 1, characterized in that, The pre-training method for the traffic prediction model includes: Obtain a training sample set of network topology graphs, wherein the training sample set includes sample feature nodes of the network topology; Obtain an initial graph convolutional network, wherein the initial graph convolutional network is based on... The feature nodes described in the layer are taken as input, and have passed through The final node features after iterative updates of the feature nodes described in the layer are the output; Obtain a training sample set for the prediction layer, which includes the final node features of the network topology and sample traffic data of multiple service flows; The initial traffic prediction model is trained using the training sample set of the prediction layer. The initial traffic prediction model takes the final node features of the network topology from time tp to time t of each service flow and the sample traffic data as input, and takes the sample traffic data at time t+1 as output. The parameters of the initial traffic prediction model are updated based on the mean square error loss function to obtain the traffic prediction model.

3. The adaptive protection path planning method for metropolitan area transmission networks according to claim 2, characterized in that, The initial graph convolutional network is randomly initialized.

4. The adaptive protection path planning method for metropolitan area transmission networks according to claim 2, characterized in that, The initial graph convolutional network includes a predetermined number of feature layers and a prediction layer. The prediction layer predicts the traffic information of each node in the next time period based on the node features output by the last feature layer and the sample traffic information of the current time period. The expression is: ； in, This represents the weights of the prediction layer. L represents the bias of the prediction layer, and L represents the number of feature layers. Indicates in The node features of the feature layer in the Lth layer of the time period. Indicates in Traffic information over a time period; ReLU represents a non-linear activation function.

5. The adaptive protection path planning method for metropolitan area transmission networks according to claim 2, characterized in that, The features of each layer are obtained by aggregating the feature information of each node and its neighbors. The expression for aggregating the feature information of each node and its neighbors is as follows: ； ； in, This represents the adjacency matrix of the metropolitan area transport network topology. Represents the identity matrix. This represents the adjacency matrix with added self-loops. yes The degree matrix, Indicates the first Nodes in each feature layer Node characteristics, Indicates the first The learnable weight matrix of each feature layer. This represents the activation function.

6. The adaptive protection path planning method for metropolitan area transmission networks according to claim 1, characterized in that, The expression for the load balancing constraint is: ； ； ； in, This represents the set of links in the metropolitan area transport network. This represents a link in the set of links. This indicates the link in the metropolitan area transport network. Traffic load situation, This represents the average traffic load of the links in the metropolitan area transport network. Indicates the business flow for the next time period Traffic load, This indicates the available capacity of link e in the next time period.

7. The adaptive protection path planning method for metropolitan area transmission networks according to claim 6, characterized in that, The expression for the minimum blocking rate constraint is: ； ； in, This indicates the proportion of blocked requests out of all requests. Define the i-th business flow Is it blocked? An equal value of 1 indicates a business flow. Unable to process due to insufficient resources A value of 0 indicates a business flow. The process was successfully completed due to sufficient resources. This indicates the total number of business flows.

8. The adaptive protection path planning method for metropolitan area transmission networks according to claim 7, characterized in that, The expression for the constraint to minimize resource consumption is: ； in, Indicates the business flow Protection path length, Indicates the business flow The number of time slots required on the protection path, Indicates the total number of business flows. The weighting coefficient represents the length of the protection path. The weighting coefficient represents the number of time slots required on the protection path.

9. The adaptive protection path planning method for metropolitan area transmission networks according to claim 8, characterized in that, The method further includes: The load balancing constraint, the minimum blocking rate constraint, and the minimum resource consumption constraint are combined to construct a comprehensive constraint, the expression of which is: ； in, Indicates the business flow The number of time slots required on the protection path, Indicates the business flow The protection path in the next time period, This indicates the load balancing constraint. This indicates the proportion of blocked requests out of all requests. The weighting coefficient represents the length of the protection path. A weighting coefficient representing the number of time slots required on the protection path. Indicates the business flow Protection path length, , and It is a weighting coefficient that balances the load balancing constraint, the blocking rate constraint, and the resource consumption constraint.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 9.

Images