Deployment Method of Software-Defined Optical Network Controller Based on Machine Learning

Abstract

The present invention discloses a deployment method for a software-defined optical network controller based on machine learning, including: traversing the SDON network topology based on a depth-first search strategy to find all possible paths between any two nodes among all nodes in the SDON network, and simultaneously counting the transmission delay values of each path; setting a delay threshold to filter out the control paths between nodes that meet the delay constraint conditions; constructing a new network topology graph of the SDON network, forming new abstract links with the control paths; dividing two nodes without a direct control path into different network regions; determining the controller positions for each network region; and selecting one from each controller as the control center to complete the deployment of the SDON network controller. The present invention reduces the control redundancy of the SDON while ensuring network survivability and improves the reliability of the SDON.

Description

Technical Field

[0001] The present invention belongs to the field of software-defined optical network controller deployment design, and specifically relates to a deployment method of a software-defined optical network controller based on machine learning. Background Art

[0002] With the large-scale deployment of 5G and B5G networks, various new applications and services, including the Internet of Things, autonomous driving, and blockchain, have developed rapidly. Correspondingly, the data traffic in the global Internet has maintained a stable and rapid growth. Since its emergence, the optical network has always been the basic bearer facility of the Internet due to its advantages such as large capacity, high transmission rate, and low loss, and has been widely used in various network application environments. However, with the exponential rapid growth of network traffic and the continuous improvement of people's requirements for network service quality, it has also brought huge challenges to the optical network. Facing the growth of network bandwidth demand and the increasingly complex business demand differences, the simple optical network evolution mode of capacity expansion can no longer meet the development needs, and an intelligent and reconfigurable optical network needs to be introduced to support services through intelligent management and control.

[0003] Since there are significant differences between the optical network and the ordinary network in the physical layer, transmission mechanism, and network architecture, if the software-defined network (SDN) method is directly used to manage the optical network, the SDN controller cannot achieve the dynamic configuration and optimization of the optical network, which will limit the flexibility and scalability of the optical network, resulting in a long fault recovery time of the optical network, reduced network reliability and availability, and other problems. Therefore, an innovative network architecture of software-defined optical network (SDON) needs to be adopted for the optical network. The software-defined optical network (SDON) can solve the problems of low resource utilization rate of the optical network management by the ordinary software-defined network (SDN) and poor scalability of the optical network. SDON can give full play to the advantages of high bandwidth, low latency, and high reliability of the optical network, and at the same time achieve flexible resource management and efficient network operation and maintenance.

[0004] The software-defined optical network (SDON) mostly uses a centralized controller for network control. Once the controller has problems or fails, it is very likely to cause the entire network to collapse, which will have a significant impact on application fields highly dependent on network resources such as cloud computing and big data, and may even exceed the acceptable range. At the same time, during the process of promoting the large-scale deployment of SDON controllers, the cost control issue has also become a key factor that cannot be ignored. How to optimize the number of deployed controllers while ensuring the efficient and reliable implementation of services has always been a highly concerned issue in the academic and industrial circles and is also a difficult point affecting the large-scale application of SDON.

[0005] Most of the SDON controller deployment work carried out so far has focused on aspects such as network transmission delay and control redundancy. However, in terms of the most critical survivability requirements of SDON and its impact on controller deployment strategies, there are still deficiencies such as the inability to effectively reduce the probability of network failures and a relatively large number of deployed controllers. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the related art to some extent.

[0007] An object of the present invention is to provide a deployment method for a software-defined optical network controller based on machine learning, which combines a graph convolutional network and a simulated annealing algorithm to reduce the control redundancy of SDON while ensuring network survivability and improve the reliability of SDON.

[0008] To achieve the above object, on the one hand, the present invention provides a deployment method for a software-defined optical network controller based on machine learning, including:

[0009] S100. Based on a depth-first search strategy, traverse the SDON network topology to find all possible paths from any starting node to any ending node in the SDON network, and simultaneously count the transmission delay values of each path ;

[0010] S200. Set a delay threshold , and use as a delay constraint condition to filter out the paths that meet the delay constraint condition as reachable paths; when there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path;

[0011] S300. Construct a new network topology graph of the SDON network. The construction rules of this new network topology graph include:

[0012] (1). Use all the control paths obtained in step S200 to form new abstract links;

[0013] (2). If there is no reachable path between and , divide the two nodes into nodes in different regions. If there is a reachable path between and , divide the two nodes into the same region. Based on this, form several network regions;

[0014] S400. Determine the controller location for each network region;

[0015] S500. Select one of the controllers as the control center to coordinate the work among the controllers and complete the deployment of the SDON network controller.

[0016] Further, the specific steps of step S100 include:

[0017] S110. Abstract the SDON network as an undirected graph , is a set of nodes, representing all nodes in the network, where is the number of nodes in the network; is the edge set, representing all edges in the network, where is the total number of edges;

[0018] S120. Use a stack structure to implement a traversal of the SDON network topology based on the depth - first search strategy; first select the starting node and push it onto the stack, then pop the top node of the stack as the current visited node, mark as visited, and add it to the current path list. Then, traverse the neighbor nodes. For each unvisited neighbor node , calculate the total path delay from to , and push onto the stack; finally, check the algorithm termination condition. If is the target node , then record the current path and its total delay value; if has no unvisited neighbor nodes, then backtrack: remove from the current path list and mark it as unvisited;

[0019] S130. Repeat the traversal process until the stack is empty to search all possible paths; after the traversal ends, obtain all feasible paths from to and their corresponding transmission delay values .

[0020] Further, the method for determining the controller location for each network area in step S400 is:

[0021] S410. Obtain the structure information and node features of the network topology through the GCN model to generate an initial solution for controller deployment;

[0022] S420. Take the initial solution generated by GCN as a heuristic solution and combine it with the simulated annealing algorithm to optimize the final controller location.

[0023] Further, in step S410, the structural information and node features of the network topology are obtained through the GCN model to generate an initial solution for controller deployment; the specific steps are as follows:

[0024] S411. Transform the controller deployment problem into an optimization problem of the network topology graph, and extract features for the nodes in the graph, including node degree, node traffic load, and clustering coefficient of the nodes;

[0025] S412. Construct a multi-layer GCN model to update the embedding representation of each node through multi-layer graph convolution; after multi-layer GCN calculation, output a controller fitness score indicating the possibility of each node as a controller;

[0026] S413. Use integer linear programming to generate the optimal controller deployment schemes for multiple SDON network topologies as the training set, and train the GCN model constructed in step S412. During the training process, GCN learns the relationship between node features and graph structure, so as to predict the optimal position of controller deployment;

[0027] S414. Use the trained GCN model to predict the new network topology graph of the SDON network constructed in step S300, and calculate the controller fitness; select the top k nodes with the highest fitness as the initial controller deployment scheme .

[0028] Further, in step S420, the initial solution generated by GCN is used as the heuristic solution, and combined with the simulated annealing algorithm to optimize the final controller position. The specific steps are as follows:

[0029] S421. Use the initial controller deployment scheme generated by GCN as the initial solution, set the initial temperature , and apply the simulated annealing algorithm for iteration. In each iteration, randomly select a node from the current solution, reverse the controller deployment state of this node, that is, change the node that originally deployed the controller to not deploy the controller or the node that did not deploy the controller to deploy the controller, and generate a new solution ; check whether the new solution satisfies the coverage constraint, that is, each network node is either a controller or adjacent to at least one controller. If the coverage constraint is not satisfied, adjust the controller position or add controllers until the coverage constraint is satisfied;

[0030] S422. Define the objective function for optimizing controller deployment by the simulated annealing algorithm as:

[0031]

[0032] where is the total number of nodes in the network; Indicates that the controller is deployed on the node, Indicates that the controller is not deployed on the node; Is the number of nodes not covered; Is the weight coefficient;

[0033] Calculate a new solution And the difference in the objective function between the new solution and the current solution , if , then directly accept the new solution ; Otherwise, accept the new solution according to the probability ;

[0034] S423. After each iteration, the temperature gradually decreases according to the set temperature reduction coefficient , that is ; When no new solution is accepted for several consecutive times, the algorithm ends and outputs the current solution as the final controller position.

[0035] Furthermore, in step S500, select the node with the minimum average delay for all controllers to deploy the control center. The specific method is as follows:

[0036] Let the position of the controller deployment node be , and the deployment position of the control center satisfies:

[0037]

[0038] Among them, Indicates taking the average value; Indicates the transmission delay between deployment nodes.

[0039] Furthermore, after selecting the node with the minimum average delay for all controllers to deploy the control center in step S500, select the node with the second lowest average delay for all controllers to deploy the auxiliary control center as a backup for the control center.

[0040] Furthermore, this method also includes deploying optical performance monitoring, using machine learning algorithms to detect SDON network anomalies and locate fault points; The specific method is as follows:

[0041] Interact with optical performance monitoring through the southbound interface SBI of the SDON network controller, collect optical performance data, and perform standardized processing on the data;

[0042] Then build and train an isolation forest model for anomaly monitoring, combine the anomaly data points with the network topology information to locate the location where the fault occurs;

[0043] Feed the detection results back to the controller of the SDON network to trigger the fault recovery operation.

[0044] On the other hand, the present invention provides a non-transitory computer-readable storage medium storing computer instructions that cause a computer to execute the above-mentioned method for deploying a machine learning-based software-defined optical network controller.

[0045] On yet another aspect, the present invention provides an electronic device, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus, and the processor invokes logic instructions in the memory to execute the above-mentioned method for deploying a machine learning-based software-defined optical network controller.

[0046] On still another aspect, the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer executes the above-mentioned method for deploying a machine learning-based software-defined optical network controller.

[0047] Advantageous effects: (1) The method for deploying a machine learning-based software-defined optical network controller of the present invention adopts a depth-first search strategy. In a software-defined optical network, the depth-first search strategy can efficiently traverse the network topology, record all possible paths and their latency values, and provide reliable support for path planning and optimization.

[0048] (2) The present invention uses a graph convolutional network (GCN) to generate an initial solution for SDON controller deployment, which can make full use of the structural information and node features of the network topology to generate a better initial solution for SDON controller deployment. The key advantage of GCN is that it can use the optimal solution of a small-scale network as training data, thereby extracting the inherent patterns and dependencies in the network topology and extending the solution to a large-scale network.

[0049] (3) The present invention uses a simulated annealing algorithm to determine the controller location. In this way, it can effectively optimize the controller deployment, minimize the number of deployed controllers, and ensure the coverage of the network at the same time.

[0050] (4) The present invention adopts a hierarchical deployment controller architecture, coordinates the work among various controllers through a control center, and at the same time selects an auxiliary location with a relatively low average latency for all controllers to deploy an additional control center as a backup. This avoids an increase in the probability of network failures caused by the failure of the selected control center.

[0051] (5) The present invention uses high-performance optical performance monitoring (OPM) combined with an isolation forest algorithm to analyze the monitoring data of SDON, which can quickly identify and locate abnormal events in the optical network, thereby improving the reliability and operation and maintenance efficiency of the network. Description of the Drawings

[0052] Figure 1 This is the overall flowchart of the deployment method of the machine learning-based software-defined optical network controller of the present invention.

[0053] Figure 2 This is the model diagram of the hierarchical deployment architecture in the embodiment of the present invention.

[0054] Figure 3 This is the pan-European COST239 network topology diagram used in the simulation experiment in the embodiment of the present invention.

[0055] Figure 4 This is the North American NSF network topology diagram used in the simulation experiment in the embodiment of the present invention.

[0056] Figure 5 This is a comparison chart of the number of controller deployments in the pan-European COST239 network topology using different algorithms.

[0057] Figure 6 This is a comparison chart of the number of controller deployments in the North American NSF network topology using different algorithms.

[0058] Figure 7 This is a graph showing the change in network failure probability in the pan-European COST239 network topology using different algorithms.

[0059] Figure 8 This is a graph showing the change in network failure probability in the North American NSF network topology using different algorithms.

[0060] Figure 9 This is a graph showing the change in network failure recovery probability in the pan-European COST239 network topology using different algorithms.

[0061] Figure 10 This is a graph showing the change in network failure recovery probability in the North American NSF network topology using different algorithms.

[0062] Figure 11 This is the graph showing the performance of optical network traffic prediction in the embodiment of the present invention. Detailed implementation manners

[0063] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention, and they should not be construed as limiting the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for the purpose of description and cannot be construed as indicating or implying relative importance.

[0064] The following combines with Figures 1 - 11 to describe the deployment method of the software-defined optical network controller based on machine learning provided by the present invention.

[0065] Embodiment 1: This embodiment provides a deployment method of a software-defined optical network controller based on machine learning.

[0066] For a software-defined optical network, the main factor affecting the survivability of the optical network is the security of fiber optic links. Data is easily lost when a link fails. Therefore, in an SDON network, according to the actual network transmission requirements, multiple controllers need to be deployed. This requires a balance between improving the survivability of the optical network and reducing the deployment cost. This embodiment provides an improved deployment method for a software-defined optical network controller. For a large-scale network structure, a hierarchical deployment architecture is adopted. The specific deployment method is as Figure 1 shown and includes:

[0067] S100. Based on the depth-first search strategy (DFS), traverse the SDON network topology to find all possible paths between any two nodes in the SDON network, and simultaneously count the transmission delay values of each path. The specific steps include:

[0068] S110. Abstract the SDON network as an undirected graph , where \(\mathcal{V}\) is the set of nodes, representing all nodes in the network, and among them \(|\mathcal{V}|\) is the number of nodes in the network; \(\mathcal{E}\) is the edge set, representing all edges in the network, and among them \(|\mathcal{E}|\) is the total number of edges;

[0069] S120. Use a stack structure to implement the traversal of the SDON network topology based on the depth-first search strategy; first select the starting node and push it onto the stack, then take out the top node of the stack as the current visited node, mark as visited, and add it to the current path list. Then, traverse the neighbor nodes. For each unvisited neighbor node of , calculate the total path delay from to , and push onto the stack; finally, check the algorithm termination condition. If is the target node , then record the current path and its total delay value; if has no unvisited neighbor nodes, then backtrack: remove

[0070] S130. Repeat the traversal process until the stack is empty to search for all possible paths. After the traversal ends, all feasible paths from to and their corresponding transmission delay values are obtained. to all feasible paths and their corresponding transmission delay values .

[0071] S200. Set a delay threshold to filter the control paths between nodes. The specific method is as follows:

[0072] Set a delay threshold , and obtain all possible paths and their corresponding transmission delay values between nodes and from the path search algorithm. If , add the corresponding path to the set of reachable paths; otherwise, add the path to the set of unreachable paths. When there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path. and all possible paths and their corresponding transmission delay values . If , add the corresponding path as a reachable path to the set of valid paths; otherwise, add the path to the set of unreachable paths. When there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path.

[0073] S300. Construct a new network topology graph for the SDON network.

[0074] The construction rules of this new network topology graph include:

[0075] (1) Form new abstract links with all the control paths obtained in step S200. In the new network topology, the abstract link between nodes and represents a path with the minimum reachable and minimum delay from to . and the abstract link between represents a path with the minimum reachable and minimum delay from to with the minimum reachable and minimum delay.

[0076] (2) If there is no reachable path between and (i.e., the path does not meet the delay constraint), it means that the two nodes belong to different network regions, then divide the two nodes into nodes in different regions; if there is a reachable path between and , then include the two nodes in the same region, and form several network regions accordingly. and if there is no reachable path between, it means that the two nodes belong to different network regions, then divide the two nodes into nodes in different regions; if there is a reachable path between and , then include the two nodes in the same region, and form several network regions accordingly.

[0077] Through the above rules, a simplified regional-level topology can be constructed, where each region represents a subnet with internal connectivity, and the connections between regions are represented only by abstract links for reachable communication paths.

[0078] S400. Determine the controller location for each network region. It includes:

[0079] S410. Obtain the structural information and node features of the network topology through a Graph Convolutional Network (GCN) model to generate an initial solution for controller deployment. The specific steps are as follows:

[0080] S411. Transform the controller deployment problem into an optimization problem of the network topology graph, and extract features for the nodes in the graph, including node degree, node traffic load, and node clustering coefficient;

[0081] S412. Construct a multi-layer GCN model to update the embedding representation of each node through multi-layer graph convolution; after multi-layer GCN calculation, output a controller fitness score indicating the possibility of each node as a controller;

[0082] S413. Use integer linear programming to generate multiple optimal controller deployment schemes for the SDON network topology as the training set, and train the GCN model constructed in step S412. During the training process, the GCN learns the relationship between node features and the graph structure, so as to predict the optimal location of controller deployment;

[0083] S414. Use the trained GCN model to predict the new network topology graph of the SDON network constructed in step S300, and calculate the controller fitness; select the top k nodes with the highest fitness as the initial controller deployment scheme , where the number of k is determined by the number of nodes in the SDON network topology.

[0084] S420. Use the initial solution generated by the GCN as a heuristic solution, and combine the simulated annealing algorithm to optimize the final controller location. When the simulated annealing algorithm optimizes the controller deployment strategy, the algorithm objective is to minimize the number of controllers and ensure the coverage of the network. The specific steps are as follows:

[0085] S421. Use the initial controller deployment scheme generated by the GCN as the initial solution, set the initial temperature , and apply the simulated annealing algorithm for iteration. In each iteration, randomly select a node from the current solution and reverse the controller deployment status of this node, that is, change the node that originally deployed the controller to not deploy the controller or change the node that did not deploy the controller to deploy the controller to generate a new solution ; check whether the new solution satisfies the coverage constraint, that is, each network node is either a controller or adjacent to at least one controller. If the coverage constraint is not satisfied, adjust the controller deployment by adding controllers or adjusting the controller location until the coverage constraint is satisfied;

[0086] S422. Define the objective function for optimizing controller deployment by the simulated annealing algorithm as:

[0087]

[0088] wherein is the total number of nodes in the network; indicates that a controller is deployed on the node, indicates that no controller is deployed on the node; is the number of nodes not covered; is the weight coefficient;

[0089] Calculate the new solution and the difference in the objective function between the new solution and the current solution , if , directly accept the new solution ; otherwise, accept the new solution according to the probability ;

[0090] S423. After each iteration, the temperature is gradually reduced according to the set temperature reduction coefficient , that is ; when no new solution is accepted for several consecutive times, the algorithm ends and outputs the current solution as the final controller position.

[0091] S500. Select the node with the minimum average delay for all controllers to deploy the control center. The specific method is as follows:

[0092] Let the position of the controller deployment node be , and the deployment position of the control center satisfies:

[0093]

[0094] wherein represents taking the average value; represents the transmission delay between deployment nodes.

[0095] S600. To avoid an increase in the probability of network failures caused by the failure of the selected control center, select an auxiliary location with a relatively low average delay for all controllers to deploy an additional control center as a backup.

[0096] Thus, a hierarchical deployment architecture of controllers for large-scale network structures is constructed. Figure 2 As shown, in this architecture, the controller is responsible for controlling the OpenFlow switch, the control center is responsible for coordinating the work of the controllers, and at the same time, a backup control center is deployed to ensure the stability of the network.

[0097] S700. Deploy optical performance monitoring, and use machine learning algorithms to detect SDON network anomalies and locate fault points. The specific method is as follows:

[0098] Interact with optical performance monitoring through the southbound interface SBI of the SDON network controller, collect optical performance data, and perform standardization processing on the data;

[0099] Then, an isolation forest model is constructed and trained for anomaly monitoring, and the anomaly data points are combined with the network topology information to locate the position where the fault occurs;

[0100] The detection result is fed back to the controller of the SDON network to trigger the fault recovery operation.

[0101] The following is to verify this embodiment through simulation experiments.

[0102] The deployment method of the software-defined optical network controller based on machine learning in this embodiment (hereinafter abbreviated as the USCD method) is applied to the Figure 3 pan-European COST239 network topology (11 nodes, 26 edges) as shown in Figure 4 and the North American NSF network topology (14 nodes, 21 edges) simulation environment as shown in

[0103] Definition 1. Probability of fault alarm occurrence It is the probability that the working route of the control channel fails, expressed as:

[0104]

[0105] Definition 2. Probability of fault recovery It is the probability that the control channel quickly recovers by relying on the protection route of the control channel, expressed as:

[0106]

[0107] Among them, represents the fiber fault probability per 100 km; represents the th OpenFlow switch node; represents the th controller node; represents the length of the working route of the control channel, represents the length of the protection route.

[0108] The USCD method proposed in this embodiment takes the survivability of the SDON control plane as a constraint condition, and the survivability of the control plane can be transformed into the control link length as a constraint condition. According to the fiber fault probability per 100 km and the maximum network fault probability provided by the user, the longest control link length is obtained as the constraint condition of the USCD method proposed in this embodiment.

[0109] As shown inFigure 5 and Figure 6 As shown in Figure 6 , the comparison of the number of controllers deployed in the pan-European COST239 network topology and the North American NSF network topology respectively under the constraint of the same longest control link length (under the user-acceptable failure probability) by using the DCD algorithm, the SCD algorithm and the USCD method proposed in this embodiment is presented. It can be seen from the simulation results that the number of controllers deployed decreases as the user's requirement for network survivability decreases, and the deployment of the number of controllers of the USCD method proposed in this embodiment is better than that of the DCD algorithm and the SCD algorithm in different parameters and different network topologies. This shows that the USCD method of this embodiment can effectively reduce the number of controllers deployed and reduce control redundancy.

[0110] As Figure 7 and Figure 8 shown in Figure 8 , the graph of the change of network failure probability in the pan-European COST239 network topology and the North American NSF network topology respectively when applying the C-MPC algorithm, the SCD algorithm and the USCD method proposed in this embodiment with the same number of controllers deployed is presented. From Figure 7 this, it can be seen that the three algorithms are basically on par in terms of network reliability. This is because in the COST239 network topology, the number of nodes is small, the number of edges is large, and the distance between nodes is short. Therefore, it is not obvious to distinguish the advantages and disadvantages between the algorithms. In the NSF network with lower connectivity and longer link length, as Figure 8 shown in Figure 8 , the USCD method proposed in this embodiment is significantly better than the other two algorithms, greatly reducing the optical network failure probability and making the optical network have good performance.

[0111] As Figure 9 and Figure 10 shown in Figure 10 , the graph of the change of the network fault recovery probability realized by the control plane when applying the MCC algorithm, the SVVR algorithm and the USCD method proposed in this embodiment in the pan-European COST239 network topology and the North American NSF network topology respectively after the fault alarm appears in the SDON network is presented. It can be clearly seen that the fault recovery probability of the USCD method proposed in this embodiment is better than that of the other two deployment models in both the COST239 and NSF network topologies. At the same time, because the network node connectivity in the COST239 network is larger than that in the NSF network, the controller can start the recovery mechanism faster and has a better protection effect on the network. Therefore, the USCD method proposed in this embodiment has greater benefits in the COST239 network.

[0112] The method of this embodiment can be applied to the prediction of optical network traffic. After verification, when it is applied to the prediction of optical network traffic, its performance is as Figure 11 shown in Figure 11 . Figure 11Among them, the blue solid line represents the simulated optical network traffic model, and the red dotted line represents the traffic prediction result of the present invention. By comparison, it can be seen that the simulated optical network traffic model presents multiple obvious peak fluctuations, indicating that the network traffic has significant burstiness and instability. However, the prediction result of the USCD method proposed in this embodiment is highly consistent with the actual situation. Even in the extreme case of sudden traffic increase, it can still accurately track the traffic change trend and demonstrate excellent prediction performance. Throughout the prediction period, the prediction curve and the real traffic curve maintain a good degree of fit, fully proving the effectiveness and reliability of the present invention in optical network traffic prediction.

[0113] Embodiment 2: This embodiment provides a non-transitory computer-readable storage medium, on which computer instructions are stored. These computer instructions cause a computer to execute a method for deploying a software-defined optical network controller based on machine learning. The method includes the following steps:

[0114] S100. Based on the depth-first search strategy, traverse the SDON network topology to find all possible paths between any two nodes among all nodes in the SDON network, and at the same time count the transmission delay values of each path.

[0115] S200. Set a delay threshold to filter out the paths that meet the delay constraint conditions as reachable paths; when there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path.

[0116] S300. Construct a new network topology graph of the SDON network, form new abstract links with the control paths; divide the two nodes without a direct control path into different network regions.

[0117] S400. Determine the controller location for each network region; and select one of the controllers as the control center to complete the deployment of the SDON network controller.

[0118] Embodiment 3: This embodiment provides an electronic device, which may include: a processor, a communications interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus. The processor can call the logical instructions in the memory to execute a method for deploying a software-defined optical network controller based on machine learning. The method includes the following steps:

[0119] S100. Traverse the SDON network topology based on the depth - first search strategy to find all possible paths between any two nodes among all nodes in the SDON network, and simultaneously count the transmission delay values of each path;

[0120] S200. Set a delay threshold to filter out the paths that meet the delay constraint conditions as reachable paths; when there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path;

[0121] S300. Construct a new network topology graph of the SDON network, form new abstract links with the control paths; divide the two nodes without a direct control path into different network regions;

[0122] S400. Determine the controller positions for each network region; and select one of the controllers as the control center to complete the deployment of the SDON network controller.

[0123] In addition, when the logical instructions in the above - mentioned memory can be implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer - readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read - only memory (ROM, Read - Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0124] Embodiment 4: What this embodiment provides is a computer program product. The computer program product includes a computer program. The computer program can be stored on a non - transitory computer - readable storage medium. When the computer program is executed by a processor, the computer can execute the method for deploying a software - defined optical network controller based on machine learning. The method includes the following steps:

[0125] S100. Traverse the SDON network topology based on the depth - first search strategy to find all possible paths between any two nodes among all nodes in the SDON network, and simultaneously count the transmission delay values of each path;

[0126] S200. Set a delay threshold, and filter out the paths that meet the delay constraint conditions as reachable paths; when there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path.

[0127] S300. Construct a new network topology graph of the SDON network, and form new abstract links with the control paths; divide the two nodes without a direct control path into different network regions.

[0128] S400. Determine the positions of the controllers for each network region; and select one of the controllers as the control center to complete the deployment of the SDON network controllers.

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

[0130] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art can be embodied in the form of a software product, and this computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A deployment method for a software-defined optical network controller based on machine learning, characterized in that Including: S100. Based on the depth - first search strategy, traverse the SDON network topology to find all possible paths from any starting node to any ending node in the SDON network, and at the same time, count the transmission delay value of each path ; S200. Set the delay threshold , and use it as the delay constraint condition to filter out the paths that meet the delay constraint condition as reachable paths; when there is only one reachable path between two nodes, this reachable path is used as the control path between the two nodes; when there are multiple reachable paths between two nodes, select the reachable path with the minimum delay as the control path; S300. Construct a new network topology diagram for the SDON network. The construction rules of the new network topology diagram include: (1). Form new abstract links with all the control paths obtained in step S200. (2) If and there is no reachable path between them, then the two nodes are classified as nodes in different regions. If and there is a reachable path between them, then the two nodes are included in the same region. Based on this, several network regions are formed; S400. Determine the controller locations for each network area. Specifically: S410. Obtain the structure information and node features of the network topology through the GCN model, and generate an initial solution for controller deployment. S420. Use the initial solution generated by the GCN as the heuristic solution and combine it with the simulated annealing algorithm to optimize the final controller position; S500. Select one of the controllers as the control center to coordinate the work among the controllers and complete the deployment of the SDON network controllers.

2. The deployment method of the machine learning-based software-defined optical network controller according to claim 1, wherein The specific steps of step S100 include: S110. Abstract the SDON network as an undirected graph , is a set of nodes, representing all nodes in the network, where is the number of nodes in the network; is the edge set, representing all edges in the network, where is the total number of edges; S120. Use a stack structure to implement the traversal of the SDON network topology based on the depth-first search strategy; first select the starting node and push it onto the stack, then pop the top node of the stack as the currently visited node, mark as visited, and add it to the current path list. Then, traverse the neighbor nodes. For each unvisited neighbor node of , calculate the total path delay from to , and push onto the stack; finally, check the algorithm termination condition. If is the target node , record the current path and its total delay value; if there are no unvisited neighbor nodes, then backtrack: remove from the current path list and mark it as unvisited; S130. Repeat the traversal process until the stack is empty to search all possible paths. After the traversal ends, all feasible paths from to and their corresponding transmission delay values are obtained. to and their corresponding transmission delay values .

3. The deployment method of the machine learning-based software-defined optical network controller according to claim 1, wherein In step S410, obtain the structure information and node features of the network topology through the GCN model, and generate an initial solution for controller deployment. The specific steps are: S411. Convert the controller deployment problem into an optimization problem of the network topology diagram, and extract features for the nodes in the diagram, including node degree, node traffic load, and clustering coefficient of the nodes. S412. Construct a multi-layer GCN model, and update the embedding representation of each node through multi-layer graph convolution. After multi-layer GCN calculation, output a controller fitness score indicating the possibility of each node as a controller. S413. Use integer linear programming to generate multiple optimal controller deployment schemes for the SDON network topology as the training set, and train the GCN model constructed in step S412. During the training process, GCN learns the relationship between node features and the graph structure, so as to predict the optimal location of controller deployment. S414. Use the trained GCN model to predict the new network topology graph of the SDON network constructed in step S300, and calculate the controller fitness; select the top k nodes with the highest fitness as the initial controller deployment plan .

4. The deployment method of the software-defined optical network controller based on machine learning according to claim 3, wherein In step S420, the initial solution generated by the GCN is used as a heuristic solution, and the simulated annealing algorithm is combined to optimize the final controller position. The specific steps are as follows: S421. Deploy the initial controller deployment plan generated by GCN As the initial solution, set the initial temperature , and apply the simulated annealing algorithm for iteration. In each iteration, randomly select a node from the current solution and reverse the controller deployment status of this node, that is, a node that originally deployed a controller will no longer deploy a controller or a node that did not deploy a controller will deploy a controller to generate a new solution ; Check whether the new solution satisfies the coverage constraint, that is, each network node is either a controller or adjacent to at least one controller. If the coverage constraint is not satisfied, add controllers or adjust the positions of controllers until the coverage constraint is satisfied; S422. Define the objective function for optimizing controller deployment by the simulated annealing algorithm as: ； wherein is the total number of nodes in the network; indicates that the node deploys a controller, indicates that the node does not deploy a controller; is the number of nodes not covered; is the weight coefficient; Calculate the new solution and the objective function difference between the new solution and the current solution , if , directly accept the new solution ; otherwise, accept the new solution according to the probability ; S423. After each iteration, the temperature is gradually reduced according to the set temperature reduction coefficient as follows ; when no new solution is accepted for several consecutive times, the algorithm ends and outputs the current solution as the final controller position.

5. The deployment method of the machine learning-based software-defined optical network controller according to claim 1, characterized in that, In step S500, select the node with the minimum average delay for all controllers to deploy the control center. The specific method is: Let the location of the controller deployment node be , and the deployment location of the control center satisfies: ； Among them, represents taking the average value; represents the transmission delay between deployed nodes, is the total number of nodes in the network.

6. The deployment method of the machine learning-based software-defined optical network controller according to claim 5, characterized in that, In step S500, after selecting the node with the minimum average delay for all controllers to deploy the control center, select the node with the second lowest average delay for all controllers to deploy the auxiliary control center as a backup for the control center.

7. The deployment method of the machine learning-based software-defined optical network controller according to claim 5, characterized in that This method further includes deploying optical performance monitoring, and using machine learning algorithms to detect SDON network anomalies and locate fault points. The specific method is: Interact with the optical performance monitoring through the southbound interface SBI of the SDON network controller, collect optical performance data, and perform standardized processing on the data. Then construct and train an isolation forest model for anomaly monitoring, combine the anomaly data points with the network topology information, and locate the location where the fault occurs. Feed back the detection results to the controller of the SDON network to trigger the fault recovery operation.

8. A non-transitory computer-readable storage medium, characterized in that, It stores computer instructions, and the computer instructions cause the computer to execute the method for deploying a software-defined optical network controller based on machine learning according to any one of claims 1-7.

9. A computer program product, characterized in that, The computer program product includes a computer program. The computer program is stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer executes the method for deploying a software-defined optical network controller based on machine learning according to any one of claims 1-7.

Images