A network controller and network switching device configuration method

Abstract

The present invention relates to the field of communication network management, and in particular to a method for configuring a network controller and a network switching device. The method comprises: generating a network state matrix based on network state data, and predicting the future network state; adjusting the weights of each dimension of the network based on the network state matrix and the predicted network state data; adjusting the network state matrix based on the adjusted weight vector; constructing an objective function, performing nonlinear optimization on the paths in the network, and selecting the optimal path in combination with nonlinear constraints; performing hierarchical optimization and global coordination on each sub-area of ​​the divided network, generating the optimal configuration strategy for each sub-area, and sending the optimal configuration strategy to each network switching device. The method solves the problems of low technical configuration efficiency, difficulty in responding to changes in network state in real time, insufficient dynamic adaptability, and inability to achieve optimal allocation of network resources, resulting in a decline in overall network performance, and thus increasing the risk of network failures and security issues.

Description

Technical Field

[0001] The present invention relates to the field of communication network management, and in particular to a method for configuring a network controller and a network switching device. Background Art

[0002] With the rapid development of information technology, networks play an increasingly important role in all walks of life. From small local area networks to large-scale wide area networks, the complexity of network infrastructure continues to increase. In this context, the configuration of network controllers and network switching devices is particularly important. The network controller is responsible for managing and controlling the transmission path of data in the network, while the network switching device is responsible for specific data packet forwarding and traffic management. In order to ensure the efficient operation of the network, it must be reasonably configured and optimized according to the real-time network status to cope with changes in network traffic, fluctuations in link status, and possible external interference.

[0003] However, with the expansion of network scale and the diversification of application scenarios, the network environment has become more complex, and changes in network status have become more frequent and difficult to predict. Therefore, traditional static configuration methods can no longer meet the needs of modern networks for efficiency, dynamic adaptability and security. In this context, there is an urgent need for an intelligent configuration method that can dynamically adjust the configuration of network controllers and network switching devices based on real-time network status and predictions of future network status, ensuring that the network always maintains optimal performance under various complex conditions.

[0004] The configuration of existing network controllers and network switching equipment still has the following technical problems: low configuration efficiency, making it difficult to respond to changes in network status in real time; insufficient dynamic adaptability, resulting in the network being unable to adjust in time to maintain stable network performance during peak load periods or when failures occur; and the inability to achieve optimal allocation of network resources, resulting in an overall decline in network performance, which in turn increases the risk of network failures and security issues. Summary of the invention

[0005] The present invention provides a method for configuring a network controller and a network switching device to solve the following technical problems existing in the configuration of the existing network controller and the network switching device: low configuration efficiency, making it difficult to respond to changes in the network state in real time; insufficient dynamic adaptability, resulting in the network being unable to adjust in time to maintain stable network performance during peak load periods or when a failure occurs; and failure to achieve optimal allocation of network resources, resulting in an overall decline in network performance, thereby increasing the risk of network failures and security issues.

[0006] The present invention provides a method for configuring a network controller and a network switching device, comprising the following steps:

[0007] S1: collect network status data in real time, generate a network status matrix based on the network status data; predict the future network status through a prediction model to obtain predicted network status data; adjust the weights of each dimension of the network based on the network status matrix and the predicted network status data to obtain an adjusted weight vector; adjust the network status matrix based on the adjusted weight vector;

[0008] S2: Based on the adjusted weight vector and network status matrix, construct the objective function, perform nonlinear optimization on the paths in the network, and select the optimal path in combination with nonlinear constraints; divide the network into sub-areas, perform hierarchical optimization and global coordination on each sub-area, generate the optimal configuration strategy for each sub-area, and send the optimal configuration strategy to each network switching device.

[0009] Through hierarchical optimization and global coordination mechanism, the present invention can realize local optimization in each sub-area while realizing optimal allocation of network resources through global coordination, thereby ensuring the consistency of overall network performance, avoiding conflicts between sub-areas, and improving the performance and resource utilization efficiency of the entire network;

[0010] The present invention introduces nonlinear constraints in the path optimization process to ensure that the link tolerance and the complexity of the network topology are fully considered in the network configuration process, to prevent network bottlenecks and potential safety hazards caused by excessive pursuit of a single optimization goal, and to improve the safety and stability of network operation.

[0011] Through the automated configuration strategy generation and adjustment mechanism, the controller can automatically issue the optimal configuration strategy to each network switching device according to the global optimization goal, reducing the complexity of manual operations, lowering the difficulty of network operation and maintenance, and making network management more efficient and convenient.

[0012] Preferably, the S1 specifically includes:

[0013] The network state data is input into the network state matrix. Each layer of the network state matrix represents a dimension of the network, and the data of all dimensions are dynamically updated and combined through nonlinear mapping functions.

[0014] Preferably, the S1 specifically includes:

[0015] A multi-level feedback mechanism is introduced to adjust the weights of each dimension of the network.

[0016] Preferably, the S1 specifically includes:

[0017] The specific implementation process of the multi-level feedback mechanism is as follows:

[0018] Combined with the current network state matrix and the predicted network state data, the weights before adjustment are initialized. The goal of weight adjustment is to minimize the loss function. Based on the minimization of the loss function and combined with the predicted network state data, the weights are adjusted by the gradient descent method to obtain the adjusted weight vector.

[0019] Preferably, the S2 specifically includes:

[0020] Based on the adjusted weight vector and the network state matrix, an objective function is constructed, and the adjusted weight vector is used to balance the optimization target.

[0021] Preferably, the S2 specifically includes:

[0022] A nonlinear path optimization model is constructed to solve the objective function. Specifically, each path is scored according to the objective function, and the path with the smallest objective function value is selected as the optimal path;

[0023] The objective function is:

[0024]

[0025] in, O ( P ) is a set of paths The objective function to be optimized; e is a link, indicating a path; W D (t+1) is the adjusted delay weight; D e Yes Link e Delays; W B (t+1) is the adjusted bandwidth weight; B e Yes Link e bandwidth; R e Yes Link e Node reliability; W R (t+1) is the adjusted reliability weight; W L (t+1) is the adjusted load weight; L e Yes Link e of load.

[0026] Preferably, the S2 specifically includes:

[0027] In the process of solving the objective function, nonlinear constraints are added to control the selection of the optimal path through the tolerance coefficient and connectivity factor of the link.

[0028] Preferably, the S2 specifically includes:

[0029] After selecting the optimal path, each sub-area is optimized hierarchically to adjust the path selection and traffic distribution in each sub-area. The global coordination mechanism integrates the hierarchical optimization results of each sub-area by using the global optimization objective function to obtain the optimal traffic distribution and path selection in the network. The formula of the global optimization objective function is:

[0030]

[0031] in, is the global optimization objective function; R is the number of sub-regions; oh r ( t +1) is the Sub-areas t +1 moment weight; P r It is r Path selection results for each sub-area; ϕ ( t +1)Yes t The weight of the global adjustment factor at the +1 moment; M is the number of global adjustment factors; It is with q Network performance indicators related to global adjustment factors; P q It is with q The path selection results related to the global adjustment factor.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The present invention provides a method for configuring a network controller and a network switching device. Through hierarchical optimization and a global coordination mechanism, local optimization can be achieved in each sub-area. At the same time, the optimal allocation of network resources can be achieved through global coordination, thereby ensuring the consistency of the overall performance of the network, avoiding conflicts between sub-areas, and improving the performance and resource utilization efficiency of the entire network.

[0034] 2. The present invention provides a method for configuring a network controller and a network switching device. By introducing nonlinear constraints in the path optimization process, the tolerance of the link and the complexity of the network topology are fully considered in the network configuration process, and the network bottleneck and potential safety hazards caused by excessive pursuit of a single optimization goal are prevented, thereby improving the safety and stability of network operation.

[0035] 3. The present invention provides a method for configuring a network controller and a network switching device. Through an automated configuration strategy generation and adjustment mechanism, the controller can automatically send the optimal configuration strategy to each network switching device according to the global optimization goal, thereby reducing the complexity of manual operations, reducing the difficulty of network operation and maintenance, and making network management more efficient and convenient. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The present invention is a flow chart of a method for configuring a network controller and a network switching device. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below in conjunction with specific embodiments. However, this should not be understood as the scope of the above subject matter of the present invention being limited to the following embodiments, and all technologies realized based on the content of the present invention belong to the scope of the present invention.

[0038] Unless otherwise specified, in the description of the specific embodiments of the present invention, the terms indicating the orientation or position relationship such as "up", "down", "left", "right", "center", "inside", "outside", etc. are all expressions based on the orientation or position relationship shown in the drawings, or are the orientation or position relationship when the invented product / equipment / device is usually used. These terms of orientation or position relationship are only for the convenience of describing the scheme of the present invention or simplifying the description in the specific embodiments, so as to facilitate the technicians to quickly understand the scheme, and do not indicate or imply that a specific device / component / element must have a specific orientation, or be constructed and operated in a specific position relationship, and therefore cannot be understood as a limitation on the present invention.

[0039] In addition, if the terms "horizontal", "vertical", "overhanging", "parallel" and the like appear, it does not mean that the corresponding devices / components / elements are required to be absolutely horizontal or vertical or overhanging or parallel, but may be slightly tilted or have deviations. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but may be slightly tilted. Alternatively, it can be simplified to mean that the corresponding devices / components / elements are set in directions such as "horizontal", "vertical", "overhanging", "parallel", etc., and can have an error / deviation of ±10% relative to the corresponding direction setting, more preferably an error / deviation within ±8%, more preferably an error / deviation within ±6%, more preferably an error / deviation within ±5%, and more preferably an error / deviation within ±4%. As long as the corresponding device / component / element is within the error / deviation range, it can still achieve its role in the scheme of the present invention.

[0040] In addition, the expressions “first”, “second”, “third”, etc., which appear in the terms, are merely used to distinguish the description of the same or similar components and should not be understood as emphasizing or implying the relative importance of specific components.

[0041] In addition, in the description of the embodiments of the present invention, "several", "plurality" and "a number" represent at least 2. It can be any number such as 2, 3, 4, 5, 6, 7, 8, 9, and even more than 9.

[0042] In addition, in the description of the technical solution of the present invention, unless otherwise clearly specified / defined / restricted, the terms "set", "install", "connect", "connected", "provided with", "laid", and "arranged" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection, and can be welding, riveting, bolting, threading, and other commonly used connection means in the field. This connection can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection, or an indirect connection through an intermediate medium, and it can be the internal connection of two elements.

[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0045] Example 1

[0046] A specific scheme of a method for configuring a network controller and a network switching device provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

[0047] See attached Figure 1 , which shows a flow chart of a method for configuring a network controller and a network switching device provided by an embodiment of the present invention, the method comprising the following steps:

[0048] S1: collect network status data in real time, generate a network status matrix based on the network status data; predict the future network status through a prediction model to obtain predicted network status data; adjust the weights of each dimension of the network based on the network status matrix and the predicted network status data to obtain an adjusted weight vector; adjust the network status matrix based on the adjusted weight vector;

[0049] In order to obtain network status information, the controller is used to collect network status data in real time, including network topology, connection status of each node, link bandwidth, delay, load status, data packet priority, historical performance of nodes, dynamic changes of links, and external interference factors.

[0050] Specifically, the controller obtains the connection status of each network node through the network detection module, and the connection status can be read in real time through the management interface on the switch or router, such as SNMP. For the bandwidth and delay information of the link, the controller sends active detection packets (such as ICMP Ping or customized detection packets) to each node, calculates the response time and the returned link status information, and evaluates the status of the network link in real time. The link load status is usually collected through the device's traffic statistics interface (such as NetFlow or sFlow). The controller periodically extracts traffic data from the switching device to obtain the current link load status.

[0051] The controller obtains the priority of the data packet by analyzing the QoS tag information in the network to ensure that data packets of different priorities can be correctly scheduled. For the historical performance of nodes and the dynamic changes of links, the controller continuously monitors and records the historical operating status of each node and link, and uses the existing time series analysis method to predict the possible future changes in the network status. External interference factors such as network attacks and physical obstacles are detected through the intrusion detection system (IDS) to ensure that the network can cope with external threats.

[0052] After collecting multi-dimensional network status data, the controller inputs the network status data into a multi-level network status matrix. Each layer of the multi-level network status matrix represents a dimension of the network, and the data of all dimensions are dynamically updated and combined through a nonlinear mapping function. The specific formula of the nonlinear mapping function is:

[0053]

[0054] in, S i ( t ) is the i The layer network state matrix is t The value of the moment; f i It is i Dimensional mapping function; X i ( t )yes t Moment i The input data vector of dimension is , and the network state data is passed through the weight W i ( t ) is weighted, where the weight valuew ij Indicates i The dimension j The degree of influence of each input data on the network state; x ij ( t ) indicates time t No. i The dimension j Input data; n i Indicates i The number of dimensions; s is an activation function with nonlinear characteristics, ensuring that the network state matrix can flexibly reflect the complex and changing network environment; the bias term b i By fine-tuning the mapping process of the multi-level network state matrix, specific network needs or abnormal situations can be addressed. The generation of the multi-level network state matrix provides a comprehensive description of the network state, so that subsequent optimization decisions can be based on more complete and accurate network state data.

[0055] After generating a multi-level network status matrix, the weight of each dimension is adaptively adjusted according to the current network status, historical data, and predictions of future network status. To achieve this goal, the controller first models the changes in network status in the past cycle through the historical data analysis module. The historical data includes all the network status data collected above, such as delay, load, packet priority, etc. Based on the historical data, the controller uses a prediction model to predict the future trend of the network status in the next time period. The prediction models used are such as the ARIMA model, the LSTM network, or the Bayesian prediction model. The predicted future trend of the network status includes the delay, bandwidth, load, and reliability of each link in the future. The prediction models all use existing technologies and will not be elaborated on here.

[0056] Based on the prediction of the future trend of the network status, the predicted network status data is obtained, and a multi-level feedback mechanism is introduced to adjust the weights of each dimension so that the weights can adapt to possible changes in the network status in the future. The specific process is as follows:

[0057] Combined with the current multi-level network state matrix and the predicted network state data, the weights before adjustment are initialized. The goal of weight adjustment is to minimize the loss function to ensure the optimal path selection under the future network state. The loss function is used to evaluate the adaptability of the current weights to the future network state, which is defined as the deviation between the current network state and the predicted network state, and the potential risks caused by the failure of the weights of each dimension to adapt to the future network state. The specific formula is:

[0058]

[0059] in, Indicates i The loss function of dimension feedback, e It is i For each link in the dimension, S e ( t ) is the link e exist t The network status value at the moment, Ŝ e ( t + n ) is the link e At prediction time t + n The network status prediction value, n is a prediction span of time, S e ( t )- Ŝ e ( t + n )│ represents the absolute deviation between the current network state and the future predicted network state, W e ( t )yes t Time Link e The corresponding weights are assigned to each link to obtain the predicted network state change value of each link at a future time point.

[0060] Combined with the predicted network status data, the controller uses the gradient descent method to adjust the weights. The weight adjustment formula is:

[0061]

[0062]

[0063] Among them, W i ( t +1)Yes t +1 moment i The weight after dimension adjustment; W i ( t )yes t Moment i The weight of the dimension; or is the learning rate, which controls the step size of the weight update, that is, the magnitude of each adjustment; ∆W i ( t )yes t Moment The instantaneous change in the weight of the dimension reflects the direct impact of the current network state; l It is the feedback adjustment coefficient, which determines the influence of historical data and future forecast data on the current weight adjustment, and is set based on expert experience or simple heuristic methods; N Represents the number of network dimensions. The feedback of each dimension is derived from the loss function to calculate the impact on the current weight, and then corrected by the gradient descent method. Through adjustment, the controller can reasonably adjust the state of each dimension in a dynamically changing network environment, ensuring that the network can be adaptively optimized and obtain the adjusted multi-level network state matrix.

[0064] By integrating the adjusted weights of various dimensions, an adjusted weight vector is obtained.

[0065] S2: Based on the adjusted weight vector and network status matrix, construct the objective function, perform nonlinear optimization on the paths in the network, and select the optimal path in combination with nonlinear constraints; divide the network into sub-areas, perform hierarchical optimization and global coordination on each sub-area, generate the optimal configuration strategy for each sub-area, and send the optimal configuration strategy to each network switching device.

[0066] After using the multi-level feedback mechanism to adjust the weights, an adjusted multi-level network state matrix and a corresponding adjusted weight vector are obtained, wherein the weight vector represents the degree of influence of different network state parameters (such as delay, bandwidth, reliability, load, etc.) on path selection; an objective function is constructed based on the adjusted weights and the multi-level network state matrix, comprehensively considering the specific state of each link in each path, including parameters extracted from the multi-level network state matrix, such as link delay, bandwidth, node reliability, load, etc., and using the adjusted weights to balance different optimization objectives.

[0067] In order to ensure the optimality of path selection, a nonlinear path optimization model is used to solve each possible path in the network as a multi-objective optimization problem. The objective function is:

[0068]

[0069] in, O ( P ) is a set of paths The objective function to be optimized; e is a link, indicating a path; W D (t+1) is the adjusted delay weight; D e Yes Link e Delays; W B(t+1) is the adjusted bandwidth weight; B e Yes Link e bandwidth; R e Yes Link e Node reliability; W R (t+1) is the adjusted reliability weight; W L (t+1) is the adjusted load weight; L e Yes Link e of load.

[0070] The decision basis for specific path selection is as follows: The controller traverses the path set All possible paths in the network are scored according to the objective function, and the path with the smallest objective function value is finally selected as the optimal path. Since the objective function contains adjusted weights, it ensures that under different network conditions, the objective function can adaptively adjust the weights of each part to ensure that the selected path can reach the optimal value in the actual network environment.

[0071] In order to further ensure the safety and stability of path selection, nonlinear constraints are added to the controller during the solution of the objective function, and control is performed through the tolerance coefficient and connectivity factor of the link. The specific formula is:

[0072]

[0073] in, k e ( t +1)Yes t +1 Time Link e The tolerance factor of the link e The ability to withstand load increases can be obtained by the ratio of the link's historical maximum load to the current load. yes t The connectivity factor at time +1 reflects the complexity of the network topology and can be obtained by the inverse of the number of links connected to the node; i e Yes Link e The angle parameter of the link e The angle between the two end nodes on the plane is used to describe the layout complexity of the physical link; It is a global resource constraint, which can set an upper limit of resource utilization manually. The controller determines whether the path meets the tolerance requirements of the current network based on the nonlinear constraints. If a path does not meet the constraints, the controller will automatically select an alternative path or dynamically adjust the traffic on other paths to avoid network bottlenecks or security risks; global resource constraints ensure that the network will not suffer a decline in overall performance due to excessive pursuit of a single optimization goal.

[0074] After the controller selects the optimal path, it divides the network into several sub-areas for hierarchical optimization, that is, local optimization is performed in each sub-area according to the optimal path, and the path selection and traffic distribution in each sub-area are adjusted; the local optimization results of each sub-area are integrated through a global coordination mechanism, and the path selection of each sub-area is dynamically adjusted according to the results of hierarchical optimization to ensure that the traffic distribution and resource utilization between each sub-area reach the global optimum. The global optimization objective function is:

[0075]

[0076] in, is the global optimization objective function; R is the number of sub-regions; oh r ( t +1) is the Sub-areas t +1 moment weight; P r It is r Path selection results for each sub-area; ϕ ( t +1)Yes t The weight of the global adjustment factor at the +1 moment; M is the number of global adjustment factors; It is with q Network performance indicators related to global adjustment factors; P q It is with q Through this hierarchical optimization and global coordination mechanism, the global optimal configuration can be achieved in a large-scale complex network environment, especially when the network conditions change dynamically, it can ensure the efficient operation of the network.

[0077] The global optimization objective function integrates the local optimization results of each sub-area to ensure the optimal traffic distribution and path selection in the entire network, thereby achieving global performance optimization of the network. Specifically, This ensures the global consistency of the network, avoids conflicts between sub-regions, and improves the overall network performance; at the same time, the weight of the sub-region oh rand the weight of the global adjustment factor ϕ The network resources are optimally allocated among the sub-areas, ensuring the efficient use of key resources such as bandwidth, delay, and load, and through dynamic feedback adjustment, ensuring that the network configuration can adapt to dynamic environments such as load changes and link failures. Finally, the controller determines the optimal configuration strategy for each sub-area based on the global optimization objective function, and sends the optimal configuration strategy to each network switching device. Each device automatically adjusts its forwarding path, bandwidth allocation, priority setting and other parameters according to the received instructions, ensuring that the entire network can maintain a global optimization state in actual operation, thereby achieving efficient, stable and secure network management goals.

[0078] In summary, a method for configuring a network controller and a network switching device is completed.

[0079] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0081] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for configuring a network controller and a network switching device, characterized in that: The following steps are involved: S1: collect network status data in real time, generate a network status matrix based on the network status data; predict the future network status through the prediction model to obtain the predicted network status data; Based on the network state matrix and the predicted network state data, the weights of each dimension of the network are adjusted to obtain an adjusted weight vector; based on the adjusted weight vector, the network state matrix is ​​adjusted; S2: Based on the adjusted weight vector and network state matrix, construct the objective function; and construct a nonlinear path optimization model to solve the objective function, specifically: score each path according to the objective function, and select the path with the smallest objective function value as the optimal path; The objective function is: in, is a set of paths The objective function to be optimized; is a link, indicating a path; is the adjusted delay weight; Yes Link Delays; is the adjusted bandwidth weight; Yes Link bandwidth; Yes Link Node reliability; is the adjusted reliability weight; is the adjusted load weight; Yes Link Load; In the process of solving the objective function, nonlinear constraints are added to control the selection of the optimal path through the tolerance coefficient and connectivity factor of the link; After selecting the optimal path, each sub-area is hierarchically optimized and globally coordinated to adjust the path selection and traffic distribution in each sub-area. The global coordination mechanism integrates the hierarchical optimization results of each sub-area by using the global optimization objective function to obtain the optimal traffic distribution and path selection in the network. The formula of the global optimization objective function is: in, is the global optimization objective function; is the number of sub-regions; It is Sub-areas The weight of the moment; It is Path selection results for each sub-area; It is for The objective function for optimizing the path selection results of each sub-region; yes The weight of the global adjustment factor at the moment; is the number of global adjustment factors; It is with Network performance indicators related to global adjustment factors; It is with The path selection results related to the global adjustment factors; Based on the global optimization objective function, the optimal configuration strategy for each sub-area is generated and sent to each network switching device.

2. A method for configuring a network controller and a network switching device according to claim 1, characterized in that: The S1 specifically includes: The network state data is input into the network state matrix. Each layer of the network state matrix represents a dimension of the network, and the data of all dimensions are dynamically updated and combined through nonlinear mapping functions.

3. A method for configuring a network controller and a network switching device according to claim 2, characterized in that: The S1 specifically includes: A multi-level feedback mechanism is introduced to adjust the weights of each dimension of the network.

4. A method for configuring a network controller and a network switching device according to claim 3, characterized in that: The S1 specifically includes: The specific implementation process of the multi-level feedback mechanism is as follows: Combined with the current network state matrix and the predicted network state data, the weights before adjustment are initialized. The goal of weight adjustment is to minimize the loss function. Based on the minimization of the loss function and combined with the predicted network state data, the weights are adjusted by the gradient descent method to obtain the adjusted weight vector.

5. A method for configuring a network controller and a network switching device according to claim 1, characterized in that: The S2 specifically includes: Based on the adjusted weight vector and the network state matrix, an objective function is constructed, and the adjusted weight vector is used to balance the optimization target.

Images