Multi-protocol cross-platform network migration management system

Abstract

The invention relates to the technical field of migration management, and provides a multi-protocol cross-platform network migration management system, which comprises a data acquisition module, a protocol identification module, a migration optimization module, an exception monitoring module, a protocol conversion module and a load scheduling module, according to the system, deep analysis and intelligent identification of different protocol fields are realized through the protocol identification module, and the protocol compatibility problem in cross-platform network migration is effectively solved; a dual-judgment strategy optimization algorithm is introduced into the migration optimization module, an optimal migration strategy is automatically generated and optimized, and reasonable allocation of resources and high efficiency of the migration process are ensured; the system can automatically adapt in a dynamically changing network environment, provides intelligent protocol identification and migration optimization, remarkably improves the reliability and efficiency of network migration, and is particularly suitable for complex and multi-platform network migration scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of migration management, and particularly to a multi-protocol cross-platform network migration management system. Background Art

[0002] In order to improve the flexibility, scalability of services and the utilization efficiency of resources, more and more organizations are beginning to face the need for network migration; network migration is not just a simple migration of devices or systems, but involves protocol incompatibilities between different platforms and devices, reconstruction of network topologies, and adjustment of different network performance requirements; these complex migration tasks often need to be carried out under the premise of high efficiency and low latency to ensure that services are not interrupted while optimizing network performance; however, there are still many deficiencies in existing network migration management systems: firstly, traditional systems lack sufficient intelligent adaptation capabilities when facing complex network topologies and ever-changing network states; the decision-making processes of traditional systems are usually based on fixed algorithms or templates, which makes it impossible for them to dynamically perceive the actual changes in the network; secondly, the intelligent capabilities of traditional systems in protocol recognition are still limited; existing protocol recognition methods are mostly based on simple rules or shallow feature matching and cannot deeply understand and model the complex structure and hierarchical relationships of protocol fields; especially in cross-platform migration scenarios, multiple unknown or non-standard protocols may be used between different platforms, and traditional methods often cannot provide sufficient intelligent support, resulting in the inability to efficiently perform dynamic recognition and automatic conversion of protocols. Summary of the Invention

[0003] The present invention provides a multi - protocol cross - platform network migration management system, aiming to solve the problem of insufficient intelligence in protocol recognition, migration optimization, performance scheduling, exception handling, etc. in the existing network migration management system. The system combines advanced technologies of Lorentz space embedding, hypersphere distance learning, and graph neural network through a protocol recognition module to construct a hypersphere Lorentz graph neural network for in - depth analysis of various protocol fields in the network. This technology enables the system to effectively handle protocol differences between different platforms through intelligent protocol recognition in complex network migration scenarios, achieving protocol compatibility optimization. Especially when facing unknown protocols or dynamically changing network environments, the protocol recognition module can analyze and adjust in real time to ensure protocol interoperability during the migration process. At the same time, the system also introduces a dual - critique strategy optimization algorithm through a migration optimization module, which combines network performance information, protocol recognition results, and network topology data to automatically generate an optimal migration strategy. This module can not only dynamically evaluate and optimize resource allocation during the migration process but also adjust the migration strategy in real time to adapt to changes in the network environment, ensuring the efficiency and stability of the migration process. Through the above technical features, the network migration management system of the present invention can provide intelligent and automated protocol recognition and migration optimization in complex and changeable cross - platform environments, significantly improving the reliability and efficiency of network migration.

[0004] The present invention provides a multi - protocol cross - platform network migration management system, which includes a data collection module, a protocol recognition module, a migration optimization module, an exception monitoring module, a protocol conversion module, and a load scheduling module;

[0005] The data collection module obtains network data packets through network monitoring technology. The network data packets include network performance information and protocol fields. By analyzing the flow of network data packets, network topology data is obtained. The data collection module transmits the protocol fields to the protocol recognition module and transmits the network performance information and network topology data to the migration optimization module;

[0006] The protocol recognition module combines Lorentz space embedding, hypersphere distance learning, and graph neural network to construct a hypersphere Lorentz graph neural network, uses the hypersphere Lorentz graph neural network to analyze the protocol fields, and generates protocol recognition results. The protocol recognition module transmits the protocol recognition results to the migration optimization module;

[0007] The migration optimization module constructs a dual - critique strategy optimization algorithm, analyzes the network performance information, protocol recognition results, and network topology data through the dual - critique strategy optimization algorithm, generates an optimal migration strategy, and performs network migration. The migration optimization module transmits the optimal migration strategy to the load scheduling module;

[0008] Anomaly monitoring module, the anomaly monitoring module is connected to the migration optimization module to monitor the network status and performance during the migration process of the migration optimization module;

[0009] Protocol conversion module, the protocol conversion module is connected to the anomaly monitoring module. When the anomaly monitoring module detects that incompatible protocols are used by different devices and platforms, protocol conversion rules are generated through a protocol analysis and matching mechanism, and then protocol mapping is performed;

[0010] Load scheduling module, the load scheduling module is connected to the migration optimization module and the anomaly monitoring module, receives the optimal migration strategy of the migration optimization module, and combines with the anomaly monitoring module to ensure reasonable load distribution and no overloading of resources.

[0011] Furthermore, for the protocol recognition module, the process of generating the protocol recognition result specifically includes the following steps:

[0012] Step S1: Construct a protocol graph, the protocol graph includes nodes and edges; define protocol fields as nodes of the protocol graph, and represent the features of protocol fields as node features; regard the relationship between protocol fields as edges;

[0013] Step S2: Embed the node features into the Lorentz space to ensure that the node features conform to the geometric constraints of the hypersurface space; this representation can capture the hierarchical structure and relationships of the features of protocol fields, and generate the Lorentz embedding representation of each node;

[0014] Step S3: Generate a rich protocol graph node representation based on the Lorentz embedding representation of each node;

[0015] Step S4: Update the protocol graph according to the rich protocol graph node representation, apply a readout function to obtain the graph embedding vector, use a multi-layer perceptron to perform protocol classification on the graph embedding vector, and output the protocol recognition result; the protocol recognition result includes the protocol type and the protocol classification confidence.

[0016] Furthermore, step S3 specifically includes: introducing a hypersphere distance learning method, constructing a hypersphere-Lorentz centroid aggregation method, for the Lorentz embedding representation of each node, using the hypersphere-Lorentz centroid aggregation method to perform weighted aggregation on the Lorentz embedding representations of its neighbor nodes to generate a weighted aggregation result; using a Lorentz transformation layer to perform non-linear update on the weighted aggregation result to further enrich the representation of nodes in the protocol graph, ensuring that the features of each node can effectively capture the hierarchical structure and relationships of the protocol; generating a rich protocol graph node representation.

[0017] Furthermore, for the migration optimization module, the process of generating the optimal migration strategy specifically includes the following steps:

[0018] Step B1: Initialize the current policy and the critic network of the dual-critic strategy optimization algorithm; the critic network includes a traditional evaluation network and a post-decision evaluation network;

[0019] Step B2: Combine the protocol recognition result, network topology data, and network performance information to construct a state space; define an action space, where the action space includes migration order, migration batch, and migration time point;

[0020] Step B3: According to the state space and the action space, use the current policy to calculate the traditional return and the post-decision return, and collect trajectory data; the trajectory data includes state, action, and return sequences;

[0021] Step B4: Combine the trajectory data, calculate the traditional state advantage function through the traditional evaluation network, calculate the post-decision state advantage function through the post-decision evaluation network, and take the larger value of the two at each time step to obtain the maximum advantage function, which serves as the basis for the optimal decision;

[0022] Step B5: Calculate the loss function of the current policy according to the maximum advantage function, update the current policy to generate an updated policy; calculate the losses of the traditional evaluation network and the post-decision evaluation network to update the critic network;

[0023] The traditional evaluation network is used to estimate the value function of the traditional state;

[0024] The post-decision evaluation network is used to estimate the value function of the post-decision state;

[0025] The relationship between the evaluation network and the current policy is as follows: the critic network provides the evaluation signal for policy optimization, that is, the value function, and the current policy adjusts its behavior through these evaluation signals, and finally selects the action that can obtain the maximum return;

[0026] Step B6: Set the maximum number of iterations, repeat Steps B2 to B5, continuously update the critic network and the current policy, and generate the optimal migration policy.

[0027] Adopting the above solution, the beneficial effects obtained by the present invention are as follows:

[0028] The multi-protocol cross-platform network migration management system provided by the present invention realizes the in-depth analysis and recognition of different protocol fields through the protocol recognition module combined with Lorentz space embedding, hypersphere distance learning, and graph neural network technology; this technology enables the system to intelligently identify and process protocol differences between different platforms, especially during the cross-platform migration process, ensuring the compatibility optimization of the protocol; compared with traditional methods, the system can automatically adapt to protocol changes in a complex network environment, improving the stability and reliability of protocol interoperability during the migration process, thus significantly increasing the success rate of cross-platform migration;

[0029] By introducing a dual-criticism strategy optimization algorithm in the migration optimization module, the present invention can intelligently generate an optimal migration strategy based on real-time network performance information, protocol recognition results, and network topology data. This optimization algorithm can not only dynamically adjust resource allocation during the migration process but also respond in real-time to changes in the network environment to ensure the optimal execution of the migration strategy. Through this automated migration optimization, the present invention significantly improves the efficiency and stability of network migration, avoids resource waste and performance bottlenecks during the migration process, and solves the problem of insufficient adaptability of static optimization algorithms in traditional migration management systems to complex network environments.

[0030] In addition, through the close combination of protocol recognition and migration optimization, the present invention demonstrates stronger adaptability when dealing with abnormal situations that occur during the network migration process. The anomaly monitoring and intelligent adjustment mechanism can detect and respond in real-time to any sudden problems in the network, such as protocol incompatibility and substandard network performance, thereby reducing the risks during the network migration process. With the above technical features, the present invention significantly enhances the system's adaptability to complex network environments, improves the security, reliability, and efficiency of cross-platform network migration, and provides a solid technical guarantee for enterprises and organizations to conduct efficient and low-risk network migration in a multi-platform environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the modules of a multi-protocol cross-platform network migration management system provided by the present invention.

[0032] Figure 2 It is a schematic diagram of the dual-criticism strategy optimization algorithm provided in Embodiment 5. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0034] Embodiment 1, according to Figure 1 , the present invention provides a multi-protocol cross-platform network migration management system, which includes a data collection module, a protocol recognition module, a migration optimization module, an anomaly monitoring module, a protocol conversion module, and a load scheduling module.

[0035] The data acquisition module obtains network data packets through network monitoring technology. The network data packets include network performance information and protocol fields. By analyzing the flow of network data packets, network topology data is obtained. The data acquisition module transmits the protocol fields to the protocol recognition module and transmits the network performance information and network topology data to the migration optimization module;

[0036] The protocol recognition module constructs a hyperspherical Lorentz graph neural network by combining Lorentz space embedding, hyperspherical distance learning, and graph neural networks. The protocol fields are analyzed using the hyperspherical Lorentz graph neural network to generate protocol recognition results. The protocol recognition module transmits the protocol recognition results to the migration optimization module;

[0037] The migration optimization module constructs a dual-critic strategy optimization algorithm. By using the dual-critic strategy optimization algorithm, it analyzes the network performance information, protocol recognition results, and network topology data to generate an optimal migration strategy for network migration. The migration optimization module transmits the optimal migration strategy to the load scheduling module;

[0038] The anomaly monitoring module is connected to the migration optimization module to monitor the network status and performance during the migration process of the migration optimization module;

[0039] The protocol conversion module is connected to the anomaly monitoring module. When the anomaly monitoring module detects that incompatible protocols are used on different devices and platforms, it generates protocol conversion rules through a protocol analysis and matching mechanism and then performs protocol mapping;

[0040] The load scheduling module is connected to the migration optimization module and the anomaly monitoring module. It receives the optimal migration strategy from the migration optimization module and combines with the anomaly monitoring module to ensure reasonable load distribution and no overloading of resources.

[0041] Embodiment 2: This embodiment is based on Embodiment 1. In this embodiment, the process of the protocol recognition module generating protocol recognition results specifically includes the following steps:

[0042] Step S1: Construct a protocol graph. The protocol graph includes nodes and edges. The protocol fields are defined as the nodes of the protocol graph, and the feature representations of the protocol fields are regarded as node features. The relationships between the protocol fields are regarded as edges;

[0043] Step S2: Embed the node features into the Lorentz space to ensure that the node features conform to the geometric constraints of the hypersurface space. This representation can capture the hierarchical structure and relationships of the features of the protocol fields to generate the Lorentz embedding representation of each node;

[0044] Step S3: Generate rich protocol graph node representations based on the Lorentz embedding representation of each node;

[0045] Step S4: Update the protocol graph according to the enriched protocol graph node representation, apply the readout function to obtain the graph embedding vector, use a multi-layer perceptron to perform protocol classification on the graph embedding vector, and output the protocol recognition result; the protocol recognition result includes the protocol type and the protocol classification confidence level.

[0046] Embodiment 3, this embodiment is based on Embodiment 2. In this embodiment, step S3 specifically includes: introducing a hyperspherical distance learning method, constructing a hypersphere-Lorentz centroid aggregation method, and for the Lorentz embedding representation of each node, using the hypersphere-Lorentz centroid aggregation method to perform weighted aggregation on the Lorentz embedding representations of its neighbor nodes to generate a weighted aggregation result; using a Lorentz transformation layer to perform non-linear update on the weighted aggregation result to further enrich the representation of the nodes in the protocol graph, ensuring that the features of each node can effectively capture the hierarchical structure and relationships of the protocol; generating the enriched protocol graph node representation, and the formula used is as follows:

[0047] Hypersphere-Lorentz centroid aggregation method:

[0048] ;

[0049] where, and represent the node indices in the protocol graph, represents the feature vector of node in the protocol graph, represents the feature vector of neighbor node in the protocol graph, represents the hyperspherical distance between node and node in the protocol graph; represents the similarity between node and neighbor node in the protocol graph, that is, the weighting coefficient, represents weighting the hyperspherical distance through an exponential function, represents the set of neighbor nodes of node in the protocol graph, represents the normalization factor;

[0050] ;

[0051] where, represents the weighted aggregation result of node in the protocol graph, represents the number of neighbor nodes, represents the weighted neighbor information aggregation of node in the protocol graph, represents the norm of the weighted neighbor information aggregation result of point in the protocol graph;

[0052] ;

[0053] Among them, represents the Lorentz space, represents the layer index of the graph neural network, represents the th layer node's feature vector, that is, the node 's feature representation in the Lorentz space, represents the th layer's Lorentz transformation layer; represents the exponential map, represents the base point on which the map depends, represents a hyperparameter that controls the feature update amplitude of the th layer, represents a scaling factor, represents the node feature vector of the previous layer, represents the th layer node's feature vector, represents the logarithmic map that maps from the Lorentz space back to the tangent space, represents the th layer's node feature after weighted aggregation, represents applying the logarithmic map to to convert the aggregation result from the Lorentz space back to the tangent space.

[0054] Example 4: This example is based on Example 2. In this example, step S3 specifically includes: constructing a Lorentz centroid aggregation method. For the Lorentz embedding representation of each node, use the Lorentz centroid aggregation method to perform weighted aggregation on the Lorentz embedding representations of its neighbor nodes to generate a weighted aggregation result; use the Lorentz transformation layer to perform non-linear update on the weighted aggregation result to further enrich the representation of nodes in the protocol graph and ensure that the features of each node can effectively capture the hierarchical structure and relationships of the protocol; generate rich protocol graph node representations.

[0055] Example 5: According to Figure 2 , this example is based on Example 3. In this example, the migration optimization module, in the process of optimizing the algorithm to analyze network performance information, protocol recognition results, and network topology data through a dual-critic strategy to generate an optimal migration strategy, specifically includes the following steps:

[0056] Step B1: Policy and critic network initialization: Initialize the current policy and critic network of the dual-critic strategy optimization algorithm; the critic network includes a traditional critic network and a post-decision critic network;

[0057] Step B2: Construction of State and Action Spaces: Combine the protocol recognition results, network topology data, and network performance information to construct the state space; define the action space, which includes the migration order, migration batch, and migration time point.

[0058] Step B3: Trajectory Data Collection: According to the state space and action space, use the current policy to calculate the traditional return and post-decision return, and collect trajectory data; the trajectory data includes states, actions, and return sequences.

[0059] Step B4: Advantage Function Calculation: Combine the trajectory data, calculate the traditional state advantage function through the traditional evaluation network, calculate the post-decision state advantage function through the post-decision evaluation network, and take the larger value of the two at each time step to obtain the maximum advantage function as the basis for the optimal decision.

[0060] Step B5: Policy and Evaluation Network Update: Calculate the loss function of the current policy according to the maximum advantage function, update the current policy to generate an updated policy; calculate the losses of the traditional evaluation network and the post-decision evaluation network to update the critic network. The formulas used are as follows:

[0061] ;

[0062] where, represents the time step, represents the current policy, represents the state at time step , represents the maximum advantage function, parameters of the current policy, represents the loss function of the current policy, represents the importance sampling ratio, represents the hyperparameter, i.e., the clipping range, represents the clipping operation, restricting within and ; represents the loss of the post-decision evaluation network, represents weight of represents the loss of the traditional evaluation network, represents weight of

[0063] The traditional evaluation network is used to estimate the value function of the traditional state;

[0064] The post-decision evaluation network is used to estimate the value function of the post-decision state;

[0065] The relationship between the critic network and the current policy is as follows: The critic network provides an evaluation signal for policy optimization, that is, the value function. The current policy then adjusts its behavior based on these evaluation signals and finally selects the action that can obtain the maximum reward.

[0066] Step B6: Iterative optimization: Set the maximum number of iterations, repeat Step B2 to Step B5, continuously update the critic network and the current policy, and generate the optimal migration policy.

[0067] Embodiment 6. This embodiment is based on Embodiment 3. In this embodiment, the process of the migration optimization module generating the optimal migration policy specifically includes the following steps:

[0068] Step R1: Initialize the current policy and the traditional critic network of the dual critic policy optimization algorithm.

[0069] Step R2: Combine the protocol recognition result, network topology data, and network performance information to construct the state space; define the action space, and the action space includes the migration order, migration batch, and migration time point.

[0070] Step R3: According to the state space and the action space, use the current policy to calculate the traditional reward, and collect the trajectory data; the trajectory data includes the state, action, and reward sequence.

[0071] Step R4: Combine the trajectory data and calculate the traditional state advantage function through the traditional critic network as the basis for the optimal decision.

[0072] Step R5: Calculate the loss function of the current policy according to the traditional state advantage function, update the current policy, and generate the updated policy.

[0073] Step R6: Set the maximum number of iterations, repeat Step R2 to Step R5, continuously update the traditional critic network and the policy network, and generate the optimal migration policy.

[0074] The above describes the present invention and its implementation manners. This description is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto; generally speaking, if those of ordinary skill in the art are inspired by it and without departing from the gist of the present invention, they design similar structural manners and embodiments to this technical solution without creative efforts, and all should belong to the protection scope of the present invention.

Claims

1. A multi-protocol cross-platform network migration management system, comprising a data acquisition module, wherein the data acquisition module collects network performance information and protocol fields to obtain network topology data; characterized in that: The system also includes a protocol identification module and a migration optimization module; The protocol identification module constructs a hypersphere Lorentz graph neural network, uses the hypersphere Lorentz graph neural network to analyze the protocol field, and generates a protocol identification result; The migration optimization module constructs a dual-critical strategy optimization algorithm, which analyzes network performance information, protocol identification results and network topology data through the dual-critical strategy optimization algorithm to generate the optimal migration strategy.

2. The multi-protocol cross-platform network migration management system according to claim 1, characterized in that: The hypersphere Lorentz graph neural network is constructed through Lorentz space embedding, hypersphere distance learning and graph neural network.

3. The multi-protocol cross-platform network migration management system according to claim 2, characterized in that: The process of generating a protocol identification result by the protocol identification module specifically includes the following steps: Step S1: construct a protocol graph, the protocol graph includes nodes; define a protocol field as a node of the protocol graph, and represent the characteristics of the protocol field as node characteristics; Step S2: embed the node features into the Lorentz space to ensure that the node features conform to the geometric constraints of the hypersurface space; generate the Lorentz embedding representation of each node; Step S3: Generate a rich protocol graph node representation based on the Lorenz embedding representation of each node; Step S4: Update the protocol graph according to the enriched protocol graph node representation and output the protocol recognition result.

4. The multi-protocol cross-platform network migration management system according to claim 3, characterized in that: Step S3 specifically includes: introducing a hypersphere distance learning method, constructing a hypersphere-Lorentz centroid aggregation method, and for each node's Lorentz embedding representation, using the hypersphere-Lorentz centroid aggregation method to perform weighted aggregation on the Lorentz embedding representation of its neighboring nodes to generate a weighted aggregation result; using the Lorentz transform layer to perform nonlinear updates on the weighted aggregation result to generate a rich protocol graph node representation.

5. The multi-protocol cross-platform network migration management system according to claim 1, characterized in that: The process of generating the optimal migration strategy by the migration optimization module specifically includes the following steps: Step B1: Initialize the current strategy and critic network of the dual critic strategy optimization algorithm; Step B2: Combining the protocol identification results, network topology data and network performance information to construct a state space; defining an action space; Step B3: Based on the state space and action space, use the current strategy to calculate the traditional reward and post-decision reward and collect trajectory data; Step B4: Combine the trajectory data, calculate the state advantage function through the critical network, take the maximum value, and obtain the maximum advantage function; Step B5: Calculate the loss function of the current strategy according to the maximum advantage function and update the current strategy; Step B6: Set the maximum number of iterations, repeat steps B2 to B5, and generate the optimal migration strategy.

6. The multi-protocol cross-platform network migration management system according to claim 5, characterized in that: The action space includes migration order, migration batches and migration time points.

7. The multi-protocol cross-platform network migration management system according to claim 5, characterized in that: Trajectory data includes state, action, and reward sequences.

Images