Network communication collaborative management method for multiple data structures

By deploying a central coordination unit in network communication, developing a deep learning recognition system, designing a dynamic routing protocol and integrating queuing theoretical model, combining graph theory and game theory algorithms, the limitations of the existing technology in multi-data structure management are solved, and the implementation of the implementation of the limitations of the existing technology is achieved by implementing the Efficient and flexible network communication management and resource allocation.

CN119071172AInactive Publication Date: 2024-12-03SITEWEI (SUZHOU) TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411278301.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art has limitations in the accuracy of data identification, flexibility of routing protocols, application efficiency of queueing theory, and strategic resource allocation, especially in real-time network environments, it is difficult to quickly adapt to the communication management of multiple data structures.

Method used

Deploy the central coordination unit, develop a data structure identification system based on deep learning, design a dynamic data routing protocol, integrate queuing theoretical models, and use graph theory and game theory strategy optimization algorithms, combine real-time state dynamic update graph theory and related algorithms, Develop resource allocation strategies for multi-users and multi-tasks.

Benefits of technology

It improves the efficiency and control capabilities of network communication management, improves the accuracy and processing speed of data identification, optimizes routing, ensures the rational allocation and use of resources, and meets the rapid changes in the network environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure SMS_2
    Figure SMS_2
  • Figure FDA0005040886120000041
    Figure FDA0005040886120000041
  • Figure FDA0005040886120000042
    Figure FDA0005040886120000042
Patent Text Reader

Abstract

The invention discloses a network communication collaborative management method for a multi-data structure. The method comprises the following steps: step 1, deploying a central coordination unit; 2, developing a data structure recognition system based on deep learning; 3, a dynamic data routing protocol is involved and implemented; 4, integrating the queuing theoretical model; and step 5, using a strategy optimization algorithm of a graph theory and a game theory, combining a network real-time state, dynamically updating the graph theory and a related algorithm, and developing a multi-user and multi-task-oriented resource allocation strategy. The invention relates to the technical field of network communication. And high quality of training data is ensured, so that the effectiveness of model training is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of network communication technology, and in particular to a network communication collaborative management method for multiple data structures. Background Art

[0002] With the rapid development of network technology, the diversity of data structures and the dynamic nature of network environments have brought new challenges to network communication management. Existing technologies have limitations in terms of the accuracy of data identification, the flexibility of routing protocols, the efficiency of queuing theory, and the strategic nature of resource allocation. Especially in real-time network environments, a communication method that can quickly adapt to and effectively manage multiple data structures is needed. Summary of the invention

[0003] In view of the deficiencies in the prior art, the technical solution adopted by the present invention to solve the technical problems is: the present invention aims to provide a network communication collaborative management method for multiple data structures to effectively solve the problems existing in the prior art.

[0004] Overall process

[0005] The present invention comprises the following steps: deploying a central coordination unit; developing a data structure recognition system based on deep learning; designing and implementing a dynamic data routing protocol; integrating a queuing theory model; using a strategy optimization algorithm based on graph theory and game theory, combining the real-time status of the network, dynamically updating graph theory and related algorithms, and developing a resource allocation strategy for multiple users and multiple tasks.

[0006] Data structure recognition system development

[0007] In the data preprocessing phase, we carefully designed the data collection and cleaning mechanism, generated high-quality training data sets, and annotated and distinguished different data structure types, clarified their sources and types, and generated corresponding sequence data in time and text order. In model design and training, we selected deep learning frameworks and algorithms, used recursive neural networks to process sequence data, trained the model on the device, adjusted the parameters to optimize recognition accuracy and speed, and finally deployed the trained model to the central coordination unit and integrated it into the data processing process to achieve real-time data classification and processing.

[0008] Design of Dynamic Data Routing Protocol

[0009] Design a dynamic routing algorithm that adapts to different network states and data priorities, define packet identification and routing selection rules, implement routing selection logic in the central coordination unit, use software-defined network technology to dynamically adjust routing, and test its performance under different network loads and conditions.

[0010] Integration of queuing theory models

[0011] Select appropriate queuing models according to network service characteristics, such as the M / M / 1 queue queuing model, analyze data flows and processing nodes to determine key deployment points, apply queuing models to optimize the work queue settings of data processing nodes, and test and adjust models through actual data to maximize processing efficiency.

[0012] Deep learning framework and algorithm steps

[0013] Select and set up a suitable deep learning framework, such as the PyTorch framework with wide support and features. Perform data preprocessing, including collecting large amounts of sequence data, cleaning invalid and erroneous data points, classifying and labeling data, performing word embedding on text data, and reasonably splitting the data into training, validation, and test sets. Design a suitable model structure, such as selecting convolutional neural networks or recurrent neural networks for different data types, defining the number of network layers, nodes, activation functions, etc., and using dropout regularization to prevent overfitting. Perform model training and optimization, set loss functions and optimizers, and iteratively adjust parameters through multiple rounds of training and validation until the expected performance is achieved. Test the model to ensure its generalization ability and deploy it to a production environment for real-time data classification and prediction.

[0014] Strategy optimization algorithms based on graph theory and game theory

[0015] In the graph theory optimization algorithm, the Dijkstra algorithm is used to calculate the shortest path between two nodes in the network. The game theory strategy optimization algorithm uses Nash equilibrium analysis to infer the strategies of each participant, and finds the best strategy combination by defining the game model, evaluating the strategy, and finding the Nash equilibrium.

[0016] The beneficial effects of the present invention are as follows:

[0017] Deploy a central coordination unit to achieve more efficient network communication management and control, and improve the coordination efficiency of the entire network. Develop a data structure recognition system based on deep learning, use advanced neural network models to improve recognition accuracy and processing speed, and adapt to a variety of data structures. Implement a dynamic data routing protocol to optimize routing selection based on real-time network status and data priority, and improve the efficiency and reliability of data transmission.

[0018] Integrate queuing theory models to optimize the work queue settings of data processing nodes and improve data processing efficiency in network communications. Use graph theory and game theory strategy optimization algorithms to develop fair and efficient multi-user and multi-task resource allocation strategies to ensure the rational allocation and use of network resources. Dynamically update graph theory models and algorithms based on real-time network status to ensure the timeliness and adaptability of resource allocation and routing selection to meet the rapidly changing needs of the network environment. Choose a deep learning framework that supports complex network structures, such as PyTorch, and use its community support and flexibility to improve model development and performance.

[0019] Through comprehensive data preprocessing steps, including collection, cleaning, labeling, and feature engineering, we ensure the high quality of training data, thereby improving the effectiveness of model training. A meticulous model training and optimization process, including the selection of loss functions and optimizers, improves the generalization ability and prediction accuracy of the model. A strict model testing and deployment process ensures the stability and reliability of the model in the actual production environment, and achieves efficient and accurate network communication management. DETAILED DESCRIPTION

[0020] The present invention is further described in detail below in conjunction with specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0021] Example:

[0022] The present invention provides a technical solution: a network communication collaborative management method for multiple data structures, comprising the following steps:

[0023] Step 1: Deploy the central coordination unit;

[0024] Step 2: Develop a data structure recognition system based on deep learning;

[0025] Step 3: Involve and implement dynamic data routing protocols;

[0026] Step 4: Integrate the queuing theory model;

[0027] Step 5: Use graph theory and game theory strategy optimization algorithms, combine with real-time network status, dynamically update graph theory and related algorithms, and develop resource allocation strategies for multiple users and multiple tasks.

[0028] In step 2, the development steps of the data structure identification system are as follows:

[0029] Data preprocessing: Design data collection and cleaning mechanisms to generate high-quality training data sets, annotate data sets, distinguish different data structure types, clarify the source and type of data, and generate corresponding sequence data in time and text order;

[0030] Model design and training: Select a deep learning framework and algorithm, use a recurrent neural network to process sequence data, and then train the model on the device to optimize recognition accuracy and speed by continuously adjusting parameters;

[0031] Deployment and integration: Deploy the trained model to the central coordination unit and integrate the model into the data processing flow for real-time data classification and processing.

[0032] In step 3, the design of the dynamic data routing protocol is as follows:

[0033] Protocol design: Design dynamic routing algorithms that adapt to different network states and data priorities, and define data packet identification and routing selection rules;

[0034] Protocol Implementation: Implement routing selection logic in the central coordination unit, use software-defined networking technology to dynamically adjust routing, and test the performance of the protocol under different network loads and conditions.

[0035] In step 4, the queuing theory model is integrated as follows:

[0036] Theoretical selection and analysis: Based on the characteristics of network services, the M / M / 1 queue model is selected, and then the data flow and processing nodes are analyzed to determine the key deployment points;

[0037] Model application and optimization: Apply the queuing model to optimize the work queue settings of data processing nodes, and maximize processing efficiency by testing and adjusting the model through actual data.

[0038] The specific steps of deep learning framework and algorithm in model design and training are as follows:

[0039] Step 1: Select and set up a deep learning framework;

[0040] Framework selection: Choose the appropriate deep learning framework PyTorch as the framework foundation, taking advantage of its support for complex network structures and its extensive community support and documentation.

[0041] Step 2: Data preprocessing;

[0042] Data collection: Collect a large amount of sequence data, including time series data and text data;

[0043] Data cleaning: clean the data and remove invalid and erroneous data points;

[0044] Data labeling: classify and label the data to ensure that each data point has an accurate label;

[0045] Feature engineering: embedding text data into words, performing feature extraction and feature conversion;

[0046] Data splitting: Divide the data into training set, validation set and test set in a ratio of 60% / 20% / 20%;

[0047] Step 3: Model design;

[0048] Select the network structure: for image data, choose convolutional neural network, for text and time series data, choose recurrent neural network;

[0049] Define the model architecture: design the number of network layers, the number of nodes in each layer, the type of activation function, and use the dropout regularization method to prevent overfitting;

[0050] Step 4: Model training and optimization;

[0051] Compile the model: set the loss function and optimizer. The loss function includes cross entropy loss and the optimizer includes Adam.

[0052] Training model: Use the training set data to train the model through multiple epochs. Each epoch includes forward propagation and backpropagation steps. Use the validation set to evaluate the performance of the model, adjust the parameters, and obtain the validation results.

[0053] Model optimization: According to the verification results, adjust the network parameters and structure, and then control the number of network layers, change the activation function, and then perform multiple rounds of training and verification iterations until the model performance reaches the expected effect;

[0054] Step 5: Testing and Deployment

[0055] Model testing: Use an independent test set to evaluate the performance of the final model to ensure that the model has generalization capabilities;

[0056] Model deployment: Deploy the trained model in the production environment to perform real-time data classification and prediction.

[0057] The strategy optimization algorithm of graph theory and game theory is as follows:

[0058] Graph optimization algorithm: In network communication, the Dijkstra algorithm is used to obtain the shortest path algorithm, which is used to find the shortest path between two network nodes, where:

[0059] Di jkstra algorithm is used to calculate the single-source shortest path problem, that is, the shortest path from the source node to all other nodes. The formula is as follows:

[0060] Let G = (V, E) be a weighted directed graph;

[0061] Where V is the set of nodes, E is the set of edges, and each edge e = (u, v) has a weight w(u, v);

[0062] The specific algorithm steps are as follows:

[0063] Initialization: Set the shortest path estimation values ​​of all nodes to infinity and the source node to 0, which is used as the initial node;

[0064] Loop through all nodes: select a node that is not permanently marked and has the smallest path estimate, mark this node as permanent, indicating that the shortest path from the source node to this node has been calculated; update the path estimate of all connected unmarked nodes: for each connected node, if the path length from the current node to the node is less than the known path estimate, update the path estimate of the node, otherwise do not update;

[0065] Repeat the above steps until all nodes are marked as permanent and the algorithm ends;

[0066] Game Theory Strategy Optimization Algorithm:

[0067] Nash equilibrium is used to analyze and infer the strategies of each participant. Nash equilibrium describes a state in which no party can obtain a better result by changing its strategy alone in a multi-participant decision-making environment. The specific definition is as follows:

[0068] Γ is a finite game with n players i, each of which has a finite strategy set S i , the payoff matrix of the game is P, where P i (j) represents the payoff when participant i chooses strategy j;

[0069] Algorithm steps:

[0070] Define the game model: identify all participants and their feasible strategy sets, that is, determine all participants N = {1, 2, ..., n};

[0071] Strategy evaluation: For each possible strategy combination of each party, calculate its corresponding benefit P i (s i );

[0072] Find Nash equilibrium: For each strategy combination, check whether there is a party that can obtain higher returns by changing its strategy, and seek the best response BR i ,Right now If for a certain strategy combination, no party can change its strategy alone to improve its own benefits, then the strategy combination constitutes a Nash equilibrium, and the calculation formula is as follows:

[0073]

[0074] Repeat the above steps until all possible Nash equilibrium points are found.

[0075] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A network communication collaborative management method for multiple data structures, Features: The following steps are involved: Step 1: Deploy the central coordination unit; Step 2: Develop a data structure recognition system based on deep learning; Step 3: Involve and implement dynamic data routing protocols; Step 4: Integrate the queuing theory model; Step 5: Use graph theory and game theory strategy optimization algorithms, combine with real-time network status, dynamically update graph theory and related algorithms, and develop resource allocation strategies for multiple users and multiple tasks.

2. A method for collaborative management of network communications for multiple data structures according to claim 1, Features: In step 2, the development steps of the data structure identification system are as follows: Data preprocessing: Design data collection and cleaning mechanisms to generate high-quality training data sets, annotate data sets, distinguish different data structure types, clarify the source and type of data, and generate corresponding sequence data in time and text order; Model design and training: Select a deep learning framework and algorithm, use a recurrent neural network to process sequence data, and then train the model on the device to optimize recognition accuracy and speed by continuously adjusting parameters; Deployment and integration: Deploy the trained model to the central coordination unit and integrate the model into the data processing flow for real-time data classification and processing.

3. A method for collaborative management of network communications for multiple data structures according to claim 1, Features: In step 3, the design of the dynamic data routing protocol is as follows: Protocol design: Design dynamic routing algorithms that adapt to different network states and data priorities, and define data packet identification and routing selection rules; Protocol Implementation: Implement routing selection logic in the central coordination unit, use software-defined networking technology to dynamically adjust routing, and test the performance of the protocol under different network loads and conditions.

4. A method for collaborative management of network communications for multiple data structures according to claim 1, Features: In step 4, the queuing theory model is integrated as follows: Theoretical selection and analysis: Based on the characteristics of network services, the M / M / 1 queue model is selected, and then the data flow and processing nodes are analyzed to determine the key deployment points; Model application and optimization: Apply the queuing model to optimize the work queue settings of data processing nodes, and maximize processing efficiency by testing and adjusting the model through actual data.

5. A method for collaborative management of network communications for multiple data structures according to claim 2, Features: The specific steps of the deep learning framework and algorithm in the model design and training are as follows: Step 1: Select and set up a deep learning framework; Framework selection: Choose the appropriate deep learning framework PyTorch; Step 2: Data preprocessing; Data collection: Collect sequence data, including time series data and text data; Data cleaning: clean the data and remove invalid and erroneous data points; Data labeling: classify and label the data to ensure that each data point has an accurate label; Feature engineering: embedding text data into words, performing feature extraction and feature conversion; Data splitting: Divide the data into training set, validation set and test set in a ratio of 60% / 20% / 20%; Step 3: Model design; Select the network structure: for image data, choose convolutional neural network, for text and time series data, choose recurrent neural network; Define the model architecture: design the number of network layers, the number of nodes in each layer, the type of activation function, and use the dropout regularization method to prevent overfitting; Step 4: Model training and optimization; Compile the model: set the loss function and optimizer. The loss function includes cross entropy loss and the optimizer includes Adam. Training model: Use the training set data to train the model through multiple epochs. Each epoch includes forward propagation and backpropagation steps. Use the validation set to evaluate the performance of the model, adjust the parameters, and obtain the validation results. Model optimization: According to the verification results, adjust the network parameters and structure, and then control the number of network layers, change the activation function, and then perform multiple rounds of training and verification iterations until the model performance reaches the expected effect; Step 5: Testing and Deployment Model testing: Use an independent test set to evaluate the performance of the final model to ensure that the model has generalization capabilities; Model deployment: Deploy the trained model in the production environment to perform real-time data classification and prediction.

6. A method for collaborative management of network communications for multiple data structures according to claim 1, Features: The strategy optimization algorithm of graph theory and game theory is as follows: Graph optimization algorithm: In network communication, the Dijkstra algorithm is used to obtain the shortest path algorithm, which is used to find the shortest path between two network nodes, where: The Dijkstra algorithm is used to calculate the single-source shortest path problem, that is, the shortest path from the source node to all other nodes. The formula is as follows: Let G = (V, E) be a weighted directed graph; Where V is the node set, E is the edge set, and each edge e = (u, υ) has a weight w(u, υ); The specific algorithm steps are as follows: Initialization: Set the shortest path estimation values ​​of all nodes to infinity and the source node to 0, which is used as the initial node; Loop through all nodes: select a node that is not permanently marked and has the smallest path estimate, and mark this node as permanent, indicating that the shortest path from the source node to this node has been calculated; update the path estimate of all connected unmarked nodes: for each connected node, if the path length from the current node to the node is less than the known path estimate, update the path estimate of the node, otherwise do not update; Repeat the above steps until all nodes are marked as permanent and the algorithm ends; Game Theory Strategy Optimization Algorithm: Nash equilibrium is used to analyze and infer the strategies of each participant. Nash equilibrium describes a state in which no party can obtain a better result by changing its strategy alone in a multi-participant decision-making environment. The specific definition is as follows: Γ is a finite game with n players i, each of which has a finite strategy set S i , the payoff matrix of the game is P, where P i (j) represents the payoff when participant i chooses strategy j; Algorithm steps: Define the game model: identify all participants and their feasible strategy sets, that is, determine all participants N = {1, 2, ..., n}; Strategy evaluation: For each possible strategy combination of each party, calculate its corresponding benefit P i (s i ); Find Nash equilibrium: For each strategy combination, check whether there is a party that can obtain higher returns by changing its own strategy, and seek the best response BR i ,Right now If for a certain strategy combination, no party can change its strategy alone to improve its own benefits, then the strategy combination constitutes a Nash equilibrium, and the calculation formula is as follows: Repeat the above steps until all possible Nash equilibrium points are found.