Dynamic communication adaptation system of virtual network terminal and implementation method

Abstract

The invention relates to a dynamic communication adaptation system and implementation method of a virtual network terminal, and relates to the technical field of communication adaptation, and the method comprises the steps: analyzing the demands of a user group for network performance, and constructing a user group network demand matrix; based on data analysis of a network measurement platform, monitoring a network condition and an equipment operation state in real time, obtaining a network resource distribution condition of user group communication equipment software, and generating a user group network behavior portrait; dynamically adjusting and optimizing a network resource allocation strategy of the virtual network terminal in combination with the user group network demand matrix and the user group network behavior portrait; and based on the network resource allocation strategy of the optimized virtual network terminal, constructing a dynamic communication adaptation model of the virtual network terminal, and realizing communication quality optimization and adaptation of the virtual network in various network environments. The method has the beneficial effects that the self-adaptive capability and reliability of the system under a high-load condition are improved, and the user experience and the communication efficiency are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of communication adaptation, and in particular to a dynamic communication adaptation system of a virtual network terminal and an implementation method thereof. Background Art

[0002] Existing virtual network terminal communication adaptation systems usually rely on fixed protocol stacks and static configurations, and lack the ability to adapt to changes in dynamic network environments. In existing technologies, many systems are unable to flexibly adjust communication strategies when faced with different network conditions, resulting in insufficient utilization of network resources or unstable communication quality. Especially in high-load and complex network environments, the adaptation efficiency of existing systems is low, and they cannot effectively guarantee the reliability and stability of communications. They lack targeted optimization strategies and are difficult to meet rapidly changing network needs. Summary of the Invention

[0003] The present invention aims at solving the technical problems in the prior art and provides a dynamic communication adaptation solution for a virtual network terminal.

[0004] The technical solution of the present invention to solve the above technical problems is as follows: A method for implementing dynamic communication adaptation of a virtual network terminal, comprising: Based on the historical network usage data of user groups, analyze the network performance requirements of user groups and build a network demand matrix for user groups; Based on data analysis from the network measurement platform, real-time monitoring of network conditions and equipment operation status is carried out to obtain the network resource allocation status of user group communication equipment and software, and generate network behavior profiles of user groups; Using reinforcement learning algorithms, combined with the user group network demand matrix and user group network behavior profile, the network resource allocation strategy of virtual network terminals is dynamically adjusted and optimized; Based on the optimized network resource allocation strategy of virtual network terminals, a dynamic communication adaptation model of virtual network terminals is constructed to achieve communication quality optimization and adaptation of virtual networks in various network environments.

[0005] Preferably, based on the historical network usage data of the user group, data preprocessing is performed on it, outliers are processed using a linear interpolation method, and the historical network usage time series data of the user group is constructed; Based on the historical network usage time series data of user groups, we divide the user group demand indicators, quantify the user group demand, use autoencoder technology to learn the user group demand characteristics, and construct the user group network demand matrix; As further content, the user group demand indicators include: delay tolerance, bandwidth demand and service priority.

[0006] Preferably, the network measurement platform acquires network status data in real time, monitors end-to-end delay, packet loss rate and bandwidth utilization in real time, generates a timestamp sequence in combination with a 5-second sampling period; Based on the device running state data in the network measurement platform, CPU occupancy and memory occupancy are collected in real time to generate a device state vector; Based on the network performance timestamp sequence, the average value of delay, the fluctuation of packet loss rate, the peak bandwidth and the average value of device load are calculated to generate a single user behavior vector.

[0007] Preferably, the K-Means clustering algorithm is used to determine the number K of single user behavior vector clusters, K single user behavior vectors are randomly selected as initial cluster centers, the distance between each point and the cluster center is calculated according to the Euclidean formula, each single user behavior vector is assigned to the nearest cluster center to form K subsets, the mean value of all single user behavior vectors assigned to the cluster is calculated, the cluster center is updated, and the iteration is performed until the cluster center is stable, and the clustering result is output. The time window technology is used to group the similarity of single user behavior vectors in different time windows, and the user group historical network demand matrix is combined to generate a user group network behavior portrait.

[0008] Preferably, based on the user group network demand matrix and the user group network behavior portrait as the state space, the adjustment of bandwidth and CPU occupancy as the action space, the reward function of the reinforcement learning model is defined, the user satisfaction, resource cost and service standard reward are combined to construct a Markov decision process model.

[0009] Preferably, the policy gradient algorithm is used, the state space is taken as the network input, and the action distribution of bandwidth and CPU adjustment is taken as the output. The parameters of the policy network and the value network are randomly initialized, the value network is used to estimate the current state value, the advantage function is calculated, the target function is updated to obtain the policy network, the policy network parameters are optimized by the gradient descent method, the action mask mechanism is used to constrain resource allocation, the reinforcement learning model is iteratively trained until convergence, and the optimal network resource allocation strategy is obtained. Dynamically adjust the network resource configuration of the virtual network terminal to realize efficient utilization of network resources.

[0010] Preferably, based on the network resource allocation strategy and real-time network status of the optimized virtual network terminal, network status data is collected, data preprocessing is performed on it, and a neural network model is trained. The network status indicator is used as the input layer, and multiple hidden layers are designed to form a deep neural network structure. The network resource allocation strategy category is used as the output layer, and the Softmax activation function is used to generate the probability distribution of each network resource allocation strategy. The cross entropy loss function is used to calculate the error between the predicted result and the actual label. The weight parameters of the neural network model are optimized by the back propagation algorithm until the loss function converges, and the dynamic communication adaptation model of the virtual network terminal is obtained.

[0011] Preferably, a dynamic communication adaptation model based on the virtual network terminal adopts a dynamic update mechanism to adjust the parameters of the neural network model to adapt to the network environment, evaluate the network status of the virtual network terminal in real time, and dynamically select the optimal network resource allocation strategy according to the current network status to achieve communication quality optimization and adaptation of the virtual network in various network environments.

[0012] Furthermore, a dynamic communication adaptation system for a virtual network terminal is provided, for implementing the above-mentioned dynamic communication adaptation method for a virtual network terminal, comprising: Network demand matrix module, network behavior profiling module, network resource allocation strategy module, and communication quality optimization and adaptation module; The network demand matrix module is used to analyze the user group's demand for network performance based on the user group's historical network usage data and construct the user group's network demand matrix; The network behavior profiling module is used to analyze network measurement platform data, monitor network conditions and device operation status in real time, obtain network resource allocation status of user group communication device software, and generate user group network behavior profiles; The network resource allocation strategy module is electrically connected to the network demand matrix module and the network behavior profile module. The network resource allocation strategy module is used to dynamically adjust and optimize the network resource allocation strategy of the virtual network terminal by using a reinforcement learning algorithm, combining the user group network demand matrix and the user group network behavior profile; The communication quality optimization and adaptation module is electrically connected to the network resource allocation strategy module. The communication quality optimization and adaptation module is used to build a dynamic communication adaptation model of the virtual network terminal based on the network resource allocation strategy of the optimized virtual network terminal, so as to realize the communication quality optimization and adaptation of the virtual network in various network environments.

[0013] The beneficial effects of the present application are: through the analysis based on user group historical network data and real-time network conditions, the resource allocation strategy of the virtual network terminal is dynamically optimized by using the reinforcement learning algorithm, which can adapt to the changing network environment in real time, improve the communication quality and stability; by constructing the dynamic communication adaptation model of the virtual network terminal, the method effectively optimizes the utilization of network resources, ensures efficient communication in complex network environment, improves the adaptive ability and reliability of the system under high load, significantly improves the user experience and communication efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The method flowchart of the present application is shown in the figure. Figure 2 The system framework diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0016] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0017] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope of principles and features disclosed in the present application.

[0018] Embodiment 1, a dynamic communication adaptation implementation method of a virtual network terminal, comprising: S1, based on user group historical network usage data, analyze the user group's demand for network performance, and construct a user group network demand matrix; The step S1 includes the following contents: Based on the user group historical network usage data, data preprocessing is carried out, abnormal values are processed by linear interpolation method, and user group historical network usage time series data is constructed; Based on the user group historical network usage time series data, the user group demand index is divided, the user group demand is quantified, the user group demand feature is learned by using the self-encoder technology, and the user group network demand matrix is constructed; As further content, the user group demand index includes: delay tolerance, bandwidth demand and service priority.

[0019] In use, in combination with the contents of the above step S1: The prior art has problems of low data processing accuracy, single demand modeling and lack of flexibility when constructing a user group network demand matrix; this step accurately processes abnormal values by linear interpolation method, introduces multi-dimensional demand indexes such as delay tolerance, bandwidth demand and service priority, and uses self-encoder technology to learn user demand features; this can more comprehensively and accurately model user demand, overcome the limitations of traditional methods, and improve the intelligence and individualization level of network services.

[0020] S2, based on network measurement platform data analysis, real-time monitoring of network status and device running state, obtaining network resource allocation status of user group communication equipment software, generating user group network behavior portrait; The step S2 includes the following contents: Based on the network measurement platform, real-time network status data is obtained, end-to-end delay, packet loss rate and bandwidth utilization rate indexes are monitored in real time, and a timestamp sequence is generated in combination with a 5-second sampling period; Based on the device running state data in the network measurement platform, CPU occupancy rate and memory occupancy rate are collected in real time, and a device state vector is generated; Based on the network performance timestamp sequence, the average value of delay, the fluctuation of packet loss rate, the peak bandwidth and the average value of device load characteristics are calculated, and a single user behavior vector is generated; Using K-Means clustering algorithm, determine the number K of single user behavior vector clusters, randomly select K single user behavior vectors as initial cluster centers, calculate the distance between each point and the cluster center according to Euclidean formula, assign each single user behavior vector to the nearest cluster center, form K subsets, calculate the mean value of all single user behavior vectors assigned to the cluster, update the cluster center, and iterate until the cluster center is stable, output the clustering result; The time window technology is used to group the similarity of the single user behavior vector in different time windows, and the user group network demand matrix is combined to generate the user group network behavior portrait.

[0021] In use, the above step is combined with the content: The existing network monitoring technology has the problems of poor real-time performance, single-dimensional analysis and static clustering, which cannot accurately capture user behavior and device changes; this step dynamically generates a more accurate user group network behavior portrait by real-time monitoring of network conditions and device status, combined with the K-Means clustering algorithm and time window technology; this method not only optimizes network resource allocation and improves accuracy, but also better adapts to changes in user demand, ultimately improving network performance and user experience.

[0022] S3, using a reinforcement learning algorithm, combining the user group network demand matrix and the user group network behavior portrait, dynamically adjusting and optimizing the network resource allocation strategy of the virtual network terminal; The step S3 includes the following content: Based on the user group network demand matrix and the user group network behavior portrait as the state space, the adjustment of bandwidth and CPU occupancy rate as the action space, the reward function of the reinforcement learning model is defined, the user satisfaction, resource cost and service standard reward are combined to build a Markov decision process model; Using the policy gradient algorithm, taking the state space as the network input and the action distribution of bandwidth and CPU adjustment as the output, randomly initializing the parameters of the policy network and the value network, using the value network to estimate the current state value, calculating the advantage function, obtaining the target function to update the policy network, optimizing the policy network parameters through the gradient descent method, using the action mask mechanism to constrain resource allocation, continuously iterating the reinforcement learning model until convergence, obtaining the optimal network resource allocation strategy, dynamically adjusting the network resource configuration of the virtual network terminal, and realizing the efficient use of network resources.

[0023] In use, the above step S3 is combined with the content: The existing domestic and foreign technologies have the following problems: on the one hand, the user demand matrix and the behavior portrait are difficult to update in real time, which makes the state space unable to accurately reflect the user demand; on the other hand, the reinforcement learning has large computational resource consumption, slow convergence speed and low optimization efficiency in large-scale complex environment; the reward function design is single and cannot balance user experience, resource cost and service quality.

[0024] This step overcomes these shortcomings by combining user demand matrix and behavior portrait, using strategy gradient algorithm and value network optimization, and through advantage function and action mask mechanism, can efficiently and dynamically adjust network resources, improve resource utilization, optimize user satisfaction and cost, and significantly improve the flexibility and efficiency of network resource allocation.

[0025] S4, based on the optimized network resource allocation strategy of the virtual network terminal, a dynamic communication adaptation model of the virtual network terminal is constructed to realize communication quality optimization and adaptation of the virtual network in various network environments. The step S4 includes the following contents: Based on the optimized network resource allocation strategy of the virtual network terminal and the real-time network status, network state data is collected, preprocessed, and a neural network model is trained, taking network status indicators as the input layer, designing multiple hidden layers to form a deep neural network structure, taking network resource allocation strategy categories as the output layer, using Softmax activation function to generate the probability distribution of each network resource allocation strategy, using cross-entropy loss function to calculate the error between the predicted result and the actual label, and optimizing the weight parameters of the neural network model through back propagation algorithm until the loss function converges, obtaining the dynamic communication adaptation model of the virtual network terminal. Based on the dynamic communication adaptation model of the virtual network terminal, a dynamic updating mechanism is adopted to adjust the parameters of the neural network model to adapt to the network environment, real-time evaluate the network status of the virtual network terminal, dynamically select the optimal network resource allocation strategy according to the current network status, and realize the communication quality optimization and adaptation of the virtual network in various network environments.

[0026] In use, the contents of the above step S4 are combined: Existing virtual network resource allocation technologies mostly rely on static or rule-based methods, which are difficult to cope with complex and dynamic network environments. Although some methods use machine learning, they have poor generalization ability and lack real-time updating mechanism, resulting in unstable communication quality and resource waste. This step builds a dynamic communication adaptation model based on deep neural network, collects network state data in real time, and dynamically selects the optimal resource allocation strategy, enhancing the adaptability and optimization ability of the virtual network, avoiding resource waste and performance bottleneck, and improving communication quality and efficiency.

[0027] Embodiment 2, a dynamic communication adaptation system of a virtual network terminal, comprising: A network demand matrix module, a network behavior portrait module, a network resource allocation strategy module, and a communication quality optimization and adaptation module. The network demand matrix module is used to analyze the demand of the user group for network performance based on historical network usage data of the user group, and to construct a network demand matrix of the user group. The network behavior portrait module is used for monitoring network conditions and device running states in real time, obtaining network resource allocation conditions of the communication devices of the user group, and generating the network behavior portrait of the user group based on network measurement platform data analysis; The network resource allocation strategy module is electrically connected with the network demand matrix module and the network behavior portrait module, and is used for dynamically adjusting and optimizing the network resource allocation strategy of the virtual network terminal by using a reinforcement learning algorithm in combination with the network demand matrix of the user group and the network behavior portrait of the user group. The communication quality optimization and adaptation module is electrically connected with the network resource allocation strategy module, and is used for constructing a dynamic communication adaptation model of the virtual network terminal based on the optimized network resource allocation strategy of the virtual network terminal, and realizing communication quality optimization and adaptation of the virtual network in various network environments.

[0028] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0029] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0030] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices produce a device that implements the flowcharts and / or block diagrams. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks. Figure 1 The device that implements the function specified in one flow or multiple flows and / or blocks.

[0031] These computer program instructions can also be stored in a computer-readable storage medium that can guide the computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable storage medium produce a manufactured product including instruction devices that implement the flowcharts and / or block diagrams. Figure 1one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0032] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide a process for implementing the flow Figure 1 one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0033] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the embodiments by those skilled in the art once they learn of the basic inventive concepts. Therefore, the appended claims are intended to encompass within their scope all possible variations and modifications of the preferred embodiments. 1

[0034] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method for implementing dynamic communication adaptation of a virtual network terminal, characterized in that: include: S1. Analyze the user group's network performance requirements based on the user group's historical network usage data and construct a user group network demand matrix; S2. Based on data analysis from the network measurement platform, real-time monitoring of network conditions and device operation status is performed to obtain the network resource allocation status of user group communication equipment and software, and generate a user group network behavior profile; S3. Utilize reinforcement learning algorithms, combined with the user group network demand matrix and user group network behavior profiles, to dynamically adjust and optimize the network resource allocation strategy for virtual network terminals. S4. Based on the network resource allocation strategy of the optimized virtual network terminal, a dynamic communication adaptation model of the virtual network terminal is constructed to achieve communication quality optimization and adaptation of the virtual network in various network environments.

2. The method for implementing dynamic communication adaptation of a virtual network terminal according to claim 1, characterized in that: Said S1 comprises: Based on the historical network usage data of user groups, data preprocessing is performed, outliers are processed using linear interpolation method, and the time series data of historical network usage of user groups is constructed; Based on the historical network usage time series data of user groups, we divide the user group demand indicators, quantify the user group demand, use autoencoder technology to learn the user group demand characteristics, and construct the user group network demand matrix; As further content, the user group demand indicators include: delay tolerance, bandwidth demand and service priority.

3. The method for implementing dynamic communication adaptation of a virtual network terminal according to claim 1, characterized in that: The S2 includes: Based on the network measurement platform, network status data is obtained in real time, end-to-end delay, packet loss rate and bandwidth utilization indicators are monitored in real time, and a timestamp sequence is generated based on a 5-second sampling period. Based on the device operation status data in the network measurement platform, the CPU usage and memory usage are collected in real time to generate the device status vector; Based on the network performance timestamp sequence, the average value of delay, fluctuation of packet loss rate, peak bandwidth and average device load characteristics are calculated to generate a single user behavior vector.

4. The method for implementing dynamic communication adaptation of a virtual network terminal according to claim 3, wherein: Said S2 further comprises: Using the K-Means clustering algorithm, the number of clusters K of single-user behavior vectors is determined. K single-user behavior vectors are randomly selected as initial cluster centers. The distance between each point and the cluster center is calculated according to the Euclidean formula. Each single-user behavior vector is assigned to the nearest cluster center to form K subsets. The mean of all single-user behavior vectors assigned to the cluster is calculated, and the cluster center is updated. The algorithm is iterated until the cluster center is stable, and the clustering result is output. The time window technology is used to group the similarities of single user behavior vectors in different time windows, and the network behavior portrait of the user group is generated by combining the historical network demand matrix of the user group.

5. The method for implementing dynamic communication adaptation of a virtual network terminal according to claim 4, characterized in that: The S3 includes: Based on the user group network demand matrix and user group network behavior portrait as the state space, the adjustment of bandwidth and CPU occupancy rate is used as the action space, and the reward function of the reinforcement learning model is defined. Combined with user satisfaction, resource cost and service compliance reward, a Markov decision process model is constructed.

6. A method for implementing dynamic communication adaptation of a virtual network terminal according to claim 5, characterized in that: Said S3 further comprises: Using the policy gradient algorithm, the state space is used as the network input, and the action distribution adjusted by bandwidth and CPU is used as the output. The parameters of the policy network and value network are randomly initialized, the value network is used to estimate the current state value, the advantage function is calculated, and the objective function is obtained to update the policy network. The policy network parameters are optimized through the gradient descent method, and the action mask mechanism is used to constrain resource allocation. The reinforcement learning model is continuously iterated and trained until convergence. The optimal network resource allocation strategy is obtained, and the network resource configuration of the virtual network terminal is dynamically adjusted to achieve efficient utilization of network resources.

7. A method for implementing dynamic communication adaptation of a virtual network terminal according to claim 6, characterized in that: The S4 includes: Based on the network resource allocation strategy and real-time network status of the optimized virtual network terminal, network status data is collected, preprocessed, and a neural network model is trained. The network status indicator is used as the input layer, and multiple hidden layers are designed to form a deep neural network structure. The network resource allocation strategy category is used as the output layer. The Softmax activation function is used to generate the probability distribution of each network resource allocation strategy. The cross-entropy loss function is used to calculate the error between the predicted result and the actual label. The weight parameters of the neural network model are optimized through the back propagation algorithm until the loss function converges, and the dynamic communication adaptation model of the virtual network terminal is obtained.

8. The method for implementing dynamic communication adaptation of a virtual network terminal according to claim 7, characterized in that: Said S4 further comprises: A dynamic communication adaptation model based on virtual network terminals adopts a dynamic update mechanism to adjust the parameters of the neural network model to adapt to the network environment, evaluate the network status of the virtual network terminals in real time, and dynamically select the optimal network resource allocation strategy according to the current network status to achieve communication quality optimization and adaptation of the virtual network in various network environments.

9. A dynamic communication adaptation system for a virtual network terminal, characterized in that: A method for implementing dynamic communication adaptation of a virtual network terminal, for implementing any one of claims 1 to 8, comprising: Network demand matrix module, network behavior profiling module, network resource allocation strategy module, and communication quality optimization and adaptation module; The network demand matrix module is used to analyze the user group's demand for network performance based on the user group's historical network usage data and construct the user group's network demand matrix; The network behavior profiling module is used to analyze network measurement platform data, monitor network conditions and device operation status in real time, obtain network resource allocation status of user group communication device software, and generate user group network behavior profiles; The network resource allocation strategy module is electrically connected to the network demand matrix module and the network behavior profile module. The network resource allocation strategy module is used to dynamically adjust and optimize the network resource allocation strategy of the virtual network terminal by using a reinforcement learning algorithm, combining the user group network demand matrix and the user group network behavior profile; The communication quality optimization and adaptation module is electrically connected to the network resource allocation strategy module. The communication quality optimization and adaptation module is used to build a dynamic communication adaptation model of the virtual network terminal based on the network resource allocation strategy of the optimized virtual network terminal, so as to realize the communication quality optimization and adaptation of the virtual network in various network environments.

Images