A system and method for establishing a VPN data flow acceleration model

Abstract

The application relates to the technical field of network communication management, and discloses a system and method for establishing a VPN data flow acceleration model, which comprises the following steps: allocating an IP address to an access device through a dynamic host configuration protocol (DHCP) and obtaining an IP address of a message queue telemetry transport protocol (MQTT) server; the access device registers identity information through the MQTT and stores the identity information; the access device is managed in batches according to a subscription mechanism of the MQTT server, and the management content comprises device configuration, a running state, terminal access and batch upgrading; meanwhile, VPN data features and historical data features of the access device are obtained and stored, a VPN data flow classification model is established based on the features, and the VPN features of the device are substituted into the model to obtain a classification result; finally, real-time performance data of a network link are obtained, an effective link is determined according to a demand relationship between the real-time data and the classification result, and a best path transmission mechanism is selected based on the effective link, so that the transmission performance of the VPN data flow is optimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network communication management, in particular to a system and method for establishing a VPN data flow acceleration model. BACKGROUND

[0002] With the rapid development of network technology, especially in the widespread application of wireless access devices and VPN data flow, the traditional centralized management mode of access controller (AC) and access point (AP) is facing increasing challenges. The traditional AC / AP centralized management scheme usually uses the CAPWAP protocol (Control and Provisioning of Wireless Access Points Protocol) to realize the access, configuration, management and monitoring of devices. However, the CAPWAP protocol has some serious limitations in the environment of high concurrency and large-scale device access. The main problem is that the protocol design is too complex, which causes high resource consumption and large delay, resulting in that the real-time performance and efficiency of device management and data flow are difficult to meet the needs of modern network environment.

[0003] In order to meet the increasing demand for bandwidth and network traffic management, VPN data flow acceleration has become one of the key technologies. However, the existing VPN acceleration technology usually relies on static link configuration and protocol scheduling mechanism, and cannot fully consider the dynamic changes between different devices and links. Especially when it comes to large-scale devices and complex network structure, the traditional VPN traffic management and scheduling method often cannot respond to the changes of network load in real time, and cannot flexibly adjust the routing path and traffic control strategy of data flow. Lack of intelligent scheduling mechanism based on real-time network performance data and device state, resulting in low transmission efficiency of VPN data flow, especially in the environment of device diversification and increasing network complexity, the problem is more obvious.

[0004] Therefore, it is urgent to invent a network management system to solve the problem that the existing VPN acceleration technology relies on static link configuration and protocol scheduling, cannot respond to the dynamic changes of network load and devices in real time, resulting in low efficiency of data flow scheduling, especially in large-scale device and complex network environment, cannot flexibly optimize the transmission path. SUMMARY

[0005] In view of this, the present application provides a system and method for establishing a VPN data flow acceleration model, which aims to solve the problem that the existing VPN acceleration technology relies on static link configuration and protocol scheduling, cannot respond to the dynamic changes of network load and devices in real time, resulting in low efficiency of data flow scheduling, especially in large-scale device and complex network environment, cannot flexibly optimize the transmission path.

[0006] The present application provides a method for establishing a VPN data flow acceleration model, comprising:

[0007] The IP addresses of the access devices are allocated based on a dynamic host configuration protocol, and the IP addresses of the message queue telemetry transport protocol server are acquired based on the dynamic host configuration protocol; identity information of each of the access devices is registered based on the message queue telemetry transport protocol, and the identity information of each of the access devices is stored;

[0008] A subscription mechanism of the message queue telemetry transport protocol server is acquired, and each of the access devices is managed in batches according to the subscription mechanism, wherein the batch management includes configuration issuing of the access devices, running state reporting of the access devices, terminal access management of the access devices, and batch upgrading of each of the access devices;

[0009] VPN data features of each of the access devices are acquired and stored, wherein the VPN data features include: traffic mode, packet size, protocol type, delay, and packet loss rate;

[0010] Historical VPN data features are acquired, and the VPN data flow classification model is established based on the historical VPN data features; the VPN data features of the access devices are substituted into the VPN data flow classification model, and a classification result of the access devices is acquired;

[0011] Real-time performance data of each link in a network link are acquired, an effective link is determined according to a relationship between the real-time performance data and performance requirement data of the classification result, and a path transmission mechanism between the effective link and the access devices is determined based on the effective link.

[0012] Further, when the VPN data flow classification model is established based on the historical VPN data features, the following steps are included:

[0013] Traffic mode, packet size, protocol type, delay, and packet loss rate of each historical VPN data in the historical VPN data features are acquired, and a historical VPN data correlation formula is established;

[0014] The distance metric between each of the historical VPN data correlations is acquired based on the Euclidean distance:

[0015]

[0016] wherein D(x, y) is the distance metric between any two historical VPN data correlations, xi and yi respectively represent the i-th data feature between the arbitrary two historical VPN data correlations, the data features include: traffic mode, packet size, protocol type, delay, and packet loss rate, and n is the total number of the data features;

[0017] A distance matrix is constructed according to the distance metric, and each of the historical VPN data correlations is iteratively clustered according to the distance matrix.

[0018] The historical VPN data association formula after iteration clustering is determined as the VPN data flow type association formula, the type of each VPN data flow is determined according to the number of each VPN data flow type association formula, and the VPN data flow classification model is established according to each VPN data flow type association formula and the type of each VPN data flow.

[0019] Further, when the VPN data flow classification model is established according to each VPN data flow type association formula and the type of each VPN data flow, it comprises:

[0020] The distance metric between each historical VPN data association formula and each VPN data flow type association formula is obtained, and each historical VPN data association formula is classified according to the distance metric and the preset distance metric;

[0021] The linear relationship axis between each historical VPN data association formula under the type of the VPN data flow is obtained, and the feature vector of the type of the VPN data flow is determined according to the linear relationship axis;

[0022] The VPN data flow classification model is established according to the feature vector of the type of each VPN data flow.

[0023] Further, when the effective link is determined according to the relationship between the real-time performance data and the performance demand data of the classification result, it comprises:

[0024] According to the relationship between the real-time performance data and the performance demand data, the similarity between the real-time performance data and the performance demand data is determined:

[0025]

[0026] Wherein L i is the similarity between the real-time performance data and the performance demand data, N is the total number of index data, G i is the i th index data in the real-time performance data, and g i is the i th index data in the performance demand data;

[0027] According to the relationship between the similarity and the preset similarity, each link is screened:

[0028] When the similarity is less than the preset similarity, the link is determined to be removed;

[0029] When the similarity is greater than or equal to the preset similarity, the link is determined to be retained, and the relationship between each retained link is determined to be the effective link.

[0030] Further, when determining the effective link according to the relationship between the reserved links, the method comprises:

[0031] Obtaining an average value between real-time performance data between the reserved links, and determining the average value as preset performance data;

[0032] Determining a performance score of each link according to the relationship between the real-time performance data and the preset performance data;

[0033]

[0034] Wherein, S is the performance score of the link, H i is the weight of the index characteristic in the real-time performance data, a i is the i th index characteristic in the preset performance data, and A i is the i th index characteristic in the real-time data;

[0035] Sorting each performance score in descending order, and determining the link corresponding to the performance score ranked first as the effective link.

[0036] Compared with the prior art, the application has the beneficial effects that: by using the DHCP protocol to dynamically allocate IP addresses for each access device and at the same time obtaining the IP address of the message queue telemetry transmission protocol server, seamless connection of the access device and the network management system is realized. Through this mechanism, the identity information of each device can be automatically registered and stored, avoiding the complexity and error risk brought by traditional static configuration, and providing an efficient and automated solution for device management. In addition, based on the subscription mechanism of the MQTT protocol, batch management of the access device can be realized. This not only includes the configuration of the device, the running state report, but also involves terminal access management and batch upgrade of the device and other key tasks. Compared with the traditional management method, this MQTT-based scheme greatly improves the flexibility and real-time performance of management, especially in a large-scale device environment, precise real-time monitoring and scheduling can be realized, greatly reducing the need for manual intervention. The state information of the device can be uploaded to the server in real time, and the administrator can obtain the latest running data of the device at any time through the subscription mechanism, and quickly respond to any changes in the network, thereby effectively improving the overall management efficiency of the network and the working stability of the device. Further, by obtaining and storing the VPN data characteristics of each access device, including traffic patterns, packet size, protocol type, delay and packet loss rate, valuable data support can be provided for subsequent network optimization and acceleration. By analyzing historical VPN data characteristics and establishing a VPN data flow classification model, the VPN traffic of the access device can be intelligently classified, and different optimization strategies can be provided for different types of traffic according to the classification results. By inputting the real-time data of the access device into the model, the data flow type of each device can be accurately judged, so that individualized acceleration solutions can be better formulated for it, and the transmission efficiency and user experience of the overall network can be improved. Finally, combined with the performance data in the real-time network link, the optimal network link can be automatically determined based on the relationship between the classification model and the real-time performance data. This mechanism not only ensures that the data flow of each access device is transmitted on the optimal path, but also adjusts the transmission path of the data flow in real time according to the changes in network load and link quality, avoiding the network congestion and performance bottleneck problems existing in the traditional static path allocation method. Through this dynamic path selection, high data transmission quality can be maintained in a complex network environment, while the flexibility and adaptability of the network are greatly improved, ensuring the acceleration effect and stability of the VPN data flow.

[0037] On the other hand, the application also provides a VPN data flow acceleration model establishing system, comprising:

[0038] An access device and an access controller;

[0039] The access controller is used for batch management of the access devices, wherein the access controller comprises a connection unit, a storage unit, a first central control unit, a second central control unit and an adjustment unit.

[0040] The connection unit is configured to assign IP addresses to the access devices based on a dynamic host configuration protocol, and obtain IP addresses of message queue telemetry transport protocol servers based on the dynamic host configuration protocol; the connection unit is further configured to register identity information of the access devices based on the message queue telemetry transport protocol, and store the identity information of the access devices;

[0041] The storage unit is configured to obtain and store VPN data features of the access devices, wherein the VPN data features comprise traffic patterns, packet sizes, protocol types, delays and packet loss rates; the storage unit is further configured to obtain historical VPN data features;

[0042] The first central control unit is configured to obtain a subscription mechanism of the message queue telemetry transport protocol servers, and batch manage the access devices according to the subscription mechanism, wherein the batch management comprises configuration issuing of the access devices, running state reporting of the access devices, terminal access management of the access devices and batch upgrading of the access devices;

[0043] The second central control unit is configured to establish the VPN data flow classification model based on the historical VPN data features, substitute the VPN data features of the access devices into the VPN data flow classification model, and obtain classification results of the access devices;

[0044] The adjustment unit is configured to obtain real-time performance data of each link in a network link, determine effective links according to a relationship between the real-time performance data and performance requirement data of the classification results, and determine a path transmission mechanism between the effective links and the access devices based on the effective links.

[0045] Further, when the second central control unit establishes the VPN data flow classification model based on the historical VPN data features, the following steps are included:

[0046] The second central control unit is further configured to obtain traffic patterns, packet sizes, protocol types, delays and packet loss rates of each historical VPN data in the historical VPN data features, and establish historical VPN data correlation formulas;

[0047] The second central control unit is further configured to obtain distance metrics between the historical VPN data correlation formulas based on Euclidean distances:

[0048]

[0049] wherein, D(x, y) is a distance metric between any two historical VPN data association formulas, xi and yi respectively refer to the ith data feature between the any two historical VPN data association formulas, the data features include: traffic pattern, packet size, protocol type, delay and packet loss rate, and n is the total number of the data features;

[0050] The second central control unit is further configured to construct a distance matrix according to the distance metric, and iteratively cluster each of the historical VPN data association formulas according to the distance matrix;

[0051] The second central control unit is further configured to determine the iteratively clustered historical VPN data association formulas as the VPN data flow type association formulas, determine the type of each of the VPN data flows according to the number of the VPN data flow type association formulas, and establish the VPN data flow classification model according to each of the VPN data flow type association formulas and the type of each of the VPN data flows.

[0052] Further, when the second central control unit establishes the VPN data flow classification model according to each of the VPN data flow type association formulas and the type of each of the VPN data flows, it includes:

[0053] The second central control unit is further configured to obtain the distance metric between each of the historical VPN data association formulas and each of the VPN data flow type association formulas, and classify each of the historical VPN data association formulas according to the distance metric and a preset distance metric;

[0054] The second central control unit is further configured to obtain the linear relationship axis between each of the historical VPN data association formulas under the type of the VPN data flow, and determine the feature vector of the type of the VPN data flow according to the linear relationship axis;

[0055] The second central control unit is further configured to establish the VPN data flow classification model according to the feature vector of the type of each of the VPN data flows.

[0056] Further, when the adjustment unit determines the effective link according to the relationship between the real-time performance data and the performance demand data of the classification result, it includes:

[0057] The adjustment unit is further configured to determine the similarity between the real-time performance data and the performance demand data according to the relationship between the real-time performance data and the performance demand data.

[0058]

[0059] Wherein Li is the similarity between real-time performance data and the performance requirement data, N is the total number of index data, Gi is the i-th index data in the real-time performance data, and gi is the i-th index data in the performance requirement data;

[0060] The adjustment unit is further configured to screen each link according to a relationship between the similarity and a preset similarity;

[0061] When the similarity is less than the preset similarity, the adjustment unit determines to eliminate the link;

[0062] When the similarity is greater than or equal to the preset similarity, the adjustment unit determines to retain the link, and determines the effective link according to a relationship between the retained links.

[0063] Further, when the adjustment unit determines the effective link according to the relationship between the retained links, the method comprises:

[0064] obtaining an average value between real-time performance data between the retained links, and determining the average value as a preset performance data;

[0065] The adjustment unit is further configured to determine a performance score of each link according to a relationship between the real-time performance data and the preset performance data;

[0066]

[0067] Wherein S is the performance score of the link, Hi is a weight of an index feature in the real-time performance data, ai is the i-th index feature in the preset performance data, and Ai is the i-th index feature in the real-time data;

[0068] The adjustment unit is further configured to sort each performance score in a reverse order, and determine a link corresponding to the performance score ranked first as the effective link.

[0069] It can be understood that the system and method for establishing a VPN data flow acceleration model in each of the above embodiments of the application have the same beneficial effects, which will not be repeated. BRIEF DESCRIPTION OF DRAWINGS

[0070] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are included to provide a description of preferred embodiments, and are not intended to limit the scope of the application. Moreover, the same reference numerals are used throughout the same figures to designate similar or equivalent components. In the drawings:

[0071] Figure 1A flow chart of a VPN data stream acceleration model establishment method provided by an embodiment of the present application is shown in the figure;

[0072] Figure 2 A structure block diagram of a VPN data stream acceleration model establishment system provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0073] Exemplary embodiments of the present disclosure will be described in detail with reference to the drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in conjunction with the embodiments.

[0074] As shown in Figure 1 In some embodiments of the present application, the present embodiment provides a VPN data stream acceleration model establishment method, which comprises the following steps:

[0075] In step S100, IP addresses are allocated to each access device based on a dynamic host configuration protocol, and the IP address of a message queue telemetry transfer protocol server is obtained based on the dynamic host configuration protocol. Identity information of each access device is registered based on the message queue telemetry transfer protocol, and the identity information of each access device is stored.

[0076] It can be understood that the IP address is allocated to the access device by using the dynamic host configuration protocol (DHCP). DHCP is an automated protocol that simplifies the access configuration process of devices by dynamically allocating IP addresses, avoiding the tedious manual configuration of IP addresses. Each access device requests an available IP address and obtains other network configuration information when accessing the network through DHCP. Through DHCP, the access device can quickly obtain the necessary configuration in the network, reduce the complexity of network management, and improve the scalability of the network and the efficiency of device access. Secondly, based on the message queue telemetry transfer protocol (MQTT), each access device registers with the MQTT server when the device obtains the IP address. MQTT is a lightweight message transmission protocol suitable for low-bandwidth, high-latency or unstable network environments. Through MQTT, the access device can send its identity information (such as device ID, model, status, etc.) to the MQTT server for registration. This process provides a unified platform for device management, which can record the identity information of each device in real time and realize two-way communication between the device and the management system. This method not only improves the efficiency of device access, but also makes the management of device information more centralized and standardized. Finally, through the MQTT protocol, the identity information of all registered access devices will be stored on the server side. This process not only realizes the unique identification of device identity, but also provides basic data support for subsequent device management, configuration issuance, status monitoring, etc. Through the storage of device identity information, the management can obtain the status data and configuration information of the device at any time, realizing dynamic monitoring and real-time management.

[0077] It can be seen that by using the dynamic host configuration protocol (DHCP) to automatically assign IP addresses to access devices, the complexity of manual configuration and the risk of errors can be greatly reduced. DHCP enables each device to quickly obtain a suitable IP address when accessing the network, thereby saving time and improving the efficiency of device access. In addition, the automatic allocation process of DHCP also provides flexibility for network managers, making device access more scalable, especially when facing large-scale device access, effectively avoiding the bottleneck of network configuration. Secondly, combined with the message queue telemetry transfer protocol (MQTT), intelligent registration and centralized management of device information can be achieved. MQTT is a lightweight and reliable message transmission protocol, through which the identity information of each access device can be uploaded to the server side in time for registration. In this way, the management platform can real-time master the basic situation, state information and interaction mode of each access device with the network, thereby improving the monitoring and management ability of the device. Centralized storage of device identity information ensures the consistency and integrity of device information, which is beneficial to subsequent dynamic scheduling, configuration management and fault diagnosis of devices. Finally, the storage of device identity information provides basic data support for subsequent device management, performance optimization and network configuration. By storing device information, management can view and retrieve the status, configuration and historical behavior of a specific device at any time, providing data support for real-time management and troubleshooting. This mechanism not only improves the automation level of network management, but also enables network operators to more accurately maintain, upgrade and dynamically adjust devices, greatly improving the manageability, maintainability and flexibility of the network.

[0078] Step S200, obtain the subscription mechanism of the message queue telemetry transfer protocol server, and perform batch management on each access device according to the subscription mechanism, wherein the batch management includes configuration delivery of the access device, running state reporting of the access device, terminal access management of the access device and batch upgrade of each access device.

[0079] It can be understood that the subscription mechanism of the message queue telemetry transport protocol (MQTT) server is acquired. The MQTT protocol adopts a publish / subscribe message delivery mode, in which a server (or called a broker) manages all client subscription relationships. Under this mechanism, each access device acts as an MQTT client and can subscribe to a specific message topic as needed to receive instructions and data related thereto in real time. The MQTT server maintains these subscription relationships and can push configuration instructions, status information, or update data according to the needs of the device. Through this mechanism, the device can obtain necessary configuration information and status updates in a timely manner without frequent polling, thereby improving the efficiency and response speed of network management. Secondly, according to the subscription mechanism of the MQTT protocol, batch management of various access devices is realized. Batch management refers to the ability of the management platform to push configuration, status information or upgrade instructions to multiple access devices at a time through the subscription mechanism. For example, the platform can issue a unified configuration file to all devices or perform differential management according to the categories and states of the devices. Through this mechanism, the administrator does not need to configure and manage each device individually, greatly reducing the complexity and labor cost of management operations. After receiving the instructions, the device will perform configuration updates, status reporting or version upgrades according to the subscribed topics, realizing batch and automated device management. Finally, batch management includes key tasks such as configuration issuance of access devices, running state reporting, terminal access management and batch upgrade. Through the MQTT protocol, the platform can perform real-time state monitoring and data collection on access devices, such as periodic reporting of device running states, to ensure that the platform can always monitor the health of the devices. At the same time, terminal access management of access devices ensures that the access process is safe and smooth, avoiding conflicts or omissions in managing access devices. In addition, with the batch upgrade mechanism, the latest firmware version or configuration parameters can be uniformly pushed to the devices to ensure that each access device can be updated synchronously and have the latest functions.

[0080] Step S300, acquiring and storing the VPN data features of each access device, wherein the VPN data features include: traffic pattern, packet size, protocol type, delay and packet loss rate.

[0081] Step S400, acquiring historical VPN data features and establishing a VPN data flow classification model based on the historical VPN data features, substituting the VPN data features of the access device into the VPN data flow classification model to obtain the classification result of the access device.

[0082] Specifically, when establishing the VPN data flow classification model based on historical VPN data features, the traffic mode, packet size, protocol type, delay and packet loss rate of each historical VPN data in the historical VPN data features are obtained, and a historical VPN data correlation formula is established. The distance metric between each historical VPN data correlation formula is obtained based on the Euclidean distance:

[0083]

[0084] wherein D(x, y) is the distance metric between any two historical VPN data correlation formulas, xi and yi respectively represent the ith data feature between the arbitrary two historical VPN data correlation formulas, the data features include: traffic mode, packet size, protocol type, delay and packet loss rate, and n is the total number of data features. A distance matrix is constructed according to the distance metric, and iterative clustering is performed between each historical VPN data correlation formula according to the distance matrix. The historical VPN data correlation formula after iterative clustering is determined as a VPN data flow type correlation formula, the type of each VPN data flow is determined according to the number of each VPN data flow type correlation formula, and a VPN data flow classification model is established according to each VPN data flow type correlation formula and the type of each VPN data flow.

[0085] Specifically, when establishing the VPN data flow classification model according to each VPN data flow type correlation formula and the type of each VPN data flow, the distance metric between each historical VPN data correlation formula and each VPN data flow type correlation formula is obtained, and each historical VPN data correlation formula is classified according to the distance metric and a preset distance metric. The linear relationship axis between each historical VPN data correlation formula under the type of the VPN data flow is obtained, and the feature vector of the type of the VPN data flow is determined according to the linear relationship axis. The VPN data flow classification model is established according to the feature vector of the type of each VPN data flow.

[0086] It can be understood that by collecting multiple features in historical VPN data such as traffic patterns, packet size, protocol type, delay, and packet loss rate, etc., each historical VPN data can be analyzed to establish the relevance between them. Then, based on these feature information, the similarity between each pair of historical VPN data is measured using the Euclidean distance. The Euclidean distance formula can effectively quantify the differences between different VPN data streams in multiple feature dimensions, thereby laying the foundation for subsequent data clustering and classification. By calculating the distance measure between historical VPN data, the similarity between these data can be quantified, and based on this, a distance matrix is constructed to provide the necessary input data for clustering analysis. Secondly, after the distance matrix is established, the historical VPN data is classified by iterative clustering method, and further generates each VPN data stream type correlation formula. The clustering algorithm can classify similar VPN data features into the same class according to the distance relationship between historical data, thereby forming multiple VPN data stream type correlation formulas. Each data stream type correlation formula represents a specific type of VPN data stream, and these classifications can define different data stream types according to their number and characteristics. For example, some data streams may have low delay and small packet size, indicating efficient real-time application traffic, while other data streams may exhibit high packet loss rate and large packet size, suitable for large-scale file transfer. After clustering, each VPN data stream type correlation formula will provide a basis for subsequent data stream classification and better identify and manage different data stream types. Finally, through the construction of the type feature vector of the VPN data stream, a specific classification model can be established for each VPN data stream type. First, based on historical VPN data and the correlation formula of each VPN data stream type, the linear relationship axis between them is calculated to determine the feature vector of each VPN data stream type. These feature vectors reflect the core characteristics of each data stream type, including traffic patterns, packet size, delay, packet loss rate, etc. These feature vectors are then used to establish a VPN data stream classification model, thereby achieving automatic classification of access device VPN traffic.

[0087] It can be seen that by deeply analyzing multiple features of historical VPN data such as traffic patterns, packet size, protocol type, delay and packet loss rate, etc., the potential regularity and trend of VPN traffic can be captured. The establishment of this classification model can more accurately identify various types of VPN data streams, optimize traffic management and scheduling strategies, and improve the performance and transmission efficiency of VPN connections, reducing network congestion and resource waste. Secondly, through the Euclidean distance measurement and clustering analysis, the precise clustering of historical VPN data association can be realized, and similar data streams are classified into the same category. This similarity-based clustering method helps to identify the commonalities and differences between different VPN data stream types. Based on this clustering result, the VPN data of the access device can be accurately classified, and different processing strategies or priorities can be assigned to each type of data stream, thereby realizing more intelligent traffic management. For example, for real-time traffic sensitive to delay, a lower delay path can be allocated, while for large data volume file transfer traffic, a larger bandwidth link can be selected. Finally, through the construction of feature vectors and the establishment of VPN data stream classification models, corresponding optimization strategies can be developed according to the characteristics of each VPN data stream type. These feature vectors not only help to identify different types of data streams, but also guide the decision-making of network routing and bandwidth allocation, thereby ensuring the efficient use of network resources. Through this classification model, changes in network load can be flexibly addressed, and transmission strategies can be adjusted in real time to maximize data stream acceleration and transmission efficiency. This is particularly important in large network environments with numerous devices and complex traffic, effectively improving the intelligent level of network management and reducing operational costs.

[0088] Step S500, real-time performance data of each link in the network link is obtained, and according to the relationship between the real-time performance data and the performance demand data of the classification result, the effective link is determined, and the path transmission mechanism between the effective link and the access device is determined.

[0089] Specifically, when determining the effective link according to the relationship between the real-time performance data and the performance demand data of the classification result, it includes: determining the similarity between the real-time performance data and the performance demand data according to the relationship between the real-time performance data and the performance demand data:

[0090]

[0091] Wherein Li is the similarity between real-time performance data and performance requirement data, N is the total number of index data, Gi is the i-th index data in real-time performance data, and gi is the i-th index data in performance requirement data. According to the relationship between the similarity and the preset similarity, each link is screened: when the similarity is less than the preset similarity, it is determined to eliminate the link. When the similarity is greater than or equal to the preset similarity, it is determined to retain the link, and according to the relationship between the retained links, the effective link is determined.

[0092] Specifically, when the effective link is determined according to the relationship between the retained links, it includes: obtaining the average value between the real-time performance data between the retained links, and determining the average value as the preset performance data. According to the relationship between the real-time performance data and the preset performance data, the performance score of each link is determined:

[0093]

[0094] Wherein, S is the performance score of the link, Hi is the weight of the index feature in the real-time performance data, ai is the i-th index feature in the preset performance data, and Ai is the i-th index feature in the real-time data. The performance scores are sorted in descending order, and the link corresponding to the first sorted performance score is determined as the effective link.

[0095] It can be understood that by defining the similarity formula, the difference between the real-time performance data and the performance requirement data is calculated, and it is determined which links meet the performance requirements. The core of this process is to identify the links that match the requirements by comparing the real-time performance data (such as bandwidth, delay, packet loss rate, etc.) of each link with the preset performance requirements (such as traffic pattern, delay requirement, etc.). Specifically, by calculating the similarity, links that do not meet the performance requirements are filtered out, and the links are classified and filtered according to the level of similarity. Secondly, after filtering out the links that meet the conditions, the performance score of the link is calculated. By obtaining the real-time performance data of the reserved link and calculating its average value, a preset performance data is formed. The gap between these links and the preset performance data is measured by the performance score, so as to judge the performance of each link. By introducing a weight coefficient, the calculation of the score can be dynamically adjusted according to the importance of different performance indicators, ensuring that important performance indicators (such as delay, bandwidth, etc.) have the greatest impact on link selection. This calculation method ensures that when selecting effective links, all performance indicators are considered to make more accurate decisions. Finally, according to the performance score of the link, the best link is selected by sorting. Among all the links that meet the conditions, based on the descending order of the performance score, the link with the highest score is selected as the effective link. Through this sorting method, the optimal link can be selected from multiple candidate links, and the best network path is allocated to the access device. This process not only improves the transmission efficiency of data flow, but also avoids delay and packet loss caused by poor network performance, thereby optimizing the acceleration effect of VPN data flow.

[0096] It can be seen that by calculating the similarity between real-time performance data and performance requirement data, it can effectively identify which links meet the predetermined performance requirements, such as bandwidth, delay and other network performance indicators. This method can dynamically evaluate the suitability of the link, thus avoiding the shortcomings of traditional static link selection, making the selection of network links more flexible and real-time, adapting to the changing network conditions. Secondly, through the performance scoring mechanism, the scheme can analyze and prioritize the reserved links in depth. The performance score of each link is calculated based on the difference between the real-time performance data and the preset data, and by introducing a weight coefficient, the importance of key performance indicators (such as delay, throughput, etc.) is highlighted. Through this scoring method, the best-performing link can be identified to provide the best transmission path for data traffic. This not only improves the efficiency of link selection, but also ensures the quality and stability of data transmission, avoiding interference from low-quality links. Finally, through the sorting and selection of effective links, the best network path and transmission mechanism can be provided for access devices. By sorting the performance scores of the links, the best-performing link can be selected as the effective link, ensuring the communication quality of the access device and the efficient transmission of data flow. This intelligent link selection strategy based on real-time data effectively improves the overall performance of the network and enhances the acceleration effect of VPN data flow, optimizes resource utilization, reduces network delay and packet loss rate, and provides an intelligent and reliable solution for complex device and link requirements in network environments.

[0097] In the above embodiment, the IP address of each access device is dynamically allocated by using the DHCP protocol, and the IP address of the message queue telemetry transport protocol server is obtained at the same time, thereby realizing seamless connection of the access device and the network management system. Through this mechanism, the identity information of each device can be automatically registered and stored, avoiding the complexity and error risk brought by traditional static configuration, and providing an efficient and automated solution for device management. In addition, based on the subscription mechanism of the MQTT protocol, batch management of the access device can be realized. This not only includes device configuration delivery and running state reporting, but also involves terminal access management and batch upgrade of devices and other key tasks. Compared with the traditional management method, this MQTT-based scheme greatly improves the flexibility and real-time performance of management, especially in a large-scale device environment, precise real-time monitoring and scheduling can be realized, greatly reducing the need for manual intervention. The state information of the device can be uploaded to the server in real time, and the administrator can obtain the latest running data of the device at any time through the subscription mechanism, and quickly respond to any changes in the network, thereby effectively improving the overall management efficiency of the network and the working stability of the device. Further, by obtaining and storing the VPN data characteristics of each access device, including traffic patterns, packet size, protocol type, delay, and packet loss rate, valuable data support can be provided for subsequent network optimization and acceleration. By analyzing historical VPN data characteristics and establishing a VPN data flow classification model, the VPN traffic of the access device can be intelligently classified, and different optimization strategies can be provided for different types of traffic based on the classification results. By inputting real-time data of the access device into the model, the data flow type of each device can be accurately determined, so that individualized acceleration solutions can be better formulated for it, improving the transmission efficiency of the overall network and user experience. Finally, based on the relationship between the classification model and real-time performance data, the optimal network link can be automatically determined based on the relationship between the classification model and real-time performance data. This mechanism not only ensures that the data flow of each access device is transmitted on the optimal path, but also adjusts the transmission path of the data flow in real time according to the changes in network load and link quality, avoiding the network congestion and performance bottleneck problems existing in the traditional static path allocation method. Through this dynamic path selection, high data transmission quality can be maintained in a complex network environment, while greatly improving the flexibility and adaptability of the network, ensuring the acceleration effect and stability of the VPN data flow.

[0098] In another preferred mode based on the above embodiment, as shown in Figure 2 , the present embodiment provides a system for establishing a VPN data flow acceleration model, comprising: an access device and an access controller.

[0099] Specifically, the access controller is used for batch management of the access devices, wherein the access controller comprises a connection unit, a storage unit, a first central control unit, a second central control unit and an adjustment unit. The connection unit is configured to assign IP addresses to the access devices based on a dynamic host configuration protocol, and obtain an IP address of a message queue telemetry transport protocol server based on the dynamic host configuration protocol. The connection unit is further configured to register identity information of the access devices based on the message queue telemetry transport protocol, and store the identity information of the access devices. The storage unit is configured to obtain and store VPN data features of the access devices, wherein the VPN data features comprise a traffic pattern, a packet size, a protocol type, a delay and a packet loss rate. The storage unit is further configured to obtain historical VPN data features. The first central control unit is configured to obtain a subscription mechanism of the message queue telemetry transport protocol server, and batch manage the access devices according to the subscription mechanism, wherein the batch management comprises configuration issuing of the access devices, running state reporting of the access devices, terminal access management of the access devices and batch upgrading of the access devices. The second central control unit is configured to establish a VPN data flow classification model based on the historical VPN data features, substitute the VPN data features of the access devices into the VPN data flow classification model, and obtain a classification result of the access devices. The adjustment unit is configured to obtain real-time performance data of each link in a network link, determine an effective link according to a relationship between the real-time performance data and performance requirement data of the classification result, and determine a path transmission mechanism between the effective link and the access devices based on the effective link.

[0100] Specifically, when the second central control unit establishes the VPN data flow classification model based on the historical VPN data features, the second central control unit is further configured to obtain the traffic pattern, the packet size, the protocol type, the delay and the packet loss rate of each historical VPN data in the historical VPN data features, and establish historical VPN data correlation formulas. The second central control unit is further configured to obtain distance metrics between the historical VPN data correlation formulas based on a Euclidean distance:

[0101]

[0102] wherein D(x, y) is a distance metric between any two historical VPN data correlation formulas, xi and yi respectively refer to the ith data feature between the any two historical VPN data correlation formulas, the data features include: traffic pattern, packet size, protocol type, delay and packet loss rate, and n is the total number of data features. The second control unit is further configured to construct a distance matrix according to the distance metric, and iteratively cluster each historical VPN data correlation formula according to the distance matrix. The second control unit is further configured to determine the iteratively clustered historical VPN data correlation formula as a VPN data flow type correlation formula, determine the type of each VPN data flow according to the number of each VPN data flow type correlation formula, and establish a VPN data flow classification model according to each VPN data flow type correlation formula and the type of each VPN data flow.

[0103] Specifically, when the second control unit establishes the VPN data flow classification model according to each VPN data flow type correlation formula and the type of each VPN data flow, the second control unit is further configured to obtain the distance metric between each historical VPN data correlation formula and each VPN data flow type correlation formula, and classify each historical VPN data correlation formula according to the distance metric and a preset distance metric. The second control unit is further configured to obtain the linear relationship axis between each historical VPN data correlation formula under the type of the VPN data flow, and determine the feature vector of the type of the VPN data flow according to the linear relationship axis. The second control unit is further configured to establish the VPN data flow classification model according to the feature vector of the type of each VPN data flow.

[0104] Specifically, when the adjustment unit determines the effective link according to the relationship between the real-time performance data and the performance requirement data of the classification result, the adjustment unit is further configured to determine the similarity between the real-time performance data and the performance requirement data according to the relationship between the real-time performance data and the performance requirement data:

[0105]

[0106] wherein Li is the similarity between the real-time performance data and the performance requirement data, N is the total number of index data, Gi is the ith index data in the real-time performance data, and gi is the ith index data in the performance requirement data. The adjustment unit is further configured to filter each link according to the relationship between the similarity and a preset similarity: when the similarity is less than the preset similarity, the adjustment unit determines to remove the link. When the similarity is greater than or equal to the preset similarity, the adjustment unit determines to retain the link, and determines the effective link according to the relationship between the retained links.

[0107] Specifically, the adjusting unit, when determining the effective link according to the relationship between the reserved links, comprises: obtaining the average value between the real-time performance data between the reserved links, and determining the average value as the preset performance data. The adjusting unit is further configured to determine the performance score of each link according to the relationship between the real-time performance data and the preset performance data:

[0108]

[0109] Wherein, S is the performance score of the link, H i is the weight of the index feature in the real-time performance data, a i is the i th index feature in the preset performance data, and A i is the i th index feature in the real-time data. The adjusting unit is further configured to sort the performance scores in descending order, and determine the link corresponding to the performance score ranked first as the effective link.

[0110] It can be understood that the system and method for establishing a VPN data flow acceleration model in each of the above embodiments of the present application have the same beneficial effects, which will not be described again.

[0111] 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 be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in 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.

[0112] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems) and computer program products 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 realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1 The means for performing the functions specified in one block or multiple blocks.

[0113] These computer program instructions can also be stored in a computer readable storage medium that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction means, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1the function specified in the one or more blocks.

[0114] 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, so that the instructions executed on the computer or other programmable devices provide processes for implementing the flow Figure 1 the flow or flows and / or blocks Figure 1 the steps of the function specified in the one or more blocks.

[0115] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, but not to limit it, although the above embodiments of the present application have been described in detail, those skilled in the art should understand: the specific embodiments of the present application can be modified or replaced by the same, without departing from the spirit and scope of the present application, any modification or equivalent replacement, which should be covered in the protection scope of the claims of the present application.

Claims

1. A method for establishing a VPN data flow acceleration model, characterized in that, The method comprises the following steps: allocating IP addresses for each access device based on a dynamic host configuration protocol, and obtaining an IP address of a message queue telemetry transport protocol server based on the dynamic host configuration protocol; registering identity information of each access device based on a message queue telemetry transport protocol, and storing the identity information of each access device; obtaining a subscription mechanism of the message queue telemetry transport protocol server, and performing batch management on each access device according to the subscription mechanism, wherein the batch management comprises configuration issuing of the access device, running state reporting of the access device, terminal access management of the access device, and batch upgrading of each access device; obtaining and storing VPN data characteristics of each access device, wherein the VPN data characteristics comprise traffic mode, packet size, protocol type, delay, and packet loss rate; obtaining historical VPN data characteristics, and establishing a VPN data flow classification model based on the historical VPN data characteristics, substituting the VPN data characteristics of the access device into the VPN data flow classification model, and obtaining a classification result of the access device; obtaining real-time performance data of each link in a network link, determining effective links according to a relationship between the real-time performance data and performance requirement data of the classification result, and determining a path transmission mechanism between the effective links and the access device based on the effective links; wherein, when determining the effective links according to the relationship between the real-time performance data and the performance requirement data of the classification result, it comprises: determining a similarity between the real-time performance data and the performance requirement data according to the relationship between the real-time performance data and the performance requirement data: ； wherein Li is the similarity between the real-time performance data and the performance requirement data, N is the total number of index data, Gi is the i-th index data in the real-time performance data, and gi is the i-th index data in the performance requirement data; screening each link according to a relationship between the similarity and a preset similarity: when the similarity is less than the preset similarity, the link is determined to be removed; when the similarity is greater than or equal to the preset similarity, the link is determined to be retained, and the effective links are determined according to a relationship between the retained links.

2. The method of claim 1, wherein the VPN data flow acceleration model is established by: when establishing the VPN data flow classification model based on the historical VPN data characteristics, it comprises: obtaining traffic mode, packet size, protocol type, delay, and packet loss rate of each historical VPN data in the historical VPN data characteristics, and establishing a historical VPN data correlation formula; obtaining distance metrics between each historical VPN data correlation formula based on Euclidean distance: ； wherein D(x, y) is the distance metric between any two historical VPN data correlation formulas, xi and yi respectively represent the i-th data characteristic between the any two historical VPN data correlation formulas, the data characteristics comprise traffic mode, packet size, protocol type, delay, and packet loss rate, and n is the total number of the data characteristics; constructing a distance matrix according to the distance metrics, and iteratively clustering each historical VPN data correlation formula according to the distance matrix. The historical VPN data association formula after iterative clustering is determined as the VPN data flow type association formula, the type of each VPN data flow is determined according to the number of each VPN data flow type association formula, and the VPN data flow classification model is established according to each VPN data flow type association formula and the type of each VPN data flow.

3. The method of claim 2, wherein the VPN data flow acceleration model is established by: When the VPN data flow classification model is established according to each VPN data flow type association formula and the type of each VPN data flow, it includes: Obtain the distance metric between each historical VPN data association formula and each VPN data flow type association formula, and classify each historical VPN data association formula according to the distance metric and a preset distance metric; Obtain the linear relationship axis between each historical VPN data association formula under the type of the VPN data flow, and determine the feature vector of the type of the VPN data flow according to the linear relationship axis; The VPN data flow classification model is established according to the feature vector of each VPN data flow type.

4. The method of claim 3, wherein the VPN data flow acceleration model is established by: When the effective link is determined according to the relationship between the reserved links, it includes: Obtain the average value between the real-time performance data between each reserved link, and determine the average value as the preset performance data; Determine the performance score of each link according to the relationship between the real-time performance data and the preset performance data: ； Where S is the performance score of the link, H i is the weight of the index feature in the real-time performance data, a i is the i th index feature in the preset performance data, and A i is the i th index feature in the real-time performance data; Sort each performance score in descending order, and determine the link corresponding to the first performance score in the sorting as the effective link.

5. A system for establishing a VPN data flow acceleration model, used in the method for establishing a VPN data flow acceleration model according to any one of claims 1-4, characterized in that, It includes: Access device and access controller; The access controller is used for batch management of each access device, wherein the access controller includes a connection unit, a storage unit, a first central control unit, a second central control unit and an adjustment unit; The connection unit is configured to assign IP addresses to each access device based on the dynamic host configuration protocol, and obtain the IP address of the message queue telemetry transfer protocol server based on the dynamic host configuration protocol; the connection unit is also configured to register the identity information of each access device based on the message queue telemetry transfer protocol, and store the identity information of each access device; The storage unit is configured to obtain and store the VPN data features of each access device, wherein the VPN data features include traffic mode, packet size, protocol type, delay and packet loss rate; the storage unit is also configured to obtain historical VPN data features; The first central control unit is configured to obtain the subscription mechanism of the message queue telemetry transfer protocol server, and to batch manage each access device according to the subscription mechanism, wherein the batch management includes configuration issuing of the access device, running state reporting of the access device, terminal access management of the access device and batch upgrading of each access device; The second central control unit is configured to obtain the performance data of each access device, and to determine the effective link according to the relationship between the reserved links; The second central control unit is configured to establish the VPN data flow classification model based on historical VPN data features, substitute VPN data features of the access device into the VPN data flow classification model, and obtain a classification result of the access device; The adjusting unit is configured to obtain real-time performance data of each link in a network link, determine an effective link according to a relationship between the real-time performance data and performance requirement data of the classification result, and determine a path transmission mechanism between the access device based on the effective link.

6. The system for establishing a VPN data flow acceleration model of claim 5, wherein, When the second central control unit establishes the VPN data flow classification model based on historical VPN data features, the following steps are included: The second central control unit is further configured to obtain traffic mode, packet size, protocol type, delay, and packet loss rate of each historical VPN data in the historical VPN data features, and establish historical VPN data correlation formulas; The second central control unit is further configured to obtain distance metrics between each historical VPN data correlation formula based on Euclidean distance: ； wherein D(x, y) is a distance metric between any two historical VPN data correlation formulas, xi and yi respectively represent the i-th data feature between the any two historical VPN data correlation formulas, the data features include traffic mode, packet size, protocol type, delay, and packet loss rate, and n is a total number of the data features; The second central control unit is further configured to construct a distance matrix according to the distance metrics, and iteratively cluster each historical VPN data correlation formula according to the distance matrix; The second central control unit is further configured to determine the iteratively clustered historical VPN data correlation formulas as the VPN data flow type correlation formulas, determine types of each VPN data flow according to a number of the VPN data flow type correlation formulas, and establish the VPN data flow classification model according to each VPN data flow type correlation formula and each VPN data flow type.

7. The system for establishing a VPN data flow acceleration model of claim 6, wherein, When the second central control unit establishes the VPN data flow classification model according to each VPN data flow type correlation formula and each VPN data flow type, the following steps are included: The second central control unit is further configured to obtain distance metrics between each historical VPN data correlation formula and each VPN data flow type correlation formula, and classify each historical VPN data correlation formula according to the distance metrics and a preset distance metric; The second central control unit is further configured to obtain a linear relationship axis between each historical VPN data correlation formula under each VPN data flow type, and determine a feature vector of the VPN data flow type according to the linear relationship axis; The second central control unit is further configured to establish the VPN data flow classification model according to each feature vector of the VPN data flow type.

8. The system for establishing a VPN data flow acceleration model of claim 7, wherein, When the adjusting unit determines an effective link according to a relationship between the real-time performance data and performance requirement data of the classification result, the following steps are included: The adjustment unit is further configured to determine a similarity between the real-time performance data and the performance demand data according to a relationship between the real-time performance data and the performance demand data: ； wherein Li is the similarity between the real-time performance data and the performance demand data, N is a total number of index data, Gi is the i-th index data in the real-time performance data, and gi is the i-th index data in the performance demand data; The adjustment unit is further configured to screen each link according to a relationship between the similarity and a preset similarity: When the similarity is less than the preset similarity, the adjustment unit determines to eliminate the link; When the similarity is greater than or equal to the preset similarity, the adjustment unit determines to retain the link, and determines the effective link according to a relationship between the retained links.

9. The system for establishing a VPN data flow acceleration model of claim 8, wherein, When the adjustment unit determines the effective link according to the relationship between the retained links, it includes: obtaining an average value between the real-time performance data of the retained links, and determining the average value as a preset performance data; The adjustment unit is further configured to determine a performance score of each link according to a relationship between the real-time performance data and the preset performance data: ； wherein S is the performance score of the link, Hi is a weight of an index feature in the real-time performance data, ai is the i-th index feature in the preset performance data, and Ai is the i-th index feature in the real-time performance data; The adjustment unit is further configured to sort each performance score in a reverse order, and determine a link corresponding to the performance score ranked first as the effective link.

Images