Network acceleration strategy optimization system and method based on data flow

Abstract

The invention relates to the technical field of network strategy optimization, and discloses a network acceleration strategy optimization system and method based on data flow, and the method comprises the steps: deploying network monitoring equipment in each network link, and collecting performance data such as communication delay, bandwidth, packet loss rate and flow; establishing a data group based on the performance data, including the link performance data of the current moment and the adjacent historical time period, and calculating a data abnormal value according to the data group; the abnormal threshold value of the abnormal value is determined through the historical performance data, after the data abnormal value and the threshold value are compared and screened, the Gaussian mixture model is used for identifying the abnormal value, and finally the abnormal data are determined and the corresponding link is removed. And evaluating the performance data of the remaining links, constructing a correlation formula, and establishing preset performance data through iterative clustering. And based on the performance score and the network performance requirement, screening effective links, and optimizing the optimal link meeting the network acceleration strategy optimization. According to the invention, the precision and reliability of network acceleration strategy optimization can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of network strategy optimization, and in particular to a network acceleration strategy optimization system and method based on data flow. Background Art

[0002] With the popularization of Internet applications, network communications play a vital role in all walks of life, especially in scenarios such as large-scale data transmission, cloud computing, video conferencing and online games, the requirements for network bandwidth, latency, stability and packet loss rate are becoming more stringent. In this context, how to optimize network performance and improve data transmission efficiency has become an important direction for network technology research. Traditional network acceleration technologies mostly rely on static configuration or preset rules, lack dynamic monitoring and real-time optimization of network status, and are difficult to cope with complex and changing network environments.

[0003] With the continuous development of network technology, the performance of network links has become more complex and changeable. A single static optimization method can no longer meet the requirements of modern networks for efficient and stable performance. In order to improve network performance and resource utilization, many studies have begun to explore the acquisition of network link performance data in real time, and analyze and model these data to optimize network acceleration strategies more dynamically and intelligently. By monitoring key performance indicators such as network link latency, bandwidth, packet loss rate, and network traffic, network administrators can evaluate the status of links in different time periods and adjust the routing and load of network traffic based on real-time changes, thereby ensuring network stability while improving data transmission efficiency. However, existing network optimization technologies often rely on static rules or preset link configurations, which can only optimize the network to a certain extent. They usually ignore the dynamic changes of network links in different time periods and under different load conditions, and fail to fully consider potential abnormal fluctuations. For example, some links may be affected by network congestion, hardware failure, or external attacks during a specific period, resulting in drastic performance fluctuations, and traditional rule-based optimization methods are difficult to respond to these emergencies in a timely and effective manner. Due to the lack of flexibility in adapting to these dynamic changes, existing technologies have certain limitations when facing complex network environments and cannot provide continuous optimization and adaptive adjustment.

[0004] Therefore, there is an urgent need to invent a network strategy optimization technology to solve the problem that the network optimization technology in the existing technology fails to effectively cope with the dynamic changes and abnormal fluctuations of network links, resulting in the inability to provide continuous, intelligent optimization and adaptive adjustment in a complex and changing network environment. Summary of the invention

[0005] In view of this, the present invention proposes a network acceleration strategy optimization system and method based on data flow, aiming to solve the problem that the current network optimization technology fails to effectively respond to the dynamic changes and abnormal fluctuations of network links, resulting in the inability to provide continuous, intelligent optimization and adaptive adjustment in a complex and changing network environment.

[0006] The present invention proposes a method for optimizing a network acceleration strategy based on data flow, comprising:

[0007] Setting a network monitoring device on each network link and acquiring performance data of each network link, wherein the performance data includes communication delay, communication bandwidth, communication packet loss rate and communication network traffic of the network link;

[0008] Establish a data group according to the performance data, the data group including the performance data of each of the network links at the current moment and the performance data of each of the network links in an adjacent historical period, and obtain data abnormal values ​​of each of the network links according to the data group;

[0009] Acquire historical performance data of each of the network links, and acquire an abnormal threshold of the data abnormal value according to the historical performance data;

[0010] Screening the data outliers according to the outlier threshold, acquiring outliers of the screened data based on a Gaussian mixture model, and determining the outliers according to the outliers;

[0011] The network links corresponding to each abnormal data are eliminated, performance evaluation is performed based on the performance data of the network links after the network links as abnormal data are eliminated, and the network links used for network acceleration strategy optimization are determined based on the evaluation results.

[0012] Further, when obtaining the abnormal threshold of the data abnormal value according to the historical performance data, it includes:

[0013] The historical normal performance data in the historical performance data is obtained, and the abnormal threshold is obtained based on the relationship between the historical normal performance data and a formula, wherein the formula is as follows:

[0014] Y=μW+k·σW;

[0015] Among them, Y represents the abnormal threshold, μW represents the difference mean of historical normal performance data, σW represents the difference standard deviation of historical normal performance data, and k represents the adjustment factor, among which:

[0016]

[0017] Wt=|X t -X t-1 |;

[0018] Where Wt represents the data difference value of historical normal performance data, X t represents the performance data of the network link at time t, X t-1 It represents the performance data of the network link at time t-1, and N represents the number of historical normal performance data.

[0019] Further, when the data abnormal value is screened according to the abnormal threshold, it includes:

[0020] Compare the data outliers with the outlier threshold and remove the difference data;

[0021] When the data abnormal value is greater than the abnormal threshold, retain the performance data corresponding to the data abnormal value;

[0022] When the data abnormal value is less than or equal to the abnormal threshold, the performance data corresponding to the data abnormal value is determined as difference data and the performance data is eliminated.

[0023] Furthermore, obtaining outliers of the filtered data based on the Gaussian mixture model and determining the abnormal data according to the outliers include:

[0024] Calculate the outlier value of each filtered data based on the Gaussian mixture model:

[0025]

[0026] Among them, p(x|Θ) represents the outlier value of data x, K is the number of Gaussian components, π k Expressed as the mixing coefficient of the kth component, It is expressed as the probability density of data x belonging to a certain Gaussian component;

[0027] When the abnormal value of the filtered data is greater than 0.8, the data is determined to be the abnormal data, and the network link corresponding to the abnormal data is identified.

[0028] Further, when the performance evaluation is performed based on the performance data of the network link after the network link as abnormal data is removed, it includes:

[0029] Establishing a correlation formula for each network link according to the performance data of each network link after eliminating the network link as abnormal data;

[0030] Acquire a distance metric between each of the network link association equations based on the Euclidean distance, and construct a distance matrix according to the distance metric;

[0031] Iteratively clustering each of the network link association equations according to the distance matrix;

[0032] Determine the performance data in the network link association formula after iterative clustering as preset performance data;

[0033] A performance score of each of the network links is determined according to a relationship between the performance data and the preset performance data.

[0034] Further, when determining the performance score of each of the network links according to the relationship between the performance data and the preset performance data, it includes:

[0035]

[0036] Among them, S is the performance score of the network link, N is the total number of indicator data, Hi is the weight coefficient of the i-th performance data, Di is the indicator data of the i-th performance data, and di is the indicator data of the i-th preset performance data.

[0037] Furthermore, when determining the performance score of each of the network links, it also includes:

[0038] Obtaining network performance requirements for data to be transmitted, wherein the network performance requirements include communication delay requirements, communication bandwidth requirements, communication packet loss rate requirements, and communication network traffic requirements;

[0039] Determine the similarity between the network performance requirement and the performance data according to the relationship between them:

[0040]

[0041] Wherein Li is the similarity between the network performance requirement and the performance data, N is the total number of indicator data, and gi is the indicator data of the i-th network performance requirement;

[0042] The effective network link is determined according to the relationship between the similarity between the network performance requirement of each network link and the performance data and a preset preset similarity.

[0043] Further, according to the relationship between the similarity between the network performance requirement of each network link and the performance data and a preset similarity, determining the valid network link includes:

[0044] Determining whether the network link is the valid network link according to the relationship between the similarity and the preset similarity;

[0045] When the similarity is less than the preset similarity, determining that the network link is not the valid network link;

[0046] When the similarity is greater than or equal to the preset similarity, the network link is determined to be the valid network link.

[0047] Furthermore, when determining the network link used for network acceleration strategy optimization based on the evaluation result, it includes:

[0048] Obtaining a performance score for each of the valid network links, and sorting the performance scores in descending order;

[0049] The performance score ranked first is obtained, and the valid network link corresponding to the performance score ranked first is determined as the network link used when optimizing the network acceleration strategy.

[0050] Compared with the prior art, the beneficial effect of the present invention is that by real-time monitoring and analyzing the performance data of network links, combining historical data and Gaussian mixture models to screen and process abnormal data, unstable or abnormal links can be effectively identified and eliminated. This process can not only eliminate performance fluctuations caused by factors such as hardware failure, network congestion or external attacks, but also dynamically adjust the optimization strategy according to the actual network status to ensure that the network performance is always in the optimal state. By eliminating abnormal links, resources and traffic can be concentrated on links that perform more stably, thereby achieving more efficient resource utilization and data transmission. Secondly, by continuously tracking and analyzing the performance data of network links, potential problems in the network can be identified in a timely manner and corresponding optimization measures can be taken. For example, when a burst load occurs in the network, the system can quickly identify abnormal indicators such as delay and bandwidth of related links, and process abnormal data to prevent these abnormal links from affecting the overall network performance. This real-time dynamic optimization method enables the network to adapt to complex and changeable working environments and provide continuous performance guarantees without relying on static rules or fixed configurations. In addition, by using Gaussian mixture models to process data outliers, not only the fault tolerance to data noise is enhanced, but also the accuracy of anomaly detection is improved. With the help of the model, links that have a negative impact on network performance can be more intelligently screened and identified, avoiding the misjudgment or missed judgment problems that may occur in traditional optimization methods. Ultimately, this network acceleration strategy based on data flow and intelligent algorithms can maximize network transmission efficiency, reduce latency, ensure business continuity, and provide users with a more stable and smooth network experience.

[0051] On the other hand, the present application also provides a data flow-based network acceleration strategy optimization system, including:

[0052] A network monitoring device, electrically connected to each network link, and configured to collect performance data of each network link;

[0053] The monitoring module is electrically connected to the network monitoring device, and the monitoring module includes a collection unit, a processing unit, a judgment unit, an adjustment unit and a storage unit;

[0054] The collection unit is configured to obtain the performance data of each of the network links collected by the network monitoring device, and establish a data group according to the performance data, wherein the data group includes the performance data of each of the network links at the current moment and the performance data of each of the network links in an adjacent historical period; the collection unit is also configured to obtain data abnormal values ​​of each of the network links according to the data group;

[0055] The processing unit is configured to obtain historical performance data of each of the network links in the storage unit, and obtain an abnormal threshold of the data abnormal value according to the historical performance data;

[0056] The judgment unit is configured to filter the data abnormal values ​​according to the abnormal threshold, obtain the abnormal values ​​of the filtered data based on the Gaussian mixture model, and determine the abnormal data according to the abnormal values;

[0057] The adjustment unit is configured to remove the network links corresponding to each abnormal data, perform performance evaluation based on the performance data of the network links after removing the network links as abnormal data, and determine the network links used when optimizing the network acceleration strategy based on the evaluation results.

[0058] It can be understood that the data flow-based network acceleration strategy optimization system and method in the above-mentioned embodiments of the present invention have the same beneficial effects and will not be described in detail. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0060] Figure 1 A flowchart of a method for optimizing a network acceleration strategy based on data flow provided by an embodiment of the present invention;

[0061] Figure 2 A functional block diagram of a data flow-based network acceleration strategy optimization system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0062] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0063] like Figure 1 As shown, in some embodiments of the present application, this embodiment provides a method for optimizing a network acceleration strategy based on data flow, including:

[0064] Step S100: set a network monitoring device on each network link and obtain performance data of each network link, wherein the performance data includes communication delay, communication bandwidth, communication packet loss rate and communication network traffic of the network link.

[0065] Step S200: Establish a data group based on the performance data, the data group includes the performance data of each network link at the current moment and the performance data of each network link in an adjacent historical period, and obtain the data abnormal value of each network link based on the data group.

[0066] It is understandable that by setting up network monitoring equipment on each network link, the performance data of the link can be obtained in real time. These data include key performance indicators such as communication delay, bandwidth, packet loss rate and network traffic of the network link. By monitoring these data, we can fully understand the current status of each link and its performance in different time periods. These performance indicators provide a basis for subsequent data analysis and optimization. Secondly, the technical principle establishes a data group by comprehensively analyzing the real-time performance data and the performance data in the adjacent historical time periods. The data group contains not only the performance data at the current moment, but also the data of the adjacent historical time periods, so as to facilitate comparison and trend analysis. This method helps to discover the law of performance changes, and then accurately identify the fluctuations and abnormal performance of the network link, and further provide a basis for subsequent anomaly detection. Finally, by analyzing the data group, the abnormal value of each network link can be obtained. The identification of the abnormal value is based on the difference between the current performance data and the historical data, so as to identify which links have performance abnormalities.

[0067] It can be seen that based on the comprehensive collection of performance data and the construction of data groups. Through real-time monitoring of key performance indicators of each network link by network monitoring equipment, including communication delay, communication bandwidth, communication packet loss rate and communication network traffic, the operation status of the network link can be accurately grasped. On this basis, the performance data at the current moment and the performance data of the adjacent historical period are combined to construct a data group to provide data support for subsequent anomaly detection. This data group not only reflects the current network status, but also retains historical trend information, which is helpful for analyzing the dynamic changes of link performance. Next, the analysis process of the data group is the key link to obtain data anomalies. By comparing the performance data at the current moment with the historical data, the link data that deviates significantly from the normal range can be found. For example, when the communication delay of a link is suddenly significantly higher than the historical average, or the packet loss rate increases abnormally, these indicators may be regarded as data anomalies. Through this comparative analysis, the system can quickly identify potential abnormal behaviors in the link and mark these abnormal data as key problem points that may affect network performance. Finally, the acquisition of data anomalies also needs to be combined with appropriate statistical methods or algorithms for accurate screening. For example, a threshold-based method can be used to calculate the normal range based on historical performance data, and values ​​outside the range are regarded as abnormal.

[0068] Step S300: Obtain historical performance data of each network link, and obtain an abnormal threshold of a data abnormal value according to the historical performance data.

[0069] Specifically, when obtaining the abnormal threshold of the data abnormal value according to the historical performance data, it includes: obtaining the historical normal performance data in the historical performance data, and obtaining the abnormal threshold based on the historical normal performance data and the formula, wherein the formula is as follows:

[0070] Y=μW+k·σW;

[0071] Among them, Y represents the abnormal threshold, μW represents the difference mean of historical normal performance data, σW represents the difference standard deviation of historical normal performance data, and k represents the adjustment factor, among which:

[0072]

[0073] Wt=|X t -X t-1 |;

[0074] Where Wt represents the data difference value of historical normal performance data, X t represents the performance data of the network link at time t, X t-1 It represents the performance data of the network link at time t-1, and N represents the number of historical normal performance data.

[0075] It can be understood that by obtaining the historical normal performance data in the historical performance data, a data basis is provided for the subsequent calculation of the abnormal threshold. The historical normal performance data is the part that meets the stable state selected from the network link performance data collected by the monitoring device. These data represent the performance characteristics of the network under normal operation. By selecting the historical normal data, the influence of abnormal values ​​on the threshold calculation can be avoided, thereby improving the accuracy of the calculation result. Secondly, the abnormal threshold is calculated by using a formula, where the formula is based on the statistical characteristics of the historical normal performance data. In the formula, the calculation of the abnormal threshold Y includes two core parts: the difference mean and the difference standard deviation of the historical normal performance data, and the sensitivity of the threshold is adjusted by the adjustment factor. The difference mean reflects the average change amplitude of the performance data between consecutive time points, while the difference standard deviation describes the fluctuation range of these changes. By combining these two statistics, a reasonable threshold range can be determined to identify abnormal values ​​that deviate significantly from the normal state. Finally, the difference value of the historical normal performance data is the key basis for calculating the difference mean and standard deviation, which is defined as the difference between the performance data at consecutive time points. This difference value can effectively capture the dynamic changes in network link performance and analyze the volatility and trend of the link. At the same time, combined with the amount of data, the mean and standard deviation can be accurately calculated, further ensuring the robustness of the anomaly threshold calculation.

[0076] Step S400: filter outliers of the data according to an outlier threshold, obtain outliers of the filtered data based on a Gaussian mixture model, and determine abnormal data according to the outliers.

[0077] Specifically, when filtering data outliers according to the outlier threshold, it includes: comparing the data outliers with the outlier threshold and eliminating the difference data. When the data outlier is greater than the outlier threshold, the performance data corresponding to the data outlier is retained. When the data outlier is less than or equal to the outlier threshold, the performance data corresponding to the data outlier is determined as difference data and the performance data is eliminated.

[0078] Specifically, obtaining the outlier value of the filtered data based on the Gaussian mixture model and determining the abnormal data according to the outlier value include: calculating the outlier value of each filtered data based on the Gaussian mixture model:

[0079]

[0080] Among them, p(x|Θ) represents the outlier value of data x, K is the number of Gaussian components, π k Expressed as the mixing coefficient of the kth component, It is expressed as the probability density of data x belonging to a certain Gaussian component. When the outlier value of the filtered data is greater than 0.8, the data is determined to be abnormal data, and the network link corresponding to the abnormal data is identified.

[0081] It can be understood that the data outliers are screened through the anomaly threshold to eliminate insignificant data differences, thereby optimizing the accuracy of subsequent calculations. Specifically, the data outliers are compared with the anomaly threshold: when the data outlier is greater than the anomaly threshold, it is considered that the data may reflect a significant performance anomaly, and its corresponding performance data is retained; when the data outlier is less than or equal to the anomaly threshold, it is regarded as a regular fluctuation, and its corresponding performance data is judged as difference data and eliminated. The core of this screening process is to divide normal fluctuations and potential anomalies through anomaly thresholds to reduce the impact of interfering data on anomaly detection. Secondly, the filtered data is further analyzed based on the Gaussian mixture model to obtain the outliers of the data. The Gaussian mixture model is a probability density estimation method that models the data distribution and calculates the probability density of each data point belonging to different Gaussian components. In the formula, p(x|Θ) represents the outlier of the data x, which is obtained by weighted mixing of the probability densities of multiple Gaussian components. Among them, π k is the mixing coefficient of the kth Gaussian component, N (x∣μk,Σk) is the probability density function of the component. By modeling the probability distribution of the data, we can accurately capture the slight differences between the data and identify outliers. Finally, by setting the threshold of the outlier (such as 0.8), we can further determine the abnormal data in the filtered data. When the outlier value of the filtered data exceeds the set threshold, the data is judged as abnormal data and its corresponding network link is identified. This process combines the probability statistical model with the set threshold to ensure that the detected abnormal data has significant deviation characteristics, thereby improving the ability to accurately identify abnormal network links and providing a reliable basis for optimizing network acceleration strategies.

[0082] Step S500: remove the network links corresponding to each abnormal data, perform performance evaluation based on the performance data of the network links after the network links as abnormal data are removed, and determine the network links used for network acceleration strategy optimization based on the evaluation results.

[0083] Specifically, when the performance evaluation is performed based on the performance data of the network links after the network links as abnormal data are removed, it includes: establishing each network link association formula based on the performance data of each network link after the network links as abnormal data are removed. Obtaining the distance metric between each network link association formula based on the Euclidean distance, and constructing a distance matrix based on the distance metric. Iteratively clustering each network link association formula according to the distance matrix. Determine the performance data in the network link association formula after iterative clustering as the preset performance data. Determine the performance score of each network link based on the relationship between the performance data and the preset performance data.

[0084] It can be understood that the correlation formula of each network link is established by removing the abnormal data from the network link performance data. This correlation formula reflects the performance connection and characteristics between different network links. In this way, the performance data of each network link can be organized into a correlation system. The establishment of the correlation formula not only helps the integration and systematic management of data, but also provides a basis for subsequent distance measurement and clustering analysis. The correlation formula can be regarded as a collection of network link performance, which is convenient for subsequent clustering and performance evaluation. Next, the distance between the correlation formulas of each network link is measured based on the Euclidean distance, and the similarity between the correlation formulas is calculated. The Euclidean distance is the most common distance measurement method and is suitable for quantitative analysis of performance data in multiple dimensions. Here, by calculating the Euclidean distance between the network link correlation formulas, the differences in performance of different links can be revealed, and then which links have similar performance characteristics can be identified. This process quantifies the similarity between each link by constructing a distance matrix, providing data support for further clustering analysis. Finally, the network link correlation formulas are iteratively clustered through the distance matrix. The clustering algorithm will gradually group similar network link associations into a group based on the Euclidean distance. The iterative process can make network links with high similarity be classified into the same category. Through the clustered performance data, the preset performance data of each group of network links can be determined, and based on the relationship between these preset performance data and the actual performance data, the performance score of each network link can be further evaluated.

[0085] Specifically, according to the relationship between the performance data and the preset performance data, the performance score of each network link is determined, including:

[0086]

[0087] Among them, S is the performance score of the network link, N is the total number of indicator data, Hi is the weight coefficient of the i-th performance data, Di is the indicator data of the i-th performance data, and di is the indicator data of the i-th preset performance data.

[0088] Specifically, when determining the performance score of each network link, it also includes: obtaining the network performance requirements of the data to be transmitted, where the network performance requirements include communication delay requirements, communication bandwidth requirements, communication packet loss rate requirements, and communication network traffic requirements. Based on the relationship between the network performance requirements and the performance data, determine the similarity between them:

[0089]

[0090] Where Li is the similarity between the network performance requirement and the performance data, N is the total number of indicator data, and gi is the indicator data of the ith network performance requirement. According to the relationship between the similarity between the network performance requirement and the performance data of each network link and the preset similarity, the effective network link is determined.

[0091] Specifically, when determining the effective network link according to the relationship between the similarity between the network performance requirements and performance data of each network link and the preset similarity, it includes: determining whether the network link is an effective network link according to the relationship between the similarity and the preset similarity: when the similarity is less than the preset similarity, it is determined that the network link is not an effective network link. When the similarity is greater than or equal to the preset similarity, it is determined that the network link is an effective network link.

[0092] Specifically, when determining the network link used in the network acceleration strategy optimization according to the evaluation result, it includes: obtaining the performance score of each valid network link, and sorting the performance scores in reverse order. Obtain the performance score ranked first, and determine the valid network link corresponding to the performance score ranked first as the network link used in the network acceleration strategy optimization.

[0093] It can be understood that the performance score of each network link is obtained by adopting a calculation method based on the deviation between the performance data and the preset performance data. The performance scoring formula combines multiple performance indicators (such as communication delay, communication bandwidth, packet loss rate and network traffic), and adjusts the weight coefficient of each indicator to ensure the fairness and accuracy of the score. Through this scoring mechanism, the comprehensive performance of each network link can be comprehensively measured, providing a scientific basis for subsequent link selection. Secondly, in order to meet the network performance requirements of the data to be transmitted, this method further introduces a similarity calculation model. The similarity formula quantifies the degree of match between the indicator data of the network performance requirements and the actual link performance data by comparing them. When the similarity is higher than or equal to the preset similarity, the link is judged to be valid, which means that the link can meet the requirements of the current data transmission task. This process ensures that the implementation of the optimization strategy not only considers the link performance, but also pays full attention to the actual needs, thereby improving the practicality and adaptability of the scheme. Thirdly, on the basis of screening out the effective links, the link selection process is further optimized by sorting the performance scores of all effective links. By arranging the performance scores in reverse order, the network link with the highest score is preferentially selected as the main transmission path. The introduction of scoring sorting ensures that the final selected link not only meets the requirements, but also provides the best performance among the links that meet the requirements. This strategy effectively avoids resource waste and improves the efficiency of the network acceleration strategy. Finally, through the dual evaluation criteria based on performance scoring and similarity, this solution achieves the ability to dynamically adapt to changes in the network environment. Even under complex and changing network conditions, the optimal link can be accurately identified and used, avoiding the limitations caused by static rules. This method provides comprehensive technical support for the optimization of network acceleration strategies, while ensuring the stability and efficiency of data transmission.

[0094] In the above embodiment, by real-time monitoring and analysis of the performance data of the network link, combining historical data and Gaussian mixture model to screen and process abnormal data, unstable or abnormal links can be effectively identified and eliminated. This process can not only eliminate performance fluctuations caused by factors such as hardware failure, network congestion or external attacks, but also dynamically adjust the optimization strategy according to the actual network status to ensure that the network performance is always in the optimal state. By eliminating abnormal links, resources and traffic can be concentrated on links with more stable performance, thereby achieving more efficient resource utilization and data transmission. Secondly, by continuously tracking and analyzing the performance data of the network link, potential problems in the network can be identified in time and corresponding optimization measures can be taken. For example, when a burst load occurs in the network, the system can quickly identify abnormal indicators such as delay and bandwidth of the relevant links, and process the abnormal data to prevent these abnormal links from affecting the overall network performance. This real-time dynamic optimization method enables the network to adapt to complex and changeable working environments and provide continuous performance guarantees without relying on static rules or fixed configurations. In addition, by using Gaussian mixture models to process data outliers, not only the fault tolerance to data noise is enhanced, but also the accuracy of anomaly detection is improved. With the help of the model, links that have a negative impact on network performance can be more intelligently screened and identified, avoiding the misjudgment or missed judgment problems that may occur in traditional optimization methods. Ultimately, this network acceleration strategy based on data flow and intelligent algorithms can maximize network transmission efficiency, reduce latency, ensure business continuity, and provide users with a more stable and smooth network experience.

[0095] In another preferred embodiment based on the above embodiment, Figure 2 As shown, this embodiment provides a data flow-based network acceleration strategy optimization system, including: a network monitoring device and a monitoring module.

[0096] Specifically, the network monitoring device is electrically connected to each network link, and the network monitoring device is configured to collect performance data of each network link. The monitoring module is electrically connected to the network monitoring device, and the monitoring module includes a collection unit, a processing unit, a judgment unit, an adjustment unit and a storage unit. The collection unit is configured to obtain the performance data of each network link collected by the network monitoring device, and establish a data group according to the performance data, and the data group includes the performance data of each network link at the current moment and the performance data of each network link in the adjacent historical period. The collection unit is also configured to obtain the data abnormal value of each network link according to the data group. The processing unit is configured to obtain the historical performance data of each network link in the storage unit, and obtain the abnormal threshold of the data abnormal value according to the historical performance data. The judgment unit is configured to filter the data abnormal value according to the abnormal threshold, and obtain the abnormal value of the filtered data based on the Gaussian mixture model, and determine the abnormal data according to the abnormal value. The adjustment unit is configured to remove the network link corresponding to each abnormal data, perform performance evaluation according to the performance data of the network link after removing the network link as abnormal data, and determine the network link used for network acceleration strategy optimization according to the evaluation result.

[0097] It can be understood that the data flow-based network acceleration strategy optimization system and method in the above-mentioned embodiments of the present invention have the same beneficial effects and will not be described in detail.

[0098] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt 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 codes.

[0099] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, 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 processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0100] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0101] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing a network acceleration strategy based on data flow, characterized in that: include: Setting a network monitoring device on each network link and acquiring performance data of each network link, wherein the performance data includes communication delay, communication bandwidth, communication packet loss rate and communication network traffic of the network link; Establish a data group according to the performance data, the data group including the performance data of each of the network links at the current moment and the performance data of each of the network links in an adjacent historical period, and obtain data abnormal values ​​of each of the network links according to the data group; Acquire historical performance data of each of the network links, and acquire an abnormal threshold of the data abnormal value according to the historical performance data; Screening the data outliers according to the outlier threshold, acquiring outliers of the screened data based on a Gaussian mixture model, and determining the outliers according to the outliers; The network links corresponding to each abnormal data are eliminated, performance evaluation is performed based on the performance data of the network links after the network links as abnormal data are eliminated, and the network links used for network acceleration strategy optimization are determined based on the evaluation results.

2. The method for optimizing a network acceleration strategy based on data flow according to claim 1, characterized in that: When obtaining the abnormal threshold of the data abnormal value according to the historical performance data, it includes: The historical normal performance data in the historical performance data is obtained, and the abnormal threshold is obtained based on the relationship between the historical normal performance data and a formula, wherein the formula is as follows: Y=μW+k·σW; Among them, Y represents the abnormal threshold, μW represents the difference mean of historical normal performance data, σW represents the difference standard deviation of historical normal performance data, and k represents the adjustment factor, among which: Wt=|X t -X t-1 |; Where Wt represents the data difference value of historical normal performance data, X t represents the performance data of the network link at time t, X t-1 It represents the performance data of the network link at time t-1, and N represents the number of historical normal performance data.

3. The method for optimizing a network acceleration strategy based on data flow according to claim 2, characterized in that: When the data abnormal value is screened according to the abnormal threshold, it includes: Compare the data outliers with the outlier threshold and remove the difference data; When the data abnormal value is greater than the abnormal threshold, retain the performance data corresponding to the data abnormal value; When the data abnormal value is less than or equal to the abnormal threshold, the performance data corresponding to the data abnormal value is determined as difference data and the performance data is eliminated.

4. The method for optimizing a network acceleration strategy based on data flow according to claim 3, characterized in that: Obtaining outliers of the filtered data based on the Gaussian mixture model, and determining the outliers according to the outliers, including: Calculate the outlier value of each filtered data based on the Gaussian mixture model: Among them, p(x|Θ) represents the outlier value of data x, K is the number of Gaussian components, π k Expressed as the mixing coefficient of the kth component, It is expressed as the probability density of data x belonging to a certain Gaussian component; When the abnormal value of the filtered data is greater than 0.8, the data is determined to be the abnormal data, and the network link corresponding to the abnormal data is identified.

5. The method for optimizing a network acceleration strategy based on data flow according to claim 1, characterized in that: When the performance evaluation is performed based on the performance data of the network link after the network link as abnormal data is removed, it includes: Establishing a correlation formula for each network link according to the performance data of each network link after eliminating the network link as abnormal data; Acquire a distance metric between each of the network link association equations based on the Euclidean distance, and construct a distance matrix according to the distance metric; Iteratively clustering each of the network link association equations according to the distance matrix; Determine the performance data in the network link association formula after iterative clustering as preset performance data; A performance score of each of the network links is determined according to a relationship between the performance data and the preset performance data.

6. The method for optimizing a network acceleration strategy based on data flow according to claim 5, characterized in that: Determining the performance score of each of the network links according to the relationship between the performance data and the preset performance data includes: Among them, S is the performance score of the network link, N is the total number of indicator data, Hi is the weight coefficient of the i-th performance data, Di is the indicator data of the i-th performance data, and di is the indicator data of the i-th preset performance data.

7. The method for optimizing a network acceleration strategy based on data flow according to claim 6, characterized in that: When determining the performance score of each of the network links, it also includes: Obtaining network performance requirements for data to be transmitted, wherein the network performance requirements include communication delay requirements, communication bandwidth requirements, communication packet loss rate requirements, and communication network traffic requirements; Determine the similarity between the network performance requirement and the performance data according to the relationship between them: Where Li is the similarity between the network performance requirement and the performance data, N is the total number of indicator data, and gi is the indicator data of the i-th network performance requirement; The effective network link is determined according to the relationship between the similarity between the network performance requirement of each network link and the performance data and a preset similarity.

8. The method for optimizing a network acceleration strategy based on data flow according to claim 7, characterized in that: Determining the effective network link according to the relationship between the similarity between the network performance requirement of each network link and the performance data and a preset similarity includes: Determining whether the network link is the valid network link according to the relationship between the similarity and the preset similarity; When the similarity is less than the preset similarity, determining that the network link is not the valid network link; When the similarity is greater than or equal to the preset similarity, the network link is determined to be the valid network link.

9. The method for optimizing a network acceleration strategy based on data flow according to claim 8, characterized in that: When determining the network link to be used for network acceleration policy optimization based on the evaluation results, the following are included: Obtaining a performance score for each of the valid network links, and sorting the performance scores in descending order; The performance score ranked first is obtained, and the valid network link corresponding to the performance score ranked first is determined as the network link used when optimizing the network acceleration strategy.

10. A data flow-based network acceleration strategy optimization system, used in a data flow-based network acceleration strategy optimization method as described in any one of claims 1 to 9, characterized in that: include: A network monitoring device, electrically connected to each network link, and configured to collect performance data of each network link; The monitoring module is electrically connected to the network monitoring device, and the monitoring module includes a collection unit, a processing unit, a judgment unit, an adjustment unit and a storage unit; The collection unit is configured to obtain the performance data of each of the network links collected by the network monitoring device, and establish a data group according to the performance data, wherein the data group includes the performance data of each of the network links at the current moment and the performance data of each of the network links in an adjacent historical period; the collection unit is also configured to obtain data abnormal values ​​of each of the network links according to the data group; The processing unit is configured to obtain historical performance data of each of the network links in the storage unit, and obtain an abnormal threshold of the data abnormal value according to the historical performance data; The judgment unit is configured to filter the data abnormal values ​​according to the abnormal threshold, obtain the abnormal values ​​of the filtered data based on the Gaussian mixture model, and determine the abnormal data according to the abnormal values; The adjustment unit is configured to remove the network links corresponding to each abnormal data, perform performance evaluation based on the performance data of the network links after removing the network links as abnormal data, and determine the network links used when optimizing the network acceleration strategy based on the evaluation results.

Images