[go: up one dir, main page]

CN118200166A - Network quality evaluation method, device, equipment, storage medium and program product - Google Patents

Network quality evaluation method, device, equipment, storage medium and program product Download PDF

Info

Publication number
CN118200166A
CN118200166A CN202211590766.0A CN202211590766A CN118200166A CN 118200166 A CN118200166 A CN 118200166A CN 202211590766 A CN202211590766 A CN 202211590766A CN 118200166 A CN118200166 A CN 118200166A
Authority
CN
China
Prior art keywords
network quality
network
industrial
quality
target index
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211590766.0A
Other languages
Chinese (zh)
Inventor
鲁效平
魏永强
高明亮
马正中
王超
孙琦
王迷珍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Cosmoplat Industrial Intelligent Research Institute Qingdao Co Ltd
Haier Cosmo IoT Technology Co Ltd
Original Assignee
Cosmoplat Industrial Intelligent Research Institute Qingdao Co Ltd
Haier Cosmo IoT Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Cosmoplat Industrial Intelligent Research Institute Qingdao Co Ltd, Haier Cosmo IoT Technology Co Ltd filed Critical Cosmoplat Industrial Intelligent Research Institute Qingdao Co Ltd
Priority to CN202211590766.0A priority Critical patent/CN118200166A/en
Priority to PCT/CN2023/116310 priority patent/WO2024124975A1/en
Publication of CN118200166A publication Critical patent/CN118200166A/en
Pending legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/142—Network analysis or design using statistical or mathematical methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0852—Delays
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • Algebra (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Pure & Applied Mathematics (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The application belongs to the technical field of communication, and particularly relates to a network quality assessment method, a device, equipment, a storage medium and a program product. The application aims to solve the problem of inaccurate network quality assessment caused by classification by adopting an empirical threshold when predicting network quality. The network quality evaluation method provided by the application is applied to network quality evaluation equipment, and comprises the following steps: collecting target index parameters of the industrial network; evaluating a network quality level of the industrial network based on the target index parameter and the quality evaluation model; the network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance. The accuracy of grading is improved through the flexibility and self-adaption grading of the network quality grade, the network quality is evaluated based on the quality evaluation model, and the accuracy of quality evaluation is improved.

Description

Network quality evaluation method, device, equipment, storage medium and program product
Technical Field
The application belongs to the technical field of communication, and particularly relates to a network quality assessment method, a device, equipment, a storage medium and a program product.
Background
With the continuous development of internet technology, the internet service of enterprises is more and more abundant. Network service quality becomes the core competitiveness of internet enterprises.
In order to improve the quality of service of the internet, the network quality needs to be evaluated. In the related art, most of the methods adopt an empirical threshold mode to evaluate the network quality, the accuracy of determining the empirical threshold is poor, and no reliable data basis exists, so that the accuracy of evaluating the network quality is poor.
Therefore, a high-precision network quality assessment method is needed.
Disclosure of Invention
In order to solve the above problems in the prior art, that is, in order to solve the problem of low accuracy of the existing network quality assessment, the present application provides a network quality assessment method, apparatus, device, storage medium and program product, which perform network quality classification by fuzzy clustering and perform network quality assessment based on a model, thereby improving accuracy of network quality classification and network quality assessment
In a first aspect, an embodiment of the present application provides a network quality assessment method, where the method includes:
Collecting target index parameters of an industrial network;
evaluating a network quality level of the industrial network based on the target index parameter and the quality evaluation model;
The network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance.
Optionally, the method further comprises:
aiming at each index parameter in a plurality of index parameters of an industrial network, calculating the association degree of the index parameter and the service response success rate;
and determining a target index parameter from the plurality of index parameters based on the association degree.
Optionally, calculating the association degree between the index parameter and the service response success rate includes:
And calculating the association degree of the index parameter and the service response success rate based on the Pearson correlation coefficient.
Optionally, the quality assessment model is a model based on an extreme gradient lifting tree XGBoost algorithm; obtaining a network quality level of the industrial network based on the target index parameter and the quality assessment model, including:
Calculating the average value of all the target index parameters acquired in a preset time period;
and inputting the average value of each target index parameter into the quality evaluation model to obtain the network quality grade of the industrial network.
Optionally, the fuzzy C-means clustering analysis algorithm searches an initial clustering center based on Xie-Beni indexes, and an objective function of the fuzzy C-means clustering analysis algorithm is a function of Euclidean distance between membership degree and service response success rate and network quality level; the membership is used for representing the degree that the service response success rate belongs to the set network quality grade.
Optionally, the method further comprises:
Predicting a network quality level of the industrial network at a future time based on a network quality level sequence and a pre-trained prediction model; wherein the network quality class sequence is a sequence of network quality classes of the industrial network assessed over time.
Optionally, the target indicator parameter includes at least one of a first HTTP packet response delay, a TCP setup link acknowledgement delay, and a last HTTP packet delay.
In a second aspect, an embodiment of the present application further provides a network quality assessment apparatus, where the apparatus includes:
the index acquisition module is used for acquiring target index parameters of the industrial network;
The quality grade evaluation module is used for evaluating the network quality grade of the industrial network based on the target index parameter; the network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance.
Optionally, the apparatus further includes:
The association degree calculation module is used for calculating association degree of each index parameter in a plurality of index parameters of the industrial network and the service response success rate;
and the target index parameter determining module is used for determining target index parameters from the plurality of index parameters based on the association degree.
Optionally, the association degree calculating module is specifically configured to:
aiming at each index parameter in a plurality of index parameters of an industrial network, calculating the association degree of the index parameter and the service response success rate based on the Pearson correlation coefficient.
Optionally, the quality assessment model is a model based on an extreme gradient lifting tree XGBoost algorithm; a quality class assessment module comprising:
Calculating the average value of all the target index parameters acquired in a preset time period; and inputting the average value of each target index parameter into the quality evaluation model to obtain the network quality grade of the industrial network.
Optionally, the apparatus further includes:
A quality level prediction module for predicting a network quality level at a future time of the industrial network based on a network quality level sequence and a pre-trained prediction model; wherein the network quality class sequence is a sequence of network quality classes of the industrial network assessed over time.
In a third aspect, an embodiment of the present application further provides a network quality assessment device, including: a memory and at least one processor;
the memory stores computer-executable instructions;
The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the network quality assessment method as provided by any embodiment corresponding to the first aspect of the present application.
In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, where computer executable instructions are stored, and when a processor executes the computer executable instructions, the network quality assessment method provided in any embodiment corresponding to the first aspect of the present application is implemented.
In a fifth aspect, embodiments of the present application further provide a computer program product, including a computer program, which when executed by a processor implements a network quality assessment method as provided in any of the embodiments corresponding to the first aspect of the present application.
As can be appreciated by those skilled in the art, in order to improve accuracy of network quality assessment, the network quality assessment method, device, equipment, storage medium and program product provided by the embodiments of the present application divide network quality levels of an industrial network according to service response success rates based on fuzzy C-means clustering analysis algorithm in advance, and implement flexible fuzzy division of network quality, and compared with a mode of threshold division, the method has high accuracy and wide application range; and carrying out network quality evaluation on the industrial network through the acquired target index parameters and the quality evaluation model of the industrial network to obtain the estimated network quality grade, thereby improving the accuracy of network quality grade estimation.
Drawings
Preferred embodiments of the network quality assessment method, apparatus, device, storage medium and program product of the present application are described below with reference to the accompanying drawings. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the application. The attached drawings are as follows:
fig. 1 is an application scenario diagram of a network quality evaluation method according to an embodiment of the present application;
FIG. 2 is a flow chart of a network quality assessment method provided by one embodiment of the present application;
FIG. 3 is a flow chart of a network quality assessment method according to another embodiment of the present application;
Fig. 4 is a schematic structural diagram of a network quality assessment device according to an embodiment of the present application;
Fig. 5 is a schematic structural diagram of a network quality assessment device according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions in the embodiments of the present application will be clearly and completely described in the following in conjunction with the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The following describes the technical scheme of the present application and how the technical scheme of the present application solves the above technical problems in detail with specific embodiments. The following embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. Embodiments of the present application will be described below with reference to the accompanying drawings.
The following explains the application scenario of the embodiment of the present application:
Fig. 1 is an application scenario diagram of a network quality evaluation method provided by an embodiment of the present application, where, as shown in fig. 1, an enterprise performs data interaction through an industrial internet (or referred to as an industrial network), for example, interaction between an enterprise server and user equipment (such as a washing machine, a refrigerator, an air conditioner, a television, a computer, a mobile phone, etc.), interaction between enterprise servers or enterprise equipment deployed in different areas corresponding to the enterprise, etc. Taking 3 enterprise servers as an example in fig. 1, the number of enterprise servers and deployment locations may be configured by specific needs.
In order to provide a user with a good quality of service, it is necessary to ensure the network quality of the industrial internet. In performing quality assessment of an industrial internet network, key performance indicators (KPIs, key Performance Indicator) of the industrial internet, such as service response times, are typically extracted, and a network quality level is determined based on a comparison of the key performance indicators with an empirical threshold.
However, because the service scene of the industrial internet is complex, the index difference is large under different services, and the network quality under different services cannot be estimated by adopting the same index. Moreover, the setting of the experience threshold value has no sufficient data base, so that the accuracy of network quality assessment is poor.
Aiming at the problems, the main conception of the network quality assessment method provided by the embodiment of the application is as follows: dividing network quality grades based on FCM (Fuzzy C-Means) algorithm in advance, wherein compared with an empirical threshold dividing mode, the grade dividing accuracy is higher, and the application range is wide; and then, based on the collected multiple target index parameters and the designed quality evaluation model, the quality evaluation of the industrial network model is carried out, the network quality grade is output, the model classification accuracy is higher, and the accuracy of the network quality evaluation is improved.
Fig. 2 is a flowchart of a network quality assessment method according to an embodiment of the present application, where the network quality assessment method according to the embodiment of the present application is applied to a network quality assessment device, and may be a server, a computer or other devices with corresponding data processing capabilities, as shown in fig. 2, and the network quality assessment method includes the following steps:
step S201, collecting target index parameters of the industrial network.
The target index parameter may be determined based on the traffic type of the industrial network. By way of example, traffic types may include high traffic types, real-time traffic types, and the like.
The target index parameter may be extracted from source data of the industrial network, which may include observable data such as log data, stream data, etc. of the industrial network.
The method can collect source data of the industrial network according to a certain period, preprocess the collected source data and extract various target index parameters in the preprocessed source data.
The preprocessing that is undertaken on the source data may include at least one of regularization, normalization, feature selection, random downsampling (Random Under Sampling, RUS), random oversampling (Random Over Sampling, ROS), synthetic minority class oversampling techniques (SYNTHETIC MINORITY OVER-sampling Technique, SMOTE), and the like preprocessing operations.
Based on the arrangement and combination of the preprocessing operations, a plurality of preprocessing scenes can be obtained, the accuracy corresponding to each preprocessing scene is determined through experiments, the preprocessing operation corresponding to the preprocessing scene with the highest accuracy and the sequence thereof are selected, and the preprocessing of the source data is performed. The preprocessing scenario may include various scenarios such as no preprocessing scenario, regularization and feature selection scenario, regularization and RUS scenario, feature selection and RUS scenario, normalization and ROS scenario, feature selection and RUS scenario, normalization and SMOTE scenario, feature selection and SMOTE scenario, normalization, feature selection and RUS scenario, normalization, feature selection and ROS scenario, normalization, feature selection and SMOTE scenario, and so forth. The naming of the preprocessing scenario is named based on the preprocessing operation employed by the preprocessing scenario.
In some embodiments, the preprocessing operations performed on the source data sequentially comprise: regularization processing, feature selection, and SMOTE.
Optionally, the target metrics include at least one of a first HTTP (Hyper Text Transfer Protocol ) packet Response delay (FIRST HTTP Response Time, FHRT), a TCP (Transmission Control Protocol ) set-up link acknowledgement delay (TCP ACK TIME, TAT), and a Last HTTP packet delay (Last Content PACKET TIME, LCPT).
Step S202, evaluating the network quality level of the industrial network based on the target index parameter and the quality evaluation model.
The network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance.
Specifically, the obtained target index parameters or the preprocessed target index parameters can be input into a quality evaluation model, and the quality evaluation model is used for carrying out feature processing and analysis on the target index parameters to output the network quality grade of the corresponding industrial network.
The fuzzy C-means clustering analysis algorithm is a flexible fuzzy dividing method, each subset of the clusters is regarded as a fuzzy set, the membership degree of each sample (service response success rate) relative to the fuzzy set is calculated, and therefore the cluster of the sample, namely the network quality level, is determined, and the clustering mode is accurate and flexible.
Optionally, the fuzzy C-means clustering analysis algorithm searches an initial clustering center based on Xie-Beni indexes, and the objective function of the fuzzy C-means clustering analysis algorithm is a function of Euclidean distance of membership degree, service response success rate and network quality level. The membership is used for representing the degree that the service response success rate belongs to the set network quality grade.
In some embodiments, the objective function of the fuzzy C-means clustering algorithm may be a function of the power p of membership and the power p of sample distance, which may be expressed in terms of Euclidean distance of the samples (traffic response success rate) relative to the network quality level. Wherein p is a fuzzy measure, and the value of p can be 2.
The specific process for dividing the network quality grade of the industrial network according to the service response success rate based on the fuzzy C-means clustering analysis algorithm comprises the following steps:
Firstly, acquiring a data set of an industrial network; the dataset includes a plurality of business response success rates for the industrial network acquired at historical times, and exemplary dataset D BISR may be represented as: d BISR＝{BISR1,BISR2,...,BISRm }, where BISR i is the service response success rate obtained at the ith time point.
The data in the data set may be collected by a sensor or other data collection device, and may also be obtained by a third party data source.
And initializing relevant parameters of a fuzzy C-means clustering analysis algorithm, wherein the relevant parameters comprise fuzzy measure p, cluster number C, iteration times t, convergence accuracy epsilon, maximum cluster number and other parameters C max.
Illustratively, the initial value of the blur measure p may be 2 and the initial value of the cluster number c may be 2.
Third, in the loop phase, c is self-increased by 1 after each loop iteration is finished, and when c < c max, the cluster center is initialized based on the subtractive clustering of the dataset D BISR.
And fourthly, calculating the membership degree of each service response success rate in the data set D BISR relative to the clustering center, and obtaining a membership degree matrix corresponding to the iteration.
The membership matrix U (t) corresponding to the t-th iteration can be expressed as:
Wherein, the element U ij in the ith row and the jth column of the membership matrix U (t) represents membership of the jth service response success rate in the dataset D BISR to the ith network quality class. v j is the j-th cluster center (or network quality level). i is a positive integer less than or equal to m, and j is a positive integer less than or equal to n.
In some embodiments, the membership value is between 0 and 1, when the membership is 0, it indicates that the service response success rate does not belong to the corresponding network quality level, and when the membership is 1, it indicates that the service response success rate completely belongs to the corresponding network quality level.
And fifthly, calculating the deviation between the membership matrix obtained by the current iteration and the membership matrix obtained by the last iteration, wherein the Euclidean distance between the two matrices can be used for representing the deviation, and if the deviation is larger than convergence precision epsilon, updating the cluster center of the dataset DBISR based on the membership matrix obtained by the current iteration.
And sixthly, calculating a value (marked as XB) of the effectiveness index Xie_Beni, and if the latest obtained XB is larger than the XB obtained in the previous iteration, updating the clustering result into a cluster number, a cluster center and a membership matrix corresponding to the current iteration number.
And adding 1 to the iteration number, and carrying out one round of iteration, and sequentially cycling until the iteration number reaches the maximum iteration number. And obtaining a classification result of the network quality level based on the cluster number, the cluster center and the membership matrix in the latest obtained clustering result.
After fuzzy classification is carried out on the industrial network of the user to obtain a plurality of network quality grades, a quality assessment model is trained based on the network quality grade corresponding to each service response success rate in the data set and the target index parameter, and the trained quality assessment model is obtained.
And evaluating the current network quality of the industrial network based on the currently acquired target index parameters of the industrial network and the trained quality evaluation model to obtain the network quality grade.
In some embodiments, the quality assessment model may be a model constructed based on XGBoost classification algorithms.
In other embodiments, the quality assessment model may be a neural network model, such as a convolutional neural network model, a recurrent neural network model, or the like.
In order to improve accuracy of network quality assessment, the network quality grade of the industrial network is divided according to the service response success rate based on the fuzzy C-means clustering analysis algorithm in advance, and flexible fuzzy division of network quality is achieved; and carrying out network quality evaluation on the industrial network through the acquired target index parameters and the quality evaluation model of the industrial network to obtain the estimated network quality grade, thereby improving the accuracy of network quality grade estimation.
Fig. 3 is a flowchart of a network quality assessment method according to another embodiment of the present application, in which step S202 is further refined based on the embodiment shown in fig. 2, a correlation step of determining a target index parameter is added before step S201, and a correlation step of predicting a network quality level is added after step S202. As shown in fig. 3, the network quality evaluation method provided in this embodiment includes the following steps:
Step S301, calculating the association degree between each index parameter and the service response success rate aiming at each index parameter in a plurality of index parameters of the industrial network.
The plurality of index parameters of the industrial network may include: the parameters of the TCP link disassembly time or the time of receiving the last data packet of UDP (User Datagram Protocol ), the uplink flow, the downlink flow, the TCP link establishment confirmation time delay, the TCP connection state indication, the first HTTP response packet time delay, the time delay of the last HTTP content packet, the ACK (Acknowledge character, confirmation character) time delay of the last HTTP content packet, the downlink duration time, the transaction type of HTTP/WAP2.0, the HTTP/WAP transaction state, the number of uplink TCP retransmission messages, the number of uplink IP packets, the number of downlink TCP retransmission messages, the number of downlink IP packets and the like. Session identification, internet content Provider (Internet Content Provider, ICP), internet service Provider (INTERNET SERVICE Provider, ISP), service name (which may include a service class name and a service subclass name), end user address, user location, user class, access server address, access domain name, flow to class, flow to subclass, etc.
The association degree of the index parameter and the service response success rate is used for representing the degree of influence of the index parameter on the service response success rate. The higher the association, the higher the impact on the success rate of service response.
In some embodiments, the association of the index parameter with the traffic response success rate may be calculated based on an average of the index parameter over a period of time.
The method can collect various index parameters of the industrial network in a set period according to the set period, and perform average operation to obtain an average value of the various index parameters.
Optionally, calculating the association degree between the index parameter and the service response success rate includes:
and calculating the association degree of the index parameter and the service response success rate based on the Pearson (Pearson) correlation coefficient.
The larger the absolute value of the pearson correlation coefficient corresponding to the index parameter is, the higher the correlation degree between the index parameter and the service response success rate is.
Step S302, determining a target index parameter from the plurality of index parameters based on the association degree.
The preset number of index parameters with higher association degree can be determined as target index parameters, and the preset number can be 3, 5 or other values. And the index parameter with the association degree higher than the preset association degree can be determined as the target index parameter, and the preset quantity and the value of the preset association degree can be configured by user definition.
Step S303, collecting target index parameters of the industrial network.
The acquisition of target index parameters of the industrial network can be performed according to a fixed period, so that the periodic network quality evaluation is realized. The fixed period may be determined based on service attributes of the industrial network, which may include parameters such as service type, service traffic, etc.
In order to improve accuracy of quality assessment, preprocessing is needed to be performed on the target index parameter of the current period, and the preprocessing can sequentially comprise regularization processing, feature selection processing and SMOTE processing.
Step S304, calculating an average value of each target index parameter acquired in a preset time period.
The preset time period may be a time period corresponding to the above-mentioned fixed period, for example, 1 hour, 4 hours, 6 hours, or other time periods.
And aiming at each target index parameter, carrying out averaging treatment on the target index parameter acquired in a preset time period to obtain an average value of the target index parameter.
Step S305, inputting the average value of each target index parameter into the quality assessment model to obtain the network quality level of the industrial network.
Through the reasonable setting of the fixed period, the periodic evaluation of the network quality can be realized, and the evaluation requirements of various services can be met through the flexible setting of the period, so that the evaluation mode is flexible and the application range is wide. The average value is used as the input of the quality evaluation model, so that the influence of individual values on network quality evaluation is avoided, and the accuracy of network quality evaluation is improved.
The quality assessment model may be a XGBoost algorithm-based model, and the XGBoost algorithm is an integrated algorithm based on a tree or linear classifier, and comprises a plurality of weak classifiers, so that a strong classifier with a good classification effect or regression effect is formed. The regular term of the objective function of the quality assessment model contains the weights of the leaf nodes and the depth of the tree, so that the complexity of the model is effectively controlled, and overfitting is prevented. The objective function of the quality evaluation model adopts a second-order Taylor expansion approximation, so that the evaluation accuracy is improved.
And a quality evaluation model is constructed based on XGBoost algorithm, the advantage of XGBoost algorithm classification is utilized, and the accuracy of quality evaluation is improved.
In some embodiments, the target indicator parameters include FHRT, LCPT, and TAT. And combining the average values of the preprocessed FHRT, LCPT and TAT to form triples { u (FHRTi), u (LCPTi), u (TATI) }, wherein u () is an averaging function for averaging elements in brackets, FHRTi is FHRT acquired in the ith period, LCPTi is LCPT acquired in the ith period, and TATI is TAT acquired in the ith period. The triplet is used as an input X of the quality assessment model, so that the corresponding network quality level Y (or Leveli) is output through the quality assessment model.
The basic idea of the quality assessment model training process is to gradually add decision trees to the quality assessment model to minimize the output value of the objective function, and form a strong classifier, i.e. the quality assessment model, from the constructed multiple decision trees. The assigned weights for the leaf nodes of each decision tree correspond to the evaluated network quality level. The construction of each decision tree is iterated based on the previous decision tree, the gain value of each leaf node is calculated each time the decision tree is constructed, and the leaf node with the highest gain value is selected for segmentation. When the gain value of the splitting leaf node is smaller than 0 or the depth of the tree reaches the set depth, the decision tree stops splitting, and the optimization of the decision tree structure is completed, so that a final quality evaluation model is obtained.
In order to avoid the problem of degradation of classification performance of the quality assessment model due to unbalanced data, during the training of the quality assessment model, the maximum step length of weight change of each tree needs to be controlled and the weight of each tree needs to be adjusted in the iterative process.
The maximum step size of the weights is used for limiting the number of iterations and loops, and the corresponding maximum step size of each tree can be the same.
In some embodiments, the objective function of the quality assessment model includes a loss function, which may be a Softmax function, and a regularization term. The regular term contains the weights of the leaf nodes and the depth of the tree. The regular term can control the complexity of the tree, avoiding overfitting.
After obtaining the evaluated network quality level, the network quality level may be sent or displayed, and when the network quality level is lower than a preset level, a prompt message may be generated, and the preset level may be set by a user or a default value may be adopted.
Step S306 predicts a network quality level at a future time of the industrial network based on the network quality level sequence and a pre-trained prediction model.
Wherein the network quality class sequence is a sequence of network quality classes of the industrial network assessed over time.
The network quality level has strong regularity in a short time, so that the prediction of the network quality level in the future time or the next period can be performed based on the time series composed of the network quality levels estimated in the history time.
A time series may be constructed using a sliding time window with a window width w, where w consecutive network quality levels assessed based on a quality assessment model corresponding to the sliding time window are included as input to a prediction model, and a network quality level corresponding to the next time (e.g., w+1 time) is output based on the prediction model.
In some embodiments, the prediction model may be a XGBoost algorithm-based model, and may also be other models for time-series prediction, such as a neural network model.
Taking a prediction model constructed based on XGBoost algorithm as an example, the expression of the prediction model is as follows: f m is the function corresponding to the mth regression tree, M is the total number of regression trees in the prediction model,/> Network quality levels predicted for the predictive model.
The training process of the prediction model mainly comprises the steps of constructing M regression trees by gradually optimizing an objective function, and combining the M regression trees into the prediction model.
The objective function of the predictive model may include a regularization term that measures the complexity of the regression tree, including the score of the leaf nodes, the number of leaf nodes, etc., and a loss function, which may be a square loss function.
Illustratively, the objective function of the predictive model may be:
Where the regularization term Ω (f m)＝γT+0.5λ||ω||2, T is the number of leaf nodes, ω represents the fraction of leaf nodes, γ and λ are weighting coefficients, The i-th network quality level output by the prediction model is Y i/>Corresponding to the actual network quality level.
The optimization process of the objective function is as follows: after the initial objective function is obtained, replacing the first derivative and the second derivative of the loss function with the loss function in the objective function according to the second-order Taylor expansion, and calculating the optimal weight of the leaf node when the objective function takes the minimum value. Based on a greedy algorithm, starting from a single leaf node, iteratively adding branches into a tree where the leaf node is located, splitting the leaf node, obtaining a split loss function, calculating the value of the loss function until the loss function after the iteration is iterated until the split loss function meets a preset condition, such as less than 0 or the iteration number reaches a maximum depth value, and finishing training to obtain a prediction model.
In some embodiments, the prompt information may also be generated based on the predicted network quality level and the network requirement of the industrial network at the corresponding time, for example, the network quality level is lower, and the requirement of the user access cannot be met.
Specifically, a threshold value of the network quality level may be determined based on network demands of the industrial network in a time period corresponding to the predicted network quality level, and when the predicted network quality level is lower than the corresponding threshold value, the prompt information is generated.
Further, resource scheduling can be performed according to the predicted network quality level, so as to meet the network requirements of the industrial network and avoid resource waste.
In this embodiment, the target index parameter for performing network quality evaluation is determined by the correlation between the index parameter and the BISR, so that the adaptivity and accuracy of index setting are improved, and the method has a wider application range than the method adopting the default index; the network quality grade evaluation is carried out based on the average value of the target index parameter in a period of time, and the influence of individual values on the accuracy of the network quality evaluation is avoided on the basis of reducing the data processing capacity; the method also provides a time sequence formed by network quality grades based on historical evaluation, and carries out a strategy for predicting the network quality grades of the industrial network in future time, so that related personnel can predict the network quality grades in advance to carry out resource scheduling in advance, the influence on the use experience of users due to poor network quality is avoided, and the timeliness and the accuracy of the quality control of the industrial network are improved.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the method embodiments described above may be performed by hardware associated with program instructions. The foregoing program may be stored in a computer-readable storage medium. The program, when executed, performs steps including the method embodiments described above; and the aforementioned storage medium includes: various media that can store program code, such as ROM, RAM, magnetic or optical disks.
Fig. 4 is a schematic structural diagram of a network quality assessment device according to an embodiment of the present application, as shown in fig. 4, where the network quality assessment device includes: an index collection module 410 and a quality level assessment module 420.
Wherein, the index acquisition module 410 is configured to acquire target index parameters of the industrial network; the quality level assessment module 420 assesses the network quality level of the industrial network based on the target indicator parameter.
Optionally, the apparatus further includes:
The association degree calculation module is used for calculating association degree of each index parameter in a plurality of index parameters of the industrial network and the service response success rate;
and the target index parameter determining module is used for determining target index parameters from the plurality of index parameters based on the association degree.
Optionally, the association degree calculating module is specifically configured to:
aiming at each index parameter in a plurality of index parameters of an industrial network, calculating the association degree of the index parameter and the service response success rate based on the Pearson correlation coefficient.
Optionally, the quality assessment model is a model based on an extreme gradient lifting tree XGBoost algorithm; a quality class assessment module comprising:
Calculating the average value of all the target index parameters acquired in a preset time period; and inputting the average value of each target index parameter into the quality evaluation model to obtain the network quality grade of the industrial network.
Optionally, the apparatus further includes:
A quality level prediction module 430 for predicting a network quality level at a future time of the industrial network based on a network quality level sequence and a pre-trained prediction model; wherein the network quality class sequence is a sequence of network quality classes of the industrial network assessed over time.
The network quality assessment device provided by the application can execute the network quality assessment method provided by any embodiment of the application, and has the corresponding functional modules and beneficial effects of the execution method.
Fig. 5 is a schematic structural diagram of a network quality assessment device according to an embodiment of the present application, and as shown in fig. 5, the network quality assessment device includes: a memory 510, and at least one processor 520.
Wherein the memory 510 stores computer-executable instructions and the at least one processor 520 executes the computer-executable instructions stored by the memory 510 such that the at least one processor 520 performs the network quality assessment method as provided by any of the embodiments corresponding to fig. 2-3.
Wherein memory 510 and processor 520 are coupled by bus 530.
The description may be understood correspondingly with reference to the description and effects corresponding to the steps of fig. 2-3, and will not be repeated here.
The present application also provides a computer readable storage medium, in which computer executable instructions are stored, which when executed by a processor, implement the network quality assessment method provided in any of the above embodiments.
The application also provides a computer program product comprising a computer program which, when executed by a processor, implements the network quality assessment method provided by any embodiment of the application.
In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules is merely a logical function division, and there may be additional divisions of actual implementation, e.g., multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.
The modules illustrated as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module. The integrated modules may be implemented in hardware or in hardware plus software functional modules.
The integrated modules, which are implemented in the form of software functional modules, may be stored in a computer readable storage medium. The software functional module is stored in a storage medium, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (english: processor) to perform some steps of the methods of the embodiments of the application. And the aforementioned storage medium includes: u disk, mobile hard disk, read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: it will be readily appreciated by those skilled in the art that the scope of the application is obviously not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present application, and such modifications and substitutions will be within the scope of the present application.

Claims (11)

1. A method for evaluating network quality, comprising:
Collecting target index parameters of an industrial network;
evaluating a network quality level of the industrial network based on the target index parameter and the quality evaluation model;
The network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance.
2. The method according to claim 1, wherein the method further comprises:
aiming at each index parameter in a plurality of index parameters of an industrial network, calculating the association degree of the index parameter and the service response success rate;
and determining a target index parameter from the plurality of index parameters based on the association degree.
3. The method of claim 2, wherein calculating the association of the indicator parameter with a traffic response success rate comprises:
And calculating the association degree of the index parameter and the service response success rate based on the Pearson correlation coefficient.
4. The method of claim 1, wherein the target metrics include at least one of a first HTTP packet response delay, a TCP setup link acknowledgement delay, and a last HTTP packet delay.
5. The method of claim 1, wherein deriving a network quality level for the industrial network based on the target metric parameters and a quality assessment model comprises:
Calculating the average value of all the target index parameters acquired in a preset time period;
and inputting the average value of each target index parameter into the quality evaluation model to obtain the network quality grade of the industrial network.
6. The method of claim 1, wherein the fuzzy C-means cluster analysis algorithm finds an initial cluster center based on Xie-Beni index, and the objective function of the fuzzy C-means cluster analysis algorithm is a function of euclidean distance with respect to membership and service response success rate and network quality level;
The membership is used for representing the degree that the service response success rate belongs to the set network quality grade.
7. The method according to any one of claims 1-6, further comprising:
Predicting a network quality level of the industrial network at a future time based on a network quality level sequence and a pre-trained prediction model;
Wherein the network quality class sequence is a sequence of network quality classes of the industrial network assessed over time.
8. A network quality assessment apparatus, comprising:
the index acquisition module is used for acquiring target index parameters of the industrial network;
The quality grade evaluation module is used for evaluating the network quality grade of the industrial network based on the target index parameter;
The network quality grades are divided according to service response success rates based on a fuzzy C-means clustering analysis algorithm in advance.
9. A network quality assessment apparatus, comprising: a memory and at least one processor;
the memory stores computer-executable instructions;
the at least one processor executing computer-executable instructions stored in the memory causes the at least one processor to perform the network quality assessment method of any one of claims 1-7.
10. A computer readable storage medium having stored therein computer executable instructions which, when executed by a processor, implement the network quality assessment method of any of claims 1-7.
11. A computer program product comprising a computer program, which when executed by a processor implements the network quality assessment method according to any one of claims 1-7.
CN202211590766.0A 2022-12-12 2022-12-12 Network quality evaluation method, device, equipment, storage medium and program product Pending CN118200166A (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202211590766.0A CN118200166A (en) 2022-12-12 2022-12-12 Network quality evaluation method, device, equipment, storage medium and program product
PCT/CN2023/116310 WO2024124975A1 (en) 2022-12-12 2023-08-31 Network quality evaluation method, apparatus and device, and storage medium and program product

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211590766.0A CN118200166A (en) 2022-12-12 2022-12-12 Network quality evaluation method, device, equipment, storage medium and program product

Publications (1)

Publication Number Publication Date
CN118200166A true CN118200166A (en) 2024-06-14

Family

ID=91411234

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211590766.0A Pending CN118200166A (en) 2022-12-12 2022-12-12 Network quality evaluation method, device, equipment, storage medium and program product

Country Status (2)

Country Link
CN (1) CN118200166A (en)
WO (1) WO2024124975A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118802615A (en) * 2024-07-02 2024-10-18 中国移动通信集团山西有限公司 Private network quality assessment method, device, equipment, medium and computer program product

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119421177A (en) * 2024-10-23 2025-02-11 武汉船舶通信研究所(中国船舶集团有限公司第七二二研究所) A communication network resource efficiency evaluation method and system
CN119536812B (en) * 2024-11-07 2025-05-27 上海大学 RISC-V based binarization neural network accelerator
CN119155760B (en) * 2024-11-14 2025-04-22 威胜集团有限公司 A communication link switching method for heterogeneous networks
CN119135555B (en) * 2024-11-15 2025-01-21 湘江实验室 Network quality evaluation method of calculation network
CN119211971B (en) * 2024-11-22 2025-02-14 唐人通信技术服务股份有限公司 5G wireless network drive test management system based on data analysis
CN119254627A (en) * 2024-12-03 2025-01-03 合肥海巍光电科技有限公司 Automatic configuration method of intelligent distributed signal source based on POST-INA integrated module
CN120614275B (en) * 2025-08-11 2025-10-03 中国铁塔股份有限公司江苏省分公司 Network quality analysis method and device, storage medium and electronic equipment

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9538401B1 (en) * 2015-12-18 2017-01-03 Verizon Patent And Licensing Inc. Cellular network cell clustering and prediction based on network traffic patterns
US10728773B2 (en) * 2018-01-26 2020-07-28 Verizon Patent And Licensing Inc. Automated intelligent self-organizing network for optimizing network performance
CN108848515B (en) * 2018-05-31 2021-06-08 武汉虹信技术服务有限责任公司 Internet of things service quality monitoring platform and method based on big data
CN113645092B (en) * 2021-07-07 2022-12-30 深圳市优克联新技术有限公司 Network quality evaluation method and device, terminal equipment and storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118802615A (en) * 2024-07-02 2024-10-18 中国移动通信集团山西有限公司 Private network quality assessment method, device, equipment, medium and computer program product

Also Published As

Publication number Publication date
WO2024124975A1 (en) 2024-06-20

Similar Documents

Publication Publication Date Title
CN118200166A (en) Network quality evaluation method, device, equipment, storage medium and program product
CN112668128B (en) Method and device for selecting terminal device nodes in a federated learning system
CN114219097B (en) Federal learning training and predicting method and system based on heterogeneous resources
CN116523079A (en) Reinforced learning-based federal learning optimization method and system
CN113379536A (en) Default probability prediction method for optimizing recurrent neural network based on gravity search algorithm
EP4002213A1 (en) System and method for training recommendation policies
CN118647092A (en) A method and system for comprehensive management of power distribution communication network
CN109754135B (en) Credit behavior data processing method, apparatus, storage medium and computer device
US20230419172A1 (en) Managing training of a machine learning model
CN113342474A (en) Method, device and storage medium for forecasting customer flow and training model
CN113516163B (en) Vehicle classification model compression method, device and storage medium based on network pruning
Manias et al. A model drift detection and adaptation framework for 5g core networks
CN117917057A (en) Server and method for obtaining key performance indicators and quickly adapting artificial intelligence models
CN115550181B (en) Network slice resource allocation method and system based on LSTM and A2C
CN118761100B (en) Privacy leakage detection method and system based on self-adaptive federal learning of terminal
CN110990548B (en) Method and device for updating reinforcement learning model
US20250005369A1 (en) Value-based action selection algorithm in reinforcement learning
CN115150335B (en) Optimal flow segmentation method and system based on deep reinforcement learning
US20230055373A1 (en) Estimating properties of units using system state graph models
CN116263889A (en) A service quality prediction method, device and equipment
CN119183141B (en) Wireless flow prediction method based on dynamic gradient sparse federal learning
CN118713916B (en) A distributed multi-agent network penetration method, device and storage medium
CN118678373B (en) Model updating method and related equipment for Internet of Things system
US20240232698A1 (en) Sustainable retraining for production machine learning
CN119089274B (en) A method and apparatus for assessing transient stability of power systems based on priority strategies.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination