Service volume prediction method based on communication network

By analyzing historical data and user behavior, generating and optimizing the traffic calculation model, the problem of large prediction errors in the existing technology is solved, and the accuracy of traffic prediction in communication networks is improved.

CN119095062APending Publication Date: 2024-12-06中国通信建设集团设计院有限公司 +1
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202410997737.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The prior art fails to reasonably combine the actual situation when predicting the traffic volume of communication networks, resulting in large errors in the prediction results, affecting subsequent management.

Method used

By obtaining historical data, calculating prediction errors and user regression values, generating a business volume calculation model, and verifying and correcting the model to optimize the prediction results and reduce the error.

Benefits of technology

It improves the accuracy of traffic volume forecasting, reduces forecast errors, and provides more accurate traffic volume distribution to support network planning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119095062A_ABST
    Figure CN119095062A_ABST
Patent Text Reader

Abstract

The invention discloses a traffic prediction method based on a communication network, relates to the technical field of traffic analysis of the communication network, and solves the technical problems that a relatively large error further exists and subsequent management is affected due to the fact that prediction analysis is not reasonably carried out by combining actual conditions and actual data. According to the method, the historical data is analyzed, the prediction error corresponding to the communication network is obtained through analysis, the user is further analyzed, the regression values of different users are obtained through analysis according to the historical data, the prediction error and the regression values are integrated to analyze and predict the corresponding service volume, and then the obtained predicted service volume is compared. According to the method, whether the data is correct or not is judged, the obtained model is corrected by analyzing the data aiming at the situation that the error is large, the prediction error is further reduced, and the overall prediction accuracy is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of communication network traffic analysis, and in particular to a communication network-based traffic prediction method. Background Art

[0002] At present, before building a mobile communication network, operators need to carry out network planning based on the network requirements of the target area, the networks of other operators and the business development, so as to reasonably build the mobile communication network. In the existing technology, the subsequent business volume distribution is mostly predicted by existing business volume data.

[0003] According to the patent application number CN201510891965.9, the patent includes: obtaining the paging volume of the first mobile communication network according to the paging information of the first mobile communication network, and obtaining the business volume of the first mobile communication network according to the business information of the first mobile communication network; obtaining the linear relationship model of the paging volume and the business volume by using the linear regression principle according to the paging volume of the first mobile communication network and the business volume of the first mobile communication network; obtaining the paging volume of the second mobile communication network according to the paging information of the second mobile communication network, the first mobile communication network is different from the second mobile communication network; estimating the estimated value of the business volume of the second mobile communication network according to the paging volume of the second mobile communication network and the linear relationship model of the paging volume and the business volume.

[0004] The above patent obtains a linear relationship model between paging volume and business volume based on the linear regression principle, and estimates the business volume of various data services in each location area of ​​other operators based on the paging volume of mobile communication networks deployed by other operators and the linear relationship between paging volume and business volume, thereby effectively improving the accuracy of the estimated value of business volume, allowing operators to accurately and conveniently obtain the business volume distribution of mobile communication networks deployed by other operators, as an important reference for network planning and construction. However, the above patent does not reasonably combine the actual situation for prediction and analysis, which will further cause large errors in the prediction results, thereby affecting subsequent management. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method for predicting business volume based on a communication network, which solves the problem that there is no reasonable combination of actual conditions and actual data for predictive analysis, which will further lead to large errors and affect subsequent management.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A traffic volume prediction method based on a communication network comprises the following steps:

[0007] Step 1: Obtain historical data and calculate the prediction error corresponding to the communication network based on the historical data;

[0008] Step 2: Obtain the user behaviors corresponding to different users based on historical data, and further calculate the regression values ​​of different users based on the user behaviors;

[0009] Step 3: Predict the traffic volume of the communication network by combining the user's regression value and prediction error, and generate a traffic volume calculation model;

[0010] Step 4: Verify and correct the generated business volume calculation model by comparing the predicted business volume with the actual business volume to verify the calculation model;

[0011] Step 5: Finally, the business volume calculation model is optimized through the obtained phased data, and the optimized model is output.

[0012] As a further solution of the present invention: the specific method of calculating the prediction error of the communication network in step 1 is:

[0013] Obtain historical data within a time period T, and divide the time period T into multiple time periods to obtain n time periods, where n=1, 2, ..., m, and select a group of time periods as analysis objects, and obtain the real-time communication volume, page views, and download volume in the historical data corresponding to the analysis objects, and record the real-time communication volume as L1, the page views as L2, and the download volume as L3;

[0014] Then substitute the acquired real-time communication volume L1, browsing volume L2 and download volume L3 into the prediction error formula: The prediction error C of the analyzed object is calculated, and the prediction error Cn corresponding to the n time periods is obtained by analogy;

[0015] At the same time, substitute the obtained prediction error Cn into the formula: The corresponding prediction error Cz within the time period T is calculated, and the calculated prediction error Cz is recorded as the prediction standard value.

[0016] As a further solution of the present invention: the specific method of calculating and analyzing the user behavior in step 2 is:

[0017] Obtain historical data within the same time period T, and label all users as i, where i=1, 2, ..., j. At the same time, select any user as the target object, and then obtain and analyze the user behavior of the target object, and the user behavior includes: the user's browsing volume and download volume;

[0018] The time period T is divided into x time series, and the views and downloads corresponding to the target object are obtained and recorded as Y1x and Y2x. Then the views and downloads corresponding to the user are substituted into the formula: Where θ is a constant, and the user value G of the target object is calculated.

[0019] As a further solution of the present invention: the specific method of calculating the regression value according to the user value is:

[0020] The number of times the same user behavior of the target object within the time period T is obtained is recorded as K, and the real-time communication volume of the target object is obtained is recorded as E. Then, the obtained user behavior number K, real-time communication volume E and user value G are substituted into the regression function: F = ∫ (K + E) + G × a2, where a2 is the preset proportional coefficient. The regression value F of the target object is calculated according to the above formula.

[0021] As a further solution of the present invention: the specific method of predicting and analyzing the traffic volume of the communication network in step 3 is:

[0022] Obtain all regression values ​​Fi, compare the user regression value Fi with the preset value Fy, and classify the users according to the regression value Fi, classify the users whose regression value Fi is greater than the preset value Fy as high-frequency users, and classify the users whose regression value Fi is less than the preset value Fy as low-frequency users, and analyze the high-frequency users and low-frequency users respectively;

[0023] All high-frequency users are obtained and the mean of the corresponding regression values ​​is calculated as Fp. Then the regression values ​​corresponding to all low-frequency users are obtained, and the regression values ​​of all low-frequency users are discretely calculated to obtain discrete values ​​Fb. At the same time, the mean Fp of high-frequency users and the discrete value Fb of low-frequency users are combined with the prediction error Cz to predict the traffic of the communication network.

[0024] As a further solution of the present invention: the specific method of predicting the traffic volume of the communication network by combining the high-frequency user mean value Fp, the low-frequency user discrete value Fb and the prediction error Cz is:

[0025] Obtain the real-time communication volume, browsing volume and download volume corresponding to the current time node, and record them as L1s, L2s and L3s respectively. Analyze the download volume, obtain the download volume corresponding to high-frequency users and the download volume corresponding to low-frequency users, and record them as Hl and Dl respectively, where HI is the download volume of high-frequency users and DI is the download volume of low-frequency users. Substitute the obtained data into the formula: Y represents the change value of the number of users corresponding to the communication network, and the predicted traffic volume W of the communication network is calculated according to the above formula.

[0026] As a further solution of the present invention: the specific method of comparing the actual business volume with the predicted business volume in step 4 is:

[0027] The actual business volume corresponding to any time node is obtained and recorded as Ws. Then the time node is analyzed to obtain the user quantity change value Y1 and various parameters within the time node. At the same time, the obtained data is substituted into the formula to calculate the predicted business volume W1. Then the actual business volume Ws is compared with the predicted business volume W1.

[0028] The difference between the actual business volume Ws and the predicted business volume W1 is recorded as the business volume difference Wc, and then the business volume difference Wc is compared with the evaluation interval. When the business volume difference Wc is not within the evaluation interval, it means that the calculated predicted business volume W1 has an error, and a correction result is generated. On the contrary, when the business volume difference Wc is within the evaluation interval, it means that the calculated predicted business volume error is within the allowable range and no processing is performed.

[0029] As a further solution of the present invention: the specific method of analyzing the correction result is:

[0030] Obtain the user change value Y1, analyze the number of users who have browsing behavior within the time node, and calculate the difference between the two. At the same time, analyze the difference between the two, calculate the ratio of the difference to the total number of users to obtain the user proportion value, then sum the user proportion value and the user change value to obtain the correction value, and then substitute the correction value into the formula to calculate the predicted business volume. At the same time, calculate the difference between the calculated predicted business volume and the actual business volume, and compare the difference with the judgment interval;

[0031] When there is a large error, a secondary correction analysis is performed on the acquired data, and the predicted business volume calculation model is iteratively updated in this way.

[0032] The present invention provides a method for predicting traffic volume based on a communication network. Compared with the prior art, it has the following beneficial effects:

[0033] The present invention analyzes historical data to first obtain the prediction error corresponding to the communication network, and then further analyzes the users. According to the historical data analysis, regression values ​​of different users are obtained, and the corresponding business volume is analyzed and predicted by combining the prediction error and regression value. Then, the obtained predicted business volume is compared to determine whether it is correct. For situations where the error is large, the obtained model is corrected by analyzing the data to further reduce the prediction error and improve the overall prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] See also Figure 1 The present application provides a method for predicting traffic volume based on a communication network, the method specifically comprising the following steps:

[0037] Step 1: Obtain historical data and calculate the prediction error corresponding to the communication network based on the historical data. The specific calculation method is as follows:

[0038] The historical data within the time period T is obtained, and the time period T is divided into multiple time periods to obtain n time periods, and n=1, 2, ..., m, and the time values ​​corresponding to the n time periods obtained here are the same, which means that the time period T is evenly divided, and a group of time periods are selected as analysis objects at the same time, and the real-time communication volume, browsing volume and download volume in the historical data corresponding to the analysis object are obtained at the same time. The specific real-time communication volume is represented by the total amount of business data processed by the instant messaging service within a certain period of time. It is an important indicator for measuring the quality of the instant messaging service, and the real-time communication volume is recorded as L1, the browsing volume is recorded as L2, and the download volume is recorded as L3;

[0039] Then substitute the acquired real-time communication volume L1, browsing volume L2 and download volume L3 into the prediction error formula: And ρ(L2, L3) is a similarity function. In the above formula, the minimum value of the similarity function is used as the standard for calculation. The prediction error C of the analysis object is calculated by the above formula, and the prediction error Cn corresponding to the n time period is obtained by analogy;

[0040] At the same time, substitute the obtained prediction error Cn into the formula: The corresponding prediction error Cz within the time period T is calculated, and the calculated prediction error Cz is recorded as the prediction standard value.

[0041] Step 2: Calculate the regression values ​​of different users based on historical data, and the specific calculation method is as follows:

[0042] Obtain historical data within the same time period T, and label all users as i, where i=1, 2, ..., j, and select any user as the target object. Then obtain and analyze the user behavior of the target object, and the user behavior here includes: the user's browsing volume and download volume;

[0043] The time period T is divided into x time series, and the page views and downloads corresponding to the target object are obtained and recorded as Y1x and Y2x. Then the page views and downloads corresponding to the user are substituted into the formula: Where θ is a constant, and the specific value is set by the operator. In this application, the value is 1.231. The user value G of the target object is calculated according to the above formula;

[0044] The regression value of the target object is analyzed and calculated based on the obtained user value G, and the specific calculation method is:

[0045] The number of times the same user behavior of the target object in the time period T is obtained is recorded as K, and the real-time communication volume of the target object is obtained is recorded as E. Then, the obtained user behavior number K, the real-time communication volume E and the user value G are substituted into the regression function: F = ∫ (K + E) + G × a2, where a2 is the preset proportional coefficient. The regression value F of the target object is calculated according to the above formula;

[0046] The regression values ​​of all users in the time period T are calculated in this way.

[0047] Step 3: Combine the user's regression value and prediction error to predict the traffic volume of the communication network, and the specific method is as follows:

[0048] Obtain all regression values ​​Fi, compare the user regression value Fi with the preset value Fy, and classify the users according to the regression value Fi, classify the users whose regression value Fi is greater than the preset value Fy as high-frequency users, and classify the users whose regression value Fi is less than the preset value Fy as low-frequency users, and analyze the high-frequency users and low-frequency users respectively;

[0049] All high-frequency users are obtained and the mean of the corresponding regression values ​​is calculated and recorded as Fp. Then the regression values ​​corresponding to all low-frequency users are obtained, and the regression values ​​of all low-frequency users are discretely calculated to obtain the discrete value Fb. At the same time, the high-frequency user mean Fp and the low-frequency user discrete value Fb are combined with the prediction error Cz to predict the traffic of the communication network. The specific prediction method is as follows:

[0050] Obtain the real-time communication volume, browsing volume and download volume corresponding to the current time node, and record them as L1s, L2s and L3s respectively. Analyze the download volume, obtain the download volume corresponding to high-frequency users and the download volume corresponding to low-frequency users, and record them as Hl and Dl respectively, where HI is the download volume of high-frequency users and DI is the download volume of low-frequency users. Substitute the obtained data into the formula: Wherein Y is the dependent variable, and in the present application, Y represents the change in the number of users corresponding to the communication network. The predicted traffic volume W of the communication network is calculated according to the above formula.

[0051] Step 4: Analyze the predicted business volume based on the actual business volume. The specific analysis method is as follows:

[0052] The actual business volume corresponding to any time node is obtained and recorded as Ws. Then the time node is analyzed to obtain the user quantity change value Y1 and various parameters within the time node. At the same time, the obtained data is substituted into the formula to calculate the predicted business volume W1. Then the actual business volume Ws is compared with the predicted business volume W1.

[0053] The difference between the actual business volume Ws and the predicted business volume W1 is recorded as the business volume difference Wc, and then the business volume difference Wc is compared with the judgment interval. The judgment interval here represents the allowable error range corresponding to the difference between the two. When the business volume difference Wc is not within the judgment interval, it means that the calculated predicted business volume W1 has an error, and a correction result is generated. On the contrary, when the business volume difference Wc is within the judgment interval, it means that the calculated predicted business volume error is within the allowable range and no processing is performed.

[0054] Then, the generated correction result is analyzed. The specific analysis method is as follows: obtain the user change value Y1, analyze the number of users with browsing behavior within the time node, and calculate the difference between the two. At the same time, analyze the difference between the two, calculate the ratio of the difference to the total number of users to obtain the user proportion value, then sum the user proportion value and the user change value to obtain the correction value, and then substitute the correction value into the formula to calculate the predicted business volume, and at the same time calculate the difference between the calculated predicted business volume and the actual business volume, and compare the difference with the judgment interval;

[0055] In the case of large errors, a secondary correction analysis is performed on the acquired data, and the predicted business volume calculation model is iteratively updated in this way.

[0056] Step 5: Output the obtained forecast business calculation model, and optimize the model according to the real-time acquired interim data. The interim data here refers to the data acquired in different time periods.

[0057] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0058] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A traffic prediction method based on a communication network, characterized in that: The following steps are involved: Step 1: Obtain historical data and calculate the prediction error corresponding to the communication network based on the historical data; Step 2: Obtain the user behaviors corresponding to different users based on historical data, and further calculate the regression values ​​of different users based on the user behaviors; Step 3: Predict the traffic volume of the communication network by combining the user's regression value and prediction error, and generate a traffic volume calculation model; Step 4: Verify and correct the generated business volume calculation model by comparing the predicted business volume with the actual business volume to verify the calculation model; Step 5: Finally, the business volume calculation model is optimized through the obtained phased data, and the optimized model is output.

2. A method for predicting traffic volume based on a communication network according to claim 1, characterized in that: The specific method of calculating the prediction error of the communication network in step 1 is: Obtain historical data within a time period T, and divide the time period T into multiple time periods to obtain n time periods, where n=1, 2, ..., m, and select a group of time periods as analysis objects, and obtain the real-time communication volume, page views, and download volume in the historical data corresponding to the analysis objects, and record the real-time communication volume as L1, the page views as L2, and the download volume as L3; Then substitute the acquired real-time communication volume L1, browsing volume L2 and download volume L3 into the prediction error formula: The prediction error C of the analyzed object is calculated, and the prediction error Cn corresponding to the n time periods is obtained by analogy; At the same time, substitute the obtained prediction error Cn into the formula: The corresponding prediction error Cz within the time period T is calculated, and the calculated prediction error Cz is recorded as the prediction standard value.

3. A method for predicting traffic volume based on a communication network according to claim 1, characterized in that: The specific method of calculating and analyzing the user behavior in step 2 is: Obtain historical data within the same time period T, and label all users as i, where i=1, 2, ..., j. At the same time, select any user as the target object, and then obtain and analyze the user behavior of the target object, and the user behavior includes: the user's browsing volume and download volume; The time period T is divided into x time series, and the page views and downloads corresponding to the target object are obtained and recorded as Y1x and Y2x. Then the page views and downloads corresponding to the user are substituted into the formula: Where θ is a constant, and the user value G of the target object is calculated.

4. A method for predicting traffic volume based on a communication network according to claim 3, characterized in that: The specific method of calculating the regression value according to the user value is: The number of times the same user behavior of the target object within the time period T is obtained is recorded as K, and the real-time communication volume of the target object is obtained is recorded as E. Then, the obtained user behavior number K, real-time communication volume E and user value G are substituted into the regression function: F = ∫ (K + E) + G × a2, where a2 is the preset proportional coefficient. The regression value F of the target object is calculated according to the above formula.

5. The method for predicting traffic volume based on a communication network according to claim 1, characterized in that: The specific method of predicting and analyzing the traffic volume of the communication network in step 3 is: Obtain all regression values ​​Fi, compare the user regression value Fi with the preset value Fy, and classify the users according to the regression value Fi, classify the users whose regression value Fi is greater than the preset value Fy as high-frequency users, and classify the users whose regression value Fi is less than the preset value Fy as low-frequency users, and analyze the high-frequency users and low-frequency users respectively; All high-frequency users are obtained and the mean of the corresponding regression values ​​is calculated as Fp. Then the regression values ​​corresponding to all low-frequency users are obtained, and the regression values ​​of all low-frequency users are discretely calculated to obtain discrete values ​​Fb. At the same time, the mean Fp of high-frequency users and the discrete value Fb of low-frequency users are combined with the prediction error Cz to predict the traffic of the communication network.

6. A method for predicting traffic volume based on a communication network according to claim 5, characterized in that: The specific method of predicting the traffic volume of the communication network by combining the high-frequency user mean value Fp, the low-frequency user discrete value Fb and the prediction error Cz is: Obtain the real-time communication volume, browsing volume and download volume corresponding to the current time node, and record them as L1s, L2s and L3s respectively. Analyze the download volume, obtain the download volume corresponding to high-frequency users and the download volume corresponding to low-frequency users, and record them as Hl and Dl respectively, where HI is the download volume of high-frequency users and DI is the download volume of low-frequency users. Substitute the obtained data into the formula: Y represents the change value of the number of users corresponding to the communication network, and the predicted traffic volume W of the communication network is calculated according to the above formula.

7. A method for predicting traffic volume based on a communication network according to claim 1, characterized in that: The specific method of comparing the actual business volume with the predicted business volume in step 4 is: The actual business volume corresponding to any time node is obtained and recorded as Ws. Then the time node is analyzed to obtain the user quantity change value Y1 and various parameters within the time node. At the same time, the obtained data is substituted into the formula to calculate the predicted business volume W1. Then the actual business volume Ws is compared with the predicted business volume W1. The difference between the actual business volume Ws and the predicted business volume W1 is recorded as the business volume difference Wc, and then the business volume difference Wc is compared with the evaluation interval. When the business volume difference Wc is not within the evaluation interval, it means that the calculated predicted business volume W1 has an error, and a correction result is generated. On the contrary, when the business volume difference Wc is within the evaluation interval, it means that the calculated predicted business volume error is within the allowable range and no processing is performed.

8. A method for predicting traffic volume based on a communication network according to claim 7, characterized in that: The specific method for analyzing the correction results is as follows: Obtain the user change value Y1, analyze the number of users who have browsing behavior within the time node, and calculate the difference between the two. At the same time, analyze the difference between the two, calculate the ratio of the difference to the total number of users to obtain the user proportion value, then sum the user proportion value and the user change value to obtain the correction value, and then substitute the correction value into the formula to calculate the predicted business volume. At the same time, calculate the difference between the calculated predicted business volume and the actual business volume, and compare the difference with the judgment interval; When there is a large error, a secondary correction analysis is performed on the acquired data, and the predicted business volume calculation model is iteratively updated in this way.

Citation Information

Patent Citations

  • Traffic prediction method and device for mobile communication network

    CN106851604A