JP2012182677A - Communication traffic prediction device, method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to predict traffic volume directly caused by new innovation communication service in a communication network.SOLUTION: A communication traffic prediction device comprises: service subscriber number prediction means for calculating a service subscriber number prediction curve by performing regression analysis processing using a structural change point corresponding to a qualitative change of service subscriber layer in a popularization process of a new communication service; subscriber behavior prediction means for predicting service utilization behavior of a subscriber at a future time point by using history information of the service utilization behavior of a new communication service subscriber obtained from time-sequence data storage means; and stream prediction means for obtaining data necessary for a stream fluctuation prediction from the time-series data storage means and for predicting the occupancy of a utilization frequency between exclusive classes of stream at the future time point by observing a utilization frequency change between the exclusive classes of the stream on the basis of improved quality and increased functions of the new communication service.

Description

  The present invention relates to a communication traffic prediction apparatus, method, and program, and more particularly, to a technology for predicting a traffic volume peak directly attributable to a new innovation communication service. Specifically, the traffic volume peak of the new innovation communication service is predicted and evaluated by the product of the number of subscribers of the new innovation communication service and the peak traffic volume per communication service subscriber. Regarding the number of innovation communication service subscribers, in particular, the structural change point, which is the point where qualitative changes are occurring in the service subscriber layer with the spread of the communication service, is detected, and the “structural theory of demand growth” Based on the structural change point, increase the medium- to long-term prediction accuracy of the number of new innovation communication service subscribers, and regarding the peak traffic volume per communication service subscriber, in particular, factors based on subscriber behavior characteristics, New innovation communication services based on high quality and multi-functionality Taking into account the variations due to bandwidth of stream, the new innovation communication service communication traffic prediction apparatus implementing increasing the prediction accuracy of peak traffic volume, to a communication service predicting method, and a program.

  In the conventional technology, prediction of conventional communication traffic such as telephone service and Internet connection service is based on various mathematical models such as simple regression model and multiple regression model. There is a technique to predict (see, for example, Non-Patent Document 1, Patent Documents 1 and 2). However, there was no prediction method that assumed application to a new innovation communication service.

  Further, there is a method for performing medium- to long-term prediction of the number of subscribers and the number of subscribers regarding innovation diffusion (see, for example, Non-Patent Documents 2, 3, and 4). However, every time extrapolation prediction is performed by adding actual data, the disadvantage that the error becomes large cannot be solved. On the other hand, there is a method that considers qualitative changes in the subscriber layer (for example, see Non-Patent Document 5 and Patent Document 3). A unique concept and theory based on this method is defined here as “structural theory of demand growth”. However, there has never been a method in which “structural theory of demand growth” is applied to the prediction of peak traffic volume, not the number of new innovation communication service subscriptions.

JP 2004-023114 A JP 2004-056273 A JP 2008-305229 A

Matsuda Jun et al. (1991), Switch traffic management technology based on daily traffic prediction method, NTT R & D Vol.40, No.12, pp.1581-1588 Bass F. M. (1969), A new product growth model for consumer durables, Management Science Theory, Vol.15, No.5, January 1969, pp.215-227. Saito H. (1987), A demand forecasting method for new telecommunication services Journal of the Operations Research Society of Japan, Vol. 30, No. 2, June 1987. Translated by Hidenori Morimura et al. (1997), Marketing Handbook, Asakura Shoten (Eliashberg, J., and Lilien, G.L., ed. (1993) Marketing, Handbooks in Operations Research and Management Science Vol.5) Shimogawa, S., and Shinno, M., (2007) .Mechanism of Diffusion of Fixed-Line Broadband Access Services in Japan and its Application to Long-Term Growth Prediction-Rapid Diffusion in Urban Areas-,

  In order for a telecommunications carrier providing a communication network to economically design the capacity of the communication network, such as the link bandwidth and node equipment, it is necessary to predict peak traffic on the communication network with high accuracy. If it cannot be predicted with a high degree of accuracy, the service capacity will be excessive and the service will be too costly, or if the capacity is too small, the service quality experienced by service subscribers will deteriorate during peak traffic. Occurs and leads to the problem that carriers may lose customers.

  However, for newly provided communication services, since past performance data is small and the length of the period necessary for prediction is insufficient, it is not possible to appropriately use the prediction method in the prior art. In addition, if the communication service is an innovation, the process of disseminating the communication service of innovation is essentially different from the process of disseminating communication services such as telephone services and Internet connection services. In the conventional technology that does not take into account, high accuracy could not be expected in the medium- to long-term prediction of the traffic of the new innovation communication service. However, the innovation communication service is defined as a communication service that provides subscribers with completely new value that is not included in the category of conventional communication services.

  The present invention is for solving the above-described problems, and provides a communication traffic prediction device, a communication traffic prediction method, and a program capable of predicting a traffic volume directly resulting from a new innovation communication service in a communication network. It is an object.

The present invention is a communication traffic prediction device for predicting a traffic volume peak directly attributable to a new communication service in a communication network,
Service subscriber number prediction means for calculating a service subscriber number prediction curve by performing a regression analysis process using a structural change point corresponding to a qualitative change of a service subscriber layer in the process of spreading the new communication service;
Subscriber behavior prediction means for predicting service usage behavior of a subscriber at a future time point while using history information of service usage behavior of the new communication service subscriber;
A stream that predicts the occupancy rate of the usage frequency between exclusive classes of streams at a future time while observing the transition of the usage frequency between exclusive classes of streams based on the improvement in quality and multifunction of the new communication service A prediction means,
The service subscriber number prediction curve is characterized in that an increase in the number of service subscriptions per unit time is a function that is proportional to the negative of the elapsed time from a predetermined reference time point.

  Further, in one configuration example of the communication traffic prediction device of the present invention, the new communication service subscriber number prediction means performs more appropriate prediction when the new communication service is an innovation communication service. It is based on the described “structural theory of demand growth”.

  According to the description in Patent Document 3, the “structural theory of demand growth” is as follows. “The demand growth phenomenon can be expressed as a single normal function until a certain period in the first half of the demand growth phenomenon. Therefore, if data fitting is performed while requiring an inertia that is not limited to an increment, From one period to the second half, it matches the negative growth that is dual as a function class with the positive growth of the first half, and “−1” is well matched as the incremental exponent of that increment. It is said that the behavior of starting use by a late user occurs on average when the accumulated amount of information from familiar surroundings reaches a certain amount, and the size of the information channel is familiar. Is based on the hypothesis that it depends on the number of people sharing

  As a simple paraphrase, “structural theory of demand growth” means that the demand growth curve of the first-time user who uses the innovation service is a downwardly convex function, and the demand growth curve of the second-time user is almost logarithmic. It can be said that it is a theory that leads to continuous growth and that both curves become continuous. Indeed, the theory data of the growth in the number of subscribers in Japan for broadband services, which are examples of innovative communication services, can be well explained by the theory. This is the content of Non-Patent Document 5.

  Therefore, the service subscriber number prediction means based on the “structural theory of demand growth” detects the structural change point, and performs data fitting of the actual data with a downward convex function before the structural change point, Since the actual data is fitted with a logarithmic function after the change point, it is possible to predict the number of subscribers of the innovation communication service with higher accuracy than in the prior art.

  Further, in one configuration example of the communication traffic prediction device of the present invention, the subscriber behavior prediction means includes a communication start time, a communication end time, a calling side NW address for each session, and a usage history of the communication service for each subscriber. The destination NW address, communication band average value, maximum value, traffic characteristic information such as burst characteristics, various attribute information of the content to be communicated, etc. are stored as time series data, and at the same time, the weather when using the service, A subscriber behavior model that evaluates periodic fluctuations of external conditions such as daily cycle, weekly cycle, monthly cycle, and yearly cycle (seasonal variation) through analysis matched with weather data such as temperature and humidity, and various event information, It builds a subscriber behavior model that correlates with certain events, so it can be combined with external conditions and event information at a future time. Thus, it is possible to accurately predict the peak coefficient with respect to the average traffic volume per subscriber.

  Further, in one configuration example of the communication traffic prediction device of the present invention, the stream prediction means appropriately performs exclusive class classification by grasping attributes for individual streams forming the traffic of the target service, The number of times of communication for each class is counted, and the correlation of the number of times of communication between these classes is extracted by multiple regression analysis, etc., so the frequency of stream class usage in the future or between classes It is possible to accurately predict the occupation ratio (share).

  As described above, according to the present invention, the number of service subscribers in the future, the average per subscriber, in predicting the peak of traffic volume directly attributable to a new communication service in a communication network, in particular, a new innovation communication service. Since it is possible to predict the peak coefficient for traffic volume and the frequency of use of each exclusive class of stream more accurately than before, the peak of traffic volume caused directly by the new communication service in the communication network can be accurately It becomes possible to predict. As a result, the peak traffic on the communication network can be predicted with high accuracy, so that the facility capacity can be made economical without causing deterioration in the quality of experience of the service user.

It is a block diagram of the communication traffic prediction apparatus in the 1st Embodiment of this invention. It is a flowchart of operation | movement of the communication traffic prediction apparatus in the 1st Embodiment of this invention. It is an example of the number of users of a certain innovation service by the subscriber number prediction function part in the 2nd Embodiment of this invention, and a prediction line. This is a change in traffic volume per subscriber of a certain innovation service in the second embodiment of the present invention.

  Embodiments of the present invention will be described below with reference to the drawings.

[First embodiment]
FIG. 1 shows the configuration of a communication traffic prediction apparatus according to the first embodiment of the present invention.

  With reference to FIG. 1, the configuration of a communication traffic prediction apparatus according to an embodiment of the present invention will be described.

  The communication traffic prediction apparatus 10 is composed of an information processing apparatus using a computer as a whole, and includes a data input / output / storage function section 20, a subscriber number prediction function section 30, a subscriber behavior prediction function section 40 as main function sections. And a stream fluctuation prediction function unit 50 and a prediction result integration processing function unit 60.

  The data input / output / storage function unit 20 includes a data input / output unit 21, an operation input unit 22, a screen display output unit 23, a main storage unit 24, a time series data storage unit 25, a processing data recording unit 26, and an output data storage unit 27. Consists of

  The data input / output unit 21 includes a dedicated communication interface circuit. The data input / output unit 21 is connected to an external device via a communication line to perform data communication, and reads various time-series data from the external device to the time-series data storage unit 25. It has a function of saving, a function of reading an output result from the output data saving unit 27 and outputting and sending it to external data, and a function of reading various commands and programs from an external device and storing them in the main storage unit 24.

  The operation input unit 22 includes an operation input device such as a keyboard and a mouse, and detects various operations such as a communication traffic prediction process start operation and a data input operation by an operator, and issues an operation command to each function unit.

  The screen display output unit 23 includes various screen display devices, and has a function of displaying a processing menu from each functional unit or output results such as a communication traffic prediction value and a communication traffic prediction graph on the screen.

  The main storage unit 24 includes a storage device such as a memory or a hard disk, and has a function of storing parameters used for processing in each functional unit, default values (initial values) for processing data such as designation requests, and programs. Yes.

  The time-series data storage unit 25 includes a storage device such as a memory or a hard disk. The communication start time, the communication end time, and the originating network (for each session) are provided for communication services for subscribers subject to communication traffic prediction. NW) Address, destination NW address, communication band average value, maximum value, traffic characteristic information such as burst characteristics, various attribute information of the content to be communicated, etc., and a function to store as time series data Yes. The session unit data is data acquired by an external device or further statistically processed data, and is stored in the time-series data storage unit 25 through a data input / output unit. In addition, various data necessary for the subscriber behavior prediction function unit 40, such as weather data such as weather, temperature, and humidity, and various event information, and content exclusive class (SD (Standard for video content) Definition) quality, HD (High Definition) quality, 3D quality, etc.) stream bandwidth, viewing frequency, future content release information, and other data required for the stream fluctuation prediction function unit 50 are also time-series data. It is stored in the time series data storage unit 25.

  The processing data recording unit 26 includes a storage device such as a memory and a hard disk, and has various functions such as a predicted number of subscribers, a subscriber behavior prediction result, and a stream fluctuation prediction result that are generated when the communication traffic prediction process is executed. It has a function of recording various data processed in the unit.

  The output data storage unit 27 includes a storage device such as a memory and a hard disk, and has a function of storing output results such as a communication traffic prediction value and a communication traffic prediction graph obtained by the communication traffic prediction process.

  FIG. 2 is a flowchart of the operation of the communication traffic prediction apparatus according to the first embodiment of the present invention.

  Hereinafter, the operation of the subscriber number prediction function unit 30, the subscriber behavior prediction function unit 40, the stream prediction function unit 50, and the prediction result integration processing function unit 60 will be described.

  The subscriber number prediction function unit 30 includes a structure change point detection function unit 31 and a subscriber number prediction calculation function unit 32.

  The structural change point detection function unit 31 reads the subscription number actual data of the service to be predicted from the time-series data storage unit 25. Next, an increase in the number of subscriptions is calculated from the actual number of subscriptions of the service given as a time series, and a point in time when the change has changed from an increase to a decrease is detected. Alternatively, after converting the time axis from the actual number of subscriptions of the service given as the time series to an elapsed time from a predetermined reference time, logarithmic conversion is performed, and the actual value of the number of service subscriptions is linearly approximated To detect the time when the change in the inclination occurs. The detected time is stored in the processing data storage unit 26 as the structure change point of the service (step 101).

  The subscriber number prediction calculation function unit 32 reads the time series data of the number of service subscriptions to be predicted from the time series data storage unit 25, converts the time axis to the elapsed time from a predetermined reference time point, and further calculates the logarithm. Apply the linear approximation extrapolation by the least square method etc. to the time series data of the number of service subscriptions that have been converted, using the actual data after the detected structural change point, and the service at the future time Calculated as the predicted number of subscribers. The calculation result is accumulated in the processing data storage unit 26 (step 102).

  The subscriber behavior prediction function unit 40 includes a subscriber behavior analysis function unit 41 and a subscriber behavior model prediction function unit 42.

  The subscriber behavior analysis function unit 41 is acquired by an external device such as a packet capture device installed in a communication network or a communication terminal (for example, a home gateway device or a set top box device) installed in a subscriber's house. , Which is situation data stored in the time-series data storage unit 25 when the subscriber uses the service, communication start time, communication end time, caller NW address, callee NW of each session of the service Analyzing traffic characteristics information such as address, average value of communication bandwidth, maximum value, burst characteristics, various attribute information of the content to be communicated, data such as weather data such as weather, temperature, humidity, and various event information (For example, there are methods such as quantification theory class I) to build a subscriber behavior model -Up 103). For example, in the communication service, the peak traffic on weekdays is around X o'clock, but increases by Y% compared to the average. On the other hand, the peak traffic on holidays is around XX, increasing YY% compared to the average. If the rain falls Umm, the traffic of the communication service will increase by V%. If the UU temperature is lower than the previous day, a subscriber behavior model supported by data such as VV% increase in traffic of the communication service can be constructed. It is possible to construct a subscriber behavior model that evaluates a periodic variation such as a daily cycle, a weekly cycle, a monthly cycle, and an annual cycle (seasonal variation). In addition, a subscriber behavior model having a correlation with a certain type of event can be constructed.

  The subscriber behavior model prediction function unit 42 is based on the subscriber behavior model constructed by the subscriber behavior analysis function unit 41 and takes into account future fluctuations, various events, and weather conditions as parameters. A peak coefficient for the average traffic volume per person is predicted and accumulated in the processing data storage unit 26 (step 104).

  The stream prediction function unit 50 includes a stream class classification / counting function unit 51, a stream class frequency analysis function unit 52, and a stream class frequency prediction function unit 53.

  The stream class classification and counting function unit 51 has a function of appropriately classifying exclusive streams by grasping attributes of individual streams forming the traffic of the target service, and counting the number of times of communication for each class. Have. Data necessary for the function is read from the time-series data storage unit 25 and processed to generate attribute information for each exclusive class of the stream (step 105). The exclusive class classification is defined as a class classification in which elements included in a plurality of classes do not exist.

  The frequency analysis function unit 52 for each stream class is a function unit that performs analysis using the attribute information for each exclusive class of the stream obtained by the stream class classification and counting function unit 51 and the use frequency data. As the quality of communication services increases and the number of functions increases, the number of communications included in the low-quality or single-function class of the exclusive class tends to decrease, and conversely, it is included in another high-quality or multi-function class. The number of communications is increasing. The correlation between the exclusive classes of such streams is extracted by multiple regression analysis or the like, and stored in the processing data storage unit 26 (step 106).

  The frequency prediction function unit 53 for each stream class combines the correlation for each exclusive class of the stream obtained by the frequency analysis function unit 52 for each stream class with information such as the content release information and the usage trend for each stream class. The usage frequency of each class at the future time point or the occupation ratio (share) between classes is predicted and stored in the processing data storage unit 26 (step 107).

  The prediction result integration processing function unit 60 includes the predicted number of subscribers at the designated future time obtained by the subscriber number prediction function unit 30 and the same future time obtained by the subscriber behavior prediction function unit 40. Prediction value such as a peak coefficient for the average traffic volume per subscriber and the usage frequency by exclusive class at the same future time point obtained by the stream prediction function unit 50, or between classes A predicted value of the occupation rate (share) is acquired from the processing data storage unit 26, and a traffic volume peak directly attributable to the new service is predicted (step 108).

  As described above, according to the present invention, the number of service subscribers in the future, the average per subscriber, in predicting the peak of traffic volume directly attributable to a new communication service in a communication network, in particular, a new innovation communication service. Since it is possible to predict the peak coefficient for traffic volume and the frequency of use of each exclusive class of stream more accurately than before, the peak of traffic volume caused directly by the new communication service in the communication network can be accurately It becomes possible to predict.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. In the present embodiment, the first embodiment will be described more specifically. Here, a specific example of application of the present invention will be described by taking VOD (Video On Demand) as a new communication service as an example.

  VOD services using optical lines have already grown significantly in Japan, but here we consider VOD services to be a new innovation communication service. Compared to conventional rental videos, there is no need to go out and it can be used immediately and there is no need to return. Compared to multi-channel broadcasting that was performed by CATV, etc., the content can be freely selected according to personal taste preferences. This is because it is considered to satisfy the definition of an innovation service that is a communication service that provides a completely new value that is not included in the conventional service category by being able to start watching immediately with a high degree of degree.

  Also in this embodiment, the apparatus configuration is the same as that of the first embodiment, and will be described with reference to FIG.

  The subscriber number prediction function unit 30 that predicts the number of subscribers in the future will be described.

  First, the structural change point detection function unit 31 reads the actual data of the number of subscribers of the VOD service from the time series data storage unit 25, and from this time series data, the net increase in the number of subscribers of the VOD service per quarter is obtained. Detect TA when it starts to increase and decreases. The point in time TA is the structural change point of the demand for the VOD service, but if it cannot be detected, it is determined that the structural change point has not yet been reached.

  Next, when the number of subscribers does not reach the structural change point, the subscriber number prediction calculation function unit 32 performs data fitting on the VOD service subscriber data with a downwardly convex function, and the subscription at a future time point. The estimated number of people. Alternatively, when the time point TA, which is a structural change point, is detected, the actual data after the time point TA is subjected to data fitting with a logarithmic function to obtain a predicted value of the number of subscribers at a future time point.

  FIG. 3 is an example of the number of users of a certain innovation service and a prediction line by the subscriber number prediction function unit according to the second embodiment of the present invention. Before the structural change point, approximation is performed with a downward convex function (dotted line), and after the structural change point, approximation is performed with a logarithmic function (solid line), and extrapolation is performed. Yes.

  Next, the subscriber behavior prediction function unit 40 that analyzes subscriber behavior and performs model construction and prediction will be described.

  First, the subscriber behavior analysis function unit 41 can perform the behavior analysis of the subscriber of the VOD service as follows. By continuously measuring the total traffic of the VOD service transmitted from the router in which the VOD server is accommodated, the peak time and peak traffic volume of the service can be extracted as samples. Further, the peak traffic volume can be calculated as the peak traffic volume per VOD service subscriber by dividing the peak traffic volume by the number of subscribers of the VOD service on that day. By analyzing the peak traffic volume per VOD service subscriber by comparing weather data such as weather, temperature, and humidity, and data such as various event information (for example, there are methods such as quantification theory class I) It is possible to construct a subscriber behavior model. Thus, as shown in FIG. 3, by analyzing the time series data of daily traffic volume per VOD service subscriber, for example, the peak traffic per VOD service subscriber is often caused by bad weather in winter. Or, by accumulating cases that doubled compared to the average day before due to a specific event, the factors that increase the peak traffic volume per VOD service subscriber and how much traffic Analyze whether it leads to an increase.

  Further, in various measuring functions of nodes installed in the communication network, packet capture devices, and external devices such as communication terminals (for example, home gateway devices and set-top box devices) installed in subscriber premises, subscribers Traffic characteristics information such as communication start time, communication end time, originating NW address, terminating NW address, average value of communication bandwidth, maximum value, burst characteristics, etc. as situation data when using the VOD service as a unit If various attribute information of the content to be communicated is acquired, a more detailed and specific subscriber behavior model can be analyzed.

  Next, the subscriber behavior model prediction function unit 42 may be assumed to be a subscriber behavior model analyzed by the subscriber behavior analysis function unit 41 and factors affecting the subscriber behavior model at a future time point, for example, It is possible to predict a peak traffic volume per subscriber or a peak coefficient with respect to an average traffic volume in consideration of a fluctuation range of a weather condition or a scheduled event. For example, as shown in FIG. 3, the upper limit value of the traffic volume per subscriber of the VOD service can be estimated.

Next, the stream prediction function unit 50 will be described.
There are two main video quality of the current VOD service: SD (Standard Definition) quality and HD (High Definition) quality, but SD quality video content has decreased relatively. There is no doubt that quality video content will increase further. In addition, stereoscopic content such as 3D video has begun to spread, and 3D video has about twice the amount of information compared to normal video, but due to the desire for higher quality and multi-functionality, It is necessary to consider that traffic per number of viewings will increase in the future based on medium- to long-term prediction of communication traffic.

  First, the stream class classification coefficient function unit 51 classifies each stream constituting the entire traffic of the VOD service into three classes, that is, an SD class, an HD quality, and a 3D video, which are exclusive classes. Measure the frequency and frequency. Further, the traffic volume of each class is measured (or stored as a default value).

  The stream class-specific frequency analysis function unit 52 performs multiple regression analysis on each communication frequency of the exclusive class as a time series, and so on, for example, SD quality, HD quality, and 3D video content that are exclusive classes. Analyzes the correlation between classes.

  Based on the correlation analysis between the exclusive classes and the like, the frequency prediction function unit 53 for each stream class is combined with information such as release information for each content class, usage trend / trend, etc. Predict the frequency of use of SD quality content, HD quality content, and 3D video content, or their share (share). Based on this, it is possible to estimate and evaluate the upper limit value of the traffic volume per subscriber for the following year, for example, as shown in FIG.

  Finally, the prediction result integration processing function unit 60 is obtained by the VOD service subscriber number prediction value at the designated future time obtained by the subscriber number prediction function unit 30 and the subscriber behavior prediction function unit 40. Furthermore, a predicted value such as a peak coefficient for the average traffic volume per VOD service subscriber at the same future time point, and exclusive contents of the VOD service at the same future time point obtained by the stream prediction function unit 50 Using the predicted values of class SD quality, HD quality, 3D video usage frequency, or occupancy (share) between classes, accurately peak traffic volume directly attributable to the VOD service, a new innovation communication service Can be predicted.

  The communication traffic prediction apparatus 10 according to the first to second embodiments can be realized by a computer having a CPU, a memory, and an external interface, and a program for controlling these hardware resources. In such a computer, a program for realizing the communication traffic prediction method of the present invention is provided in a state of being recorded on a recording medium such as a flexible disk, a CD-ROM, a DVD-ROM, or a memory card. The CPU writes the program read from the recording medium into the memory, and executes the processes described in the first and second embodiments according to the program.

  Although the present invention has been specifically described above based on the above-described embodiment, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the claims.

10 Communication Traffic Prediction Device 20 Data Input / Output / Storage Function Unit 21 Data Input / Output Unit 22 Operation Input Unit 23 Screen Display Output Unit 24 Main Storage Unit 24
25 Time Series Data Storage Unit 26 Processing Data Storage Unit 27 Output Data Storage Unit 30 Subscriber Number Prediction Function Unit 31 Structure Change Point Detection Function Unit 32 Subscriber Number Prediction Calculation Function Unit 40 Subscriber Behavior Prediction Function Unit 41 Subscriber Behavior Analysis Function unit 42 Subscriber behavior model prediction function unit 50 Stream prediction function unit 51 Stream class classification coefficient function unit 52 Frequency analysis function unit by stream class 53 Prediction function unit by stream class 60 Prediction result integration processing function unit

Claims (7)

A communication traffic prediction device for predicting a traffic volume peak directly attributable to a new communication service in a communication network,
Time-series data storage means for storing time-series data to be subjected to communication traffic prediction;
Reads the actual number of subscriptions of the service to be predicted from the time series data storage means, and performs a regression analysis process using structural change points corresponding to qualitative changes in the service subscriber layer in the process of spreading the new communication service Service subscriber number prediction means for calculating a service subscriber number prediction curve,
Subscriber behavior prediction means for predicting service usage behavior of a subscriber at a future time point while using history information of service usage behavior of the new communication service subscriber acquired from the time series data storage means;
While acquiring data necessary for stream fluctuation prediction from the time-series data storage means, and observing the transition of the frequency of use between exclusive classes of streams based on higher quality and multi-functionality of the new communication service, Stream predicting means for predicting the occupancy rate of usage frequency between exclusive classes of streams,
The communication traffic prediction apparatus, wherein the service subscriber number prediction curve is a function in which an increase in the number of service subscriptions per unit time is proportional to a negative of an elapsed time from a predetermined reference time point. The time series data storage means includes
As time series data of the forecasted service,
Includes traffic characteristics information for each session, various attribute information of the content being communicated,
As the history information of the service usage behavior of the new communication service subscriber,
Includes weather data and various event information,
As data necessary for the stream prediction,
The communication traffic prediction apparatus according to claim 1, which includes a stream band for each exclusive class of content, a viewing frequency, and resource information of future content. The service subscription number prediction means includes:
As the structural change point,
When the number of service subscribers has decreased from an increase,
Or
From the service subscription number actual data, the time axis is converted from the predetermined reference time to the elapsed time, logarithmic conversion is performed, linear conversion of the service subscriber actual value is performed, and a change occurs in the slope,
The communication traffic prediction apparatus according to claim 1, wherein A communication traffic prediction method for predicting a traffic volume peak directly attributable to a new communication service in a communication network,
The service subscription number predicting means reads the subscription number actual data of the prediction target service from the time series data storage means storing the time series data to be subjected to communication traffic prediction, and in the process of spreading the new communication service, the service subscriber A service subscriber number prediction step of calculating a service subscriber number prediction curve by performing a regression analysis process using a structural change point corresponding to a qualitative change of the layer;
Subscriber behavior predicting means for predicting a subscriber's service usage behavior at a future time while using historical information of service usage behavior of the new communication service subscriber acquired from the time-series data storage means A prediction step;
Stream prediction means obtains data necessary for stream fluctuation prediction from the time-series data storage means, and observes changes in the frequency of use between exclusive classes of streams based on higher quality and multi-functionality of the new communication service While performing a stream prediction step for predicting the occupancy rate of usage frequency between exclusive classes of streams at a future time point,
The communication traffic prediction method, wherein the service subscriber number prediction curve comprises a function in which an increase in the number of service subscriptions per unit time is proportional to a negative of an elapsed time from a predetermined reference time point. In the service subscriber number prediction step,
From the time series data storage means, as the time series data of the service to be predicted, traffic characteristic information for each session, various attribute information of the content to be communicated,
In the subscriber behavior prediction step,
From the time-series data storage means, as the history information of the service usage behavior of the new communication service subscriber, obtain weather data, various event information,
In the stream prediction step,
The communication traffic prediction method according to claim 4, wherein the stream bandwidth for each exclusive class of content, the viewing frequency, and the resource information of future content are acquired from the time-series data storage unit as data necessary for the stream prediction. In the service subscription number prediction step,
As the structural change point,
When the number of service subscribers has decreased from an increase,
Or
From the service subscription number actual data, the time axis is converted from the predetermined reference time to the elapsed time, logarithmic conversion is performed, linear conversion of the service subscriber actual value is performed, and a change occurs in the slope,
5. The communication traffic prediction method according to claim 4, wherein Computer
The communication traffic prediction program for functioning as each means of the communication traffic prediction apparatus of any one of Claims 1 thru | or 3.

Images