JP2016034079A - Communication speed selection device, communication speed selection method, and communication speed selection program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To appropriately select the communication speed of soon-to-be-performed data communication.SOLUTION: A terminal 1 includes: a web browser 11 that requests a communication page 30 of a web server 3 at a communication speed based on the prior probability of a bandwidth determination database 12; and a Bayesian update unit 15 that calculates likelihood referring to a probabilistic causal relation table 16 from the traffic of the web server 3 received by the web browser 11, calculates posterior probability by Bayes' theorem on the basis of the calculated likelihood and the prior probability from the bandwidth determination database 12, and updates the bandwidth determination database 12 using the posterior probability as the next prior probability.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technology of a communication speed selection device, a communication speed selection method, and a communication speed selection program.

  Some network services for data communication for mobile terminals as of July 2014 have restrictions on the amount of communication. For example, there is a service in which an upper limit communication amount (7 GB [Gigabytes], etc.) is provided for each terminal, and when the communication amount per month exceeds the upper limit communication amount, the excess is switched to low-speed communication with a maximum communication rate of 128 kbps.

  Non-Patent Document 1 describes a system that suppresses the amount of communication to be used by setting switching between high-speed communication and low-speed communication by an end user's own terminal operation as an IIJmio coupon switch (Miopon). ing. Thereby, since the end user who has not reached the upper limit communication amount can make only his / her desired communication high-speed communication, the stress of waiting for communication can be reduced.

IIJ (Internet Initiative Japan Inc.), “Internet Topic: IIJmio High-Speed Mobile / D Service and Online Charging”, [online], [Search July 22, 2014], Internet <URL: http: // www. iij.ad.jp/company/development/report/iir/pdf/iir_vol20_topic.pdf>

  On the other hand, in the method in which the end user directly operates switching between high-speed communication and low-speed communication, the switching operation is troublesome and the speed is selected according to the end user's expectation, so it cannot be said that the speed selection accuracy is good. Therefore, a system that automatically switches speed without making the end user aware of switching is desired.

  Accordingly, the main object of the present invention is to appropriately select the communication speed of data communication to be performed in the future.

In order to solve the above-mentioned problem, a communication speed selection device according to the present invention is provided for each data server in communication speed for accessing communication data managed by the data server and Bayesian update for obtaining the communication speed. Access to a speed discrimination database that correlates prior probabilities,
Storage means for storing a probabilistic causal relationship table for associating likelihood in Bayesian update for each communication speed used for the communication for each communication amount of the communication data;
A data requesting unit that requests the data for communication from the data server at a communication speed based on the prior probability of the speed determination database;
The likelihood is obtained by referring to the probabilistic causal relationship table from the traffic of the data server received by the data request unit, and based on the obtained likelihood and the prior probability of the speed discrimination database. And a Bayesian update unit that calculates the posterior probability according to Bayes' theorem and updates the posterior probability as the next prior probability in the speed determination database.

  As a result, the posterior probability calculated from the communication amount of the data server received this time is Bayes-updated in the speed determination database as the next prior probability, so that the communication speed of data communication to be performed can be selected appropriately.

  The present invention is characterized in that the communication speed selection device has communication means for accessing the external speed discrimination database instead of having the speed discrimination database in its storage means.

  Thereby, even if the communication speed selection device is a low-performance mobile terminal, it is possible to reduce the processing load and storage capacity required for Bayesian updating.

  The present invention is characterized in that the Bayesian update unit obtains the likelihood by referring to the probabilistic causal relationship table from the traffic received in a predetermined period from the start of reception of the communication data.

  As a result, an appropriate communication speed can be set according to the waiting time allowed by humans such as the 8-second rule.

  In the present invention, the data request unit adds information indicating a communication speed based on the prior probability of the speed determination database to a message requesting the communication data, so that the requesting data server The communication data corresponding to the communication speed is selected as a reception target.

  As a result, it is possible to avoid an event in which a packet loss occurs when trying to receive a large amount of communication data even though low-speed communication is selected. Furthermore, when low speed communication is selected, content with sufficient quality can be browsed.

  In the present invention, when the prior probability of the speed determination database is not set, the Bayes update unit sets the prior probability according to the requested media type of the communication data as the initial prior probability. It writes out to the speed discrimination database.

  Thereby, even if it is the data for communication without a communication history, an appropriate communication speed can be selected before a Bayesian update.

  According to the present invention, it is possible to appropriately select the communication speed of data communication to be performed in the future.

It is a block diagram which shows the communication speed selection system regarding one Embodiment of this invention. It is a block diagram which shows the form which changed the position of the zone | band discrimination | determination database in the communication speed selection system regarding one Embodiment of this invention. It is a flowchart which shows the download process of the web page in the form of FIG. 1 or FIG. 2 regarding one Embodiment of this invention. It is a flowchart which shows the process which acquires a communication speed from the zone | band discrimination | determination database regarding one Embodiment of this invention. It is a block diagram which shows the zone | band discrimination | determination database (FIG.5 (a)) and probabilistic causal relationship table (FIG.5 (b)) regarding one Embodiment of this invention. It is a block diagram which shows the form which prepared the page according to speed in the communication speed selection system regarding one Embodiment of this invention. It is a flowchart which shows the download process of the web page in the form of FIG. 6 regarding one Embodiment of this invention.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a configuration diagram showing a communication speed selection system.
In the communication speed selection system, a terminal 1 and a Web server 3 are connected by a communication carrier network 2. Although FIG. 1 illustrates two servers, Web servers 3 (A, B), the number of Web servers 3 is not particularly limited.
Each device of these communication speed selection systems is configured as a computer having a CPU (Central Processing Unit), a memory, a hard disk (storage means), and a network interface, and this computer executes a program read on the memory by the CPU. Thus, each processing unit is operated.

Each device is assigned the following combination of IP address and port number in order to receive the communication page 30.
Web server 3 (A): IP address “α1” and port number “α2”.
Web server 3 (B): IP address “β1” and port number “β2”.
Terminal 1: IP address “γ1” and port number “γ2”.

The terminal 1 includes a Web browser 11, a bandwidth determination database 12, a terminal-side speed switching unit 13, a packet measurement unit 14, a Bayes update unit 15, and a probabilistic causal relationship table 16.
The web browser 11 is an application for referring to the communication page 30. The Web browser 11 transmits an http (Hypertext Transfer Protocol) request to the Web server 3 in order to request the communication page 30 indicated by the input URL (Uniform Resource Locator). Then, the Web browser 11 causes the user to browse the communication page 30 that is responded (downloaded) from the Web server 3.
Note that the IP address of the Web server 3 indicated by the host name included in the URL can be specified by referring to DNS (Domain Name System).

The bandwidth determination database 12 uses a combination of the IP address and port number of the Web server 3 as a key, and a communication speed determination result (high speed communication or low speed communication) for accessing the communication page 30 of the Web server 3. The prior probability for obtaining the discrimination result is held.
The terminal side speed switching unit 13 switches the communication speed of the terminal 1 between high speed and low speed based on the determination result of the band determination database 12. The low-speed communication is, for example, a maximum of 2 Mbps (2 Mbytes in 8 seconds), and the high-speed communication is a band larger than the low-speed communication.

The packet measurement unit 14 is a measurement tool such as pcap (Packet Capture). The packet measuring unit 14 measures the communication amount of the downloaded communication page 30 by counting the number of IP packets of the communication page 30 requested by the Web browser 11 and the data size per packet.
The packet measuring unit 14 counts the amount of communication downloaded for 8 seconds from the start of download for each communication page 30 based on, for example, the 8-second rule.
Here, the 8-second rule means that if it takes more than 8 seconds for a page to be displayed after a user visits a site, the user cannot go to another site without waiting. It is a rule that is very rare to come back again.

The Bayesian update unit 15 calculates the posterior probability based on the prior probability of the bandwidth discrimination database 12 and the likelihood acquired from the probabilistic causality table 16 using the measurement result of the packet measurement unit 14 as a key, and the posterior The probability is updated to the band discrimination database 12 as the next prior probability (that is, the prior probability is updated by Bayes).
The probabilistic causal relationship table 16 is a likelihood indicating the plausibility (likelihood) of the communication speed (low speed communication or high speed communication) used for the communication for each measurement result (8-second communication amount) of the packet measurement unit 14. It is a table for making a degree search.

The telecommunications carrier network 2 (or a control server provided in the network) includes a network side speed switching unit 23 and a network side measuring unit 24.
The network side measurement unit 24 counts the amount of high-speed communication for each terminal 1 (or for each user of the terminal 1).
The network-side speed switching unit 23 is a speed designated by the terminal-side speed switching unit 13 for the user terminal 1 whose communication volume measured by the network-side measuring unit 24 exceeds the limit of the contracted capacity (for example, 7 GB (gigabyte)). Regardless of whether or not, forcibly switch to low-speed communication.

FIG. 2 is a configuration diagram showing a form in which the position of the band discrimination database 12 is changed in the communication speed selection system. Compared with FIG. 1, the bandwidth discrimination database 12 of FIG. 2 is provided on the side of the carrier network 2. This bandwidth discrimination database 12 is configured as, for example, a network-attached hard disk (NAS). That is, the terminal 1 has a communication unit for accessing the external bandwidth determination database 12.
With this configuration, it is possible to reduce the load (processing capacity and storage capacity) on the terminal 1 side.

FIG. 3 is a flowchart showing download processing of the Web page (communication page 30) in the form of FIG.
As S11, the Web browser 11 accepts designation of a URL indicating a Web page to be downloaded.
As S <b> 12, the terminal-side speed switching unit 13 acquires the communication speed of the Web page designated in S <b> 11 from the bandwidth determination database 12. Details of the processing of S12 will be described later with reference to FIG.
As S <b> 13, the terminal-side speed switching unit 13 switches to the communication speed acquired in S <b> 12 and requests the specified Web page from the Web server 3. Thereby, the communication speed can be automatically switched.
As S <b> 14, the web browser 11 downloads the requested web page, and the packet measuring unit 14 measures the downloaded packet amount.
As S15, the Bayesian update unit 15 performs Bayesian update of the bandwidth determination database 12 based on the measured packet amount. Accordingly, it is possible to select an appropriate communication speed following the change in the data size of the Web page.

FIG. 4 is a flowchart showing a process for acquiring the communication speed from the bandwidth discrimination database 12.
As S111, the terminal side speed switching unit 13 refers to the DNS from the address (URL) of the Web page specified in S11, and specifies a Web server that transmits the Web page.
As S112, the terminal-side speed switching unit 13 searches the bandwidth determination database 12 for the Web server specified in S111.
In S113, the terminal-side speed switching unit 13 determines whether the Web server specified in S111 exists in the bandwidth determination database 12 as a search result in S112. If Yes in S113, the process proceeds to S121, and if No, the process proceeds to S114.

  As S114, the terminal-side speed switching unit 13 determines whether or not manual input of prior probabilities has been accepted. The manual input here indicates that a numerical value of the prior probability is directly input from the outside. In addition to the input from the user via an input device such as a keyboard, the numerical value of the prior probability is input from another device to the terminal 1. Arbitrary input forms, such as transmitting to, are included. If Yes in S114, the process proceeds to S115, and if No, the process proceeds to S116.

In S115, the Bayesian update unit 15 adopts the manually input value in S114 as the prior probability, and writes the adopted prior probability in the bandwidth determination database 12 in association with the Web server specified in S111. For example, when the user knows in advance that the media type of the Web page to be accessed is a moving image, the user may input a manual input value that prioritizes high-speed communication such as prior probability = 70%.
Or, based on the media type of the web page accessed by the terminal 1, refer to the initial value for each preset media type (eg, prior probability = 70% for video) and determine the prior probability to adopt May be.

  In S116, the Bayesian update unit 15 writes the prior probability = 50% (default value) in the bandwidth determination database 12 in association with the Web server specified in S111. This default value can be derived from the principle of insufficient reason.

As S121, the terminal side speed switching unit 13 determines whether or not the prior probability ≧ 50% is satisfied. If Yes in S121, the process proceeds to S122, and if No, the process proceeds to S123.
In S122, the terminal-side speed switching unit 13 sets the communication speed used for downloading the Web page designated in S11 as “high-speed communication”.
In S123, the terminal side speed switching unit 13 sets the communication speed used for downloading the Web page designated in S11 to “low speed communication”.

FIG. 5A is a configuration diagram showing the bandwidth determination database 12. FIG. 5A illustrates the contents of the three-stage band discrimination database 12 (before update → after the first Bayes update → after the second Bayes update).
The bandwidth determination database 12 uses a combination of the IP address and port number of the Web server 3 as a key, and a communication speed determination result (high speed communication or low speed communication) for accessing the communication page 30 of the Web server 3. The prior probability for obtaining the discrimination result is held.
For example, the communication speed for accessing the Web server B before the Bayes update is the high-speed communication (S122) specified from the prior probability input in S116 = 50% (default value) (Yes in S121).

FIG. 5B is a configuration diagram showing the probabilistic causal relationship table 16.
The probabilistic causal relationship table 16 is a likelihood indicating the plausibility (likelihood) of the communication speed (low speed communication or high speed communication) used for the communication for each measurement result (8-second communication amount) of the packet measurement unit 14. It is a table for making a degree search. Since each Web page is continuously updated, the data amount also changes every moment. Therefore, the likelihood is defined in the range of 0 to 100% as well as the probability.
Note that the terminal 1 may allow the user to manually input the probabilistic causal relationship table 16 or from an access history (probability of accessing by high-speed communication and probability of accessing by low-speed communication) for a sufficient time in advance. The probabilistic causality table 16 may be created automatically.

Hereinafter, the Bayes update process by the Bayes update part 15 is demonstrated. Suppose that result B occurs from one of K multiple causes A1, A2, ..., Ak. Under the precondition that the cause Ai has occurred, the probability that the result B will occur is defined as the conditional probability P (B | Ai). The following equation is Bayes' theorem.
P (Ai | B) = P (Ai) x P (B | Ai) / (P (A1) x P (B | A1) + P (A2) x P (B | A2) + ... + P (Ak) x P (B | Ak))

Prior probability P (Ai) is a probability that event Ai will occur. Prior probabilities are mutually exclusive, and A1, A2,..., Ak do not occur simultaneously (P (A1 | B) + P (A2 | B) + ... + P (Ak | B) = 1).
Likelihood P (B | Ai) is a conditional probability that event B will occur under the precondition that event Ai has occurred. In other words, the likelihood P (B | Ai) is the likelihood (probability) of estimating that the cause is Ai when the result B is observed.
The posterior probability P (Ai | B) is a conditional probability that the event Ai occurs under the precondition that the event B has occurred. In other words, the posterior probability P (Ai | B) is a probability that the result B is generated from the cause Ai.

The Bayesian update unit 15 performs the first Bayesian update for the Web server B in FIG.
Assume that the packet measuring unit 14 measures 1.5 MB as the communication amount of the Web page received using high-speed communication selected from the prior probability before Bayesian update = 50% (default value). The Bayesian update unit 15 calculates the Bayesian theorem, assuming that A1 is low-speed communication, A2 is high-speed communication, and B is 1.5 MB (= “1.2 M or more and less than 1.6 M” in the probabilistic causal relationship table 16).
P (A1 | B) = P (A1) × P (B | A1) / (P (A1) × P (B | A1) + P (A2) × P (B | A2))
= 0.5 x 0.15 / (0.5 x 0.15 + 0.5 x 0.08) = 0.65
P (A2 | B) = 1-P (A1 | B) = 1-0.65 = 0.35 (= This posterior probability and next prior probability)

The Bayesian update unit 15 performs the second Bayesian update for the Web server B in FIG.
Assume that the packet measuring unit 14 measures 2 MB as the communication amount of the Web page received using the low-speed communication selected from the prior probability after the first Bayes update = 35%. The Bayesian update unit 15 calculates the Bayesian theorem, assuming that A1 is a low-speed communication, A2 is a high-speed communication, and B is 2 MB (= “2.0M or more and less than 2.4M” in the probabilistic causality table 16).
P (A1 | B) = P (A1) × P (B | A1) / (P (A1) × P (B | A1) + P (A2) × P (B | A2))
= 0.65 x 0.07 / (0.65 x 0.07 + 0.35 x 0.14) = 0.48
P (A2 | B) = 1-P (A1 | B) = 1-0.48 = 0.52 (= this posterior probability, next prior probability)
Thereby, next time, high-speed communication can be selected.

FIG. 6 is a configuration diagram showing a mode in which pages for each speed are prepared in the communication speed selection system. The communication page 30 in FIG. 1 is divided into a high-speed communication page 31 and a low-speed communication page 32 in FIG. The high-speed communication page 31 and the low-speed communication page 32 are both pages accessed by the same URL, but the high-speed communication page 31 accessed by high-speed communication is more than the low-speed communication page 32 accessed by low-speed communication. The amount of data is large.
For example, the same video material data may be configured as a high-speed communication page 31 encoded with high image quality and a low-speed communication page 32 encoded with low image quality, or the same video material data may be configured as a high-speed communication encoded with a video. It may be configured as a page for low speed communication and a page for low speed communication 32 cut out as a still image. The high image quality is, for example, high resolution or data encoded at a low compression rate.

FIG. 7 is a flowchart showing Web page download processing in the form of FIG. S11, S12, and S15 are as described in FIG.
As S13b, similarly to S13, the terminal-side speed switching unit 13 switches to the communication speed acquired in S12, and requests the specified Web page from the Web server 3. Here, in S13b, the terminal 1 adds a low-speed communication flag (lowSpeed) or a high-speed communication flag (highSpeed) to the User-Agent header of the message (http request) requesting the specified Web page, Send. This flag indicates the switched communication speed.
Note that the User-Agent header is originally a header for designating the type of the Web browser 11 (Mozilla / 5.0 or the like), and the header for adding the communication speed may be another header.

As S14b, the Web browser 11 downloads the requested Web page, and the packet measurement unit 14 measures the downloaded packet amount, as in S14. Here, in S <b> 14 b, the Web server 3 responds to the Web browser 11 of the terminal 1 with the high-speed communication page 31 or the low-speed communication page 32 at the specified speed in accordance with the User-Agent header of the received http request.
As a result, an amount of Web pages suitable for the communication speed and Web pages that require communication quality (broadband and low delay) suitable for the communication speed are responded.

In the present embodiment described above, the main feature is that the terminal 1 automatically determines the communication band (low speed communication or high speed communication) necessary for referring to the continuously updated Web page by utilizing Bayesian theory. And The discrimination result is stored in the band discrimination database 12. Each time the Web page is referred to, the Bayesian update unit 15 automatically updates the determination result (prior probability of the bandwidth determination database 12) by Bayesian update.
Here, the Bayesian update unit 15 refers to the probabilistic causal relationship table 16 for obtaining the likelihood. This probabilistic causal relationship table 16 assumes the time (for example, 8 seconds) when a response is returned after an end user accesses for Web reference, and the likelihood and low speed of high-speed communication so that the response is returned within that range. The likelihood of communication is defined.

As a result, the terminal-side speed switching unit 13 can automatically determine the communication speed with reference to the Bayes-updated band determination database 12, so that the end user switching can be performed rather than the end user subjective switching manually. In addition to reducing the time and effort of operation, the selection accuracy for selecting an appropriate speed can be improved.
That is, the data amount of the Web page is updated as the content is updated, but the prior probability of the bandwidth discrimination database 12 that is a probability expression is also updated by the Bayesian update, so that the selection accuracy can be improved. .

1 terminal (communication speed selection device)
2 Telecommunications carrier network 3 Web server (data server)
11 Web browser (data request part)
12 Bandwidth discrimination database (speed discrimination database)
DESCRIPTION OF SYMBOLS 13 Terminal side speed switching part 14 Packet measuring part 15 Bayes update part 16 Probabilistic causal relationship table 23 Network side speed switching part 24 Network side measuring part 30 Communication page (data for communication)
31 Page for high-speed communication 32 Page for low-speed communication

Claims (7)

For each data server, it is possible to access a speed determination database that associates a communication speed for accessing communication data managed by the data server and a prior probability in Bayesian update for obtaining the communication speed,
Storage means for storing a probabilistic causal relationship table for associating likelihood in Bayesian update for each communication speed used for the communication for each communication amount of the communication data;
A data requesting unit that requests the data for communication from the data server at a communication speed based on the prior probability of the speed determination database;
The likelihood is obtained by referring to the probabilistic causal relationship table from the traffic of the data server received by the data request unit, and based on the obtained likelihood and the prior probability of the speed discrimination database. And a Bayesian update unit that calculates the posterior probability according to Bayes' theorem and updates the posterior probability to the speed determination database as the next prior probability. The communication speed selection apparatus according to claim 1, wherein the communication speed selection apparatus has the speed determination database in its storage means. The said Bayesian update part calculates | requires the said likelihood with reference to the said probabilistic causal relationship table | surface from the traffic received in the predetermined period from the reception start of the said communication data, The Claim 1 or Claim 2 characterized by the above-mentioned. The communication speed selection device described. The data request unit adds the information indicating the communication speed based on the prior probability of the speed determination database to the message requesting the communication data, thereby setting the communication speed to the data server that is the request destination. The communication speed selection device according to any one of claims 1 to 3, wherein the communication data corresponding to the selected data is selected as a reception target. When the prior probability of the speed determination database is not set, the Bayes update unit sets the prior probability according to the media type of the requested communication data as the initial prior probability as the speed determination database. The communication speed selection device according to any one of claims 1 to 4, wherein the communication speed selection device is written. The communication speed selection device includes a storage unit, a data request unit, and a Bayes update unit.
For each data server, it is possible to access a speed determination database that associates a communication speed for accessing communication data managed by the data server and a prior probability in Bayesian update for obtaining the communication speed,
The storage means stores a probabilistic causality table for associating likelihood in Bayesian update for each communication speed used for the communication for each communication amount of the communication data,
The data request unit requests the data for communication from the data server at a communication speed based on the prior probability of the speed determination database,
The Bayesian update unit obtains the likelihood by referring to the probabilistic causal relationship table from the communication amount of the data server received by the data request unit, and the obtained likelihood and the event in the speed determination database. A communication speed selection method characterized in that a posteriori probability is calculated according to Bayes' theorem based on a prior probability, and the posteriori probability is updated to the speed discrimination database as the next prior probability.   A communication speed selection program for causing the communication speed selection device, which is a computer, to execute the communication speed selection method according to claim 6.

Images