JP6640067B2 - Delivery control device, delivery control method and program - Google Patents

Description

  The present invention relates to a distribution control device, a distribution control method, and a program.

  It is important for a service provider to predict communication quality and perform appropriate control in order to improve the quality of experience of the provided service. For example, in a moving image distribution service, if high-quality moving image data is transmitted / received (downloaded) in an environment with a low transfer rate, an event occurs in which reproduction is interrupted halfway due to insufficient data. However, there are situations where the quality of experience is increasing.

  As one method of predicting and controlling communication quality, the past communication quality is stored in a DB, and the communication quality is predicted and controlled using the DB information and the communication environment information of a user who is about to communicate. Technology (Non-Patent Document 1) exists. When the transfer rate is handled in the DB in this technology, past transfer rates (goodput) measured in the past are collected, and a line transfer rate (throughput) in communication to be performed is predicted.

[Invited Talk] Introduction of initiatives for realizing QoE-centric operation, Koji Yamamoto, IEICE Technical Report, vol. 114, no. 299, ICM2014-24, pp. 49-52, November 2014.

  However, when the transmission control protocol (TCP) is adopted as a communication transport layer protocol and the data size transmitted / received at one time is small due to intermittent division transmission or the like, a congestion window (estimating the congestion state of the network) There is a possibility that the congestion window is set small for the control (adjusting the data transmission amount), and the goodput is not sufficient for the throughput. Also, in the case of a mobile network, if the size of data to be transmitted and received is small, the allocation of radio wave resources in base station control decreases, and goodput may not be sufficient.

  Therefore, if the prediction is performed by treating the measured good put as it is as a throughput, there is a disadvantage that the throughput is predicted to be lower. Also, even if the throughput prediction is correct, when actually downloading data, since the goodput changes depending on the communication control situation described above, there is a disadvantage that it is not accurate to calculate the throughput value as it is as the goodput. Occurs.

  As described above, in the related art, when predicting the throughput of the communication to be performed based on the past communication quality, the actually measured goodput is used as the throughput of the network. In addition, since the predicted throughput is used as the predicted goodput, the difference between the goodput and the throughput due to the influence of the communication control is not considered. For this reason, there have been cases where downloading of moving image data is performed under control conditions that are not suitable for communication quality.

  The present invention has been made in view of the above points, and an object of the present invention is to enable prediction of throughput in consideration of the influence of communication control and control of communication based on the prediction.

Therefore, in order to solve the above-described problem, the distribution control device includes, for each distribution of moving image data, communication environment information indicating a communication environment of a distribution destination, communication control-related information related to communication control in the distribution, and measurement of the distribution. A storage unit for storing the put in association with the first communication environment information relating to the distribution destination of the first distribution of the moving image data, and an actual measurement good stored in the storage unit in association with the communication environment information having a common part with the first communication environment information A first calculating unit that calculates a predicted value of the throughput based on the put, a function indicating a relationship between the predicted value of the throughput and the communication control-related information, and the good put, the first calculating unit calculating the function; A second calculating unit configured to calculate a goodput predicted value by applying the predicted value and the communication control related information related to the first distribution, and Possess a selection unit for selecting a control condition in accordance with the predicted value of Ddoputto, and the storage unit, for each delivery chunk of video data, a communication environment information, the communication control related information, and the measured value of goodput The communication control-related information includes an accumulated data amount from the start of distribution of the moving image data, a chunk size, and a distribution interval with a previous chunk .

  It is possible to predict the throughput in consideration of the influence of the communication control and control the communication based on the prediction.

FIG. 2 is a diagram illustrating a network configuration example according to the embodiment of the present invention. FIG. 3 is a diagram illustrating an example of a hardware configuration of a quality control server 10 according to the embodiment of the present invention. FIG. 2 is a diagram illustrating a functional configuration example according to the embodiment of the present invention. It is a flow chart for explaining an example of a processing procedure performed in a quality collection process. FIG. 3 is a diagram illustrating a configuration example of a quality DB; It is a flow chart for explaining an example of a processing procedure performed in a quality control process.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating a network configuration example according to an embodiment of the present invention. In FIG. 1, the moving image distribution server 30 and the client terminal 20 are connected via a network such as the Internet. Further, the client terminal 20 and the quality control server 10 are connected via a network such as the Internet.

  The moving image distribution server 30 is one or more computers that provide a moving image data distribution service (hereinafter, referred to as “moving image distribution service”) to the client terminal 20. The moving image distribution server 30 intermittently divides the moving image data into chunk units and transmits the divided data to the client terminal 20. For example, distribution of moving image data follows ABR (Adaptive Bitrate Streaming). That is, for each moving image data, data re-encoded at several patterns of bit rates is stored in the moving image distribution server 30 for each chunk, which is an example of a division unit at a fixed time interval.

  The client terminal 20 is a terminal that is a distribution destination of the moving image data in the moving image distribution service and reproduces the distributed moving image data. For example, a PC (Personal Computer), a smartphone, a tablet terminal, or the like may be used as the client terminal 20.

  The quality control server 10 is one or more computers that predict the throughput related to the distribution of moving image data when the client terminal 20 receives the moving image distribution service.

  FIG. 2 is a diagram illustrating a hardware configuration example of the quality control server 10 according to the embodiment of the present invention. The quality control server 10 in FIG. 2 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, and the like, which are interconnected by a bus B.

  A program for realizing the processing in the quality control server 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100, the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100. However, the program need not always be installed from the recording medium 101, and may be downloaded from another computer via a network. The auxiliary storage device 102 stores installed programs and also stores necessary files and data.

  The memory device 103 reads out the program from the auxiliary storage device 102 and stores it when there is an instruction to start the program. The CPU 104 executes a function related to the quality control server 10 according to a program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

  Note that the client terminal 20 and the moving image distribution server 30 may also have the same hardware as in FIG. Desirably, the client terminal 20 has a display device, an input device that receives an input from a user, and the like.

  FIG. 3 is a diagram illustrating an example of a functional configuration according to the embodiment of the present invention. 3, the client terminal 20 includes a reproducing unit 21, a measured goodput calculating unit 22, a control condition selecting unit 23, and the like. These units are realized by a process in which one or more programs installed in the client terminal 20 are executed by the CPU of the client terminal 20. The client terminal 20 also has a control condition DB 24. The control condition DB 24 can be realized using, for example, an auxiliary storage device or the like of the client terminal 20.

  The reproducing unit 21 receives (downloads) the moving image data from the moving image distribution server 30 and reproduces the moving image data.

  The measured goodput calculation unit 22 calculates the measured value of the goodput based on the time required for receiving the moving image data and the data size of the moving image data, and associates the calculation result with the communication environment information and the communication control related information. Record. Hereinafter, the measured value of the good put calculated by the measured good put calculation unit 22 is referred to as “measured good put”.

  The communication environment information is information on the communication environment of the client terminal 20 at the time of distributing the moving image data. The communication environment information includes, for example, a communication date and time, a communication method (Wi-fi (registered trademark), mobile NW), a communication carrier, cell information of the mobile NW, a moving image distribution server 30, a location information, a radio wave intensity, and the like. It is composed of items.

  The communication control related information is information (information that can estimate a communication control state) related to communication control regarding distribution of moving image data. The communication control-related information includes items such as an accumulated data amount from the start of communication (distribution start) of the moving image data, a chunk size, and a free time from the previous download. Note that the measured goodput, communication environment information, and communication control-related information are used by the quality control server 10 to calculate a predicted value of a throughput in a future moving image distribution service. The predicted value of the throughput calculated by the quality control server 10 is called “predicted throughput”.

  The control condition selection unit 23 selects, for example, a control condition for maximizing or improving the perceived quality in the predicted throughput from a plurality of control conditions prepared in advance. The selected control condition is used for controlling the distribution of the moving image data by the moving image distribution server 30. The control condition refers to a parameter specified to the moving image distribution server 30 when receiving the moving image distribution service. For example, in the case of moving image data for which a plurality of bit rates are prepared, the bit rate corresponds to the control condition.

  The control condition DB 24 stores a plurality of control conditions prepared in advance.

  The quality control server 10 includes an actual measurement goodput collection unit 11, a throughput prediction unit 12, and the like. These units are realized by a process that causes the CPU 104 to execute one or more programs installed in the quality control server 10. The quality control server 10 also uses the quality DB 13. The quality DB 13 can be realized using, for example, a storage device that can be connected to the auxiliary storage device 102 or the quality control server 10 via a network.

  The measured goodput collection unit 11 collects measured goodput, communication environment information, and communication control-related information from the client terminal 20. The collected information is stored in the quality DB 13.

  The throughput prediction unit 12 calculates a predicted throughput related to the distribution of the moving image data based on the communication environment information when the client terminal 20 receives the distribution of the moving image data, the information stored in the quality DB 13, and the like.

  Note that the arrangement state of each functional unit shown in FIG. 3 is an example. Other arrangement states may be adopted as long as information can be exchanged between the functional units. For example, the throughput prediction unit 12 may be arranged in the client terminal 20.

  Hereinafter, a processing procedure executed in the client terminal 20 or the quality control server 10 will be described. In the present embodiment, a quality collection process for accumulating communication quality related to the distribution of moving image data in the past in the quality DB 13, and communication between the information accumulated in the quality DB 13 and the client terminal 20 to communicate (reproduce moving image data) from now on It is roughly classified into a quality control process of predicting communication quality using environment information and controlling a video distribution service.

  FIG. 4 is a flowchart illustrating an example of a processing procedure executed in the quality collection process. The processing procedure of FIG. 4 is executed for each chunk of moving image data to be reproduced. Hereinafter, a chunk to be processed is referred to as a “target chunk”.

  In step S101, the reproduction unit 21 downloads a target chunk from the video distribution server 30. Next, the measured goodput calculation unit 22 calculates the measured goodput by dividing the data size (chunk size) of the target chunk by the time required for download (download time) (S102). Subsequently, the measured goodput calculation unit 22 transmits the calculated measured goodput, the communication environment information of the client terminal 20, and the communication-related information related to the distribution of the target chunk to the quality control server 10 (S103).

  When receiving the measured goodput, the communication environment information, and the communication-related information (S104), the measured goodput collection unit 11 of the quality control server 10 stores the received information in the quality DB 13 (S105).

  FIG. 5 is a diagram illustrating a configuration example of the quality DB 13. As shown in FIG. 5, the communication environment information, the communication control information, and the measured goodput are stored in the quality DB 13 for each chunk.

  The communication environment information includes items such as a communication date and time, a communication method, a communication carrier, cell information, a distribution server, position information, radio wave intensity, and the like. The communication date and time is the date and time when the chunk was downloaded. The communication method is a communication method used for downloading a chunk. The telecommunications carrier is the telecommunications carrier used to download the chunk. The cell information is identification information of a cell to which the client terminal 20 belonged when the chunk was downloaded. The distribution server is address information of the moving image distribution server 30 that is the distribution source of the chunk. The position information is information (for example, latitude and longitude) indicating the position of the client terminal 20 at the time of downloading the chunk. The radio wave intensity is the radio wave intensity at the time of downloading the chunk. Note that each item constituting the communication environment information can be determined based on information observed or measured by the client terminal 20 at the time of downloading the chunk, information stored in the client terminal 20, or the like.

  On the other hand, the communication control related information includes items such as an accumulated data amount, a chunk size, and a free time. The cumulative data amount is a cumulative value of the data amount downloaded from the start of reproduction of the moving image data to which the chunk belongs. The chunk size is the data size of the chunk. The idle time is a time interval from when the chunk before the chunk is downloaded to when the download of the chunk starts. That is, the idle time is a chunk distribution interval.

  According to the above-described processing procedure, a record is stored in the quality DB 13 for each result of the distribution of the moving image data.

  By executing the processing procedure shown in FIG. 4 for a plurality of moving image data, a record for the plurality of moving image data may be stored in the quality DB 13. In addition, by executing the processing procedure illustrated in FIG. 4 with respect to the plurality of client terminals 20, records regarding the plurality of client terminals 20 may be stored in the quality DB 13.

  Steps S103 and S104 may be executed collectively for a plurality of chunks. For example, after the reproduction of the moving image data is completed, all the chunks of the moving image data may be collectively executed.

  FIG. 6 is a flowchart illustrating an example of a processing procedure executed in the quality control process.

  When an instruction to start viewing moving image data is input to the client terminal 20 (S201), the control condition selection unit 23 acquires current communication environment information regarding the client terminal 20 (S202). Subsequently, the control condition selection unit 23 transmits the communication environment information to the quality control server 10 (S203).

  Upon receiving the communication environment information, the throughput prediction unit 12 of the quality control server 10 searches the quality DB 13 for a record group including communication environment information similar to the communication environment information (S204). The communication environment information similar to the communication environment information refers to communication environment information having a common part with the communication environment information. For example, the communication environment information having the same combination of the communication environment information and the communication time information, the time zone of the communication date and time, the communication method, the communication carrier, and the cell information as the communication environment information is similar to the communication environment information. It may be. The same time zone of the communication date and time may be that the difference between the communication date and time to be compared is within a predetermined time (for example, within 30 minutes).

  Subsequently, the throughput prediction unit 12 refers to the communication control-related information of the searched record group, and sets a record in which it is assumed that the goodput and the throughput are deviated due to the influence of the communication control as an outlier. It is removed from the group (S205). For example, a record that satisfies one of the conditions that the accumulated data amount is equal to or less than a predetermined value, the chunk size is equal to or less than a predetermined value, and the idle time is equal to or more than a predetermined value is removed.

  The record in which the accumulated data amount is equal to or less than the predetermined value is regarded as an outlier because the communication control has a characteristic that the communication speed is suppressed at the start of communication. Records with chunk sizes smaller than a predetermined value are regarded as outliers because, in the communication of small-sized data, the bandwidth is not sufficiently used, and the measured goodness is lower than the actual performance of the network. This is because there is a possibility that a put has been obtained. Also, a record in which the idle time is equal to or more than the predetermined value is regarded as an outlier because, when the idle time is equal to or more than the predetermined value, the communication speed may be suppressed by the same control as at the start of communication. Because there is. Note that each of the predetermined values may be obtained by an experiment or the like.

  Subsequently, the throughput prediction unit 12 calculates a weighted average of the measured goodputs of the remaining record groups as the predicted throughput (S206). That is, the measured good put which is stored in the quality DB 13 in association with the communication environment information similar to the received communication environment information, and which is not related to the predetermined value of the communication control related information. Is calculated based on the estimated throughput. The weight in the weighted average may be increased for new records, for example. Alternatively, the weights for all records may be the same. That is, a simple average value may be used as the predicted throughput. The throughput prediction unit 12 transmits the calculated predicted throughput to the client terminal 20 (S207).

  Upon receiving the predicted throughput, the control condition selection unit 23 of the client terminal 20 calculates a predicted value (predicted good put) of a good put related to the current distribution of the moving image data based on the predicted throughput (S208).

  The predicted goodput indicates the correspondence between the communication control-related information (accumulated data amount from the start of communication, chunk size, free time since the previous download, etc.), the predicted throughput, and the predicted goodput regarding the current distribution. Is calculated by applying the communication control related information and the predicted throughput to the corresponding relational expression (function).

すなわち、グッドプットとスループットとが乖離していることが推測される状態においては、（１−ａ／（チャンクサイズ＋ａ））×予測スループットに基づいて、予測グッドプットが算出される。一方、グッドプットとスループットとが乖離していることが推測されない状態においては、予測スループットがそのまま予測グッドプットとされる。

That is, in a state where it is estimated that the goodput and the throughput are different, the predicted goodput is calculated based on (1−a / (chunk size + a)) × the predicted throughput. On the other hand, in a state where it is not estimated that the goodput and the throughput are different from each other, the predicted throughput is directly used as the predicted goodput.

  Note that the chunk size of the chunk to be downloaded from now on is substituted for the chunk size in the above-mentioned correspondence relational expression. The accumulated data amount from the start of the current reproduction to the current time is substituted for the accumulated data amount to be compared with b. Further, the free time from the download of the previous chunk is assigned to the free time to be compared with c for the chunk to be downloaded from now on. Therefore, when FIG. 6 is executed for each chunk, the chunk size, the accumulated data amount, and the free time in the correspondence expression change as the chunk to be downloaded advances.

  The above-mentioned correspondence expression is a function that receives a predicted throughput, which is fitted to the relationship between the predicted throughput and the communication control-related information obtained in advance by an experiment and the measured goodput at the time of the experiment.

  From the relationship between the predicted throughput and communication control related information obtained by the experiment and the measured goodput at the time of the experiment, for example, a function as described above is determined as a mapping function, and the fitting accuracy is determined using a least square method or the like. The coefficient a is derived so as to obtain a good function shape. The coefficients b and c are values determined by the area where the throughput and the goodput are separated by the communication control. Note that the above mapping function is an example. For example, a polynomial function may be used as the mapping function.

  Subsequently, the control condition selection unit 23 acquires a plurality of control conditions stored in the control condition DB 24 (S209). The control condition is information indicating an image quality, and each control condition is associated with a good put (required good put) required for reproduction of the image quality. The image quality of each control condition is different from each other. In other words, the required goodput is stored in the control condition DB 24 for each image quality option.

  The information indicating the image quality is, for example, a bit rate. The resolution from which the bit rate is calculated, the frame rate of the moving image data, the compression ratio of the moving image data, and the like may be information indicating the image quality.

  Subsequently, the control condition selection unit 23 compares the required goodput of each control condition with the predicted goodput and extracts some control conditions (S210). Specifically, a control condition associated with a required goodput smaller than the predicted goodput is extracted.

  Subsequently, the control condition selection unit 23 selects a control condition having the highest image quality from among the extracted control conditions as an adoption target (S211). Subsequently, the reproducing unit 21 controls distribution of the moving image data by the moving image distribution server 30 according to the selected control condition (S212). For example, the reproducing unit 21 downloads the data of the bit rate according to the selected control condition from the moving image distribution server 30 and reproduces the data.

  Note that the required goodput associated with the control condition may be set according to the purpose of moving image distribution. For example, in the case where the purpose of maximizing the perceived quality is to enhance the image quality without interrupting the reproduction in moving image distribution, a good put that does not interrupt the reproduction may be set as a necessary good put. If the interruption of the reproduction is permitted, a lower goodput may be set as a necessary goodput as a goodput for the same image quality.

  Note that the quality control process may be executed only at the start of viewing, or may be executed each time a chunk is downloaded. Further, the predicted throughput for each communication environment information may be calculated in advance, and may be stored in the quality control server 10 in association with the communication environment information.

  As described above, according to the present embodiment, it is possible to predict the throughput in consideration of the influence of the communication control and control the communication based on the prediction.

  In the present embodiment, the quality control device and the client terminal 20 are an example of a distribution control device. The quality DB 13 is an example of a storage unit. The throughput prediction unit 12 is an example of a first calculation unit. The control condition selection unit 23 is an example of a second calculation unit and a selection unit.

  As mentioned above, although the Example of this invention was described in full detail, this invention is not limited to such a specific embodiment, A various deformation | transformation is carried out within the range of the gist of this invention described in the claim.・ Change is possible.

10 Quality Control Server 11 Measured Goodput Collection Unit 12 Throughput Prediction Unit 13 Quality DB
Reference Signs List 20 client terminal 21 playback unit 22 measured goodput calculation unit 23 control condition selection unit 24 control condition DB
30 moving image distribution server 100 drive device 101 recording medium 102 auxiliary storage device 103 memory device 104 CPU
105 Interface device B bus

Claims (6)

For each delivery of video data, a communication unit that stores communication environment information indicating a communication environment of a delivery destination, communication control related information related to communication control in the delivery, and actual measurement goodput of the delivery in association with each other,
Based on the measured goodput stored in the storage unit in association with the communication environment information having the common part with the first communication environment information related to the distribution destination of the first distribution of the moving image data, the predicted value of the throughput is calculated. A first calculator for calculating,
Applying the predicted value calculated by the first calculator and the communication control related information related to the first distribution to a function indicating the relationship between the predicted value of throughput and the communication control related information and the goodput. A second calculator for calculating a predicted value of Goodput,
A selecting unit that selects a control condition according to the predicted value of the good put with respect to the first distribution;
Have a,
The storage unit stores the communication environment information, the communication control related information, and the measured value of the goodput for each delivery of the chunk of the moving image data,
The communication control related information includes a cumulative data amount from the start of the distribution of the moving image data, a chunk size, and a distribution interval with the previous chunk,
A distribution control device, characterized in that: The first calculation unit is a measured goodput stored in the storage unit in association with communication environment information having a common part with the first communication environment information, and the first calculation unit satisfies a predetermined condition. Based on the measured goodput not associated with the control-related information, calculate the predicted value of the throughput,
The distribution control device according to claim 1, wherein: The selection unit, from among a plurality of control conditions each associated with a required good put, a predicted value of the good put, based on a comparison of the required good put associated with each control condition, Select the control conditions,
The distribution control device according to claim 1 or 2, wherein: The predetermined condition is a condition indicating that it is estimated that the goodput and the throughput are separated.
3. The distribution control device according to claim 2, wherein: For each distribution of video data, with reference to the communication environment information indicating the communication environment of the distribution destination, communication control related information related to communication control in the distribution, and the storage unit that stores in association with the measured goodput of the distribution,
Based on the measured goodput stored in the storage unit in association with the communication environment information having a common part with the first communication environment information related to the distribution destination of the first distribution of the moving image data, the predicted value of the throughput is calculated. A first calculating procedure for calculating;
Applying the predicted value calculated by the first calculation procedure and the communication control-related information related to the first distribution to a function indicating the relationship between the predicted value of throughput and the communication control-related information and the goodput. , A second calculation procedure for calculating the predicted value of Goodput,
A selection step of selecting a control condition according to the predicted value of the good put with respect to the first distribution;
Is executed by the computer ,
The storage unit stores the communication environment information, the communication control related information, and the measured value of the goodput for each delivery of the chunk of the moving image data,
The communication control related information includes a cumulative data amount from the start of the distribution of the moving image data, a chunk size, and a distribution interval with the previous chunk,
A distribution control method, characterized in that: Program for causing a computer to function as claims 1 to 4 each unit according to any one claim.

Images