JP5749693B2 - User pause time estimation apparatus, method and program - Google Patents

Description

  The present invention relates to an apparatus and method for estimating a user pause time in a video service that estimates a data playback state at a receiving terminal, and in particular, estimates a user pause operation (video playback pause) based on capture data, and The present invention relates to a user pause time estimation apparatus, method, and program in a video service for estimating player parameters in a video receiving terminal.

  With the progress of broadband communication networks in recent years, progressive download type content distribution systems in which large-capacity content data are downloaded and processed in parallel at the client terminal are becoming widespread. In such a content distribution form, transmission quality degradation caused by the network affects content reproduction on the terminal side, so it is a user for content data providers and circuit providers that provide access lines. Implementation of remote quality monitoring is an important issue, such as isolation when reporting quality deterioration from

  On the other hand, a mechanism that prepares a playback buffer in the terminal and conceals the influence of the network from higher-level applications is widely used as the conventional technology, and also reduces the influence of the network and shortens the waiting time until the start of data processing. In order to optimize the balance, a technique for dynamically changing the initial buffer amount based on the packet transfer delay time is also disclosed. Furthermore, a given capture is applied to a playback stop event that occurs when the network throughput temporarily drops below the content data processing speed and the content data storage amount in the playback buffer falls below a certain amount (playback stop threshold). A service quality estimation technique for deriving an estimated value of a playback stop state by grasping a time series of the received byte amount of a terminal based on data and inputting it to a modeled playout buffer is disclosed (for example, non Patent Literatures 1 to 3).

  Progressive download type applications are known to be roughly classified into two types according to their playback methods, and the above technique can be applied to both of them. By using these techniques, the playback state of the session can be estimated based on given capture data.

Daisuke Ikegami, Yasunori Honda, Koji Yamamoto et al .: "Progressive download service stop time estimation method" IEICE Technical Report, vol. 111, no. 278, CQ2011-59, pp. 91-96, November 2011 . Daisuke Ikegami, Yasunori Honda, Koji Yamamoto: "Examination of playback state estimation method for chunked video distribution service", IEICE General Conference, B-11-24, March 2012. Yasutoshi Honda, Daisuke Ikegami, Koji Yamamoto: “Study of video playback state estimation method by Ack observation”, IEICE General Conference, B-11-25, March 2012.

  In recent years, the spread of services for smartphones on 3G lines and tablet terminals in a wireless LAN environment is particularly remarkable. In general, these terminals are characterized in that the playout buffer length is shorter than that of the PC due to memory capacity constraints. As a result, the capacity of the downloaded unreproduced data reaches the buffer upper limit, and the situation where the download is temporarily stopped is confirmed more frequently than the PC. In particular, the sequential progressive download type service is characterized in that the download is continued even after the pause operation by the user, and as a result, the data often reaches the buffer upper limit after a predetermined time.

  As a service provider or a line provider, it is desirable to objectively grasp the quality of experience of video services provided to users from the viewpoint of maintenance. Non-Patent Documents 1 to 3 are techniques for estimating a video playback state based on capture data. However, in this document, the reproduction stop time due to the user's pause operation is not included in the estimation target. For this reason, if the user performs a pause operation for a long time, there is a concern that the output stop time estimated value includes the pause operation time in addition to the actual playback stop time.

  The present invention has been made in view of the above points, and in a tablet terminal, a smart phone, etc., when the arrival state of the buffer upper limit can be confirmed from the capture data, the pause time estimation is realized based on predetermined information, It is an object of the present invention to provide a highly accurate user pause time estimation apparatus, method, and program.

In order to solve the above problems, the present invention is a user pause time estimation device for estimating a user's pause operation in data reproduction,
A user pause time estimation device for estimating a user's pause operation in data reproduction,
Buffer upper limit arrival time detecting means for detecting the buffer upper limit arrival time based on given capture data;
Continue watching without performing a pause operation in the viewing experiment, and estimate the amount of data in the buffer when a session disconnection packet appears, and use it as the buffer upper limit (B).
B = B1 − ((t3−t1) −p) * dec
(Where B1 is the cumulative amount of received bytes between the playback start operation time and the buffer upper limit arrival time, t3 is the buffer upper limit arrival time, t1 is the playback start time, and dec is a predetermined decoding rate)
The estimated time calculating value p is obtained by calculating the pause time estimated value p and adding to the time when the data amount in the buffer stays at the upper limit .

  According to the technique of the present invention, the pause operation time by the user included in the playback stop time is estimated based on the capture data for the received data of the service provided via the network such as the sequential progressive download video service. Thus, it is possible to improve the accuracy of reproduction state estimation. Although limited to the pause operation, the user operation can be estimated from the captured data.

1 is a configuration diagram of a user pause time estimation device in a first embodiment of the present invention. FIG. 3 is a playout buffer image according to the first embodiment of the present invention. 3 is a buffer length transition image according to the first embodiment of the present invention. It is the relationship (the 1) of the pause operation and reproduction | regeneration state estimated value in the 2nd Embodiment of this invention. It is the relationship (the 2) of the pause operation and reproduction | regeneration state estimated value in the 2nd Embodiment of this invention. It is an example of an estimation in the position example of the present invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  A user pause time estimation device of the present invention is provided in a terminal that reproduces data received via a network, and is based on a parameter estimation value calculated by an external parameter estimation value calculation device, an estimated buffer model, and capture data. A pause operation time estimated value is calculated.

[First Embodiment]
FIG. 1 shows a configuration of a user pause time estimation apparatus according to an embodiment of the present invention.

  The user pause time estimation device 10 includes a pause operation time estimation unit 2 and a buffer upper limit arrival time detection unit 3, and the pause operation time estimation unit 2 is connected to the parameter estimated value calculation device 1.

  An outline of the operation is shown.

  The parameter estimation value calculation device 1 connected to the user pause time estimation device 10 models the player, calculates a parameter estimation value based on the model, and outputs the parameter estimation value to the pause operation time estimation unit 2. The processing can be realized by various conventional techniques that can be modeled (for example, Non-Patent Documents 1 to 3).

  The parameter estimated value calculation device 1 outputs a parameter estimated value to the pause operation time estimation unit 2. The parameter estimation device 1 models the player by decoding data in a finite-length buffer at a fixed rate. The parameter estimation value output by the model is shown in FIG. In the present embodiment, the decoding rate of the player model (player buffer model) for calculating the parameter estimation value is a fixed value set in advance.

  1) Reproduction start threshold: This is the remaining amount of data in the buffer required for the first reproduction after the application is started.

  2) Playback restart threshold: This is the remaining amount of data in the buffer required for the second and subsequent playback after playback is stopped.

  3) Playback stop threshold: This is a threshold for the amount of data in the buffer that transitions from the playback state to the playback stop state.

  4) Buffer length: Playout buffer capacity. The present invention allows a finite length.

  5) Decoding rate: Decoding rate of data stored in the buffer.

  6) The amount of data in the buffer.

  The parameter estimation value device 1 outputs the playback start threshold, the playback restart threshold, the buffer length, the decode rate, and the playback stop threshold among the parameter estimation values to the pause operation time estimation unit 2. However, if the playback stop has not occurred before reaching the buffer upper limit, the playback stop threshold is not necessary.

  The buffer upper limit arrival time detection unit 3 calculates the buffer upper limit arrival time from the input capture data and outputs it to the pause operation time estimation unit 2.

  The pause operation time estimation unit 2 outputs a pause operation estimation value from the parameter estimation value acquired from the parameter estimation value device 1 and the buffer upper limit arrival time acquired from the buffer upper limit arrival time detection unit 3.

  Hereinafter, the operation of each component will be described in detail.

  The buffer upper limit arrival time detection unit 3 detects the buffer upper limit arrival state (s) by detecting a specific packet such as a TCP RST packet, FIN packet, or Dup Ack packet in the capture data, and outputs it to the pause operation time estimation unit 2 To do. For example, when the buffer limit is reached due to continuous data accumulation in the playout buffer, the terminal closes the session by sending a TCP RST packet or FIN packet, or sets the receive window size to 0 (TCP Zero Window) Prevents buffer overflow. When the TCP session is closed, it is distinguished from session close due to timeout because there are no packets indicating transfer quality degradation such as Dup Ack and retransmission in N packets before and after the capture data.

  FIG. 3 shows a buffer length transition image in the first embodiment of the present invention.

  The buffer length transition image shown in the figure is obtained by converting captured data in time series.

  In the figure, it is clear that the time (t4−t3) during which the data amount in the buffer remains at the upper limit is the pause operation time. In this case, the pause operation has continued from [a3]. The buffer upper limit arrival time detection unit 3 outputs the time (t4−t3) to the pause operation time estimation unit 2 as the buffer upper limit arrival time.

  The pause operation time estimation unit 2 estimates the time (p: = t3−t2) until the amount of data in the buffer reaches the upper limit after the pause operation is started ([a2]).

  The cumulative received byte amount B1 [byte] between [a0] and [a3] is obtained from the captured data, the upper limit of the buffer is B [byte], and the pause time estimated value p [sec] (in t3-t2 in FIG. 3) (Equivalent) is calculated by equation (1).

B = B1 − ((t3−t1) −p) * dec (1)
The buffer upper limit B is an estimated data amount in the buffer when a packet for disconnecting a TCP session appears without performing a pause operation by a viewing experiment or the like.

  The pause operation time estimation unit 2 estimates the overall pause operation time P by adding p obtained by the equation (1) to the time (t4−t3).

  Also, the buffer upper limit B is the accumulated amount of received bytes when the buffer upper limit is reached, for example, when there is no pause time or a known operation in the parameter estimation value apparatus 1, the playback time and the decoding rate up to that point. It is also possible to calculate by considering the above.

  If playback is resumed when the buffer limit is reached, [a3] and [a4] are simultaneous events. Note that in this case, the pause operation time is output as a cumulative value in the time interval [t1, t3].

[Second Embodiment]
The apparatus of the present invention can also be realized as an apparatus for pause time estimation using the buffer data amount estimation value of the final output of the reproduction state estimation techniques in Non-Patent Documents 1 to 3, and for improving the estimation accuracy. . Note that the buffer data amount estimation value is obtained from the parameter estimation value apparatus 1.

  FIG. 4 shows an output image of the reproduction state estimated value and the actual data amount in the buffer in Non-Patent Documents 1 to 3 when the user performs a pause operation. In the figure, when the buffer upper limit is reached for the first time after execution of the pause operation ([a3]), the buffer data amount estimation value starts to decrease with a value less than the actual buffer length. This is because the pause length (t2 to t2) is not taken into consideration and the data download time (b) is determined to be in the playback state, and the buffer length during that time increases with the difference between the download speed and the estimated decode rate. In addition, after the buffer upper limit arrival time, data download is stopped. As long as the buffer length is equal to or greater than the playback stop threshold, it is estimated as the “playback state”. 4 and 5 are images of performing a pause operation ([a2]) after starting playback ([a1]) and restarting playback ([a4]) after reaching the buffer upper limit ([a3]). In addition, the data amount in the estimated buffer is assumed to be equal to or greater than the constant reproduction stop threshold. At this time, as an estimation result using existing techniques such as Non-Patent Documents 1 to 3, in FIG. 4, all of t1 and after are estimated as playback states, but in reality, the time interval [t2, t3] is paused. State.

  In the present embodiment, a process for estimating [t2 to t3] of an actual pause state will be described.

The pause operation time estimation unit 2 first calculates the pause time estimated value p,
p = (B−B2) / dec
Calculated by However, B is the buffer upper limit value at the buffer upper limit arrival time t3 acquired from the buffer upper limit arrival time detection unit 3, dec is a preset decoding rate (fixed value), and B2 [byte] is the parameter estimated value This is the estimated buffer data amount at time t3 calculated from the capture data using the buffer data amount estimated value output from the apparatus 1. As a result, the pause state is estimated between the time (t3−p) and the buffer upper limit arrival time t3.

  The pause operation time estimation unit 2 can correct the operation after the time (t3-p) and correct the estimated reproduction time. When playback is resumed when the buffer upper limit is reached [a3], [a3] and [a4] are simultaneous events. Note that in this case, the pause operation time is output as a cumulative value in the time interval [t1, t3].

  FIG. 5 assumes a case where the estimated buffer data amount is below the reproduction stop threshold in the same state as described above.

  In this case, the estimated playback time is the section sum a shown in the figure, but actually the time sections [t1, t2] and after t4 are playback times. In this case as well, the pause operation time can be estimated in the same manner as in FIG. 4, and the estimated reproduction time can be corrected by estimating the subsequent data amount transition in the buffer based on the estimated pause time.

  As described above, the pause time can be estimated using the data amount in the estimation buffer, and can be applied to the refinement of the reproduction state estimation.

  In this embodiment, it is assumed that the buffer parameters are given from the parameter estimation value device 1 as follows.

Playback stop threshold: 100 kbytes;
Buffer length: 700kbyte;
Decode rate: 100kbyte / sec.
Next, the pause operation time estimation unit 2 sets the accumulated number of received bytes from [a0] to [a3] to 2M [byte], the time from [a1] to [a3] (the time from playback start to reaching the buffer upper limit). ) To 15 [sec] and substituting into the above equation (1),
700 [kbyte] = 2000− (15−p) * 100 [kbyte / sec].
From this, the estimated pause time until [a3] can be calculated as 2 seconds by p = 2 [sec]. The cumulative pause time is
(t4−t3) + 2
Can be calculated.

  Further, in FIG. 6, the reproduction stop time estimation in the case of a pause state after 11.7 seconds after the start of reproduction is shown. By using the present invention, it can be seen from the figure that there is almost no difference from the actual state, the pause time is accurately estimated, and the playback stop time can be estimated with high accuracy.

  The operation of each component of the user pause time estimation device shown in FIG. 1 is constructed as a program, installed and executed on a computer used as the user pause time estimation device, or distributed via a network. It is possible.

  The present invention is not limited to the above-described embodiments and examples, and various modifications and applications are possible within the scope of the claims.

DESCRIPTION OF SYMBOLS 1 Parameter estimated value calculation apparatus 2 Pause operation time estimation part 3 Buffer upper limit arrival time detection part 10 User pause time estimation apparatus

Claims (5)

A user pause time estimation device for estimating a user's pause operation in data reproduction,
Buffer upper limit arrival time detecting means for detecting the buffer upper limit arrival time based on given capture data;
Continue watching without performing a pause operation in the viewing experiment, and estimate the amount of data in the buffer when a session disconnection packet appears, and use it as the buffer upper limit (B).
B = B1 − ((t3−t1) −p) * dec
(Where B1 is the cumulative amount of received bytes between the playback start operation time and the buffer upper limit arrival time, t3 is the buffer upper limit arrival time, t1 is the playback start time, and dec is a predetermined decoding rate)
An estimated value calculating means for calculating a pause time estimated value p and calculating the user pause operation time estimated value by adding to the time when the data amount in the buffer remains at the upper limit ;
A user pause time estimation device characterized by comprising: The estimated value calculating means includes
The pause time estimate p is
p = (B−B2) / dec
(Where B is the buffer upper limit value, B2 is the estimated buffer data amount at the buffer upper limit arrival time calculated from the capture data using the parameter estimate value of the player buffer model, and dec is a predetermined decoding rate)
2. The user pause time estimation device according to claim 1, further comprising means for obtaining the above. A user pause time estimation method for estimating a user's pause operation in data reproduction,
In the apparatus having the buffer upper limit arrival time detecting means and the estimated value calculating means,
The buffer upper limit arrival time detecting means for detecting the buffer upper limit arrival time based on given capture data;
The estimated value calculation means continues viewing without performing a pause operation in a viewing experiment , obtains an estimated data amount in the buffer when a session disconnection packet appears, and sets it as a buffer upper limit (B),
B = B1 − ((t3−t1) −p) * dec
(Where B1 is the cumulative amount of received bytes between the playback start operation time and the buffer upper limit arrival time, t3 is the buffer upper limit arrival time, t1 is the playback start time, and dec is a predetermined decoding rate)
An estimated value calculation step of calculating a pause time estimated value p by calculating the user pause operation time estimated value by adding to the time when the data amount in the buffer remains at the upper limit ,
A user pause time estimation method characterized by: In the estimated value calculating step,
The pause time estimate p is
p = (B−B2) / dec
(Where B is the buffer upper limit value, B2 is the estimated buffer data amount at the buffer upper limit arrival time calculated from the capture data using the parameter estimate value of the player buffer model, and dec is a predetermined decoding rate)
The method of estimating a user pause time according to claim 3 obtained by : Computer
User pause time estimation program for causing to function as each unit of the user pause time estimation apparatus according to claim 1 or 2.

Images