WO2025150096A1 - Selectable bit rate determination device and selectable bit rate determination method - Google Patents

Abstract

A selectable bit rate determination device disclosed herein comprises: a perceived quality estimation unit that calculates an estimated value of perceived quality for a bit rate in a range designated for a Web conference service; a power consumption estimation unit that calculates an estimated value of power consumption for the bit rate in the range on the basis of the characteristic that the increase in power consumption per unit bit rate decreases as the bit rate increases; a normalization unit that normalizes each of the estimated value of perceived quality and the estimated value of power consumption; an index calculation unit that, on the basis of the normalized estimated values, calculates an index that increases as perceived quality increases and power consumption decreases; and a selection unit that selects, as a bit rate that can be selected in the Web conference service, a bit rate related to the estimated value of perceived quality corresponding to the maximum point of the index in the relationship between the index and the estimated value of perceived quality calculated for the bit rate in the range. The selectable bit rate determination device thus supports bit rate control capable of suppressing power consumption while increasing perceived quality.

Description

SELECTABLE BIT RATE DETERMINATION DEVICE AND SELECTABLE BIT RATE DETERMINATION METHOD

The present invention relates to a selectable bitrate determination device and a selectable bitrate determination method.

Due to the increase in remote work, web conferencing services are now widely used on a variety of devices. Web conferencing service providers are required to maintain quality for users so that they can use the service comfortably. In order to improve user quality, it is necessary to increase the video bit rate of web conferences, but increasing the bit rate excessively causes an increase in the amount of data transferred, which leads to increased operational costs. For this reason, it is necessary to suppress the amount of data transferred while maintaining the quality required by users.

To address this issue, a video bitrate control method has been proposed that reduces the amount of data transferred while maintaining the user's quality at a target value (Non-Patent Document 1).

In Non-Patent Document 1, the quality of experience of each user is estimated from the audio and video information received by each user. Based on that information, a prediction is made as to what future video bit rate is required to achieve a pre-set target quality of experience, and a bit rate that will achieve the target quality of experience while minimizing the amount of data transferred is selected, and the bit rate for transmitting video is indicated to each user.

On the other hand, environmental protection efforts have become more important in recent years, and there is also a growing demand for reducing ICT power consumption. Therefore, it is necessary to consider not only the user's quality of experience and operational costs, but also power consumption. Furthermore, from the user's perspective, with the promotion of remote work, participation in web conferences can be expected in a variety of environments. When using web conferences while away from the office, etc., high power consumption means that the time spent participating in the web conference will be shorter, so it is important to be aware of power consumption from the perspective of convenience as well.

M. Yokota and K. Yamagishi, "Quality-based video bitrate control for WebRTC-based tele-conference services", Electronic imaging 2022, IQSP-333, Jan. 2022 P. Lebreton and K. Yamagishi, "Transferring adaptive bit rate streaming quality models from H.264/HD to H.265/4K-UHD ", IEICE Transactions on Communications, vol. E102-B, no. 12, pp. 2226-2242, Dec. 2019, 10.1587/transcom.2019EBP3045

In Non-Patent Document 1, control that takes into account the quality of experience and the amount of data transferred is realized, but power consumption is not taken into account, so control that increases user convenience is not necessarily achieved.

The present invention has been made in consideration of the above points, and aims to support bit rate control that can reduce power consumption while improving the quality of experience.

In order to solve the above problem, the selectable bitrate determination device is characterized by having a quality of experience estimation unit configured to calculate an estimated value of quality of experience for a bitrate in a range specified for a Web conference service, a power consumption estimation unit configured to calculate an estimated value of power consumption for the bitrate in the range based on the characteristic that the increase in power consumption per unit bitrate decreases as the bitrate increases, a normalization unit configured to normalize each of the estimated value of quality of experience and the estimated value of power consumption, an index calculation unit configured to calculate an index that increases as the quality of experience increases and increases as the power consumption decreases based on the normalized estimate, and a selection unit configured to select a bitrate related to the estimated value of quality of experience that corresponds to the maximum point of the index in the relationship between the estimated value of quality of experience calculated for the bitrate in the range and the index as a selectable bitrate in the Web conference service.

It can assist with bitrate control that can reduce power consumption while improving the quality of experience.

1 is a diagram illustrating an example of a hardware configuration of a selectable bit rate determination device 10 according to an embodiment of the present invention. 1 is a diagram illustrating an example of a functional configuration of a selectable bit rate determination device 10 according to an embodiment of the present invention. 10 is a flowchart illustrating an example of a process performed by a selectable bit rate determination device 10. FIG. 13 is a diagram showing the relationship between QoE _scaling and index. FIG. 13 is a diagram for explaining QoE _scaling corresponding to a maximum point of the index. FIG. 13 is a diagram for explaining selection of QoE _scaling that maximizes the nearest neighbor distance.

The reason why conventional technology is unable to control the video bit rate while considering power consumption (hereinafter referred to as "Problem 1") is that, for web conferencing services, the relationship between the transmission bit rate and the power consumption of the sending client, taking into account the encoding process in the web browser, etc., is unclear, making it impossible to control the power consumption while capturing its characteristics (hereinafter referred to as "Problem 2").

In this embodiment, therefore, in order to solve problem 2, a technology is realized that estimates power consumption from the transmission bit rate.

In addition, in the technology for controlling bit rates as in Non-Patent Document 1, the bit rate to be controlled is selected from a selectable group of bit rates. Therefore, in this embodiment, in order to solve problem 1, the selectable group of bit rates is determined using a technology for estimating power consumption from the transmission bit rate to solve problem 2, and a technology for estimating the quality of experience (the quality of experience of the video viewer (participant in the web conference)) from the bit rate (Non-Patent Document 2, etc.).

Below, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a diagram showing an example of the hardware configuration of a selectable bitrate determination device 10 in an embodiment of the present invention. The selectable bitrate determination device 10 in Fig. 1 has a drive device 100, an auxiliary storage device 102, a memory device 103, a processor 104, and an interface device 105, which are all interconnected by a bus B.

The program that realizes the processing in the selectable bit rate determination device 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 via the drive device 100 into the auxiliary storage device 102. However, the program does not necessarily have to be installed from the recording medium 101, but may be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program as well as necessary files, data, etc.

When an instruction to start the program is received, the memory device 103 reads out and stores the program from the auxiliary storage device 102. The processor 104 is a CPU or a GPU (Graphics Processing Unit), or a CPU and a GPU, and executes functions related to the selectable bitrate determination device 10 in accordance with the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

FIG. 2 is a diagram showing an example of the functional configuration of a selectable bitrate determination device 10 in an embodiment of the present invention. In FIG. 2, the selectable bitrate determination device 10 has an input unit 11, a quality of experience estimation unit 12, a power consumption estimation unit 13, a normalization unit 14, an index calculation unit 15, a selection unit 16, and an output unit 17. Each of these units is realized by a process in which one or more programs installed in the selectable bitrate determination device 10 are executed by the processor 104.

The processing procedure executed by the selectable bit rate determination device 10 will be described below. FIG. 3 is a flowchart for explaining an example of the processing procedure executed by the selectable bit rate determination device 10. The processing procedure shown in FIG. 3 only needs to be performed once before a certain Web conference service is provided, asynchronously with the Web conference service. The selectable bit rate obtained as the processing result can be set as Config (setting information) of a control technology such as that described in Non-Patent Document 1 before the Web conference service is provided.

In step S101, the input unit 11 receives from the user a specification of the range (minimum value, maximum value) of the bit rate (br) for the Web conference service for which the selectable bit rate is to be calculated. The input unit 11 assigns the minimum value to the variable br _min and the maximum value to the variable br _max . For current Web conference services, a video of 2 Mbps is sufficient to achieve high quality, so for example, br _min may be set to 0 Mbps and br _max to 2 Mbps.

Next, the quality of experience estimator 12 calculates the minimum quality of experience estimate (QoE _{min ) for the minimum bit rate (br min} ) and calculates the maximum quality of experience estimate (QoE _max ) for the maximum bit rate (br _{max )} (S102).

In this case, the quality of experience estimator 12 calculates an estimate of the quality of experience using a model that can estimate the quality of experience of a user who receives video at a given bit rate from the bit rate, as formulated in Non-Patent Document 2 and other documents. In Non-Patent Document 2, in order to estimate the quality of experience, information on the resolution and frame rate in addition to the bit rate is required, so the quality of experience estimator 12 is given in advance the values of the resolution and frame rate that are set in each bit rate section.

Next, the power consumption estimation unit 13 calculates the minimum power consumption estimate value (pc _min ) for the minimum bit rate (br min ), and calculates the maximum power consumption estimate value (pc _max ) for the maximum bit rate (br _{max )} (S103).

At this time, the power consumption estimation unit 13 calculates an estimated value of power consumption using a model that can estimate power consumption from the transmission bit rate (br). As described in Problem 2, such an estimation model does not exist, so in this embodiment, a specific estimation model and its characteristics are constructed based on the following idea.

In a client holding a web conference, the main process during the web conference is thought to be encoding video and audio, and transmitting the encoded video. The higher the bit rate of the transmitted video, the more processing is required for encoding, and the amount of data transferred increases, so power consumption increases. In addition, by taking into account the characteristic that the increase in power consumption per unit bit rate decreases as the transmission bit rate increases, it is possible to estimate power consumption (pc) from the transmission bit rate using the model described in formula (1).

That is, the power consumption estimation unit 13 substitutes br _min for br in the formula (1) to calculate pc _min , and substitutes br _max for br in the formula (1) to calculate pc _max .

Here, β ₀ , β ₁ , β ₂ , and β ₃ are parameters of the model.

In this embodiment, it is assumed that such a model will be used, but other models can also be used as long as they are capable of estimating power consumption from the transmission bit rate.

Next, the input unit 11 initializes a variable br to br _min (S104).

Then, the quality of experience estimator 12 calculates an estimated value of the quality of experience for br and assigns the calculation result to the variable QoE (S105). The method of calculating the estimated value of the quality of experience may be the same as in step S102.

Then, the power consumption estimation unit 13 calculates an estimated value of the power consumption for br and assigns the calculation result to the variable pc (S106). The method of calculating the estimated value of the power consumption may be the same as in step S103.

In the following steps S107 and S108, the normalization unit 14 performs processing to align (unify) the scales of the power consumption and the quality of experience, so that the power consumption estimated by the power consumption estimation unit 13 and the quality of experience estimated by the quality of experience estimation unit 12 can be compared.

In step S107, the normalization unit 14 calculates a QoE _scaling with a unified scale (normalized) by substituting the calculated QoE, QoE _min , and QoE _max into the following formula (2).

That is, the normalization unit 14 normalizes the QoE so that the minimum value of the QoE is 0 and the maximum value is 1.

Next, the normalization unit 14 calculates pc _scaling with a unified scale (normalized) by substituting the calculated pc, pc _min , and pc _max into the following equation (3).

That is, the normalization unit 14 normalizes pc so that the minimum value of pc is 0 and the maximum value of pc is 1.

Next, the index calculation unit 15 calculates an index for evaluating the Web conference service that takes into account both the quality of experience (QoE _scaling ) and the power consumption (pc _scaling ) that have been made comparable by unifying (normalizing) the scale (S109).

In a Web conferencing service, it is desirable to have a high quality of experience and low power consumption (requirement 1). Taking these factors into account, an index that considers both quality of experience and power consumption is shown in formula (4).

Here, α is a parameter (coefficient) that changes the weighting given to quality of experience and power consumption. The smaller the value of α, the more importance is placed on power consumption, and the larger the value, the more importance is placed on quality of experience.

That is, the index is an example of an index that increases as the quality of experience increases and as the power consumption decreases. The index calculation unit 15 calculates the index by substituting the QoE _scaling and the pc _scaling into formula (4).

Next, the index calculation unit 15 records the set of (QoE _scaling , index) in the memory device 103 (S110).

Next, the index calculation unit 15 judges whether the value of br+Δbr exceeds br _max (S111). Here, Δbr is the interval of br for obtaining (QoE _scaling , index) for multiple brs in the range from br _min to br _max . Therefore, the smaller Δbr is, the more accurately the relationship between QoE _scaling and index can be derived. For example, Δbr=1 may be used.

If the value of br+Δbr is equal to or smaller than _brmax (No in S111), the index calculation unit 15 updates br by substituting br+Δbr for br (S112). Then, steps S105 and after are repeated based on the updated br.

If the value of br+Δbr exceeds br _max (Yes in S111), the selector 16 executes a process for selecting a pre-specified number of selectable bit rates based on the index in order to select a bit rate that can improve the quality of experience while reducing power consumption and achieving a wide range of quality of experience in the Web conference service. In this process, the selector 16 uses the set of (QoE _scaling , index) recorded in step S110 for each step size of Δbr. The set indicates the relationship between QoE _scaling and index (relationship based on formula (4)) that also takes power consumption (pc _scaling ) into consideration, as shown in FIG. 4 . In Fig. 4, a curve showing the relationship between QoE _scaling and index is shown in a coordinate system in which QoE _scaling is assigned to the x-axis and index is assigned to the y-axis, but in reality it is a set of points with an interval of Δbr. However, the selection unit 16 may derive an approximation curve based on the set of points. The "relationship between QoE _scaling and index" in the following description may be a set of points or an approximation curve.

When selecting a selectable bit rate in a web conferencing service, it is required to select a bit rate that will give a high index.

In addition, the selectable bitrates must be set to accommodate fluctuating network bandwidth conditions. For example, in situations where the network bandwidth is narrow (e.g., while traveling on a train or in an enclosed environment), selecting a high bitrate will cause the video to freeze and the quality of experience to deteriorate, so a low bitrate must be prepared according to the bandwidth conditions. In this way, it is necessary to prepare bitrates that do not deteriorate the quality of experience in response to various fluctuating network conditions, so it is also necessary to select bitrates that allow for a wide range of quality of experience (Requirement 2).

In order to select selectable bitrates from the number (N) of selectable bitrates set in advance while satisfying the above two requirements, requirement 1 and requirement 2, the selection unit 16 selects bitrates as follows.

In step S113, the selection unit 16 identifies the QoE _scaling corresponding to each maximum point of the index in the relationship between the QoE _scaling and the index (FIG. 4).

Fig. 5 is a diagram for explaining QoE _scaling corresponding to the maximum point of the index. In Fig. 5, the maximum point of the index is shown by a dashed line with respect to the relationship between the QoE _scaling and the index shown in Fig. 4. The value of the x-axis on the dashed line is the QoE _scaling corresponding to the maximum point of the index.

Here, the number of maximum points is k (k≦N), so k QoE _scalings are selected in step S113.

Next, the selection unit 16 acquires the bit rates associated with each of the k QoE _scalings as selectable bit rates (S114). Thus, k bit rates are acquired. Note that the bit rate associated with a certain QoE _scaling is the br that was the basis for calculating the QoE _scaling in steps S105 and S107. Therefore, in steps S105 and S107, the br and the QoE _scaling may be stored in association with each other.

Note that steps S113 and S114 correspond to requirement 1 above.

Next, the selection unit 16 judges whether k is less than N (S115). If k is less than N (Yes in S115), the selection unit 16 selects N-k QoE scalings so that the closest distance of the quality of experience (QoE _scaling ) is maximized in order to meet the above requirement 2 (S116). Here, the closest distance refers to the minimum value of the intervals between adjacent QoE _scalings in the N QoE _scalings corresponding to the selected br. Therefore, selecting N-k QoE _scalings so that the closest distance is maximized means selecting _{N-k QoE scalings so} that the minimum value is maximized. Such selection of QoE _scalings can be performed, for example, by logic as shown in FIG. 6. Note that adjacent QoE _scaling refers to adjacent QoE _scaling when a set of QoE _scaling corresponding to a selected br is sorted in ascending or descending order.

Fig. 6 is a diagram for explaining the selection of QoE _scaling that maximizes the nearest neighbor distance. The selection of QoE _scaling in step S116 is repeated until N-k QoE _scalings are selected. Since one QoE _scaling is selected in one repetition, the selection is repeated N-k times. Fig. 6 shows specific examples of the first to fourth repetitions.

In the first round, the selection unit 16 calculates the interval between adjacent QoE _scalings in a set of QoE _scalings corresponding to the minimum value of QoE _scaling , the maximum value of QoE _scaling , and the maximum point of index. In the first round in FIG. 6, the interval is indicated by a double-headed arrow. The selection unit 16 selects the median value (the value indicated by the perpendicular line L1 in the figure) of the longest interval (the second double-headed arrow from the left in the first round in FIG. 6) from among these intervals as one QoE _scaling out of N-k. Here, if k+1=N, step S116 ends, and if k+1<N, proceed to the second round.

In the second round, the selection unit 16 calculates the interval between adjacent QoE _scalings in the set of the minimum value of QoE _scaling , the maximum value of QoE _scaling , the QoE _scaling corresponding to the local maximum point of index, and the QoE _scaling selected in the first round (perpendicular line L1). In the second round in FIG. 6, the interval is indicated by a double-headed arrow. The selection unit 16 selects the median value (the value indicated by the perpendicular line L2 in the figure) of the longest interval (the double-headed arrow at the right end in the second round in FIG. 6) from among these intervals as one QoE _scaling out of N-k pieces. Here, if k+2=N, step S116 ends, and if k+2<N, proceed to the third round.

In the third round, the selection unit 16 selects the value indicated by the perpendicular line L3 as one of the N-k QoE _scalings by the same process as in the second round. If k+3=N, step S116 ends, and if k+2<N, the process proceeds to the fourth round.

In the fourth iteration, the selection unit 16 calculates the minimum value of QoE _scaling , the maximum value of QoE _scaling , the QoE _scaling corresponding to the local maximum point of index, and the interval between adjacent QoE _scalings in the set of QoE _scalings (perpendicular lines L1, L2, L3) selected in the first to third iterations. In the fourth iteration in FIG. 6, the intervals are indicated by double-headed arrows. The selection unit 16 selects the longest interval (the third double-headed arrow from the left in the fourth iteration in FIG. 6). Here, one of the third arrow from the left (the left end) is the QoE _scaling (perpendicular line L1) added in step S116 (i.e., L1 is a QoE _scaling not included in the k items). In this case, the selection unit 16 does not select a new QoE _scaling at the center of the third arrow from the left, but shifts the perpendicular line L1 to one of the two QoE _scalings that divide the interval between the QoE _scalings corresponding to the maximum points of the index (the dashed double arrows in the figure) into thirds, containing the two arrows, to select perpendicular line L1', and selects the other (perpendicular line L4) as one of the N-k QoE _scalings .

Therefore, at the end of the fourth round, the perpendicular lines L1', L2, L3, and L4 become four QoE _scalings out of Nk.

In other words, the selection unit 16 basically repeats the same process as the first time, and in the same situation as the fourth time, executes the same process as the fourth time. This is repeated N-k times.

In the above, it was explained that the minimum and maximum values of QoE _scaling are set as the boundaries of the values that QoE _scaling can take, and N-k QoE _scalings are selected between those boundaries. However, depending on the quality of experience model used, there may be quality of experience values that cannot be achieved no matter what bit rate is selected, and therefore those values may also be set as boundaries of the values that QoE _scaling can take.

Next, the selection unit 16 acquires the bit rate (br) related to the QoE _scaling selected in step S116 (S117).

If the answer is No in step S115, or following step S117, the output unit 17 outputs the acquired N bit rates as candidates for bit rates that can be selected in bit rate control in the Web conference service (S118).

As described above, according to this embodiment, it is possible to support bit rate control that can reduce power consumption while improving the quality of experience. In other words, when controlling the bit rate in a Web conference service, a user can select a selectable bit rate according to this embodiment, thereby enabling bit rate control that can reduce power consumption while improving the quality of experience.

The above describes in detail the embodiments of the present invention, but the present invention is not limited to such specific embodiments, and various modifications and variations are possible within the scope of the gist of the present invention as described in the claims.

REFERENCE SIGNS LIST 10 Selectable bit rate determination device 11 Input unit 12 Quality of experience estimation unit 13 Power consumption estimation unit 14 Normalization unit 15 Index calculation unit 16 Selection unit 17 Output unit 100 Drive device 101 Recording medium 102 Auxiliary storage device 103 Memory device 104 Processor 105 Interface device B Bus

Claims (4)

a quality of experience estimator configured to calculate an estimate of the quality of experience for a range of bit rates specified for the Web conferencing service;
a power consumption estimating unit configured to calculate an estimated value of power consumption for the bit rate in the range based on a characteristic that the increase in power consumption per unit bit rate decreases as the bit rate increases;
a normalization unit configured to normalize each of the quality of experience estimate and the power consumption estimate;
an index calculation unit configured to calculate an index based on the normalized estimated value, the index increasing as the quality of experience increases and as the power consumption decreases;
a selection unit configured to select, as a selectable bit rate in the Web conference service, a bit rate related to the estimated value of quality of experience that corresponds to a maximum point of the index in a relationship between the estimated value of quality of experience calculated for the bit rate in the range and the index;
13. A selectable bit rate determination device comprising: the selection unit is further configured to select, as the selectable bit rate, a bit rate associated with the estimated quality of experience corresponding to a center of a maximum interval among the intervals between the plurality of maximum points.
2. The apparatus for determining a selectable bit rate according to claim 1. the selection unit is further configured to select, as the selectable bit rate, a bit rate associated with the estimated quality of experience that corresponds between the center and the maximum point.
3. The apparatus for determining a selectable bit rate according to claim 2. a quality of experience estimation step for calculating an estimate of the quality of experience for a range of bit rates specified for the Web conferencing service;
a power consumption estimation step of calculating an estimated value of power consumption for the bit rate in the range based on a characteristic that the increase in power consumption per unit bit rate decreases as the bit rate increases;
a normalization step for normalizing each of the quality of experience estimate and the power consumption estimate;
an index calculation step of calculating an index that increases as the quality of experience increases and as power consumption decreases based on the normalized estimated value;
a selection step of selecting, as a selectable bit rate in the Web conference service, a bit rate associated with the estimated value of quality of experience that corresponds to a maximum point of the index in a relationship between the estimated value of quality of experience calculated for the bit rate in the range and the index;
2. A method for determining a selectable bit rate, comprising: