JPH06152648A - Data packet communication method - Google Patents

Abstract

PURPOSE:To make the setting of a fluctuation absorption delay time proper by counting number of received packets stored in a buffer, number of overrun packets and number of underrun packets for each timing and calculating the count according to a prescribed arithmetic operation equation. CONSTITUTION:Number of packets stored in a fluctuation absorption buffer is measured at a reproduction timing of voice packets. Then an averaged storage packet number is calculated for each talk spurt by the following arithmetic equation; average value = sum of packet numbers/K at the reproduction timing from a head of talk spurt to the K-th talk spurt, where K=number of packets of talk spurt-(object packet number-1). Then the calculated average storage packet number is used to calculate a new fluctuation absorption delay time; new fluctuation absorption delay time = (object value - average value)X packet processing period time + current fluctuation absorption delay time. The delay time is set properly with a simple configuration by providing steps in this way to decide the new fluctuation absorption delay time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for absorbing voice packet delay fluctuations.

[0002]

2. Description of the Related Art FIGS.
[Fig. 36] Fig. 36 is a diagram for describing a conventional method for absorbing delay fluctuation of a voice packet shown in 3641. In the figure, 301-0,
Reference numeral 301-1 designates first and second voiced sections (hereinafter referred to as talk spurts), and coded voice data in which the talk spurts 301 are sectioned at regular time intervals 320 and 3, respectively.
21, 322, 323 are the first, second, third, and
This is the fourth encoded voice data. 330 is a voice packet in which encoded voice data for a certain period of time is packetized
333 is encoded voice data 320, 321, respectively.
322 and 323 are first, second, third, and fourth voice packets packetized. 304 is encoded voice data 3
It is a packetization cycle for assembling 20 to 323 into voice packets 320 to 333. Voice packets 320-333
In order to absorb the difference in transfer delay time which differs for each packet when a packet is transferred, there is a fluctuation absorption delay time waiting from the arrival of the first packet of the talk spurt to the start of reproduction, and 350 is the first fluctuation absorption delay time, Reference numeral 351 is a second fluctuation absorption delay time in which the first fluctuation absorption delay time is changed. 306 is a reproduction from the reproduction start point in the underrun state (a packet to be reproduced has not arrived at the reproduction timing at the reproduction start time) generated because the difference in the transfer delay time due to the fluctuation absorption delay time cannot be absorbed. It is the underrun time until the arrival of the packet to be reproduced at the timing.

FIG. 21 is a communication sequence chart when an underrun state occurs, and FIG.
Is the second fluctuation absorption delay time 3 newly calculated from the first fluctuation absorption delay time 350 and the underrun time 306.
It is a communication sequence chart which shows the case where 51 is added. Further, FIG. 23 is a diagram showing a method of obtaining an average transmission delay time of packets, which is necessary when calculating the fluctuation absorption time in the conventional example. In the figure, 307 is the average transmission delay time of the packet calculated from the previous talk spurt. FIG. 24 is a flowchart showing a conventional method for changing the delay fluctuation absorption time.

Next, the operation will be described. On the sending side,
The first talk spurt 301-0 is packetized with a packetization period 30.
The encoded voice data 320 to 323 divided into 4 are assembled and transmitted into voice packets 330 to 333, respectively. While being transmitted in the packet switching network, the receiving side
An operation of correctly reproducing the first talk spurt 301-0 from a voice packet which has a different transmission delay for each packet depending on the traffic state and transmission route of the network is performed. That is, after the first voice packet 330 of the first talk spurt 301-0 is received, the first fluctuation absorption delay time 350 =
After the lapse of T2 time, the reproduction of the first encoded audio data 320 included in the first audio packet 330 is started. When the reproduction of the first encoded voice data 320 is completed, the second encoded voice data 321 included in the second voice packet 331 is subsequently reproduced.

On the other hand, according to the example of the figure, the third voice packet 332 suffers a large delay during the transfer, and the voice packet 3
The third encoded voice data 322 is included in the third encoded voice data 322 included in 32 since it arrived at the receiving side after the underrun time 306 = T3 has elapsed from the reproduction start time.
Is discarded without being recycled. In FIG. 21, since the value of the first fluctuation absorption delay time 350 = T2 is small, the delay fluctuation of the voice packet 332 cannot be absorbed, and a part 22 of the encoded voice data in the talk spurt 301-0 is lost. ing. Therefore, the second talk spurt 321-1
On the other hand, the fluctuation absorption delay time is changed according to the flowchart of the method for changing the delay fluctuation absorption time in FIG.

In step S61, it is determined whether or not the delay fluctuation is absorbed, that is, whether or not the voice packet 330 or the like is discarded due to a short delay fluctuation absorption time. When the delay fluctuation is absorbed, the process proceeds to step S62, and when the delay fluctuation is not absorbed, the process proceeds to step S63. In step S63, an underrun time 30 is added to the current fluctuation absorption delay time in order to match the current packet transfer delay fluctuation.
The value obtained by adding 6 is set as a new fluctuation absorption delay time, and the delay fluctuation absorption time 351 is increased. For example, in the case of FIG. 21, since the delay fluctuation is not absorbed, as the second fluctuation absorption delay time 351, a value obtained by adding the value of the underrun time 306 to the value of the first fluctuation absorption delay time 350, T2 + T3. To set. As shown in FIG. 22, the delay fluctuation can be absorbed by using the second fluctuation absorption delay time 351 shown in step 3,
1 and the delay fluctuation in the network are the same, it is possible to reproduce all the encoded voice data 320 and the like in the talk spurt (1-1).

Further, when the process proceeds to step S62 in FIG. 24, the number of talk spurts in which the fluctuation is absorbed is counted (S62), and the value is compared with a predetermined threshold value. (S64) If the count value> the threshold value,
It is determined that the delay fluctuation is sufficiently absorbed, and the average transmission delay time calculated from this talkspurt is multiplied by the maximum number of talkspurt packets among the talkspurt from the present time to a certain time ago. Then, the new fluctuation absorption delay time (7) is changed (S64). If the count value ≦ threshold value, the processing is ended without changing the delay time. The method of calculating the average transmission delay time (7) is shown in FIG.
As shown in. That is, the arrival intervals of the voice packets 20 to 24 are measured, all the differences from the packetization cycle time (4) = T1 are added, and the average transmission delay time (7) calculated from the previous talk spurt (1) is used. Some T0
In addition, the sum of transmission delay times at the time of talk spurt (1) transfer is obtained. A new value of the average transmission delay time (7) is obtained by dividing the sum by the number of packets forming the talk spurt (1). Therefore, step S6
By the processing of 4, the fluctuation absorption delay time can be changed in the state where the delay fluctuation is absorbed, and when the average transmission delay time (7) is almost constant, the talks received up to a certain time in the past The fluctuation absorption delay time can be decreased or increased depending on the length of the part, and when the talkspurt length is almost constant, the fluctuation absorption delay time can be decreased or increased in proportion to the average transmission delay time (7).

[0008]

In the conventional voice packet communication method, it is necessary to measure the average transmission delay time and the underrun time, and there is a problem that the apparatus becomes complicated due to the measurement, and the average There was also a problem that the initial value of the transmission delay time had to be measured in advance. Also, when the transmission delay increases or decreases rapidly in response to changes in the network load, it is difficult to quickly change the fluctuation absorption delay time following the increase or decrease of the transmission delay. was there.

The present invention has been made in order to solve the above-mentioned problems, and does not require measurement of the average transmission delay time and underrun time, and has a simple structure and can be adaptively changed in delay fluctuation absorption. Aim to get a way. Another object of the present invention is to obtain a voice packet communication device capable of changing the fluctuation absorption delay time in response to changes in the load on the network.

[0010]

Data according to the present invention
The packet communication method is as follows:
A step of counting the number of accumulated packets that are accumulating received data, a step of counting the number of overrun packets that the buffer is full and discarded, and a step of counting the underrun packets that arrived after reproducing the immediately preceding accumulated packet and were discarded. , From the number of each packet and packetization cycle time,
And a step of determining a fluctuation absorption delay time from packet reception to the start of accumulation in the accumulation buffer.

In the data packet communication method according to the second aspect of the present invention, the encoded data is MSP (Mos) according to a predetermined importance.
t Significant Part) bit and LSP (Least Significa)
nt Part) bit into a data block, and when the packet of the data block is received in a significant unit, counting the number of packets discarded according to the predetermined importance in units of importance, From the packet reception to the start of accumulation in the accumulation buffer.

[0012]

In the data / packet communication method according to the present invention, the number of packets accumulated in the buffer, the number of overrun packets, and the number of underrun packets of the packets received in significant units are determined at each timing. The fluctuation absorption delay time is determined by a predetermined arithmetic expression. In the data packet communication method according to the second aspect of the present invention, the number of discarded packets is counted for each importance of MSP and LSP of the received packet, and the fluctuation absorption delay time is determined by comparing with the threshold value.

[0013]

【Example】

Example 1. An embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a flowchart when the present invention is applied to a fluctuation absorption delay method of a voice packet receiving device. Here, the fluctuation absorption delay time is the time from the reception of a packet to the start of storage in the storage buffer. Figure 2
7 is a communication sequence chart in the case where an underrun state occurs in which the next packet arrives after reproducing the immediately preceding stored packet. In the figure, 8 is the number of accumulated packets in the fluctuation absorption buffer, which is obtained by counting the encoded voice data 20 and the like at the start of reproduction. FIG. 3 is a communication sequence chart when the first fluctuation absorption delay time 50 is changed to the second fluctuation absorption delay time 51 calculated from the number of accumulated packets 8 in FIG. FIG. 4 is a communication sequence chart when an overrun state occurs in which the voice packet 30 and the like are discarded because the fluctuation buffer has no space. this is,
5 is a communication sequence chart in the case where the original first fluctuation absorption delay time (50) in FIG. 4 is changed to the second fluctuation absorption delay time (51) calculated based on the number of accumulated packets 8 in FIG. 4.

FIG. 6 is a communication sequence chart when an underrun state occurs. This is a diagram showing a process of calculating the fluctuation absorption delay time when the number of accumulated packets in the fluctuation absorption buffer is 8 is counted as -1 when an underrun is detected. FIG. 7 is a communication sequence chart using the second fluctuation absorption delay time (51) obtained from the method of counting the number of storage buffers (8) in FIG. FIG. 8 is a communication sequence chart when an overrun state occurs. In this case, assuming that the buffer is infinite, and regarding the packet discarded due to overrun, it is assumed that the packet is accumulated until the reproduction start of the packet, and the number of accumulated packets in the fluctuation absorption buffer is counted 8. FIG. 9 shows the second fluctuation absorption delay time (51) using the number of accumulated packets calculated in the chart of FIG.
It is a figure which shows the communication sequence chart which set. FIG. 10 is a communication sequence chart showing a case where the second fluctuation absorption delay time calculated from the counted number of accumulated packets 8 and the target number of accumulated buffers is the same as the current state when the overrun state occurs. FIG. 11 shows the fluctuation absorption delay time so that the new fluctuation absorption delay time calculated according to the flowchart of FIG. 1 is equal to or more than a certain number of overrun packets (in this case, 1 or more) when the same as the current state. It is a communication sequence chart when it is changed.

Next, the operation will be described with reference to FIG.
First, at the reproduction timing of the audio packet (3), the number of packets stored in the fluctuation absorbing buffer is measured. Then, for each one talk spurt (voiced section), the average number of accumulated packets is calculated using the equation (1) (step S1). Average number of stored packets = N / k ...
(1) However, N = sum of the number of stored packets at the reproduction timing from the head of the talk spurt to the k th k = the number of packets forming the talk spurt- (the target number of stored packets-1) In the equation (1), the target number of accumulated packets is the number of packets accumulated in the fluctuation absorption buffer, which is the fluctuation absorption delay time in which the overrun and underrun states do not occur. K in the equation (1) also indicates the number of reproduction timings at which the target number of accumulated packets may be accumulated.

The meaning of the equation (1) is as follows. For example, consider the case where the target number of accumulated packets is two. At the reproduction timing of the last packet of the talk spurt, only one packet is stored at the maximum, so when the average value is calculated by adding the number of stored packets at this timing, the average value becomes lower than the actual value. Therefore, in order to obtain a correct average value, the average value of the number of accumulated packets is calculated only for the reproduction timing at which the target number of accumulated packets may be accumulated, that is, for the k kths from the head of the talk spurt. Next, the second fluctuation absorption delay time (51) is calculated using Equation 2 using the average number of accumulated packets calculated by the above method. (Step S2) New fluctuation absorption delay time (Tnew) = ΔN × T + Told
(2) However, ΔN = (target average number of accumulated packets−average number of accumulated packets) T = packeting cycle time 4 Told = current fluctuation absorption delay time The meaning of equation (2) is The difference between the target average accumulated packet count and the current average accumulated packet count and the packetization cycle time 4 is added to the current fluctuation absorption delay time to make a new fluctuation absorption delay time. Is.

The processing of steps S1 and S2 in FIG. 1 will be described with reference to FIGS. FIG. 2 shows that the target average number of accumulated packets = 1 and the first fluctuation absorption delay time 5
The case where the value of 0 is Told is shown. In the figure, considering the number of accumulated packets in which the encoded voice data 20 to 23 are stored in the fluctuation absorbing buffer at the reproduction start timing, in this case, the third voice packet 32 including the third encoded voice data 22 is assumed. Since only one arrives after the reproduction start timing, the numbers are 1, 1, 0 and 1, respectively. As a result,
The average number of accumulated packets is as follows according to the equation (1). Average number of accumulated packets = (1 + 1 + 0 + 1) / 4- (1-
1) = 3/4. As a result, the new fluctuation absorption delay time is the target average number of accumulated packets = 1, so that the new fluctuation absorption delay time is (1−3 / 4) × T + Tol from equation (2).
d = 1/4 × T + Told. FIG. 3 shows the case where the delay fluctuation amount of the voice packets 20 to 23 is the same as that of FIG. 2, and fluctuation absorption delay time = 1/4 ×
It is shown that when changing to T + Told, the underrun state is eliminated and the number of packets stored in the fluctuation absorption buffer becomes the target average accumulated packet number = 1.

Further, in FIG. 4, the target average number of accumulated packets = 1 and the value of the first fluctuation absorption delay time 50 is Told.
, The case where the maximum number of packets stored in the fluctuation absorbing buffer = 3 is shown. At this time, the number of accumulated packets 8 for the encoded audio data 20 to 24 in the fluctuation absorbing buffer at the reproduction start timing is allowed up to 3. First
In this example, when the fourth voice packet 33 arrives, the voice packets 30 to 32 are too large because the value Told of the fluctuation absorption delay time 50 is too large compared to the delay fluctuation amount in the network.
Is stored in the fluctuation absorption buffer and is in the state of being reproduced or in the state of waiting for reproduction, and is discarded because it cannot be completely stored in the buffer (hereinafter, this state is referred to as an overrun state). In this case, the number of accumulated packets at each reproduction timing of the fluctuation absorption buffer is 3, 2, 2, 1, and 1, respectively, and the average number of accumulated packets 8 is calculated by the equation (1) as follows: 3 + 2 + 2 + 1 + 1) / 5-
(1-1) = 9/5, and as a result, the second fluctuation absorption delay time 51 is the second average fluctuation absorption delay time = (1- 9/5) x T + Tol
d = −4 / 5 × T + Told. FIG. 5 shows the case where the amount of delay fluctuation of the voice packets 20 to 24 is the same as that of FIG. 4, and fluctuation absorption delay time = −4 / 5.
It indicates that the overrun state is resolved when the value is changed to × T + Told.

When the number of packets accumulated in the fluctuation absorbing buffer is counted as -1 instead of 0 when the underrun condition occurs, the fluctuation absorbing delay time is larger than that in the case of FIG. 2 when the underrun time 6 is large. It can grow quickly. FIG. 6 shows a case where the number of accumulated packets is counted as -1 when the underrun state occurs. The number of packets 8 of the encoded audio data 20 to 23 stored in the fluctuation absorbing buffer at the reproduction start timing is as follows. Since only the third voice packet 32 carrying the third encoded voice data 22 arrives after the reproduction start timing and is discarded, the number of packets becomes 1, 1, -1, 1 respectively, and the average accumulated packet Equation 8 is given by the following equation using equation (1). Average number of accumulated packets = (1 + 1-1 + 1) / 4- (1-
1) = 2/4 As a result, the second fluctuation absorption delay time 51 is the second fluctuation absorption delay time = (1-2 / 4) × from the target average accumulated packet number = 1 to the equation (2). T + Tol
d = 1/2 × T + Told. This sets the number of accumulated packets during underrun to 0
Compared with the case of FIG. 3 which is counted as (1 / 2-1 /
4) The fluctuation absorption delay time can be increased by (T) = 1/4 × T. FIG. 7 shows the fluctuation absorption delay time from Told to 1/2 × T + Tol according to the above calculation result.
When the value is changed to d, the underrun state is eliminated, and the number of packets stored in the fluctuation absorbing buffer becomes the target average accumulated packet number = 1.

Further, the number of packets accumulated in the fluctuation absorbing buffer after the occurrence of the overrun state is discarded by the overrun until the reproduction of the encoded voice data 2 included in the voice packet 3 discarded by the overrun is started. When it is considered that the voice packet 3 is virtually stored in the fluctuation absorption buffer, counting is performed as compared with the case where the voice packet discarded due to overrun is not considered.
The fluctuation absorption delay time can be shortened more quickly. In FIG. 8, the capacity of the fluctuation absorption buffer is set to 3 packets. Therefore, when the fourth voice packet 33 arrives, the first voice packet 30 is reproducing the first voice coded data 20 included in the packet, and the second voice packet 31 and the third voice packet 30 are being reproduced. The packet 32 is waiting for reproduction and there is no free space in the fluctuation absorption buffer. However, it is considered that the voices are stored in the fluctuation absorbing buffer, and at the reproduction start timing of the second voice encoded data 21, the fourth voices actually discarded in the voice packets 31 and 32 stored in the fluctuation absorbing buffers. The packet 33 is added and counted as 3. Also, at the reproduction start timing of the third voice coded data 22, the packets 32 and 34 and the discarded fourth voice packet 3
3 is added to count 3 and similarly, at the reproduction start timing of the fourth voice coded data 23, it is counted as 2.

As described above, assuming that the size of the fluctuation absorbing buffer is infinite, and the number of accumulated packets is counted even if an overrun state occurs, the average number of accumulated packets is calculated as 8 According to (1), the average number of accumulated packets = (3 + 3 + 3 + 2 + 1) / 5-
(1-1) = 12/5. From the equation (2), the second fluctuation absorption delay time is calculated as follows: Second fluctuation absorption delay time = (1-12 / 5) × T + T
old = −7 / 5 × T + Told, and the fluctuation absorption delay time can be greatly shortened compared to the case where the voice packets discarded due to overrun are not counted (FIG. 5).

Now, returning to FIG. 1, the processing after step S3 will be described. Step S1 and step S2
It is checked whether or not the fluctuation absorption delay time has been changed by the process (step S3), and if not, step S
If it is changed to 4, the processing is terminated. Step S4
Then, it is checked whether the number of overrun packets in one talk spurt is equal to or more than a threshold value. If it is equal to or more than the threshold value, the process proceeds to step S5, and the value of the fluctuation absorption delay time is reduced according to the equation (3). New fluctuation absorption delay time = current fluctuation absorption delay time−packeting cycle time (T) × α ... (3) where α is a predetermined value. Check if the number of underrun packets in one talk spurt is above a threshold.

If it is detected in step S6 that the number of underrun packets is equal to or greater than the threshold value, the value of the fluctuation absorption delay time is increased according to the equation (4) (S
7) If it is detected that it is less than the threshold value, the process ends. New fluctuation absorption delay time = current fluctuation absorption delay time + packeting cycle time (T) × β (4) However, β is a predetermined value. Therefore, when overrun occurs, , Formula (3)
Therefore, the fluctuation absorption time can be shortened, and when the underrun occurs, the fluctuation absorption time can be increased by the formula (4). Then, the probability of occurrence of overrun and underrun can be suppressed for the subsequent talk spurts.

The processes of steps S3 to S6 will be described with reference to the charts of FIGS. 10 and 11.
In FIG. 10, since the fourth voice packet 33 is discarded, the number of packets stored in the fluctuation absorption buffer at the reproduction start timing of each voice coded data 2 is
3, 2, 2, 1, 1. From equation (1), the average number of accumulated packets = (3 + 2 + 2 + 1) / (5- (2
-1)) = 8/4 = 2. Therefore, from the equation (2), the second fluctuation absorption delay time = 2-2 + Told = Told, and the fluctuation absorption delay time does not change. If the threshold value of the number of overrun packets in one talk spurt = 1, the fluctuation absorption delay time does not change.
The process proceeds in the order of 3, step S4, and step S5. Formula (3)
Therefore, the second fluctuation absorption delay time = Told−T × α = Told−T
× α. Here, if α = 1/2, the second fluctuation absorption delay time = Told−T × 1/2, and the second fluctuation absorption time calculated based on the average number of accumulated packets in the fluctuation absorption buffer is 8 Even if it is the same as the fluctuation absorption time, the fluctuation absorption delay time can be shortened if the number of overrun packets is a certain value or more. As a result, if the delay fluctuations in the network are the same in FIGS. 10 and 11, the fluctuation absorption delay time becomes an appropriate value by the above processing, and the delay fluctuations can be absorbed as shown in FIG.

Example 2. Another embodiment of the present invention will be described. The transmitter side embeds the voice, accumulates the encoded voice data for a predetermined number of samples, collects the bits at the same bit positions for all the samples, creates a data block for each same bit position, and collects the MSP bits. There is a method of arranging and transmitting the LSP bits from the data blocks in the order of the data blocks. Then, when congestion occurs at the intermediate switching center or the like, the data block in which the LSP bits are collected is discarded to alleviate the congestion. On the receiving side, encoded voice data for each sample is restored from the received data block and reproduced. At this time, the number of discarded packets of the data block in which the LSP lower bits are collected and the number of discarded data blocks of the LSP upper bits and the LSP lower bits in the received voice packet are measured for each talk spurt. In this embodiment,
The fluctuation absorption delay time added to the first packet of the subsequent talk spurt is changed according to the number. In this way, an appropriate fluctuation absorption delay time corresponding to the change in the transmission delay of the voice packet due to the load change in the network can be obtained.

FIG. 12 shows a method of constructing a voice packet used in the data packet communication method of claim 2. In the figure, 9 is the encoded voice data that has been embed-coded,
Reference numeral 10 indicates each bit of the encoded voice data 9. 1
00 is MSP (Most Significant Part) upper bit, 1
01 is an MSP lower bit, which is a bit whose voice quality is significantly deteriorated when these bits are discarded. 10
2 is the upper bit of LSP (Least Significant Part), 1
Reference numeral 03 is an LSP lower bit, which is a bit with little deterioration in voice quality when these bits are discarded. Reference numeral 11 is a payload for storing the encoded voice data 9 of the voice packet. 110 is an MSP upper data block created by collecting MSP upper bits 100, 111 is an MSP lower data block created by collecting MSP lower bits 101, 112 is an LSP upper data block created by collecting LSP upper bits 102, and 113 is It is an LSP lower data block created by collecting LSP lower bits 103. 12 is a voice packet header, 120 is a header 12
Is a discard indication bit indicating whether or not the data block included in is discarded.

FIG. 13 is a configuration block diagram of a communication system using the data packet communication method according to the second aspect.
In the figure, 130 is a voice packet communication terminal on the transmitting side, 1
31 is a voice packet communication terminal on the receiving side, and 140 and 141 are first and second intermediate node devices, respectively. Also, 1
50 and 151 are first and second terminal lines, respectively, and 16 is a relay line. FIG. 14 shows the first intermediate node device 14
It is a figure which shows the processing flow of 0. FIG. 15 is a diagram showing the relationship between the load in the first intermediate node device 140 and the end-to-end delay amount of the voice packet 3, and at the same time, shows the relationship between the LSP upper or LSP lower data block discard and the delay amount. FIG.

FIG. 16 and FIG. 17 are diagrams showing an algorithm of a method of determining a fluctuation absorption delay time in a reception example used in the embodiment of the invention of claim 2. That is, in a voice packet forming one talk spurt, the number of packets in which the data block in which the LSP lower bit 103 is collected and the data block in which the LSP upper bit 102 and the LSP lower bit 103 are collected are discarded. It is a figure which shows the processing flow at the time of determining a new fluctuation absorption delay time based on the result of having measured the number of the existing packets. Figure 1
FIG. 8 is a diagram showing an algorithm for providing the fluctuation absorption delay time when the first packet of the next talk spurt arrives after the fluctuation absorption delay time is determined by the processing flows of FIGS. 16 and 17. FIG. 19 is a diagram showing a communication sequence when the fluctuation absorption delay time is changed using the processing flows of FIGS. 16 to 18. In the figure, 17 is the elapsed time from the arrival of the last packet of the first talk spurt 1-0 to the arrival of the first packet of the second talk spurt 1-1.

Next, the operation will be described. The method of generating the voice packet 30 and the like in the transmitting side voice packet communication terminal 130 is as shown in FIG. That is, the MSP high-order bit 10 of the encoded voice data 9 that has been embedded-coded
0, MSP lower bit 101, LSP upper bit 10
2, the LSP lower bit 103 is collected for each same bit for a plurality of samples, and the MSP upper data block 110, the MSP lower data block 111, the LSP upper data block 112, and the LSP lower data block 1 are collected.
Create 13. Furthermore, payload 1 of voice packet 3
The header 12 in which 110, 111, 112 and 113 are inserted in that order and the discard indication bit 120 is set to no discard is added to one part.

The voice packet created by the above method is transmitted to the voice packet communication terminal 130 on the transmitting side as shown in FIG.
To the receiving side voice packet communication terminal 131, the first terminal line 150, the first intermediate node device 140, the relay line 1
6, the second intermediate node device 141, and the terminal line 151. The intermediate node device 130 compares the number of voice packets waiting for transmission to the relay line 16 with a threshold value. Then, when it is equal to or more than the threshold value, it is determined that the congestion occurs, and the LSP lower data block 113 or the LSP lower data block 113 which has a lesser degree of quality deterioration.
Also, the LSP upper data block 112 is discarded to reduce congestion. The processing flowchart of the first intermediate node device 140 is as shown in FIG. 14, and it is determined whether the number of packets (queue length) waiting to be transmitted to the relay line 16 is equal to or greater than the threshold value 1 (step S11). In the case, if the threshold value is less than 1, the process of discarding the data block is not performed. In step 12, it is determined whether or not the number of packets waiting to be transmitted (queue length) is greater than or equal to a threshold 2 (threshold 2> threshold 1). When the threshold value is 2 or more, the trunk line 1
6, the LSP lower data block 113 and the LSP upper data block 112 included in the payload 11 of the voice packet 30 and the like waiting to be transmitted to
Discard. Then, a code indicating the discard of two data blocks is set in the discard display bit 120 (step S1).
4). When the threshold value is less than 2, the LSP lower data block 113 is discarded for the voice packet 30 and the like waiting for transmission to the relay line 16, and the discard indication bit 12
A code indicating the discarding of one data block is set to 0 (step S14).

On the other hand, as shown in FIG. 15, the end-to-end delay time of the voice packet is larger when the number of voice channels to be multiplexed is larger than the line capacity of the relay line (16).
That is, if the required line speed exceeds the line capacity when all voice channels are sending packets, the LSP
Even if data blocks are discarded, the number will increase as the number of voice channels increases. When the number of voice channels increases, discarding of the LSP lower data block 113 starts at point A, and LSP lower data block 113 and LSP upper data block 11 at point B.
When the discard of 2 starts and the number of voice channels decreases on the contrary, the discard of the LSP lower data block 113 and the LSP upper data block 112 is stopped at the point B,
At point A, discarding of the LSP lower data block 113 is stopped. Therefore, the receiving side voice packet communication terminal 131
Then, in the received voice packet 30 etc., the LSP lower data block 113 and the LSP upper data block 11
By measuring the number of discarded packets of 2, it is possible to predict the delay variation when the transmission delay time of the voice packet 30 and the like and the load of the first intermediate node device 140 vary. When the load changes within the range of point A or less where the data block is not discarded, the delay variation is the smallest, and the load changes between point A and point B where the LSP lower data block 113 is discarded. Or, when the load changes from the point A to the point B to the point A or less or the point B or more, the delay variation becomes larger than the variation in the range of the point A or less, and from the point B or more to the point A or less. When the load changes, the delay variation is the largest.

Therefore, the underrun state and the overrun state can be prevented by changing the fluctuation absorption delay time according to the magnitude of the delay fluctuation. LS
A processing flow when the fluctuation absorbing delay time is changed at the receiving terminal using the measurement results of the number of discarded packets of the P lower data block 113 and the LSP upper data block 112 will be described with reference to FIGS. 16 and 17. In the figure, X0, X1, and X2 are values of fluctuation absorption time.
0 <X1 <X2, and the threshold values A and B are 1 or more. First, the value of the fluctuation absorption time at the present time is checked (step S21), and if X0, that is, less than X1,
LSP lower data block 1 in one talk spurt
The number of voice packets 30 etc. in which only 13 or both LSP lower data block 113 and LSP upper data block 112 are discarded is measured. It is determined whether or not the number is equal to or more than the threshold value A (step S22). If the number is less than the threshold value A, it is determined that the delay variation is the smallest, and the fluctuation absorption time is set to X0 (step S23). If it is equal to or larger than the threshold A, the number of voice packets in which both the LSP lower data block 113 and the LSP upper data block 112 are discarded is compared with the threshold B (S24). When it is equal to or more than the threshold value B, the fluctuation absorption time is set to X2 (S25), and when it is less than the threshold value B, the fluctuation absorption time is set to X1 (step S24).

The value of the current fluctuation absorption time is X1.
In the case of (that is, X0 or more and less than X2), step S
In step S27, the number of voice packets in which both the LSP lower data block 113 and the LSP upper data block 112 are discarded is counted. It is determined whether the number is equal to or more than the threshold C (S27), and the threshold C
If less than, the fluctuation absorption time is left as X1 (S2
8) If it is equal to or more than the threshold value C, the fluctuation absorption time is set to X2 (step S29). On the other hand, when the current value of the fluctuation absorption time is X2, steps S21 to S30
In step S30, the number of voice packets 30 and the like in which only the LSP lower data block 113 is discarded is measured, and it is determined whether the number is equal to or more than the threshold value D (S30). When it is less than the threshold value D, the fluctuation absorption time is left as X2 (S31), and when it is more than the threshold value D, the fluctuation absorption time is set as X1 (step S32). However, if the occurrence interval of the talk spurt is large, even if the fluctuation absorption time is determined by the above processing flow, the load of the intermediate node device changes between the time of the decision and the arrival of the packet belonging to the next talk spurt, The optimal fluctuation absorption time may change. Therefore, as shown in the flowchart of FIG. 18, when the first packet of the talk spurt arrives, the elapsed time 17 from the arrival of the last packet of the previous talk spurt is compared with the threshold Z (S41). If it is less than or equal to the threshold value Z, the fluctuation absorption time determined by the flow of FIGS. 16 and 17 is set. If the elapsed time 17 exceeds the threshold value Z, the fluctuation absorption time is set to the smallest X0 (step S43). The values of X0, X1, and X2 are shown in FIG.
The appropriate value that can absorb the fluctuation of the packet transfer delay may be selected with reference to the graph of FIG.

A communication sequence using the processing flows of FIGS. 16 to 18 will be described with reference to FIG. In FIG. 19, the LSP lower block 113 of the second voice packet 231-0 of the first talk spurt 201-0 is discarded. Also, the first voice packet 230-1 of the second talk spurt 201-1 is the LSP lower block 11
3 has been discarded and the second voice packet 231-1 has LSP lower block 113 and LSP upper block 1
12 are discarded, and the second voice packet 231-1
Transfer delay due to congestion of the first intermediate node device 140. Threshold A = B = of the processing flow of FIG.
If it is 1, the current fluctuation absorption time is X0, so
Step S for the first talk spurt 201-0
The process proceeds to 21, S22, S24 and S25. Fluctuation absorption time for the second talk spurt 201-1 from X0 to X
It becomes 1. Furthermore, since the elapsed time 17 from the arrival of the final packet 231-0 of the first talkspurt 201-0 to the arrival of the head packet 230-1 of the second talkspurt 201-1 is Z0 <Z, the second The fluctuation absorption time for the talk spurt 201-1 is finally set to X1. Therefore, the voice packets 230-1 and 231-
Even if the transfer delay difference (fluctuation) of 1 becomes large, the fluctuation absorption time becomes longer from X0 to X1, so that the fluctuation can be absorbed and the occurrence of the underrun state can be prevented.

In the above description, the processing in the case of one channel is described, but the number of discarded packets of the LSP lower data block 113 and the LSP upper data block 112 in the talk spurt of a certain channel is measured. Then, the value of the fluctuation absorption addition delay time is determined according to the processing flowcharts of FIGS.
This decision value may be applied to the talkspurt of another channel that arrives within a certain time passing through the same route in the network.

Example 3. In the above, LSP lower bit 1
The number of packets in which the data blocks collecting 03 are discarded and the LSP upper bit 102 and the LSP lower bit 1
The number of packets in which the data blocks of 03 are discarded is measured for each talk spurt, and the fluctuation absorption delay time is changed based on the result. Instead, the number of packets in the talk spurt in which the data blocks in which the LSP lower bits 103 are collected and the packet in which both the LSP upper bits 102 and the data blocks in which the LSP lower bits 103 are collected are discarded. The fluctuation absorption delay time may be changed by monitoring whether or not there is a part. FIG. 20 shows that for the packet of the last part of the talk spurt, the number of packets in which the data block in which the LSP lower bit 103 is collected and the data block in which the LSP upper bit 102 and the LSP lower bit 103 are collected are both discarded. It is a figure which shows the processing flow which monitors the presence or absence of the packet which exists, and changes a fluctuation absorption delay time.

Next, the operation will be described. It is checked for each talk spurt whether or not the data block is discarded for the last packet of the talk spurt or the packets from the last to a plurality of packets before (S51). If there is no discard, the fluctuation absorption delay time is set to X0 ( S52). LS
If the P lower data block 113 is discarded, the fluctuation absorption delay time is set to X1 (S53), and the LSP
When both the lower data block 113 and the LSP upper data block 112 are discarded, the fluctuation absorption delay time is set to X2 (S54). Note that when packets for discarding the LSP lower data block 113 and discarding both the LSP lower data block 113 and the LSP upper data block 112 are mixed, the LSP lower data block 113 and the LSP upper data block 112 are both discarded. Processing is performed assuming that there are only packets. The processing flow shown in FIG. 18 is used as the algorithm for providing the fluctuation absorption delay time when the first packet of the next talk spurt arrives. The operation will be described by taking the communication sequence of FIG. 19 as an example. Since the LSP lower block 113 of the final voice packet 231-0 of the first talkspurt 201-0 is discarded, the fluctuation is absorbed in step S53 of the flowchart of FIG. Delay time is X0
Changed to X1. Since the elapsed time 17 = Z0 is equal to or less than the threshold value Z, the fluctuation absorption delay time is set to X1.
Therefore, even if the transfer delay difference (fluctuation) between the voice packets 230-1 and 231-1 becomes large, the fluctuation absorption time X
Since the length increases from 0 to X1, the fluctuation can be absorbed and the occurrence of the underrun state can be prevented.

In the above description, the processing in the case of one channel is described, but the number of discarded packets of the LSP lower data block 113 and the LSP upper data block 112 in the talk spurt of a certain channel is measured. The value of the fluctuation absorption delay time 5 may be determined according to the processing flowcharts of FIGS. 16 and 17, and this determined value may be applied to the talkspurt of another channel that arrives within a certain time and passes through the same route in the network.

Example 4. In the above embodiment, the voice packet is described, but the packet transmission of the moving image can be applied to both the first and second embodiments.

[0040]

As described above, according to the present invention, the step of counting accumulated packets, overrun packets, and underrun packets, or the number of discarded packets of low importance among packets grouped according to importance. Since the step of counting and the step of determining the fluctuation absorption delay time from these are provided, there is an effect that the fluctuation absorption delay time can be appropriately determined with a relatively simple configuration.

[Brief description of drawings]

FIG. 1 is a flow chart of a fluctuation absorption delay time setting of a voice packet according to an embodiment of the present invention.

FIG. 2 is a communication sequence diagram when an underrun state occurs.

3 is a communication sequence diagram after changing the fluctuation absorption delay time in the underrun state of FIG. 2 using the processing flow of FIG.

FIG. 4 is a communication sequence diagram when an overrun state occurs.

5 is a communication sequence diagram when the fluctuation absorption delay time is changed in the overrun state of FIG. 4 by using the processing flow of FIG.

FIG. 6 is a communication sequence diagram when an underrun state occurs.

7 is a communication sequence diagram after the fluctuation absorption delay time is changed using the method for counting the number of accumulated packets in the fluctuation absorption buffer of FIG. 6 and the processing flow of FIG.

FIG. 8 is a communication sequence diagram when an overrun state occurs.

9 is a communication sequence diagram after the fluctuation absorption delay time is changed using the method for counting the number of accumulated packets in the fluctuation absorption buffer in FIG.

FIG. 10 is a communication sequence diagram in the case where the fluctuation storage delay time of the count storage packet and the target storage buffer number becomes the same as the current state when an overrun occurs.

FIG. 11 is a communication sequence diagram after changing the fluctuation absorption delay time so that the number of overrun packets becomes a certain number or more when the calculated fluctuation absorption delay time is the same as the current state.

FIG. 12 is a diagram showing the structure of a voice packet used in the data packet communication method according to claim 2 of the present invention.

FIG. 13 is a block diagram of a communication system that realizes the data packet communication method of claim 2;

FIG. 14 is a diagram showing a packet transmission processing flow to the relay line of the intermediate node device.

FIG. 15 is a diagram showing the relationship between the load and the end-to-end delay amount of voice packets, and the relationship between the discard of LSP upper or LSP lower data blocks and the delay amount.

FIG. 16 is a diagram showing a fluctuation absorption delay time determination processing flow based on the number of discarded packets of a data block of LSP lower bits and the number of discarded packets of a data block of LSP upper bits and LSP lower bits.

FIG. 17 is a diagram showing a fluctuation absorption delay time determination processing flow based on the number of discarded packets of a data block of LSP lower bits and the number of discarded packets of a data block of LSP upper bits and LSP lower bits.

18 is a diagram showing an algorithm for assigning the fluctuation absorption delay time of the next head packet after the fluctuation absorption delay time is determined in the processing flow of FIG. 17;

FIG. 19 is a communication sequence diagram according to the fluctuation absorption delay time in the processing flows of FIGS. 16 to 18;

FIG. 20 is a diagram illustrating a fluctuation absorption delay time change processing flow according to the third embodiment.

FIG. 21 is a communication sequence diagram for explaining a method of absorbing delay fluctuation of a voice packet in an underrun state in a conventional example.

22 is a communication sequence diagram after setting the new fluctuation absorption delay time of FIG. 21.

FIG. 23 is an explanatory diagram of a conventional packet average transmission delay time calculation method.

FIG. 24 is a flowchart showing a method of changing the fluctuation absorption delay time in the conventional example.

[Explanation of symbols]

1-0,1-1 Talk spurt 20, 21, 22, 33, 24 Coded voice data 30, 31, 32, 33, 34 Voice packet 4 Packetization period 50, 51 Fluctuation absorption delay time 6 Underrun time 7 packets Average transmission delay time 8 Number of accumulated packets 9 Embed-coded encoded voice data S1 Average value calculation step S2 Fluctuation absorption delay time setting step S4 Overrun packet measurement step S5 Fluctuation absorption delay time setting step S6 Underrun packet measurement step S7 Fluctuation absorption delay time setting step S21 Fluctuation absorption time measurement step S22 LSP lower, LSP lower + LSP upper drop packet number comparison step S23 Fluctuation absorption time setting step S24 LSP lower + LSP upper drop packet number comparison step Step S25 Fluctuation absorption time setting step S26 Fluctuation absorption time setting step S27 LSP lower + LSP higher discard packet number comparison step S28 Fluctuation absorption time setting step S29 Fluctuation absorption time setting step S30 LSP lower discard packet number comparison step S31 Fluctuation absorption time setting Step S32 Fluctuation absorption time setting step

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] March 24, 1993

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] Claim 1

[Correction method] Change

[Correction content]

[Procedure Amendment 2]

[Document name to be amended] Statement

[Name of item to be corrected] Claim 2

[Correction method] Change

[Correction content]

[Procedure 3]

[Document name to be amended] Statement

[Name of item to be corrected] 0002

[Correction method] Change

[Correction content]

[0002]

2. Description of the Related Art FIGS.
[Fig. 36] Fig. 36 is a diagram for describing a conventional method for absorbing delay fluctuation of a voice packet shown in 3641. In the figure, 301-0,
Reference numeral 301-1 designates first and second voiced sections (hereinafter referred to as talk spurts), and coded voice data in which the talk spurts 301 are sectioned at regular time intervals 320 and 3, respectively.
21, 322, 323 are the first, second, third, and
This is the fourth encoded voice data. 330 is a voice packet in which encoded voice data for a certain period of time is packetized
333 is encoded voice data 320, 321, respectively.
322 and 323 are first, second, third, and fourth voice packets packetized. 304 is encoded voice data 3
It is a packetization cycle for assembling 20 to 323 into voice packets 320 to 333. Voice packets 320-333
In order to absorb the difference in transfer delay time that differs for each packet when a packet is transferred, the fluctuation absorption delay time that waits from the arrival of the first packet of the talk spurt to the start of playback
It is necessary, 350 is the first fluctuation absorption delay time, 3
Reference numeral 51 is a second fluctuation absorption delay time in which the first fluctuation absorption delay time is changed. 306 is a reproduction from the reproduction start point in the underrun state (a packet to be reproduced has not arrived at the reproduction timing at the reproduction start time) generated because the difference in the transfer delay time due to the fluctuation absorption delay time cannot be absorbed. It is the underrun time until the arrival of the packet to be reproduced at the timing.

[Procedure amendment 4]

[Document name to be amended] Statement

[Name of item to be corrected] 0005

[Correction method] Change

[Correction content]

On the other hand, according to the example of the figure, the third voice packet 332 suffers a large delay during the transfer, and the voice packet 3
The third encoded voice data 322 is included in the third encoded voice data 322 included in 32 since it arrived at the receiving side after the underrun time 306 = T3 has elapsed from the reproduction start time.
Is discarded without being recycled. In FIG. 21, since the value of the first fluctuation absorption delay time 350 = T2 is small, the delay fluctuation of the voice packet 332 cannot be absorbed and a part 322 of the encoded voice data in the talk spurt 301-0 is lost. ing. Therefore, the second talk spurt 301-
For No. 1 , the fluctuation absorption delay time is changed according to the flowchart of the method for changing the delay fluctuation absorption time in FIG.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0006

[Correction method] Change

[Correction content]

In step S61, it is determined whether or not the delay fluctuation is absorbed, that is, whether or not the voice packet 330 or the like is discarded due to a short delay fluctuation absorption time. When the delay fluctuation is absorbed, the process proceeds to step S62, and when the delay fluctuation is not absorbed, the process proceeds to step S63. In step S63, an underrun time 30 is added to the current fluctuation absorption delay time in order to match the current packet transfer delay fluctuation.
The value obtained by adding 6 is set as a new fluctuation absorption delay time, and the delay fluctuation absorption time 351 is increased. For example, in the case of FIG. 21, since the delay fluctuation is not absorbed, as the second fluctuation absorption delay time 351, a value obtained by adding the value of the underrun time 306 to the value of the first fluctuation absorption delay time 350, T2 + T3. To set. As shown in FIG. 22, the delay fluctuation can be absorbed by using the second fluctuation absorption delay time 351 shown in step 3,
1 and the delay fluctuation in the network are the same, it is possible to reproduce all the encoded voice data 320 and the like in the talk spurt ( 301-1 ).

[Procedure correction 6]

[Document name to be amended] Statement

[Correction target item name] 0009

[Correction method] Change

[Correction content]

The present invention has been made in order to solve the above-mentioned problems, and does not require measurement of the average transmission delay time and underrun time, and has a simple structure and can be adaptively changed in delay fluctuation absorption. Aim to get a way. In addition, a data packet communication method that can change the fluctuation absorption delay time in response to changes in network load
The purpose is to get the law .

[Procedure Amendment 7]

[Document name to be amended] Statement

[Correction target item name] 0010

[Correction method] Change

[Correction content]

[0010]

Data according to the present invention
The packet communication method is as follows:
A step of counting the number of accumulated packets that are accumulating received data, a step of counting the number of overrun packets that the buffer is full and discarded, and a step of counting the underrun packets that arrived after reproducing the immediately preceding accumulated packet and were discarded. , From the number of each packet and packetization cycle time,
And a step of determining a fluctuation absorption delay time from the packet reception to the start of reproduction .

[Procedure Amendment 8]

[Document name to be amended] Statement

[Correction target item name] 0011

[Correction method] Change

[Correction content]

In the data packet communication method according to the second aspect of the present invention, the encoded data is MSP (Mos) according to a predetermined importance.
t Significant Part) bit and LSP (Least Significa)
nt Part) bit into a data block, and when the packet of the data block is received in a significant unit, counting the number of packets discarded according to the predetermined importance in units of importance, To determine the fluctuation absorption delay time from the packet reception to the start of reproduction .

[Procedure Amendment 9]

[Document name to be amended] Statement

[Correction target item name] 0013

[Correction method] Change

[Correction content]

[0013]

EXAMPLES Example 1. An embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a flowchart when the present invention is applied to a fluctuation absorption delay method of a voice packet receiving device. Here, the fluctuation absorption delay time is the playback after the packet is received.
It is time to start . FIG. 2 is a communication sequence chart when an underrun state occurs in which the next packet arrives after reproducing the immediately preceding stored packet. In the figure, 8 is the number of accumulated packets in the fluctuation absorption buffer, which is obtained by counting the encoded voice data 20 and the like at the start of reproduction. FIG. 3 shows a second fluctuation absorption delay time 5 obtained by calculating the first fluctuation absorption delay time 50 from the number of accumulated packets 8 in FIG.
It is a communication sequence chart when changing to 1. FIG. 4 is a communication sequence chart when an overrun state occurs in which the voice packet 30 and the like are discarded because the fluctuation buffer has no space. 5 is the original first of FIG.
The fluctuation absorption delay time (50) of the second fluctuation absorption delay time (51) calculated based on the number 8 of accumulated packets in FIG.
It is a communication sequence chart when it is changed.

[Procedure Amendment 10]

[Document name to be amended] Statement

[Correction target item name] 0014

[Correction method] Change

[Correction content]

FIG. 6 is a communication sequence chart when an underrun state occurs. This is a diagram showing a process of calculating the fluctuation absorption delay time when the number of accumulated packets in the fluctuation absorption buffer is 8 is counted as -1 when an underrun is detected. FIG. 7 is a communication sequence chart using the second fluctuation absorption delay time (51) obtained from the method of counting the number of storage buffers (8) in FIG. FIG. 8 is a communication sequence chart when an overrun state occurs. In this case, assuming that the buffer is infinite, and regarding the packet discarded due to overrun, it is assumed that the packet is accumulated until the reproduction start of the packet, and the number of accumulated packets in the fluctuation absorption buffer is counted 8. FIG. 9 shows the second fluctuation absorption delay time (51) using the number of accumulated packets calculated in the chart of FIG.
It is a figure which shows the communication sequence chart which set. FIG. 10 is a communication sequence chart showing a case where the second fluctuation absorption delay time calculated from the counted number of accumulated packets 8 and the target number of accumulated buffers is the same as the current state when the overrun state occurs. FIG. 11 shows a new fluctuation absorption delay calculated according to the flowchart of FIG.
This is a communication sequence chart when the fluctuation absorption delay time is changed because the number of overrun packets becomes a certain number or more (in this case, one or more) when the time is the same as the current state .

[Procedure Amendment 11]

[Document name to be amended] Statement

[Name of item to be corrected] 0015

[Correction method] Change

[Correction content]

Next, the operation will be described with reference to FIG.
First, at the reproduction timing of the audio packet (3), the number of packets stored in the fluctuation absorbing buffer is measured. Then, for each one talk spurt (voiced section), the average number of accumulated packets is calculated using the equation (1) (step S1). Average number of stored packets = N / k ...
(1) However, N = sum of the number of stored packets at the reproduction timing from the head of the talk spurt to the k th k = the number of packets forming the talk spurt- (the target number of stored packets-1) In the equation (1), the target number of accumulated packets is the number of packets accumulated in the fluctuation absorption buffer, which is the fluctuation absorption delay time in which the overrun and underrun states do not occur. K in the equation (1) indicates the number of reproduction timings at which the target number of accumulated packets may be accumulated.

[Procedure Amendment 12]

[Document name to be amended] Statement

[Correction target item name] 0019

[Correction method] Change

[Correction content]

When the number of packets accumulated in the fluctuation absorbing buffer is counted as -1 instead of 0 when the underrun condition occurs, the fluctuation absorbing delay time is larger than that in the case of FIG. 2 when the underrun time 6 is large. It can grow quickly. FIG. 6 shows a case where the number of accumulated packets is counted as -1 when the underrun state occurs. The number of packets 8 of the encoded audio data 20 to 23 stored in the fluctuation absorbing buffer at the reproduction start timing is as follows. Since only the third voice packet 32 carrying the third encoded voice data 22 arrives after the reproduction start timing and is discarded, the number of packets becomes 1, 1, -1, 1 respectively, and the average accumulated packet Equation 8 is given by the following equation using equation (1). Average number of accumulated packets = (1 + 1-1 + 1) / 4- (1-
1) = 2/4 As a result, the second fluctuation absorption delay time 51 is the second fluctuation absorption delay time = (1-2 / 4) × from the target average accumulated packet number = 1 to the equation (2). T + Tol
d = 1/2 × T + Told. As a result, compared with the case of FIG. 3 in which the number of accumulated packets during underrun is counted as 0, (1 / 2−
The fluctuation absorption delay time can be increased by 1/4) × T = 1/4 × T. FIG. 7 shows the fluctuation absorption delay time from Told to 1/2 × T + T according to the above calculation result.
When changing to old, the underrun condition is resolved,
It shows that the number of packets stored in the fluctuation absorption buffer is the target average accumulated packet number = 1.

[Procedure Amendment 13]

[Document name to be amended] Statement

[Correction target item name] 0028

[Correction method] Change

[Correction content]

16 and 17 are diagrams showing an algorithm of a method of determining a fluctuation absorption delay time on the receiving side, which is used in the embodiment of the invention of claim 2. That is, in a voice packet forming one talk spurt, the number of packets in which the data block in which the LSP lower bit 103 is collected and the data block in which the LSP upper bit 102 and the LSP lower bit 103 are collected are discarded. It is a figure which shows the processing flow at the time of determining a new fluctuation absorption delay time based on the result of having measured the number of the existing packets. Figure 1
FIG. 8 is a diagram showing an algorithm for providing the fluctuation absorption delay time when the first packet of the next talk spurt arrives after the fluctuation absorption delay time is determined by the processing flows of FIGS. 16 and 17. FIG. 19 is a diagram showing a communication sequence when the fluctuation absorption delay time is changed using the processing flows of FIGS. 16 to 18. In the figure, 17 is the elapsed time from the arrival of the last packet of the first talk spurt 1-0 to the arrival of the first packet of the second talk spurt 1-1.

[Procedure Amendment 14]

[Document name to be amended] Statement

[Name of item to be corrected] 0032

[Correction method] Change

[Correction content]

Therefore, the underrun state and the overrun state can be prevented by changing the fluctuation absorption delay time according to the magnitude of the delay fluctuation. LS
A processing flow when the fluctuation absorbing delay time is changed at the receiving terminal using the measurement results of the number of discarded packets of the P lower data block 113 and the LSP upper data block 112 will be described with reference to FIGS. 16 and 17. In the figure, X0, X1, and X2 are values of fluctuation absorption time.
0 <X1 <X2, and the threshold values A and B are 1 or more. First, the value of the fluctuation absorption time at the present time is checked (step S21), and if X0, that is, less than X1,
LSP lower data block 1 in one talk spurt
The number of voice packets 30 etc. in which only 13 or both LSP lower data block 113 and LSP upper data block 112 are discarded is measured. It is determined whether the number is equal to or more than the threshold A (step S22), and the threshold A is not
If it is full, it is determined that the delay variation is the smallest, and the fluctuation absorption time is set to X0 (step S23). LSP lower data block 113 and LSP when the threshold value is A or more
The threshold B is compared with the number of voice packets in which both upper data blocks 112 are discarded (S24). When it is equal to or more than the threshold value B, the fluctuation absorption time is set to X2 (S25), and when it is less than the threshold value B, the fluctuation absorption time is set to X1 (step S24).

[Procedure Amendment 15]

[Document name to be amended] Statement

[Correction target item name] Fig. 16

[Correction method] Change

[Correction content]

FIG. 16 is a diagram showing a fluctuation absorption delay time determination processing flow based on the number of discarded packets of a data block of LSP lower bits and the number of discarded packets of a data block of LSP upper bits and LSP lower bits.

[Procedure Amendment 16]

[Document name to be amended] Statement

[Name of item to be corrected] Fig. 17

[Correction method] Change

[Correction content]

FIG. 17 is a diagram showing a fluctuation absorption delay time determination processing flow based on the number of discarded packets of data blocks of LSP lower bits and the number of discarded packets of data blocks of LSP upper bits and LSP lower bits.

[Procedure Amendment 17]

[Document name to be amended] Statement

[Name of item to be corrected] Fig. 19

[Correction method] Change

[Correction content]

FIG. 19 is a communication sequence diagram according to the fluctuation absorption delay time obtained by the processing flows shown in FIGS.

[Procedure 18]

[Document name to be corrected] Drawing

[Correction target item name] Fig. 22

[Correction method] Change

[Correction content]

FIG. 22

Claims (2)

[Claims] 1. A step of counting the number of accumulated packets that are accumulating received data at each packet reproduction timing, a step of counting the number of overrun packets that are full in a buffer and discarded, and arrive after reproducing the immediately preceding accumulated packet. And a step of counting the number of discarded underrun packets, and a step of determining a fluctuation absorption delay time from the packet reception to the start of accumulation in the accumulation buffer based on the number of each packet and the packetization cycle time. 2. The encoded data is M according to a predetermined importance.
SP (Most Significant Part) bit and LSP (Least S)
ignificant Part) divided into data blocks into data blocks, and when receiving packets of the data blocks in a significant unit, counting the number of discarded packets according to the predetermined importance in units of importance; From the packet reception to the start of accumulation in the accumulation buffer, the step of determining a fluctuation absorption delay time.