JP2003264583A - Packet shaping method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a packet shaping method and device capable of transferring a packet string with an almost fixed rate by reducing jitter without deteriorating any service property due to delay. <P>SOLUTION: This packet shaping device is provided with a means (6) for measuring the successive packet intervals τ<SB>i</SB>or packet rates of a received packet string, a means (8) for calculating the mean packet interval α<SB>i</SB>or mean packet rate of the received packet string from the measured successive packet intervals τ<SB>i</SB>, a means (8) for calculating the jitter of the received packet string by using the successive packet intervals τ<SB>i</SB>and the mean packet interval α<SB>i</SB>, a statistical processing means (8) for executing statistical processing to the calculated jitter, and for calculating the usage of a packet buffer based on the result of the statistical processing, a means (8) for setting the buffer usage of the packet buffer as the calculated usage, and an output control means (9) for executing the output control of packets stored in the packet buffer by using the calculated mean packet interval or mean packet rate. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to an IP (Internet).
TECHNICAL FIELD The present invention relates to a packet interval shaping method in a network using a communication protocol such as et Protocol), and more particularly to a method and apparatus for performing shaping so as to reduce jitter of stream contents.

[0002]

2. Description of the Related Art In a content distribution service by IP,
A service for delivering compressed video data such as MPEG2 is conceivable. In such services, MPEG2
A video server or a video encoder that converts video data into IP packets and transmits the IP data can be used for distribution such as.

However, when the output characteristics of commercially available video servers and video encoders are examined, the actual situation is that the packet intervals are not maintained constant. That is, the packet sequence often has jitter or is output in bursts. Jitter and burst characteristics to some extent
Although it can be absorbed by a buffer provided in a router or a video decoder in the network, if the jitter becomes large or the burst output time becomes long, the buffer will be broken and eventually the video will be disturbed.

FIG. 1 shows an output example of a packet output from a general server. In this example, an example in which burstiness occurs in the packet sequence and the packet interval is not constant, that is, an example in which there is jitter is schematically shown. To solve this, a packet shaping device can be used to reduce jitter and burstiness and smooth the packet interval. Techniques for realizing this packet shaping device include a leaky packet system, a token packet system, and a method of using a combination thereof.

FIG. 2 shows an example in which a packet shaping device, which is a conventional technique, is used. Upper limit output rates and buffer capacities are set in this known packet shaping device. In this known example, the packet sequence input from the server to the packet shaping device is
It is transmitted as a packet with jitter below the set upper limit output rate.

[0006]

When the conventional packet shaping device is applied to shape the output from the video server or the video encoder, there are the following three problems. The first problem is that when the output rate from the video server or the video encoder is changed, the setting corresponding to each output rate must be performed every time the output rate is changed. For example, since the content rate that requires packet shaping during a certain time period is 6 Mbps, the upper limit rate is set to 7 Mbps, and the content rate during the next time period is 2 Mbps.
Since it is ps, it is necessary to change the setting such as setting the upper limit rate to 2.5 MbpS.

The second problem is that, with regard to the setting of the packet buffer capacity of the packet shaping device, only the upper limit capacity is set. The purpose of the conventional packet shaping device is to limit the packet rate below the rate contracted with the carrier of the leased line, for example, to prevent packet discard in the network. Therefore, the problem is how much the burstiness of the packet can be absorbed, and it is only necessary to specify the upper limit capacity of the buffer. However, in the case of content such as video, it is desirable that the transmission rate be as constant as possible. To achieve a constant transmission rate, control is performed so that buffer underflow does not occur, that is, data is transferred after it has accumulated to some extent. There is a need to. However, the conventional shaping device has not been able to sufficiently meet such a request.

A third problem is a problem relating to the buffer capacity control method mentioned in the second problem. Underflow can be prevented by increasing the capacity of the buffer, but if it is too large, the absolute delay time increases and the serviceability deteriorates. In order to shorten the absolute delay time, it is necessary to set the optimum buffer capacity, but since the optimum buffer capacity differs depending on the type of content, video server and video encoder, the transmission rate of content is changed. It is necessary to reset the buffer capacity every time.

As described above, the conventional packet shaping device has a drawback that the output rate of the packet shaping device must be changed every time the output rate from the video server or the video encoder is changed. Furthermore, there is a problem that buffer underflow occurs because there is no function of storing data in the buffer and then outputting the data. Also, if you manually set the buffer capacity,
There is a problem that the setting work of the buffer capacity is complicated because the setting must be performed again every time the type of the video server or the video encoder, the type of content, the rate, etc. are changed.

Therefore, an object of the present invention is to reduce the jitter and burst characteristics when the packet sequence output from the video server or the video encoder has jitter or burst characteristics, so that the transmission rate is almost constant. It is to realize a packet shaping method and apparatus capable of delivering video contents.

Further, another object of the present invention is to provide a packet shaping method and apparatus which do not need to reset the output rate of the packet shaping apparatus even if the output rate from the video server or the video encoder is changed. Especially.

Further, another object of the present invention is to be able to automatically set an appropriate buffer capacity, and to prevent buffer underflow while suppressing deterioration of serviceability due to increase of delay time as much as possible. A packet shaping method and apparatus are provided.

[0013]

In the packet shaping method of the present invention, the packet rate output from the packet shaping device is used as an average rate of packet intervals of packets received by the packet shaping device or an average packet rate. The packet buffer usage set in the packet shaping device is measured by measuring the jitter of the packet intervals of received packets, statistically processing the jitter, and setting the packet buffer usage based on the statistical processing result. It has made it possible to solve the problems of the prior art.

First, the packet interval of the packet sequence received by the packet shaping device is measured and stored in the packet shaping device. Next, the average packet interval is calculated from the packet intervals of the packets received within the preset time period from the current time. Next, the packet jitter is calculated using the packet interval of the packets received within the preset time period from the current time. Statistical processing is performed on the calculated jitter of each packet, and the usage amount of the packet buffer is calculated based on the statistical processing result. In this statistical processing, for example, the distribution pattern of jitter may be assumed to be a normal distribution, or a pattern that can be most approximated to a distribution pattern prepared in advance may be estimated by a test and the distribution pattern may be used. . Next, the packet output from the packet buffer is controlled and transferred so that the average packet interval calculated earlier is obtained.

Further, the time period used when calculating the average packet interval and the time period used when calculating the jitter of the packet may use preset values, Focusing on the average value of the time derivative of the packet rate calculated from the average packet interval within a certain time, if the average value exceeds a certain threshold, the time range used when calculating the average packet interval, and the packet It is also possible to perform an operation such as changing the time range used when calculating the jitter of. By performing this operation, it is possible to improve the followability when the rate of packets input to the packet shaping device is changed, rather than continuing to use the preset value. Here, the time range used when calculating the average packet interval and the time range used when calculating the jitter of the packet may be defined by the number of packets instead of time.

Here, regarding the relation between the packet interval and the packet rate, the interval of each sequential packet is defined as the packet interval, and the number of packets within a certain unit time is defined as the packet rate. The reciprocal of the packet interval gives the instantaneous packet rate at the arrival of the packet. Conversely, if the reciprocal of the packet rate is taken, the average packet interval within a certain unit time is obtained. Therefore, in the present specification, the packet interval and the packet rate are terms having the same meaning, and processing using the packet interval and processing using the packet rate have the same technical meaning. I shall.

[0017]

FIG. 3 is a block diagram of a packet shaping apparatus having a configuration common to all the embodiments described below. A packet shaping device according to the present invention comprises a packet receiving means 1 functioning as an input interface, an input buffer 2, and a packet buffer 3.
It has an output buffer 4, a packet transfer means 5, an input packet rate measuring means 6, an input packet rate storage means 7, a statistical processing means 8 and an output control means 9.

The sequential packets received by the packet receiving means 1 are stored in the input buffer 2. At this time, the input packet rate measuring means 6 measures the time intervals of the arriving sequential packets. It is also possible to measure the packet rate of the input packet. The parameters obtained by the input packet rate measuring means 6 are stored in the input packet rate storage means 7. The measured sequential packet interval and / or packet rate is the statistical processing means 8.
Are supplied to the packet buffer 3, various statistical processes are performed, and the buffer usage B of the packet buffer 3 is calculated and set. The statistical processing means 8 obtains the average packet interval α of the packet sequence from the sequential packet intervals τ. Furthermore, as described below,
The jitter of the packet sequence is also obtained from the sequential packet interval τ and the average packet interval α. Furthermore, the statistical processing means 8
Performs statistical processing on the obtained jitter and also obtains the standard deviation σ and average value M of the jitter, calculates the buffer usage B of the parameter buffer 3 using these parameters, and sets the buffer usage of the packet buffer 3. To do.
The parameters obtained by the statistical processing means 8 are supplied to the output control means 9. Therefore, the statistical processing means 8 comprises means for measuring the average packet interval or average packet rate, means for obtaining jitter, means for statistically processing jitter,
And functions as means for calculating and setting the buffer usage of the packet buffer based on the result of the statistical processing. The output control unit 9 reads the packets accumulated in the packet buffer 3 at the average packet rate or the average packet interval using the average packet rate or the average packet interval of the input packets sent from the statistical processing unit, and outputs the output buffer. Transfer to 4. The packet transferred to the output buffer 4 is transferred from the packet transfer output 5 to a network device such as a router. Therefore, even if the rate of the packet sequence sent from the video encoder or the video server is different or changed for each data, it can be automatically transferred to another network device at the received packet rate.

FIG. 4 is a diagram for explaining the packet jitter common to all the embodiments described below. In FIG. 4, reference numeral 10 indicates a packet sequence that sequentially arrives. Reference numeral 11 is a calculation range y1 for calculating the average packet interval α.
The packet average interval calculated in the range of 5 is shown. Code 1
2 indicates each packet interval τ measured by the input packet rate measuring means 6 of FIG. Reference numeral 14 indicates a calculation range x for performing statistical processing of packet jitter. Reference numeral 13 is
A reference point for statistical processing of packet jitter is shown.

A method of calculating the average packet interval α will be described. The number of packets included in the range y for calculating the average packet interval is calculated. Here, y may be a unit time or the number of packets may be set in advance. Next, it is calculated by taking the difference between the arrival time of the first packet and the arrival time of the last packet included in y and dividing it by the number of packets included in y. For example,
The average packet interval calculated upon arrival of packet i + j is: α _{i + j} = (τ _a + ... + τ _{i + j-1} ) ÷ (i + j-a)

[0021]

[Embodiment 1] FIG. 5 is a flow chart of a first embodiment of the present invention. This flowchart shows a flow when the packet i + j is received after a sufficient time has elapsed since the packet was input to the packet shaping device. In the first embodiment, statistical processing is performed assuming that the jitter distribution pattern is a normal distribution.

FIG. 6 is a diagram schematically showing a jitter distribution pattern according to the normal distribution as a reference. The packet i + j is received by the packet receiving means 1 of FIG. 3 (step Sl). Next, the received packet i + j is transferred to the input buffer 2 of FIG. At this time, the input packet rate measuring means 6 uses the sequential packet intervals τ _{i + j-1.}
To measure. This sequential packet interval τ is stored in the input packet rate storage means 7 (step, S2). Next, the sequential packet intervals τ are supplied to the statistical processing means 8. The statistical processing means 8 uses the average packet interval α shown in FIG.
The average packet interval α _{i + j} of the packets i + j from the packet a included in the range y when calculating is calculated (step S3). Next, the jitters J _{i + 1} to J _{i + j} from the packet i to the packet i + j included in the statistical processing range x shown in FIG. 4 are calculated (step S4). Here, the jitter is a value obtained by subtracting a value obtained by multiplying the average packet interval by the number of packets from the sum of a predetermined number of sequential packet intervals, and is calculated as follows. J _{i + n} = (τ _{i + 1} + τ _{i + 2} + ... τ _{i + n} ) −α _{i + j} × n
(N = 1 to j) These parameters are stored in the input parameter rate storage means 7.

Next, the statistical processing means 8 of FIG. 3 performs statistical processing on the obtained jitter. In this example, the standard deviation σ and the average value M of the jitter J _{i + n} (n = 1 to j) are calculated as statistical processing based on the assumption that the jitter is normally distributed (step S5). Next, the statistical processing means of FIG. 3 calculates the packet buffer usage B (step S6). Here, the usage amount B of the packet buffer is calculated as follows, for example. B = | M + q × σ | + | M−q × σ | where q is a constant calculated by the preset failure probability (two-sided probability) of the buffer. Next, the output control means 9 of FIG. 3 sets the packet buffer usage B for the packet buffer 3. In this way, by determining the usage of the packet buffer based on the concept that the jitter is normally distributed, it is possible to statistically guarantee that the buffer will not fail and to minimize the delay. The advantage that can be achieved is achieved.

Next, the output control device 9 uses the average packet interval α to control the output from the packet buffer 3 so that the average packet interval α _{i + j} is obtained for the packet buffer 3 in FIG. The packet is transferred to the output buffer 4 of No. 3 (step S7). The packet transferred at this time is the oldest packet among the packets stored in the packet buffer 3 of FIG. The output control unit 9 can also control the transfer from the packet buffer 3 using the average packet rate instead of the average packet interval.

Next, the packet accumulated in the output buffer 4 is transferred from the packet shaping device to the network device by the packet transfer means 5 (step S).
8).

By the above operation, it becomes possible for the packet shaping device to transfer the packet sequence having the jitter as the packet sequence in which the jitter is greatly reduced.

[Second Embodiment] FIG. 7 is a flowchart of the second embodiment. This flowchart shows the flow when the packet i + j is received after a lapse of time from the input of the packet to the packet shaping device. In the second embodiment, the jitter distribution pattern is considered to be a normal distribution.

FIG. 8 shows an example in which the packet rate input to the packet shaping device changes abruptly, and also shows an example of the amount of change in the packet rate, which is the time derivative of the change in the packet rate.

In the packet receiving means 1, the packet i
+ J is received (step Sl0). Next, the input packet i + j is transferred to the input buffer 2. At this time, the input packet rate measuring means 6 calculates the packet interval τ _{i + j− 1} of the successive packets, and the input packet rate storing means 7 stores the packet rate (step Sll). Next, the statistical processing means 8 of FIG. 3 calculates the average packet interval α _{i + j} of the packet i + j from the packet a included in the range y when the average packet interval shown in FIG. 2 is calculated (step S12). next,
The statistical processing means of FIG. 1 calculates the average packet rate P and the amount of change D thereof (step S1).
3). Here, as the change amount D of the average packet rate,
The average value of the time derivative of the packet rate within a fixed time is used. Next, in the statistical processing means 8, D is compared with the preset threshold value A of the amount of change (step 1
4).

If D is smaller than A, then in the statistical processing means of FIG. 1, the jitter J from the packet i to the packet i + j included in the statistical processing range x shown in FIG.
_{From i + 1} to jitter J _{i + j} is calculated (step S1)
5). Then, the jitter J _{i + n} in the statistical processing unit of Fig. 1
The standard deviation σ and the average value M of (n = 1 to j) are calculated (step S16). Next, the statistical processing means of FIG. 1 calculates the packet buffer usage B (step S1).
7). Next, in the output control means of FIG. 1, the packet buffer usage amount B is set, and output control from the packet buffer is performed on the packet buffer of FIG. 1 so that the average packet interval α _{i + j} is obtained. The packet is transferred to the output buffer of FIG. 1 (step S18). The packet transferred at this time is the oldest packet stored in the packet buffer of FIG.
Next, the packet transfer means of FIG. 1 transfers the packet from the packet shaping device (step S1).
9).

In step S14, when D is larger than A, the range y for calculating the average packet interval shown in FIG. 2 and the statistical processing range x shown in FIG. 2 are changed (step 20). Next, the statistical processing means of FIG. 1 calculates the average packet interval α _{i + j} of the packet i + j from the packet a ′ included in y set in step 20 (step S21). Next, the statistical processing means of FIG. 1 calculates the average packet rate P and the amount of change D thereof (step S22). Next, in the statistical processing means of FIG. 1, D is compared with a preset threshold value A of the amount of change (step 23). When D is larger than A, the process returns to step 20 and y and x are changed again. This operation is repeated until D becomes smaller than A.

After D becomes smaller than A, step 15
Move on to. The subsequent operation is the same as the processing when D is smaller than A in step 14. By the above operation, the packet shaping device can transfer the packet interval in which the jitter is added as the packet interval in which the jitter is reduced. Further, when the rate of the packet received by the packet shaping device is changed, by changing the values of y and x, it is possible to realize even faster followability.

[Third Embodiment] FIG. 7 is a flow chart of the third embodiment. This flowchart shows the flow when the packet i + j is received after a lapse of time from the input of the packet to the packet shaping device. In the third embodiment, a plurality of jitter distribution patterns are prepared in advance in the packet shaping device.

The packet i + j is received by the packet receiving means 1 of FIG. 3 (step S30). Next, the input packet i + j is transferred to the input buffer 2. Next, the input packet rate measuring means of FIG. 1 calculates the packet interval τ _{i + j−1,} and the input packet rate storing means of FIG. 1 stores the packet rate (step S31). Next, the range y when the average packet interval shown in FIG. 2 is calculated by the statistical processing means of FIG.
The average packet interval α _{i + j} of the packet i + j from the packet a included in is calculated to obtain the jitter distribution pattern (step S32). Next, the statistical processing means of FIG. 1 calculates the jitters J _{i + 1} to J _{i + j} from the packet i to the packet i + j included in the statistical processing range x shown in FIG. 2 (step S33). Next, the statistical processing means of FIG. 1 tests the jitters J _{i + 1} to J _{i + j} and the distribution pattern of the jitter prepared in advance in the packet shaping device (step S3).
4). In the operation after step S34, the statistical processing is performed using the jitter distribution pattern that can be most approximated among them. Next, the statistical processing means of FIG. 1 calculates the packet buffer usage B (step S35). Next, the output control means 9 of FIG. 3 sets the packet buffer usage B. Further, the output control means 9 uses the average packet interval α to perform output control from the packet buffer to the packet buffer 3 so that the average packet interval α _{i + j,} and the output buffer 4 in FIG. Transfer the packet. The packet transferred at this time is the oldest packet among the packets stored in the packet buffer 3 of FIG. Next, the packet transfer means 5 of FIG.
Thus, the packet is transferred from the packet shaping device. Also in this example, instead of the average packet interval, the average packet rate can be used for processing and control.

With the above operation, the packet shaping device can transfer the packet interval in which the jitter is added as the packet interval in which the jitter is reduced. In addition, by preparing a plurality of distribution patterns and performing a test, it is possible to perform control with higher accuracy than in the first embodiment.

By combining the third embodiment with the method of calculating the packet rate change amount shown in the second embodiment and changing the values of y and x according to the result,
When the rate of packets received by the packet shaping device is changed, it is possible to realize fast followability. In the above three embodiments, the calculation of the packet buffer usage amount and the calculation of the average packet interval α used for output packet control are performed for each packet input, but the same effect can be obtained by performing this for each plurality of packets. It goes without saying that you can get

The present invention is not limited to the above-described embodiments, and various changes and modifications can be made. For example, the packet shaping device according to the present invention can be installed not only as an independent device between the video encoder or the video server and the router, but also installed or mounted in the router, for example. In this case, it is possible to realize a router having both a route control function and a packet shaping function.

[0038]

By applying the present invention, it becomes possible to automatically set the output rate of the packet shaping device according to the output rate from the video server or the video encoder in the packet shaping device. Also,
It is possible to provide a function of storing data in the buffer and then outputting the data. It also becomes possible to provide a function of automatically setting an appropriate buffer capacity.

[Brief description of drawings]

FIG. 1 is a diagram for explaining packet burstiness and packet interval jitter that occur in a conventional device.

FIG. 2 is a diagram showing an example of packet shaping by a packet shaping device which is a conventional technique.

FIG. 3 is a block diagram showing a basic form of a packet shaping device according to the present invention.

FIG. 4 is a diagram illustrating each parameter obtained from the measurement result of the packet interval, for explaining the processing steps of the packet shaping method according to the present invention.

FIG. 5 is a diagram showing a flow of the packet shaping device in the first embodiment.

FIG. 6 is a diagram showing a packet jitter histogram for explaining the first embodiment.

FIG. 7 is a diagram showing a flow of a packet shaping device in the second exemplary embodiment.

FIG. 8 is a diagram illustrating an example of a packet rate change and an example of a packet rate change amount as a differential value of the packet rate for explaining a second embodiment.

FIG. 9 is a diagram showing a flow of a packet shaping device in the third exemplary embodiment.

[Explanation of symbols]

1 Packet receiving means 2 input buffer 3 packet buffer 4 output buffer 5 Packet transfer means 6 Input packet rate measuring means 7 Input packet rate storage means 8 statistical processing means 9 Output control means 10 packets 11 Average packet spacing α 12 Measured packet interval τ 13 Reference point S when calculating jitter 14 Statistical processing range x 15 Range y when calculating the average packet interval α 20 communication lines 21 server 22 packets 30 communication lines 31 servers 32 packets 33 Packet shaping device

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Satoshi Shiozaki             2-3-1, Otemachi, Chiyoda-ku, Tokyo             Inside Telegraph and Telephone Corporation F-term (reference) 5K030 GA11 HA08 HB29 KX11 LC02                       MB09 MB12 MC07

Claims (10)

[Claims] 1. A packet shaping method for shaping and transferring a received packet sequence so that the packet interval is substantially constant, and a step of measuring a sequential packet interval or packet rate of input packets, and a sequence of input packets. Calculating an average packet interval or an average packet rate of the input packet from the packet intervals or the packet rates of the input packets, and calculating a jitter of the input packet using the sequential packet interval of the input packet and the calculated average packet interval. Then, statistical processing is performed on the calculated jitter, the required buffer usage of the packet buffer that temporarily stores the input packet is calculated based on the statistical processing result, and the usage of the packet buffer is calculated using the calculated buffer. The process of setting the usage and the calculated packet Using the average interval or average packet rate, packet shaping method characterized by a step of controlling the output of packets accumulated in the packet buffer. 2. A packet shaping method which shapes and transfers a received packet sequence so that the packet intervals are substantially constant, and a step of measuring a sequential packet interval or a packet rate of input packets, and a step of sequentially input packets. Calculating the average packet rate of the input packet from the packet interval or packet rate of the packet, calculating the packet rate change amount from the calculated average packet rate, and averaging the packet rate over a fixed time of the time differential network. The step of comparing the value with a preset threshold, and the range of the packet sequence when calculating the preset average packet interval when the average value of the time derivative of the packet rate within a certain time exceeds the threshold and Packets when calculating the received packet interval jitter The step of updating the range of, the step of calculating the jitter of the received packet by using the sequential packet interval of the input packet and the average packet interval, and the statistical processing of the calculated jitter. Calculating the required buffer usage of the packet buffer for temporarily accumulating the input packet based on the setting, and setting the buffer usage of the packet buffer to the calculated buffer usage; and the average packet of the calculated packets. Transferring packets stored in the buffer using a rate or an average packet interval. 3. The packet shaping method according to claim 1, wherein in the statistical processing step, statistical processing is performed with a distribution pattern of jitter as a normal distribution. 4. The method according to claim 1 or 2, wherein in the statistical processing step, a distribution pattern obtained by comparing the obtained jitter distribution pattern with a jitter distribution pattern prepared in advance is used. A packet shaping method, characterized in that the statistical processing is performed. 5. The method according to any one of claims 1 to 4, wherein a preset buffer overflows when calculating a necessary buffer usage of a packet buffer based on the statistical processing result. A packet shaping method characterized by calculating a buffer usage so as to be smaller than a probability of underflow. 6. A packet shaping device that shapes and transfers a received packet sequence so that packet intervals are substantially constant, and means for receiving the packet sequence and a packet for temporarily accumulating the received packet sequence. A buffer, a packet transfer means for transferring the packets output from the packet buffer, a means for measuring the sequential packet interval τ _i or the packet rate of the received packet sequence, and a reception from the measured sequential packet interval τ _i Means for calculating the average packet interval α _i or average packet rate of the packet sequence, and means for calculating the jitter of the received packet sequence using the sequential packet interval τ _i and the average packet interval α _i , Statistical processing is performed on the obtained jitter, and based on the result of the statistical processing, the packet buffer Processing means for calculating the usage amount of the packet buffer, means for setting the buffer usage amount of the packet buffer to the calculated usage amount, and using the calculated average packet interval or average packet rate to the packet buffer. A packet shaping device, comprising: output control means for controlling output of accumulated packets. 7. The packet shaping device according to claim 6, wherein an input buffer is arranged between the packet receiving means and the packet buffer, and an output buffer is arranged between the packet buffer and the packet transfer device. A packet shaping device. 8. The packet shaping device according to claim 6 or 7, wherein the statistical processing means calculates a jitter average value M and a standard deviation σ from the calculated jitter.
And the average value M and standard deviation σ of these jitters and a constant q calculated from the preset failure probability of the buffer, the usage amount B of the packet buffer is calculated by the formula B = | M + q × σ | + A packet shaping device characterized in that it is calculated based on | M−q × σ |. 9. The packet shaping device according to claim 6 or 7, further comprising a storage unit that stores various distribution patterns of jitter in advance, and the statistical processing unit has a jitter distribution obtained by calculation. A packet shaping device characterized by comparing a pattern and a jitter distribution pattern stored in the storage means, selecting a closest jitter distribution pattern, and statistically processing using the selected jitter distribution pattern. 10. A packet shaping device that shapes and transfers a received packet sequence so that packet intervals are substantially constant, and means for receiving the packet sequence and a packet for temporarily accumulating the received packet sequence. A buffer, a packet transfer means for transferring the packet output from the packet buffer, a means for measuring the packet rate of the received packet string, a means for calculating an average packet rate from the measured packet rate, and a measured Means for performing a time derivative process on the packet rate and calculating an average value of the time derivative value of the packet rate within a fixed time, and comparing the average value of the time derivative value of the packet rate within a constant time with a preset threshold value And the average value of the time derivative of the packet rate within a certain time. Based on the result of comparison with the threshold set for the above, the means for updating the range of the packet sequence when calculating the preset average packet interval and the range of the packet sequence when calculating the jitter, Means for calculating jitter, means for performing statistical processing on the calculated jitter, and calculating the buffer usage of the packet buffer based on the statistical processing result, and setting the calculated buffer usage for the packet buffer Means for controlling the output of the packets accumulated in the packet buffer by using the calculated average packet rate.