JP3694451B2 - Gateway and data transfer method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a gateway that connects a network having delay jitter in data transmission to another network, and a data transfer method therefor.
[0002]
[Prior art]
A gateway is an indispensable member for performing multimedia communication between terminals connected to different networks having different protocols and network characteristics. This gateway receives data according to the protocol of the network to which the transmission source terminal belongs, and transmits this data according to the protocol of the network to which the transmission destination terminal belongs. As a result, data transmitted from the transmission source terminal passes through the network to which the transmission source terminal belongs, the gateway, and the network to which the transmission destination terminal belongs.
[0003]
For multimedia, in packet switched networks such as the Internet, ITU-T Recommendation H.264 is recommended. A terminal conforming to H.323 is used, and in a circuit switching network such as a PSTN (public switched telephone network), ITU-T Recommendation H.323. A terminal conforming to H.324 is used. Recommendation H. H.323 and H.H. In 324, the communication protocol used is different (RTP / UDP in Recommendation H.323 and Multiple Transmission Protocol specified in Recommendation H.223 in Recommendation H.324). become. Furthermore, to achieve high quality multimedia communications, it is necessary to compensate for differences in network characteristics such as delay, delay jitter and error characteristics.
[0004]
[Problems to be solved by the invention]
In the network, a delay related to data transmission is caused by various factors, and the delay of each data unit may be different. Such fluctuation in transmission delay is called delay jitter. Factors that cause delay jitter include, for example, storage switching in a packet switching network, waiting at nodes on a path, or the fact that each data unit arrives at a gateway via a different path.
[0005]
Even in a circuit switched network, delay jitter is caused by simultaneously transmitting different types of data. For example, a multimedia terminal such as a videophone multiplexes image, audio and / or application data in one bit stream. In such a case, some data is transmitted immediately, and some data is not transmitted until transmission of currently transmitted data is completed. This type of delay jitter occurs at both the terminal and the gateway, and increases as the transmission rate decreases.
[0006]
When delay jitter occurs, the time interval between data units at the source terminal is not maintained at the destination terminal. For example, it is desirable to be able to eliminate such delay jitter when transmitting highly accurate data such as voice.
In order to eliminate delay jitter, some communication terminals absorb some delay jitter. For example, a certain communication terminal temporarily stores data units received at various delay times, and continuously outputs them after a series of data units is accumulated. However, since the cause and characteristics of delay jitter differ depending on the network, it is difficult to know in advance how large the delay jitter of data received by a communication terminal that can be a transmission destination is. Therefore, the delay jitter allowed in the destination terminal is often based on the maximum delay jitter that can occur in the network to which the terminal belongs. This increases end-to-end delay, which can be a serious problem in real-time communications.
[0007]
The present invention provides a gateway and a data transfer method capable of minimizing an increase in delay by reducing jitter.
[0008]
[Means for Solving the Problems]
According to one aspect of the invention, a gateway connecting two networks is
A receiving unit for receiving a plurality of data units of at least one form from a first network;
Temporarily storing the data unit received by the receiving unit; Determine a certain time based on the delay jitter reduction capability of the destination terminal, and output the first data unit after the determined time has elapsed from the time when the first data unit was received in the receiver, Subsequent data units are sequentially output in the same cycle as the transmission cycle in the first network. A control unit;
A transmission unit that transmits the data unit corresponding to the data unit output by the control unit to the second network in order to deliver the data unit to the transmission destination terminal.
In the present invention, first and second networks are connected to the gateway. The destination terminal may be served directly by the second network. Alternatively, the destination terminal may be served indirectly by the second network. That is, one or a plurality of networks and another gateway that connects these adjacent networks may be interposed between the transmission destination terminal and the second network.
Preferably, the gateway is the destination terminal. delay A jitter reduction capability acquisition unit that acquires information on the jitter reduction capability may be further provided.
[0010]
Preferably, the receiving unit can receive data units of different types, and when the data units of different types are not transmitted at the same time, the control unit You may determine the priority in the said transmission part according to the delay time which arose when transmitting. In this case, the transmission unit may multiplex data units of different formats output from the control unit according to the priority and transmit the multiplexed data units to the destination terminal through the second network.
[0011]
Preferably, the control unit detects an initial output time when the first data unit in each format is output from the control unit, and the first data unit in each format based on the first output time And a delay time calculation unit for calculating the delay time of each data unit. In this case, the control unit may control the transmission unit so that the data unit determined to have a long delay time is transmitted with priority over other data units.
[0012]
The control unit obtains the time by reducing a maximum delay time that can occur when the data unit is transmitted in the first network according to a delay jitter allowed in the transmission destination terminal. Also good.
[0013]
According to another aspect of the invention, a data transfer method includes receiving a plurality of data units of at least one format from a first network, temporarily storing the data units, Determining a certain time based on the jitter reduction capability of the destination terminal acquired in advance, outputting the leading data unit after the determined time has elapsed since the time when the leading data unit was received, Sequentially output subsequent data units in the same cycle as the transmission cycle in the first network; Transmitting a data unit corresponding to the data unit to a second network for delivery to the destination terminal.
[0014]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment according to the present invention will be described with reference to the accompanying drawings.
As shown in FIG. 1, in the embodiment according to the present invention, the gateway 10 interconnects the network 20 and the network 30, so that data is served to the network 30 from the source terminal serving the network 20. In contrast, data is transmitted from a transmission source terminal serving the network 30 to a transmission destination terminal serving the network 20. Each of the networks 20 and 30 serves a large number of terminals. In FIG. 1, for convenience, only the terminal 21 serving the network 20 and the terminal 31 serving the network 30 are illustrated. In both networks 10 and 20, at least one type of data unit is exchanged.
[0015]
In this embodiment, the network 20 is a packet switching network in which data units are included in packets and transmitted and received, and the network 30 is a circuit switched network in which data units are transmitted and received as frames. The frame includes, for example, an audio data frame and an image data frame. However, the present invention is not intended to be limited to this, and changes and modifications are possible. The network 20 or 30 may be a mobile network or a fixed network.
[0016]
The gateway 10 includes a network interface 12 for the network 20, a network interface 13 for the network 30, a transmission sequence AB for transmitting data from the network 20 to the network 30, and a transmission for transmitting data from the network 30 to the network 20. A series BA is provided.
The transmission sequence AB includes a reception unit 14A, a protocol entity 15A, a media transform coding unit 16, a protocol entity 17B, a transmission unit 18B, and a control unit 100. The transmission sequence BA includes a reception unit 14B, a protocol An entity 15B, a media transform coding unit 16, a protocol entity 17A, a transmission unit 18A, and a control unit 100 are provided.
[0017]
The receiving unit 14A receives a bit stream from the network 20 via the network interface 12. This bit stream may include a plurality of data units (packets) of at least one format created by a transmission source terminal (for example, terminal 21) serving the network 20. On the other hand, the receiving unit 14 </ b> B receives a bit stream from the network 30 via the network interface 13. This bit stream may include a plurality of data units (frames) of at least one format created by a transmission source terminal (for example, terminal 31) serving the network 30.
[0018]
The protocol entity 15A is a termination for the protocol used in the network 20, and extracts payload information and related information from each packet. If RTP / UDP is used as a communication protocol, the related information includes a data type, a time stamp (transmission time), a sequence number, and the like. Similarly, protocol entity 15B is a termination for the protocol used in network 30.
[0019]
The media conversion encoding unit 16 performs code conversion of voice and images as necessary. For example, the code conversion is performed when the audio or image code systems used in the network 20 and the network 30 are different. .
[0020]
The control unit 100 includes a buffer 110, which buffers a data unit input from the protocol entity 15A or 15B. The control unit 100 reads the data unit from the buffer 110 according to the jitter reduction capability (specifically, allowable delay jitter) provided in the transmission destination terminal (for example, the terminal 31) acquired in advance, and corresponds to the transmission destination. The data unit is output to a protocol entity (eg, 17B) for transmission to a network (eg, network 30).
[0021]
Regarding the jitter reduction capability of the transmission destination terminal, for example, ITU-T recommendation H.264. It can be acquired using the ability exchange means described in H.245. ITU-T recommendation H.245 defines various control protocols used in multimedia communication. 324 and H.264. The multimedia terminal conforming to H.323 and other recommendations conforms to this. ITU-T recommendation The control protocol defined in H.245 includes a procedure for exchanging terminal capabilities, and by using this procedure, the communication capability between nodes performing communication can be exchanged. This capability exchange is the ITU-T recommendation H.264. This is realized by transmitting various control messages defined in H.245 to each other. In this embodiment, the transmission sequence AB and the transmission sequence BA jointly acquire the jitter reduction capability information of the transmission destination terminal, and the corresponding control unit 100 stores the information.
[0022]
When the gateway 10 receives data in a plurality of formats related to multimedia such as voice and images from the network (for example, the network 20), the buffer 110 stores data units in one format in the other format as described later. Store in a location independent of the format data unit. Further, the control unit 100 reads out one type of data unit from the buffer, distinguishing it from other types of data units, outputs it to a protocol entity (eg, 17B), and transmits it to the network (eg, 30). .
[0023]
Each control unit 100 further includes a priority determination unit 120. As will be described in detail later, the priority determination unit 120 is effectively used in a plurality of types of data communication related to multimedia.
[0024]
Each of the protocol entities 17A and 17B generates a packet or a frame conforming to the protocol of the network connected to the destination terminal from the data unit read from the buffer 110. Each of the transmission units 18A or 18B transmits the packet or frame to a network serving the transmission destination terminal.
[0025]
Under the above configuration, a data unit to be transmitted from a source terminal serving the network 20 to a destination terminal serving the network 30 arrives at the gateway 10 via the network 20. The gateway 10 transmits the data unit to the network 30 while reducing the delay jitter generated in the network 20. Similarly, the delay jitter generated in the network 30 is reduced for the data unit to be transmitted from the transmission source terminal serving the network 30 to the transmission destination terminal serving the network 20.
[0026]
Hereinafter, it is assumed that the transmission source terminal is the terminal 21 and the transmission destination terminal is the terminal 31. Each packet transmitted from the terminal 21 is received by the gateway 10 with a different delay time by passing through the network 20. The gateway 10 reduces such delay jitter in the control unit 100 and then transmits it according to the protocol that the network 30 follows.
[0027]
With reference to FIG. 2, the buffering of the received single-format data unit in the control unit 100 will be described.
In this embodiment, the buffer 110 includes a predetermined number (9 in FIG. 2) of buffer registers # 1 to # 9. Each buffer register of the buffer 110 has a capacity substantially corresponding to the length of one packet, and stores the contents in one packet. The data in the buffer register is sequentially read out and output to the control unit 100 at regular time intervals. Let INT be the length of the time interval. The time tn (t1, t2, t3...) Shown in FIG. 2 is separated every time interval INT.
[0028]
In the buffer 110 shown in FIG. 2, the data stored in the buffer register # 9 is output at time t1. At time t2, the data stored in the buffer register # 1 is output. At time t3, the data stored in buffer register # 2 is output. At time t4, the data stored in buffer register # 3 is output. Thus, the data units stored in the adjacent buffer registers are sequentially output at a constant time interval. However, nothing is output if there is no data stored in the buffer register.
[0029]
In the buffer 110, as shown in FIG. 2, it is assumed that the data unit of the packet P1 is stored in the buffer register # 2 at time t1. According to the above method, the data unit of the packet P1 stored in the buffer register # 2 is output at time t3. Therefore, the data unit of the packet P1 is output after time 2 × INT after being stored in the buffer 110.
[0030]
Unlike FIG. 2, it is assumed that the data unit of packet P1 is stored in buffer register # 1 at time t1. The data unit of the packet P1 stored in the buffer register # 1 is output at time t2 after time 1 × INT after being stored in the buffer 110 according to the above-described method.
[0031]
As can be understood from the above, in the buffer 110, by storing the data unit of the packet P1 in the buffer register # 2, the output time is equivalent to one buffer register compared to storing in the buffer register # 1. Will be delayed by INT.
[0032]
In the present embodiment, after the control unit 100 determines the buffer register to store the data unit of the first packet in one type of packet received from the network 20, the data unit of the first packet is stored. The delay time generated in the network 20 is absorbed by adjusting the time until output. More specifically, the control unit 100 sets a buffer register in which the time from when the data unit of the first packet is stored to when it is output is equal to the width of the delay jitter to be absorbed. The initial writing position is selected (as shown by steps S1 and S2 in FIG. 2).
[0033]
The control unit 100 stores the data units of the second and subsequent packets in the buffer register to be read immediately after the buffer register at that time (as shown in step S3 in FIG. 2). Furthermore, the control unit 100 refers to the header of the packet when the reception order of the packets in the reception unit 14A does not match the transmission order at the transmission source terminal due to delay jitter, and after aligning the order, Store in buffer 110.
[0034]
Returning to FIG. 2, the data unit of the first packet P1 received at time t1 is stored in the buffer register # 2 which is the initial write position. Assume that no packet is received at time t2, and that the second packet P2 is received at time t3. If the packets P1 and P2 are continuously transmitted from the transmission source, the delay of the packet P1 and the delay of the packet P2 are considered to be different by the time INT because of the network 20. At the time t3, the data unit of the packet P2 is stored in the buffer register # 3 to be read out immediately after the buffer register at that time in accordance with the method described above. At the same time, the data unit of the packet P1 stored in the buffer register # 2 is read from the buffer 110 at time t3. At time t4 after time INT from time t3, the data unit of the packet P2 stored in the buffer register # 3 is read from the buffer 110.
[0035]
As described above, even when the packet P2 arrives at the gateway 10 after being delayed by the time INT longer than the delay of the packet P1, the output timing of the data unit of the leading packet P1 is delayed by the time INT. Therefore, the data units of the packets P1 and P2 are output at continuous timing, and the delay jitter is absorbed. As described above, the order of the packets is aligned by the control unit 100. Therefore, if the transmission time relationships of the packets P1 and P2 are aligned, the delay jitter is similarly absorbed in the subsequent data units.
[0036]
The above buffering is based on the assumption that the allowable delay jitter is zero in the network 30 to which the transmission destination belongs. However, let us assume that the destination terminal 31 can absorb delay jitter corresponding to the maximum delay jitter assumed to occur in the network 30. In this case, the gateway 10 may not need to absorb all of the delay jitter generated in the network 20. However, if the delay jitter generated in the other network 20 is larger than that of the network 30, the terminal 31 may not be able to absorb all of the delay jitter generated in the other network 20. Therefore, the controller 100 of the gateway 10 determines how much of the actual delay jitter is absorbed according to the allowable delay jitter that is the jitter reduction capability of the terminal 31 acquired in advance. The operation of the gateway 10 in this case will be described in detail later.
[0037]
In FIG. 2, the delay jitter in the network 20 is 1 × INT, but in an actual network, the delay jitter may be larger. For example, FIG. 9 shows a graph showing the occurrence of delay, the upper graph shows the occurrence of delay at the time of transmission at the transmission source terminal 21, and the lower graph shows the packet delay at the time of reception at the gateway 10 after passing through the network 20. Indicates occurrence. As shown by the straight line D0 in the upper graph of FIG. 9, at the time of transmission from the terminal 21, of course, all the packets have no delay, and there is no delay jitter at this stage. However, by passing through the network 20, each packet arrives at the gateway 10 with various delay times. As indicated by a curve D1 in the lower graph of FIG. 9, maximum delay time = 4 × INT, minimum delay time = 0, delay jitter = maximum delay time = 4 × INT.
[0038]
In order to absorb all the 4 × INT delay jitter generated in the network 20 as shown in FIG. 9 by the buffer 110, the buffer register # 5 becomes the initial writing position as shown in FIG.
[0039]
However, in the present embodiment, as described above, the delay jitter width (buffer absorption width) to be absorbed in the buffer 110 is the delay jitter generated in the network 20 and the delay jitter allowed in the network 30 acquired in advance. Based on the above, the control unit 100 determines. If the allowable delay jitter in the network 30 is 3 × INT, the buffer 110 may absorb 1 × INT worth of delay jitter. As shown in FIG. 4, the initial write position in this case may be buffer register # 2.
[0040]
Hereinafter, an operation of absorbing a part of the delay jitter generated in the network 20 by the buffer 110 will be described in comparison with an operation of absorbing all of the delay jitter.
As shown in FIG. 3, the first packet P1 transmitted from the terminal 21 is received by the gateway 10 with the minimum delay time (0), and the next packet P2 is received by the gateway 10 with the maximum delay time (4 × INT). Assume that. That is, as shown in FIG. 3, the first packet P1 is transmitted from the terminal 21 at time t1, and the second packet P2 is transmitted from the terminal 21 at time t2. The first packet P1 passes through the network 20 with the minimum delay time 0s and arrives at the gateway 10 at the time t1, and the second packet P2 passes through the network 20 with the maximum delay time 4s and reaches the gateway 10 at the time t6. Has arrived.
[0041]
As described above, as shown in FIG. 4, when the allowable delay jitter is 3INT in the network 30 to which the transmission destination belongs, the data unit of the packet P1 is stored in the buffer register # 2 at time t1. Therefore, the data unit of the packet P1 is output from the buffer 110 at time t3 2 × INT after t1.
[0042]
On the other hand, in the comparative example in which all of the delay jitter generated in the network 20 is absorbed by the buffer 110, as described above, the data unit of the packet P1 should be stored in the buffer register # 5 as shown in FIG. . Therefore, in the comparative example, the data unit of the packet P1 will be output from the buffer 110 at time t6 after 5 × INT from t1.
[0043]
As shown in FIGS. 4 and 5, the data unit of the second packet P2 arriving at the gateway 10 at the time t6 delayed by the maximum delay time 4s is sent to the control unit 100 immediately after storage in the open buffer register. It is stored in the buffer register # 6 to be read and output at time t7.
[0044]
When the allowable delay jitter is 3INT in the network 30 to which the transmission destination belongs, the data unit of the packet P2 is output 4INT after the data unit of the packet P1 is output as shown in FIG. 1INT which is a part of the delay jitter 4INT generated in step 1 is absorbed in the buffer 110, and the delay jitter of 3INT remains. The residual delay jitter is absorbed by the destination terminal 31. In this case, as shown in FIG. 3, the delay time from the time when the packet P1 is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110 = 2INT, and the time when the packet P2 is transmitted from the terminal 21 Delay time from when the data unit is output from the buffer 110 to 5 INT = 5INT.
[0045]
In contrast, in the comparative example in which all of the delay jitter generated in the network 20 is absorbed by the buffer 110, the data unit of the packet P2 is output after 1 INT from the output of the data unit of the packet P1, as shown in FIG. Therefore, the delay jitter 4INT generated in the network 20 is all absorbed in the buffer 110. In this case, as shown in FIG. 3, the delay time from the time when the packet P1 is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110 = 5INT, and the time when the packet P2 is transmitted from the terminal 21 Delay time from when the data unit is output from the buffer 110 to 5 INT.
[0046]
Next, the buffering operation of the gateway 10 when the first packet arrives with the maximum delay time will be described with reference to FIGS.
As shown in FIG. 6, it is assumed that the leading packet P1 is transmitted from the terminal 21 at time t1, and the second packet P2 is transmitted from the terminal 21 at time t2. The first packet P1 arrives at the gateway 10 at time t5 with the maximum delay time 4INT that can occur in the network 20, and the second packet P2 arrives at the gateway 10 at time t6 with the maximum delay time 4INT. Even if the arrival order is disturbed, the control unit 100 arranges the order again, so that the subsequent packet is not processed before the previous packet.
[0047]
When the allowable delay jitter is 3INT in the network 30 to which the transmission destination belongs, as shown in FIG. 7, the data unit of the packet P1 is stored in the buffer register # 2 at time t5, and at time t7 2INT after t5. Output from the buffer 110. Then, the data unit of the second packet P2 that arrives at the gateway 10 at the time t6 delayed by the maximum delay time 4INT from the transmission time t2 is output to the buffer register # 3 that is output most quickly among the buffer registers at that time. Stored and output at time t8 after 2INT. As shown in FIG. 6, for both packets P1 and P2, the delay time from the time when the data is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110 is 6INT.
[0048]
On the other hand, in the comparative example in which all of the delay jitter generated in the network 20 is absorbed by the buffer 110, as shown in FIG. 8, the data unit of the packet P1 is stored in the buffer register # 5 at the time t5, and the packet P1 Are output from the buffer 110 at time t10, which is 5 INT after t5. The data unit of the second packet P2 that arrives at the gateway 10 at the time t6 delayed by the maximum delay time 4INT from the transmission time t2 is stored in the buffer register # 6 that is output most quickly among the buffer registers at that time. It is output at time t11 after 5 INT. As shown in FIG. 6, for both packets P1 and P2, the delay time from the time transmitted from the terminal 21 to the time when the data unit is output from the buffer 110 is 9INT.
[0049]
As can be understood from the above, using the buffer 110 provided in the gateway 10 connecting the network, the data unit of the first packet is output after a certain time from the reception, and the subsequent data units are sequentially output. By doing so, all or part of the delay jitter generated in the network 20 can be absorbed.
[0050]
As described above, when the buffer 110 absorbs all of the delay jitter occurring in the network 20, even when the leading packet arrives with the minimum delay, the data unit is changed from the time when each packet is transmitted from the terminal 21. The delay time until the time output from the buffer 110 = 5INT (see FIG. 3). When the first packet has the maximum delay, the delay time is 9 INT (see FIG. 6). The total delay time between the terminals 21 to 31 is the total of the delay time from the time when the packet P1 is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110 and the maximum delay time generated in the network 30. is there. Accordingly, the total delay time increases due to an increase in the delay time from the time when the packet P1 is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110.
[0051]
On the other hand, when a part of the delay jitter is absorbed, when the leading packet has the minimum delay, the time from the time when the packet P1 is transmitted from the terminal 21 to the time when the data unit is output from the buffer 110. When the delay time = 2INT and the leading packet has the maximum delay, the delay time = 6INT (see FIGS. 3 and 6). Thus, by adjusting the delay to be absorbed by the buffer 110 based on the delay jitter allowed in the network 30, the total delay time between the terminals 21 to 31 can be further reduced. As described with reference to FIGS. 3, 4, and 7, when a part of the delay jitter is absorbed, the delay jitter generated in the network 20 partially remains. However, this residual delay jitter is absorbed by the destination terminal 31.
[0052]
Next, the application of this embodiment to a plurality of types of data communication related to multimedia will be described.
In the multimedia communication, the protocol entity 15A or 15B (see FIG. 1) transmits a data unit of one format among the data of a plurality of formats received by the corresponding receiving unit 14A or 14B and data of another format. Distinguish from units.
[0053]
Moreover, the priority determination part 120 of each controller 100 determines the priority of a data unit. Specifically, the priority determination unit 120 determines the order of transmitting data units to the network 30 according to the delay jitter of each data unit.
[0054]
The corresponding protocol entity 17A or 17B maps the output signal from the buffer 110 to a channel for multiplexing transmission based on the priority information output from the priority determination unit 120. The corresponding transmitter 18A or 18B multiplexes the data unit to the network 30 according to the protocol entity mapping.
[0055]
FIG. 10 shows the internal functional entities of each control unit 100 in detail. In the following description, as shown in FIG. 10, the protocol entity 15 </ b> A or 15 </ b> B is referred to as the protocol entity 15, and the protocol entity 17 </ b> A or 17 </ b> B is referred to as the protocol entity 17.
[0056]
As shown in FIG. 10, the buffer 110 includes an audio buffer 111 and an image buffer 112.
[0057]
The audio buffer 111 stores data units of audio data packets output from the protocol entity 15, and the image buffer 112 stores data units of image data packets output from the protocol entity 15. Data units read from each buffer are supplied to the protocol entity 17 separately.
[0058]
In addition, the data unit output from each of the buffers 111 or 112 is also supplied to the priority determination unit 120. The priority determination unit 120 includes an output time detection unit 122, a delay time calculation unit 123, and a determination unit 124.
The output time detection unit 122 detects when the first audio data unit is output from the buffer 110 and when the first image data unit is output from the buffer 110 based on the internal clock of the gateway 10, and outputs this to the delay time calculation unit 123. . The output time of the first audio data unit output from the audio buffer 111 is referred to as Bs (A), and the output time of the first image data unit output from the image buffer 111 is referred to as Bs (V).
[0059]
The delay time calculation unit 123 calculates the standard output time Ss (A) of the audio data unit that has not been transmitted from the protocol entity 17. The standard output time Ss (A) is the time when the audio data unit should be output from the buffer 110 if there is no delay, and the audio buffer output time Bs (A) of the first audio data unit is the starting point. . Further, the time difference between the current time Ct and the standard output time Ss (A) of the voice data unit that has not been transmitted from the protocol entity 17 is calculated as the voice delay time D (A) and output to the determination unit 124. In addition, the delay time calculation unit 123 calculates the standard output time Ss (V) of the image data unit that has not been transmitted from the protocol entity 17. The standard output time Ss (V) is the time when the image data unit should be output from the buffer 110 if there is no delay, and the image buffer output time Bs (V) of the first image data unit is the starting point. . Further, the time difference between the current time Ct and the standard output time Ss (V) of the image data unit that has not been transmitted from the protocol entity 17 is calculated as the image delay time D (V) and output to the determination unit 124. The audio delay time D (A) and the image delay time D (V) are sufficient indexes for comparing the delay time of each data unit with respect to the first audio or image data unit.
[0060]
For example, when an audio data unit and an image data unit are simultaneously read from the audio buffer 111 and the image buffer 112, an audio data unit and an image data unit that have been output from the buffer 110 but have not been transmitted from the protocol entity 17 are simultaneously displayed. If present, the determination unit 124 compares the audio delay time D (A) and the image delay time D (V), and based on the comparison, transmission control for controlling the priority of transmission in the protocol entity 17 Create a command and output it to the protocol entity 17.
[0061]
The priority determination process in the priority determination unit 120 will be described with reference to the flowchart shown in FIG. This illustrated process is repeated every 1 INT.
First, in step S11, when there is an audio data unit and an image data unit that are output from the buffer 110 but not transmitted from the protocol entity 17, the delay time calculation unit 123 outputs the audio buffer of the first audio data unit. Based on the time Bs (A), the standard output time Ss (A) of the voice data unit not yet transmitted from the protocol entity 17 is calculated according to the equation (1).
Ss (A) = Bs (A) + (N (A)-1) x Cy (A) (1)
Here, N (A) is a number indicating the order of the voice data units, and Cy (A) indicates a voice packet transmission cycle by the terminal 21.
[0062]
Also, in step S11, the delay time calculation unit 123, based on the image buffer output time Bs (V) of the first image data unit, the standard output time Ss (V) of the audio data unit that has not been transmitted from the protocol entity 17. Is calculated according to equation (2).
Ss (V) = Bs (V) + (N (V)-1) x Cy (V) (2)
Here, N (V) is a number indicating the order of image data units, and Cy (V) indicates a transmission cycle of image packets by the terminal 21.
[0063]
In step S12, the delay time calculation unit 123 calculates the delay time D (V) of the data unit according to the equation (3) based on the standard output time Ss (V) of the image data unit and the current time Ct.
D (V) = Ct-Ss (V) (3)
[0064]
In step S13, the delay time calculation unit 123 calculates the delay time D (A) of the data unit according to the equation (4) based on the standard output time Ss (V) of the audio data unit and the current time Ct.
D (A) = Ct − Ss (A) (4)
[0065]
Next, the determination unit 124 determines whether or not the audio delay time D (A) is larger than the image delay time D (V) in step S14. If the determination in step S14 is affirmative, in step S15, the determination unit 124 supplies the determination unit 17 with a transmission control command instructing to transmit the audio data unit with priority. On the other hand, if the determination in step S14 is negative, in step S16, the determination unit 124 supplies the determination unit 17 with a transmission control command instructing to transmit the image data unit with priority. In this way, the process ends and is then repeated.
[0066]
Next, packet processing in multimedia communication will be specifically described with reference to the time charts shown in FIGS. 12 and 13. In FIGS. 12 and 13, characters A and V indicate data units corresponding to voice packets and image packets, respectively, and suffixes indicate the transmission order from the terminal 21.
[0067]
First, it is assumed that the transmission mode of each packet from the terminal 21 and the output mode of each packet from the buffer 110 are as shown in FIGS. That is, the data unit relating to each type of packet from the terminal 21 is buffered by the gateway 10 in such a manner that only a part of the delay jitter generated in the network 20 is absorbed by the buffer 111 or 112 as described above. 110. For this reason, as shown in FIGS. 12 and 13, although the transmission cycle of the audio packet A is 3INT, the data unit of the packet A2 is output from the buffer after the packets A1 to 4INT, and the transmission cycle of the image packet V is Even though 5INT, the data unit of packet V2 is output from the buffer after packets V1 to 6INT.
[0068]
In the audio buffer 111 and the image buffer 112, when all the delay jitters of audio and image data units are absorbed, all the audio data units and image data units are the same cycle as the transmission cycle from the transmission source terminal 21 to the network 20. Should be output. Accordingly, since only a part of the delay jitter is absorbed by the buffer, 1 INT delay jitter remains in the second and subsequent audio data groups, and 1 INT delay jitter remains in the second and subsequent image data groups. It can be understood from FIG. 12 and FIG.
[0069]
In addition, because the protocol entity 17 multiplexes the voice and image data units into a single channel and transmits them to the network 30, this multiplexing also causes a delay. Therefore, the delay jitter due to multiplexing may be superimposed on the residual delay jitter at the time of output from the buffer 110, and the delay jitter of the output data unit from the protocol entity 17 may spread.
[0070]
Delay jitter in multiplexing by the protocol entity 17 is greatly affected by the transmission order of data units belonging to different types. More specifically, as shown in FIGS. 12 and 13, voice packets are transmitted from the terminal 21 at the same data length and at equal intervals 3INT, whereas image packets have different lengths. Moreover, their length is larger than the length of the voice packet, and they are transmitted at an equal interval of 5 INT. In such a case, the gateway 10 cannot transmit the audio data unit even when it is preferable to output the audio data unit during transmission of the long image data unit. Therefore, if no contrivance is added to the multiplexing method, the audio data unit may be delayed due to multiplexing. Then, the delay jitter may further spread due to the delay jitter of the remaining audio data unit and the delay at the time of multiplexing.
[0071]
Various multiplex systems can be considered. For example, when audio data units output from the buffer 110 but not transmitted from the protocol entity 17 and image data units exist at the same time, the image data unit may always be given priority over the audio data unit during multiplexing. Can be. Conversely, in such a case, it may be considered that the audio data unit is always prioritized over the image data unit during multiplexing. These methods have drawbacks as described below.
[0072]
In FIG. 12, at time t13, the data unit of the audio packet A4 is output from the audio buffer 111, and the data unit of the image packet V3 is output from the image buffer 112. At this time, the data unit of the previous voice packet A3 is already output from the voice buffer 111, but is not transmitted from the protocol entity 17. At another time t18, the data unit of the image packet V4 is output from the image buffer 112, and there is a data unit of the untransmitted audio packet A5.
[0073]
In the method in which the voice packet is always prioritized, as shown in FIGS. 12 and 13, the protocol entity 17 transmits the data unit of the voice packet A3 at time t13 and the data unit of the voice packet A4 at time t14. Thus, the data unit of the image packet V3 is finally output until time t15. Therefore, the data unit of the image packet V3 is transmitted to the network 30 with a delay of 4 INT from the time t11 transmitted from the terminal 21. Similarly, the data unit of the voice packet A5 is prioritized at time t18, and the data unit of the new voice packet A6 is prioritized at time t19. Therefore, the data unit of the image packet V4 is delayed by 4INT from the time t16 transmitted from the terminal 21. Then, it is transmitted to the network 30 at time t20. However, the data unit of the first image packet V1 is output to the network 30 at time t2 with a delay of 1INT from time t1 transmitted from the terminal 21, as shown in FIG.
[0074]
In the method in which the image data unit is always prioritized, as shown in FIGS. 12 and 13, the transmission of the data unit of the image packet V3 is started from the protocol entity 17 at time t13, and the image packet V3 is started at times t14 and t15. Continue sending data units. The data unit of the audio packet A3 output from the audio buffer 111 at time t10 and the data unit of the audio packet A4 output from the audio buffer 111 at time t13 are transmitted after the transmission of the data unit of the image packet V3 is completed. The Therefore, the data unit of the voice packet A3 is transmitted to the network 30 at time t16 with a delay of 9INT from time t7 transmitted from the terminal 21, and the data unit of the voice packet A4 is delayed by 7INT from time t10 transmitted from the terminal 21. Then, it is transmitted to the network 30 at time t17. Similarly, since the data unit of the image packet V4 is given priority at time t18, the data unit of the audio packet A5 is transmitted to the network 30 at time t22 with a delay of 9INT from time t13 transmitted from the terminal 21, and the audio packet The data unit A6 is transmitted to the network 30 at time t23 with a delay of 7 INT from the time t16 transmitted from the terminal 21. However, the data unit of the first voice packet A1 is output to the network 30 at time t6 with a delay of 5INT from the time t1 transmitted from the terminal 21.
[0075]
As described above, when any one of the data units in the group is always prioritized, the delay jitter generated by the multiplexing in the protocol entity 17 is further superimposed on the delay jitter generated in the network 20, and the result is sent to the network 30. Delay jitter spreads during transmission.
[0076]
FIG. 14 shows the occurrence of delay in multimedia communication. A curve D1 indicates the delay jitter caused by passing through the network 20, and is equivalent to the curve D1 in the lower graph of FIG. A curve D3 shows the remaining delay jitter partially absorbed by the buffer 110, and this remaining delay jitter is smaller than the delay jitter after passing through the network 20 of the curve D1 by the amount absorbed by the buffer. A curve D21 represents a delay jitter at the time of multiplexing transmission to the network 30 when the data format of any one group is always prioritized. As can be seen from the curve D21, even when a part of the delay jitter is absorbed by the buffer 110, the delay jitter due to multiplexing is superimposed, and the delay jitter spreads during transmission to the network 30.
[0077]
In the present embodiment, the priority of each data unit in multiplex transmission is determined according to the delay time in order to reduce the delay jitter of the packet. More specifically, when an audio data unit and an image data unit output from the buffer 110 but not transmitted from the protocol entity 17 exist at the same time, data with a large delay time for the first similar data unit at that time. Prioritize units. Next, the processing of this embodiment will be specifically described.
[0078]
12 and 13, at time t13, the delay time D (A3) of the data unit A3 of the untransmitted voice packet is 4INT, and the delay time D (V3) of the data unit V3 of the untransmitted image packet is 1INT. It is. The delay times D (A) and D (V) in FIGS. 12 and 13 are calculated by the above-described equations (1) to (4). In the present embodiment, at time t13, the priority determination unit 120 outputs a transmission control signal indicating that the priority of the data unit of the voice packet A3 having a large delay time is high to the protocol entity 17.
[0079]
At time t14, the delay time D (A4) of the data unit of the untransmitted voice packet A4 is 2INT, and the delay time D (V3) of the data unit of the image packet V3 is the same as 2INT. In the present embodiment, the priority determination unit 120 outputs a transmission control signal indicating that the priority of the data unit of the voice packet A4 is high to the protocol entity 17 in accordance with the rule represented in step S14 of FIG. Therefore, the data unit of the voice packet A3 is transmitted at time t13, the data unit of the voice packet A4 is transmitted at time t14, and the data unit of the image packet V3 is transmitted at time t15. This state is the same as the method in which the audio data unit is always prioritized. However, in the time between time t13 and time t17, the increase in delay jitter is less with respect to the data units of the audio packets A3 and A4 than in the method in which the image data unit is prioritized.
[0080]
At time t18, the delay time D (A5) of the data unit of the untransmitted voice packet A5 is 3INT, and the delay time D (V4) of the data unit of the image packet V4 is 1INT. Therefore, in this embodiment, the priority determination unit 120 outputs a transmission control signal indicating that the priority of the data unit of the voice packet A5 having a large delay time is high to the protocol entity 17.
[0081]
At time t19, the delay time D (A6) of the data unit of the untransmitted audio packet A6 is 1INT, and the delay time D (V4) of the data unit of the image packet V4 is 2INT. Therefore, in the present embodiment, the priority determination unit 120 outputs a transmission control signal indicating that the priority of the data unit of the image packet V4 having a large delay time is high to the protocol entity 17. At time t23, the data unit of the voice packet A6 is transmitted.
[0082]
Therefore, the data unit of the voice packet A5 is transmitted to the network 30 at time t18 with a delay of 5INT from the time t13 transmitted from the terminal 21, and the data unit of the image packet V4 is delayed with 3INT from the time t16 transmitted from the terminal 21. Is transmitted to the network 30 at time t19, and the data unit of the voice packet A6 is transmitted to the network 30 at time t23 with a delay of 7 INT from the time t16 transmitted from the terminal 21. Therefore, the increase in delay jitter is less with respect to the data unit of the image packet V4 than the method in which the audio data unit is prioritized.
[0083]
A curve D22 in FIG. 14 represents the delay jitter in the multiplex transmission to the network 30 when the priority of the packet is determined according to the delay time according to the present embodiment. As is apparent from a comparison between the curves D21 and D22, in this embodiment, an increase in delay jitter can be further suppressed, and the delay jitter at the time of multiplex transmission to the network 30 is the residual delay jitter (the width of the curve D3). )
[0084]
The delay jitter in the multiplex transmission is preferably equal to or less than the allowable delay jitter that can be absorbed by the destination terminal 31. Therefore, in multimedia communication, the amount of delay jitter absorbed by the buffer determined based on the allowable delay jitter of the destination before multiplexing is larger than the amount of delay jitter in the case of transferring data in a single format. Is effective.
[0085]
While the invention has been described with reference to the preferred embodiments, those skilled in the art will recognize that various changes in form and detail can be made without departing from the scope of the invention as defined in the claims. Like. Such modifications, alternatives, and modifications are also included in the claims as equivalents. An example of such an equivalent will now be described.
[0086]
With respect to the network 20 that is a packet switching network, due to network delay, the order of packets of different types in multimedia communication is likely to be disturbed. However, with respect to the network 30 that is a circuit switching network, network delay is unlikely to occur. Disorders between different types of frames are unlikely to occur. Therefore, unlike the embodiment of FIG. 1, it is also possible not to provide the priority determination unit 120 in the transmission sequence BA that handles frames transferred from the circuit switched network 30 having the above-described advantages.
[0087]
The element for absorbing the delay jitter is not limited to the buffer 110. Other elements that can hold the first data unit for a time according to the allowable delay jitter of the destination terminal and sequentially output subsequent data units in a normal order can be used.
[0088]
In the above embodiment, in multimedia communication, a high priority is simply given to a data unit having a large delay time. However, when transmitting three or more types of data, a high priority is given to one or more types of data in advance, and for other types of data, the data unit with the longest delay time is given next. High priority may be given.
[0089]
Instead of the delay time, the data unit to be given priority may be determined based on the length of the data unit.
If the output timing of a short data unit with a large delay time is reached during transmission of a long data unit, interrupt transmission may be performed. In some cases, such interrupt transmission may further reduce delay jitter. For this purpose, for example, ITU-T recommendation H.264. As described in H.223, long data units may be segmented. According to this, even after the transmission of the long image data unit is interrupted and the short audio data unit is transmitted, the remaining segments can be transmitted without any trouble.
[0090]
In the above embodiment, the network 20 is a packet switching network, and the network 30 is a circuit switching network. However, the present invention is not intended to be limited to this, and changes and modifications are possible. Causes of delay jitter include network factors and multiplexing in terminals and gateways. In a packet switching network, delay jitter may occur due to various network factors. On the other hand, in a circuit switching network, delay jitter due to various network factors is unlikely to occur, but frames are exchanged with the delay jitter remaining due to multiplexing in a terminal or gateway. The delay jitter in the packet switching network is generally larger than the delay jitter caused by multiplexing in the terminal or gateway. Generally, a terminal serving each network can absorb the maximum delay jitter that can occur in that network when it becomes a transmission destination.
[0091]
As shown in FIG. 15, it is assumed that the network 20 and the network 300 are both packet-switched networks, and the maximum delay jitter that can occur in the packet-switched network 20 is larger than the maximum delay jitter that can occur in the packet-switched network 300. When the terminal serving the network 20 is a transmission source and the terminal serving the network 300 is a transmission destination, the gateway 10 is configured according to the jitter reduction capability of the transmission destination terminal 31 as in the above embodiment. It is effective to absorb a part of the delay jitter generated in the network 20 on the transmission source side.
[0092]
However, when a terminal serving the network 300 having a small possible delay jitter is a transmission source, a terminal serving the network 20 having a large possible delay jitter is a transmission destination, and the jitter reduction capability of the transmission destination terminal is high. Therefore, it may be less necessary for the gateway 10 to absorb the delay jitter generated in the network 300 on the transmission source side. In such a case, an element for de-jitter need not be provided in the transmission sequence BA for transferring data from the network 300 to the network 20.
[0093]
As described with reference to the above embodiment, when the network 20 on the transmission source side is a packet switching network and the network 30 on the transmission destination side is a circuit switching network, the jitter reduction capability of the terminal 31 on the transmission destination side Accordingly, it is effective for the gateway 10 to absorb a part of the delay jitter generated in the network 20 on the transmission source side. However, the terminal serving the circuit-switched network 30 with a small possible delay jitter is the transmission source, the terminal serving the packet-switched network 20 with the large possible delay jitter is the transmission destination, and jitter reduction of the transmission destination terminal When the capability is high, it may be less necessary for the gateway 10 to absorb the delay jitter generated in the network 30 on the transmission source side. In such a case, it is not necessary to provide an element for dejittering in the transmission sequence BA.
[0094]
As described above, in a circuit switching network, delay jitter due to network factors is unlikely to occur, but frames are exchanged with the delay jitter remaining due to multiplexing in a terminal or gateway. Therefore, the delay jitter that is allowed in the destination terminal belonging to the circuit switching network and is a parameter for determining the amount of absorption in the buffer may be equal to the maximum delay time that can occur due to multiplexing in the network.
[0095]
In the above embodiment, the destination terminal is served directly by the second network 30 connected to the gateway 10. However, the destination terminal may be served indirectly by the network 30. That is, between the transmission destination terminal and the network 30, one or a plurality of networks and another gateway connecting the adjacent networks may be interposed.
[0096]
【The invention's effect】
As described above, according to the present invention, an increase in delay can be minimized by reducing jitter.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a gateway according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an operation of a buffer in the gateway of FIG.
FIG. 3 is a time chart showing processing of a data unit in the gateway of FIG. 1 when the first packet arrives with a minimum delay time.
FIG. 4 is a diagram illustrating processing of a data unit in the buffer when the first packet arrives with a minimum delay time.
FIG. 5 is a diagram showing another processing of the data unit in the buffer when the first packet arrives with the minimum delay time.
6 is a time chart showing processing of a data unit in the gateway of FIG. 1 when the first packet arrives with the maximum delay time.
FIG. 7 is a diagram illustrating processing of a data unit in the buffer when the first packet arrives with the maximum delay time.
FIG. 8 is a diagram showing another processing of the data unit in the buffer when the first packet arrives with the maximum delay time.
FIG. 9 has a graph showing the occurrence of network delay.
10 is a block diagram showing an internal function entity of each control unit shown in FIG. 1. FIG.
FIG. 11 is a flowchart showing processing of a priority determination unit in the control unit in multimedia communication.
12 is a part of a time chart showing processing of a data unit in the gateway of FIG. 1 in multimedia communication.
FIG. 13 constitutes the time chart in cooperation with FIG.
FIG. 14 is a diagram for explaining the effect of application of the above embodiment to multimedia communication;
FIG. 15 is a diagram showing a modified embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... Gateway, 14A, 14B ... Reception part, 18A, 18B ... Transmission part, 20, 30 ... Network (network), 100 ... Control part, 122 ... Output time detection part, 122 ... Delay time calculation part, AB, BA ... Transmission sequence (acquisition part)

Claims (7)

A gateway connecting two networks,
A receiving unit for receiving a plurality of data units of at least one form from a first network;
The data unit received by the receiving unit is temporarily stored, a certain time is determined based on the delay jitter reduction capability of the destination terminal, and the determined time from the time when the first data unit is received by the receiving unit A controller that outputs the first data unit after elapse of time, and then sequentially outputs subsequent data units in the same cycle as the transmission cycle in the first network ;
A gateway comprising: a transmission unit that transmits a data unit corresponding to the data unit output by the control unit to a second network in order to deliver the data unit to the transmission destination terminal. The gateway according to claim 1, wherein
A gateway further comprising an acquisition unit for acquiring information related to the delay jitter reduction capability of the destination terminal. The gateway according to claim 1 or 2, wherein
The receiving unit can receive data units of different formats;
When different types of data units are not transmitted at the same time, the control unit determines a priority in the transmission unit according to a delay time caused when the data unit is transmitted in the first network,
The transmission unit is a gateway that multiplexes data units of different formats output from the control unit according to the priority and transmits the multiplexed data units to the destination terminal through the second network. The gateway according to claim 3, wherein
The controller is
An output time detection unit that detects the first output time when the first data unit in each format is output from the control unit;
A delay time calculation unit for calculating the delay time of each data unit based on the first output time,
The gateway, wherein the control unit determines the priority so that a data unit determined to have a long delay time is transmitted with priority over other data units. The gateway according to claim 1, wherein
The control unit obtains the time by reducing a maximum delay time that can occur when the data unit is transmitted in the first network according to a delay jitter allowed in the destination terminal. And gateway. Receiving a plurality of data units of at least one form from a first network;
Temporarily storing the data unit;
Determining a certain time based on the jitter reduction capability of the destination terminal acquired in advance;
Outputting the first data unit after the determined time has elapsed from the time of receiving the first data unit;
Sequentially output subsequent data units in the same cycle as the transmission cycle in the first network.
Transmitting a data unit corresponding to the data unit to a second network for delivery to the destination terminal;
A data transfer method comprising: The method of claim 6, comprising:
The receiving comprises receiving different types of data units;
This method further
When different types of data units are not simultaneously transmitted, the control unit determines priority in the transmission unit according to a delay time caused when the data unit is transmitted in the first network. With
The method, wherein the longer the delay time generated in the first network, the higher the corresponding data unit is transmitted to the second network.

Images