JPH022270A - Data exchange and multiplex system - Google Patents

Abstract

PURPOSE:To reduce the hardware quantity rand to reduce the average delay time of data by including information controlling exchange order in a header and applying continuously exchange and multiplex to the end of a series of mini-packet string without interrupting other call number into the header by means of the information. CONSTITUTION:The information generated from terminal equipments 31-33 is sent in a form of cell to signal lines 61-63. Moreover, a cell (d) including the voice information exchanged by a PBX and generated from a telephone set 35 exists in a signal line 64. In an exchange 71, after the cell a1 is changed on the signal line 61 and since the M of the cell a1 is logical 1 and the M of the cell b1 is logical 1, the cell a2 is exchanged successively, Since M=0 exists in the cell a3, the cell b1 is exchanged and then the exchange order is discriminated and the result is multiplexed after exchange. Thus, the hardware quantity for a reproduction buffer of the user frame and for order confirming logic at reproduction is saved remarkably.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a data exchange and multiplexing method, and in particular to a data exchange method that performs multiplexing and exchange using mini packets and can significantly reduce the amount of hardware. and multiplexing methods.

[Conventional technology]

Conventional Local Area Ne
The length of data packets used in twork) varies from several tens of bytes to four bytes, but recently,
The length of a mini-packet used in a high-speed switching multiplex communication network is a fixed length, and either 32 bytes or 64 bytes are used (hereinafter, this mini-packet is referred to as a cell).

Regarding conventionally used cell switching methods, for example,
"I.C.C. 8711's 'Prelude'"
(ICC, '87 22.2.1-22.2.6
”Pr1ELUD[E: An Asynchronous
ous Time-Division S wi
ched Network”).

A conventional cell switching system as described in Section 22 will be explained with reference to FIG.

In the cell switching system described in the above document, the cell length is 16 bytes, and the number of incoming and outgoing lines in the unit switch 901 of the exchange is also 16.

Therefore, cell lines used by 16 users can be input into the exchange. A serial-to-parallel converter 902 is provided at the input section of the switch 901, and the serial cell group (AI, A2, . . .
・, Bl, l'32. ...C1, C2, ...
・, DI, D2. ) 903 is converted into 8-bit serial/parallel data by this serial/parallel converter 902. Since the cells input to the serial/parallel converter 902 have different phases for each of the 16 input lines, the phases are adjusted by the phase adjustment unit 904 so that the beginnings of the cells on each input line are sequentially shifted by 1 byte. In other words, cell B1 is delayed by 1 byte from cell A1.
, cell c1 1 byte later than cell B1, cell C1
The cells D1 are adjusted one byte later and inputted to the serial/parallel converter 902.

Each switch 901 exchanges these cells in the order of arrival and then multiplexes them onto the five switch outgoing lines. As shown in FIG. 9, the multiplexed cell column 905 includes At, 131 . C1
, Di, A2. B2. C, 2, B2°A3. ...
・Output in order. This cell row goes in a specific direction
Cell rows not shown are sent out to different routes Y, respectively.

[Problem to be solved by the invention]

In this way, in the conventional cell switching system, when cells arrive simultaneously on each incoming line of a cell switch, the switching process is performed in the order of incoming lines. Therefore, if we focus on adjacent cells on the same incoming line (for example, Al and A2 in FIG. 9), we can see that the previous cell A
The 16th subsequent cell A2 after cell 1 is replaced will be replaced.

In other words, 16 incoming lines contain cell string A from 16 users.
-P is input, and the phases of these cell strings are shifted by one byte by the phase adjustment unit, so when they are exchanged by a switch and multiplexed, if all of these cells are destined for the same route, After 16 bytes A1 to P1 are arranged, 16 bytes 1- of A2 to P2 are arranged. In other words, there are 15 cells from other incoming lines between A1 and A2 of the outgoing line.

By the way, the packets used by LAN use a connectionless communication method (in other words, data is sent directly to the other party without sending the call to the other party in advance), but as in i\i talk communication, the call is first sent to the other party. Using a connection communication method in which information is sent after receiving a call has the advantage that the other party can prepare to receive data after accepting the call. A method of transmitting and receiving calls and control signals using a channel different from the channel used to transfer data in connection communication is called an outband call control method.

The cell switching and statistical multiplexing method using outband call control is called ATM (A5ynchronous T
transferMode), and currently C
It is being discussed at CTTT.

Currently, in the communication network using the above-mentioned ATM, the conventional LA
Let us consider the case where N is accommodated.

That is, in an ATM switching device that accommodates this LAN, the LAN is connected to one input line of a cell switch within the switching device.

As mentioned above, the LAN frame generally has a variable length of about 4 bytes at the maximum, and is divided into lengths that can be transferred in cells at the connection part. in this case,
It will be divided into 4000÷32=125 cells.

On the other hand, the AT that accommodates the destination LAN of this LAN frame
In the M switching equipment, the cells generated from the same LAN frame are buffered, and the LAN frame is reproduced while checking the position in the original LAN frame using the sequence number assigned to the cell. Transferring to LAN. To explain this using the example of FIG. 9, At and Bl are exchanged at the same time by the switch 901 of the exchange device.

C1,...A2. B2. Since it is multiplexed and output in the order of C2,..., in order to pass it to the destination LAN,
Again, the first Al, A2. A3. ..., Bl
, B2. B3. ..., CI, C2゜C3,...
It is necessary to convert the data into the order of ... and recreate it for each LAN frame. In this case, if not multiplexed, A1
- After buffering A4, while transferring A1-A4 to the destination, buffering 81 to B4 to another buffer memory allows you to alternately provide two buffer memories. Just use it. However, in the case of FIG. 9, since the data is multiplexed, buffer memories 906 are required as many as the number of LANs input at the same time.

Now, if we consider a system in which one ATM switching equipment accommodates 16 LANs, in the conventional system, cells from 16 routes are multiplexed in order on the outgoing line, so
In the TM switching equipment, 1
Requires 6 buffer memories. Furthermore, control for confirming cell sequence numbers becomes complicated, which naturally results in an increase in hardware.

As described above, in the conventional system, there were problems to be solved in order to solve these problems.

The purpose of the present invention is to solve these problems and to provide a communication network using cells (fixed length minislots) that can significantly reduce the amount of hardware and reduce the average data delay time. The objective is to provide a possible data exchange and multiplexing method.

[Means to solve the problem]

In order to achieve the above object, the data exchange and multiplexing method of the present invention includes (i) exchanging fixed-length mini-packets containing a pre-assigned call number in the header through outband call control; In a data exchange and multiplexing method that performs statistical multiplexing, information for controlling the exchange order is included in the header, and this information allows a series of mini-packet strings generated from specific user data to be The feature is that the mini-packet including the call number in the header is continuously exchanged and multiplexed until the end of the series of mini-packets without interrupting the mini-packet. Also, (if
) More data (M) for requesting continuous exchange and multiplexing of mini-packets generated from the same user data as information for controlling the exchange order included in the header.
Using bits, set only M of the last mini-packet in a specific mini-packet string to 1Ol, set M of mini-packets other than the last to '1', and set exceptional mini-packets such as audio information to '0'. , respectively, and when an exceptional mini-packet with M of IO+ exists during the exchange of a specific mini-packet string with M of '1', the exceptional mini-packet is exchanged first, and M is '] ′, even if mini-packets with different call numbers exist, continuous exchange is performed until the end of the specific mini-packet string, and after the final exchange of the specific mini-packet string is completed, Another feature is that mini-packets with different call numbers are exchanged. (iii) When exchanging a mini-packet consisting of a header including the more data (M) bit, a call number, and user information, a call number for removing the mini-packet according to the call number and the value of M. A filter is provided, and the call number filter temporarily stores the call number of each first mini-packet in the series, and exchanges mini-packets where M is '1' and the call number matches the stored call number. After continuously allowing multiplexing and exchanging mini-packets with M as '0', if the call number of the mini-packet matches the above-mentioned stored call number, all incoming lines are Another feature is that the exchange and multiplexing of mini-packets of ``1'' from other incoming lines are permitted, and when there is a mismatch, the exchange of mini-packets with M of ``1'' from other incoming lines and multiple interrupt processing are prohibited.

and (iv) instead of the call reference included in the header.

A bit R indicating that the mini-packet string is requested to be continuously exchanged and multiplexed is included in the cell header in addition to M bits 1~ (for example, requested when R is '1'), and M After exchanging and multiplexing a mini-packet with R = '1' and R = '1', the exchange of mini-packets with R = '1' and multiple interrupt processing from the incoming line to which the mini-packet belongs is prohibited, and 1M After exchanging and multiplexing mini-packets with R set to '1' at O', allow mini-packet exchange and multiplex interrupts from all incoming lines, and when M is 1'0' and R is 'O'. Another feature is that after the mini-packet exchange and multiplexing of ', the process follows the processing content when the latest R is '1'. Furthermore, (v) consecutive inputs are made from the same input line, and M is '1' or M is '
1', when exchanging and multiplexing mini-packets with R of '1', the maximum number of consecutive mini-packets calculated from the maximum user data length and mini-packet length is memorized, and the input mini-packet string is Count the number of.

The counted value is constantly compared with the number stored above, and if the result of the comparison exceeds the stored maximum number of continuous mini-buckets, it is determined that there is an abnormality, and mini-packets from all incoming lines are exchanged and multiple interrupts are performed. Another feature is that it allows

[For production]

In the present invention, the cell switch is controlled so that cells generated from the same LAN frame are not interrupted and multiplexed by other cells on one outgoing line of the final destination ATM switching device. Therefore, by providing an identifier in the header of each cell that indicates that a cell generated from the same LAN frame follows, the cell switch can respond to a series of cell columns on the same input line indicated by this identifier. The exchange process should be performed continuously.

In other words, packets are originally intended to be multiplexed and transmitted in large quantities in exchange for allowing for data delays, but if packets for which real-time performance is important, such as in audio bucket 1, are delayed too much, the sound quality will deteriorate. Therefore, in this case, processing is performed by interrupting the exchange order of cell strings generated from the same user data to prevent delays as much as possible.

At the same time, in order to prevent the increase and complexity of hardware due to multiplexing, cells generated from the same LAN frame are made to follow each other as much as possible, eliminating the need to check for missing cells and reducing the need for buffer memory. be.

If there are consecutive cells generated from the same LAN frame on the outgoing line, the final destination ATM that receives them
The switching device is capable of sequentially reproducing LAN frames regardless of the number of LAN systems to be multiplexed, so in principle, the reproducing buffer memory is the buffer memory that is being transferred to the destination LAN, and the buffer that is being reproduced. It is only necessary to provide two sides of memory.

If a cell indicating that the last cell generated from the same LAN frame is missing, it may become impossible to process cells from other incoming lines.

In this case, since a maximum frame length is set for the LAN frame, when cells from all other incoming lines are received consecutively, exceeding the maximum number of consecutive cells calculated from this cell length, Treat the receiver as if it were attached.

〔Example〕

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

FIG. 3 is a configuration diagram of a communication network to which the present invention is applied.

In FIG. 3, 71 to 75 are switching devices that perform cell switching and statistical multiplexing using call numbers assigned by Ara 1-band call control. Of these, 74.75 are relay switching devices. LAN41 to 44 which are user networks
．． and its LAN nodes 41a-44a, PBX-45
, 46 are connected to protocol conversion devices 51 to 56 for converting each user network protocol into an exchange/network protocol. This protocol conversion device may actually be housed in the same housing as a part of the exchange device.

In this embodiment, the data of the LAN terminals 31 to 34 and the ? !
! Although each medium of voice information from 35 to 38 voice reduction terminals is handled, it is also possible to accommodate other media such as packet switching information having CCITT Recommendation X and 25 protocols. That is, since the type of storage media is not the essence of the present invention, the case of only the above two media will be described here.

FIG. 4 is a diagram showing the configuration of the protocols of the LAN terminal, protocol conversion device, and switching device in FIG. 3.

Now, in the communication network shown in FIG. 3, from terminal 31 to terminal 34
LAN terminal protocol 3'1 when transferring data to
FIG. 4 shows the AN protocol 4'1', the protocol 5'1' of the protocol conversion device 51, the protocol 7'1' of the switching device 71, and the flow of information.

LAN terminal protocol 3'1' is OS I (Op.
nSystems Interconnectio
n) M A C (Med
ia Access Control) Layer 4
01 and P II I (Physical) Layer 4
02 and an upper layer 403.

Furthermore, the protocol 5'1' of the protocol conversion device is L
MAC layer 401 and P of AN protocol 4'1'
HI layer 402 and protocol converter 407 of the switching device
It is composed of.

Upon receiving LAN data 404 from the terminal 31, the protocol conversion device 51 determines whether the received data is data that should be exchanged at the exchange device 71, and if it is data that should be exchanged, the protocol conversion device 51 transmits the data to the exchange device 71. The outbound call control information 406 is converted into cell information, which is an exchange unit within the network, and is transferred along with outband call control information 406. This call control information 406 includes
LAN address of each terminal 31 and terminal 34 and terminal 3
4 and a logical call number of the cell used in subsequent data transfers.

The call control information 406 is received and processed by all the switching devices 71, 74, and 73 indicated by the above-mentioned switching device address string. When a response to this control information is returned by the switching device 73 accommodating the destination terminal, the switching device 71 that has received this response allows the protocol conversion device 51 to transfer data.

As a result, the protocol conversion device 51 sends out the LAN data 405 changed into the cell format.

FIG. 5 is a diagram showing the cell format used in the present invention.

A cell is composed of an inter-switch header and user data. The header is a 2-byte LCA (logical call number)5
01 and 1 byte SI (service information) 502, and the user data consists of 1 byte SN (division number) 502.
03 and 31 bytes of T (information) 504. L.C.A.
501 is assigned in advance based on call control information. Also,
5I502 is the priority and more data (M) of that cell
bit (when M=1, indicates the existence of a subsequent cell generated from the same user data) and CR for error detection of LCA 501
Consists of C. In addition, LAN data is stored in multiple devices (information) 5
04 and is sent, but the divided division number is written in 5N503 and sent before ■ (information).

If the transfer information is audio information, the audio information is set to 1504, and the information occurrence time, generally called a timestamp, is set to 5N.
503 respectively. And call control information is also l.
504 for transfer.

FIG. 6 is a block diagram of the protocol conversion device in FIG. 3.

In FIG. 6, the protocol is shown in correspondence with the protocol diagram in FIG. That is, it is composed of a MAC layer 401, a PHI layer 402, a switching device protocol converter 407, and a protocol 7'1' of the switching device 71. A LAN access unit 601 in the MAC layer and PHI layer 401 and 402 executes LAN protocols 401 and 402. Call initiator 60 in exchange protocol converter 407
2 is.

Upon receiving the LAN control information 404 in FIG. 4, it generates call control information and starts transmitting it. Cell generation/reproduction section 603 divides call control information or LAN data into cells, generates/adds cell headers, and reproduces LAN data from a group of received cells.

The cell control unit 605 executes outband call control processing, and the cell switching unit 1 switches cells based on this information and LCA. The link support unit 604 connects the trunk line 6
5, frame assembly, and synchronization processing.

Call control information from the LAN 41 side to the trunk line 65 side.

Since there are data, call control information and data going from the trunk line 65 side to the LAN 4.1 side, in the former case, a cell is generated, and in the latter case, a LAN frame is reproduced from the cell.

FIG. 1 is an operation time chart of a cell switching system showing an embodiment of the present invention.

In FIG. 1, a signal time chart of signals LMA61-67 in FIG. 3 is shown. Time t is taken in the horizontal direction, and cell arrangement on each signal line is shown in the vertical direction. Time t is progressing from left to right.

The signal lines 61 to 67 are all connected to the protocol conversion device 51 to
55, each information block represents a cell. On the signal lines 61.62.63, information generated from the terminal 31.32.33 is transmitted in cell format (al, a2.a3 and bl, b2, respectively).
．． b3 and cl.

c2. c3). Further, on the signal line 64, there is a cell d containing voice information generated at the telephone terminal 35 and exchanged at the PBX. Here, a, b, c, d, . . . written in the cell indicate logical call numbers, and 1, 2, 3, . . . are added to those numbers. ...
...indicates the division number. Furthermore, cell a
3. b3. c3. d is more data (M) bit OI
In other cells, the more data bit is set to ``1''. In other words, a3. b3. c3. d indicates that there is no subsequent cell generated from the same user data, and other cells at, a2 . bl.

b2. Each cell cl and c2 indicates that there is a subsequent cell generated from the same user data.

In the switching device 71, after replacing the cell a1 on the signal line 61, since the M of the cell a1 is ``1'' and the M of bl is also 111, the cell a2 is subsequently replaced. In a3, since M='O', bl is exchanged next, and thereafter, the exchange order is determined in the same manner, and after the exchange, multiplexing is performed.

FIG. 10 is a diagram showing criteria for determining the replacement order in the present invention.

In Figure 10, the horizontal axis shows the more data of your cell.
The value of 1 bit is plotted on the vertical axis as more data of other cells.
' bit values are shown respectively, and the cell to be exchanged under that condition is shown at their intersection. For example, when your own cell is ``1'', in this case a cell generated from the same LAN frame will follow, so your own cell will not be replaced unless the other cell is ``O''. In addition, when the own cell is ``0'', in this case, the same -L
Since the cell generated from the AN frame does not follow and ends with its own cell, another cell is replaced. As a result, the own cell is replaced when the more data amount bit of the own cell is '1' and the M bit of the other cell is '1'.
It's time. At other times, all other cells are replaced. Thereby, data delay can be reduced.

The exchange promises are as follows:

■If both are '1', prioritize yourself.

■If it is '0'', you can enter others.

■If it is '1' but the others are 1'0', you can enter it.

In FIG. 1, according to the principle of this exchange order, on the signal line 65 there are a l to a3. It will be multiplexed in the order of b 1 to b 3.

In the exchange device 72, after exchanging 01 on the signal line 63,
Since M of cl is '1', M of d is '0', d is exchanged in the next order.

After that, c2. The exchange of c3 takes place. Therefore, on the signal line 66, as shown in FIG.

d, c2. It is multiplexed in the order of c3.

Furthermore, in the relay switching device 74, the signal a65 and the signal line 6
The input cells from 6 are exchanged, but after exchanging al,
Since there is a cell d whose call number is different from cl and whose M is ``0'', this cell d is exchanged first. After that, a2.
The order of a3 is exchanged, and after a3 with M='O', cl is exchanged.

Note that 101 in FIG. 1 is the processing time in the exchange 'jA device 1, and 102 is the processing time in the exchange device 74.

In this way, the multiplexed cell string on 67 in FIG. 1 is sent out to the output side trunk line 67 of the transit switching device 74 in FIG. 3.

FIG. 2 is a detailed block diagram of the cell switching unit in FIG. 6. In Figure 6, the L sent from the LAN41 side
After the AN frame is divided into cells by the cell generation/reproduction unit 603, it is input to the cell switching unit 1 through the signal line 11(a), where it is exchanged and output to the destination I and the signal line 91(a) on the AN side. On the other hand, the signal line 11(b) on the destination LAN side in the opposite direction
The multiplexed cell string sent from the cell exchanger 1 is exchanged by the cell exchanger 1, and then outputted to the signal line 91(b).The cell generation/reproduction unit 603 reproduces the multiplexed cell string into a LAN frame and connects it to the signal line 93 for LAN access. It is sent to the LAN 41 side via the section 601.

In FIG. 2, these opposite directions are shown as the same direction. This cell switching unit 1 is configured such that for cells that arrive successively, M with another LCA becomes ``1'' until a cell with the same LCA and M with ``0'' passes through.
The cells input from each input line 11 are connected to the LCA converter 2.
1 performs LCA conversion and is stored in the input FII"031.Furthermore, in the routing header adder 41,
A routing header corresponding to the outgoing line 91 is added based on the LCA of the cell and output to the corresponding outgoing line through the space switch 50. If two cells are output to the same outgoing line, only one of them is output and the other cell is discarded. The cells output from the space switch 50 have their routing headers removed by a routing header remover 61, and are filtered by LCA and M bits by an LCA filter 71.The cells that have passed through this filter are output 1? It is accumulated in IFO81. At this time, if the cell arrives at the output 1;', IFO8]-, the output P I FO 81 returns ΔACK I3 16 to the input FIFO 31 through the space switch 5o. The ACK signal 16 passes through the space switch 50 and becomes the ACK signal 18.
It is fed back to the input FTFO 31 as a signal. Input F'0
3L, if the ACK signal 18 is returned, the next cell is output, but if it is not returned, the same cell is output again.

The L CA filter 71 in this cell exchange section 1 has 1M as ``1'' when successive cells pass through it, and its L
Cells with CA are passed through, but consecutive cells with other LCAs are filtered out.

This LCA filter 71 temporarily stores the call number of each first cell in the series, and exchanges cells that match the stored call number when M is ``1'' and the call number is ``Y''. After continuously permitting multiplexing and exchanging cells with 1M of ``0'', if the call number of that cell matches the above-mentioned stored call number, cells from all incoming lines are If there is a mismatch, the exchange and multiple interrupt processing of cells with M of ``1'' from other incoming lines are prohibited. For example, the exchange of bl after a3 in FIG. 1 and the exchange of c2 after d are also based on the above principle.

Here, the routing header added to the cell is valid only within the network, and the header is added only when passing through an exchange switch. Therefore, it is no longer needed after being replaced, so it is removed.

In this way, the specific processing method in the cell exchange unit 1 is as follows:
Even if the input side sends out all the signals, the output side receives and stores only one of the signals that arrive at the same time, and discards the other. When the cells arrive at the output side, they are confirmed by sending a response back to the input side, and the discarded cells are recovered by retransmitting only those cells for which no response is received on the input side.

This cell switching section 1 has the same -LCA, and M is ''0
If the cell ' has passed, the LCA that arrives first is newly stored. Further, the LCA filter 71 is provided with a counter for counting the number of identical cells that pass through continuously, and the counter value of this counter is always compared with the maximum setting value 133. This maximum setting value is a value obtained by dividing the maximum LAN data length of 4096 bytes by the cell length of 32 bytes, 128, and adding 5 to allow for a margin.

And when this counter overflows, the same
The same processing as when a cell with M of IO+ passes through with LCΔ is performed.

That is, the "double process" is a process when the cell switching unit 1 is abnormal.

Note that instead of the same-LCA comparison process, it is also possible to use the R bit that is set when requesting continuous exchange. In this case, the R bit is the fifth
It is included in SI (service information) 502 in the figure.

Figure 7 shows details of only the reproduction part of the cell generation/reproduction section in Figure 6! II is a logical configuration diagram.

As shown in FIG.
It is composed of an AN access control section 706.

The received cell information 91 is sorted by call number by a logical call number switch 702 and then stored in a logical call number processing section 703 provided for each call number.

At this time, from the logical call number switch 702, the same -LC
A cell column 708 (thick line) and final cell information 707 (thin line)
is output. When the logical call number processing unit 703 receives the final cell information 708, it outputs a user data reproduction completion signal 711 to the LAN access control unit 706. L
At the same time as receiving the user data reproduction completion signal 711, the AN access control unit 706 controls the LAN access 601 (sixth
At the same time, it sends out a transmission request signal 714 to the client (see figure), controls the selector 704, and uses a transmission completion signal 715 from the LAN access section 601 to send out a transmission request signal 714 for each logical call number via control numbers i 712 and 713. Controls write/read operations of the processing unit 703.

Note that 710 is a signal that outputs the processing result of the logical call number processing section 703, and output cuff 05 selected by the selector 704 is sent out.

FIG. 8 is a detailed logical configuration diagram of the logical call number processing section in FIG. 7.

Reception octet counter 8 of logical call number processing unit 703
01 counts the input cell length according to the write control signal 712 from the LAN access control unit 706. Next, this count value is decoded by the decoder 803, and when this value indicates 31, the received expected SN counter 802
Add +1 to

That is, each time the cell length of 32 bytes is counted, the number of cells is counted by counting up. Expecting to receive this
Output 8]1 of N counter 802 and AN812 of the receiving cell
are compared in the minus number circuit 807. In the transmission request determination unit 806, the output of the minus number circuit 807 and the decoder 80
3's output '0'' and the final cell information 707 are all '1'
', the user data reproduction completion signal 711 is
N access control unit 706. As a result, LA
N access control unit 706 makes a transmission request to the LAN access unit 601.

The input data 708 is input to a buffer memory 805. At this time, the address stored in the buffer memory 805 uses the output of the reception octet counter 801, the WE multiplication signal, and the control signal 712 from the LAN access control section 706. Also, buffer memory 8
05, the LAN access control unit 70
A read address register 808 is controlled by a control signal 713 from 6, and its output is used as a read address.

As described above, in this embodiment, since the continuous exchange control of consecutive cells is performed by the LCA filter 71, any switch such as a matrix switch, a bus switch, or a Banyan switch can be used as the space switch. .

Furthermore, if an error occurs within the ATM network, the erroneous cell is discarded, and if there is even one missing cell when reproducing a user frame, the user frame is invalidated and discarded. Thereafter, the upper level protocol of the terminal 2 detects the loss of the user frame and takes steps to retransmit it. In general, errors occur in the transmission path toward berth 1, but when multiplexing is performed in units of user frames as in the present invention, the number of user frames discarded statistically decreases, and the overall number of discarded user frames decreases. As a result, the average delay time due to retransmission can be reduced.

〔Effect of the invention〕

As explained above, according to the present invention, cells generated from other user frames are not interrupted and multiplexed in a cell string generated from the same user frame, so that the playback buffer of the user frame 11 and the The amount of hardware for the orderliness check logic can be significantly reduced. Furthermore, since multiplexing is performed in units of user frames, the number of user frames discarded statistically is reduced, and the average delay time due to retransmission can be reduced.

[Brief explanation of the drawing]

FIG. 1 is an operation time chart of a cell switching and multiplexing system showing an embodiment of the present invention, FIG. 2 is a detailed block diagram of the cell switching section in FIG. 3, and FIG. 3 is a communication to which the present invention is applied. The overall configuration of the network, Figure 4 is the LA in Figure 33.
FIG. 5 is a diagram of the format of the cell used in the present invention. FIG. 6 is a block diagram of the protocol converter and switching device in FIG. 3. The figure is a detailed block diagram of the reproduction part of the cell generation/reproduction unit in Figure 6, Figure 8 is a detailed logical configuration diagram of the logical call number processing unit in Figure 7, and Figure 9 is the operation of the conventional cell switching system. Explanatory diagram. FIG. 10 is a diagram showing the exchange order principle of the present invention. 1: Cell exchange section, 21: LCA converter, 31: Input F4
F○, 41 Nilting header adder, 50: Space switch, 61 Nilting header remover, 71:
LCA filter, 81: Output FIF0.31-38: Terminal, 41-44: LAN, 45.46: PBX, 5
1 to 56: Protocol conversion device, 71 to 75: Switching device, 601: LAN access unit, 602: Call activation unit, 60
3: Cell generation/reproduction section, 604: Link correspondence section, 605
: Cell control unit, 702: Logical call number switch, 703:
Logical call number processing unit, 706: LAN access control unit,
801: Reception octet counter, 802: Reception expectation counter, 803: Decoder, 805: Bathophore memory,
806: Transmission request determination section, 807: - number comparison section, 808
: Read address counter. Patent applicant: Hitachi, Ltd.

Claims (1)

[Claims] 1. Exchange of fixed-length mini-packets containing call numbers in their headers that are allocated in advance by outband call control and have the same value when generated from the same user data; In a data exchange and multiplexing method that performs statistical multiplexing of the mini-packets, information for controlling the exchange order is included in the header, and the information is used to control a series of mini-packet strings generated from specific user data. A data exchange and multiplexing system characterized in that the series of mini-packets is continuously exchanged and multiplexed until the end of the series of mini-packets without interrupting the mini-packets containing other call numbers in the header. 2. As information for controlling the exchange order included in the header, more data (M
) bit, set M only in the last mini-packet of a specific mini-packet string to '0', set M in mini-packets other than the last to '1', and set exceptional mini-packets such as audio information to '0'. ', and when an exceptional mini-packet with M of '0' exists during the exchange of a specific mini-packet string with M of '1', the exceptional mini-packet is exchanged first, and M is '1' and even if there are mini-packets with different call numbers, the above-mentioned specific mini-packet string is continuously exchanged until the end, and after the final exchange of the specific mini-packet string is completed, the above-mentioned M is '1'. 2. The data exchange and multiplexing system according to claim 1, wherein mini-packets with different call numbers are exchanged at the . 3. When exchanging a mini-packet consisting of a header including the more data (M) bit, a call number, and user information, a call number filter is provided to remove the mini-packet according to the call number and the value of M. , the call number filter temporarily stores the call number of each first mini-packet in the series, and exchanges and multiplexes mini-packets where M is '1' and the call number matches the stored call number. After continuously permitting and exchanging mini-packets with M as '0', if the call number of the mini-packet matches the above-mentioned stored call number, mini-packets from all incoming lines are accepted in the next order. The first aspect of the present invention is characterized in that when there is a mismatch, the exchange of mini-packets with M of '1' from other incoming lines and multiple interrupt processing are prohibited. Data exchange and multiplexing method according to item 1 or 2. 4.Instead of the call number included in the header, include in the cell header a bit R indicating that it is a mini-packet string that requires continuous exchange and multiplexing (for example, R is '1'), after exchanging and multiplexing a mini-packet where M is '1' and R is '1', R from the incoming line to which the mini-packet belongs is '1'.
1' mini-packet exchange and multiple interrupt processing are prohibited, and after exchanging and multiplexing mini-packets where M is '0' and R is '1', mini-packet exchange and multiplexing from all incoming lines are disabled. Enable interrupts and set M to '0'
In this case, after exchanging and multiplexing the mini-packet with R being '0', the processing content when the latest R is '1' is followed. Data exchange and multiplexing method as described in Section 3. 5. Continuously input from the same entry line, and the first
M in the second and third terms is '1' or M in the fourth term is '1'
', when exchanging and multiplexing mini-packets with R of '1', the number of consecutive mini-packets calculated from the maximum user data length and mini-packet length is memorized, and the input mini-packet string is The number of mini-packets is counted, and the counted value is constantly compared with the number stored above. If the result of the comparison exceeds the stored maximum number of continuous mini-packets, it is determined that there is an abnormality, and all incoming lines are 5. A data exchange and multiplexing system according to claim 1, 2, 3 or 4, characterized in that mini-packet exchange and multiple interrupts are permitted.