JP2875559B2 - Routing method and node circuit for burst data transfer - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a routing method and a communication processing node circuit used in a packet switching system. In particular, the present invention relates to a node circuit for processing burst data transmission and a routing method applied thereto.

(Prior art) An example of a conventional packet communication system that absorbs burst data is “Statistical Switching Architecture for Futur”.
e Services "ISS'84 (May 1984). This system employs a simple communication protocol and a self-routing scheme.

The term "burst data transmission" generally means a part in which significant information is continuously transmitted as an attribute of transmitted data,
This is a data transmission consisting of a combination of paused parts without significant information.

For example, in the case of transmitting a moving image, the entire image is transmitted first, and then a small amount of data representing rewritten data of a changed portion is transmitted. These are used to improve transmission efficiency.

For example, in the case of multimedia communication composed of audio data and image data, particularly in the case of transmitting moving images, such burst data transmission is often used to reduce the total amount of information to be transmitted.

9 (a), (b) and (c) are diagrams for explaining two different types of packet formats in burst data transmission. As shown in FIG. 9 (a), a large amount of large data is generated first, and subsequently, rows of small data are generated at intervals. When these data are transmitted in a variable-length packet format, a special flag packet indicating that there is no data to be transmitted is transmitted during the interval as shown in FIG. 9 (b). . If the data is transmitted in a fixed-length packet format,
An empty packet having no valid information as shown in FIG. 9 (c) "E" is transmitted during the interval. Note that, in the figure, a packet with a half hatched portion represents a packet in which significant information is not included in the entire length of the packet.

(Problems to be Solved by the Invention) However, according to this conventional system, the system itself is merely a routing device according to the address information included in the packet. Therefore, since the system does not take into account the state of the transmission line or the information held by the packet, it cannot correct an error generated during packet transmission, nor can it handle loss of a packet due to overflow of the exchange buffer due to congestion.

Generally, since burst data transmission in multimedia communication has the special properties described above,
This causes a buffer overflow, and further, a packet loss which deteriorates communication quality.

In general, the characteristics of the transmission path are determined by the maximum transmission speed, and in most cases, it depends on the transmission speed of the large amount of data. If the transmission path is set in consideration of the average transmission rate, a buffer overflow may occur when the transmission path is significantly lower than that at the time of the initial transmission of a large amount of data.

However, on the other hand, since there are many intervals during data transmission, the transmission efficiency is wasted if the transmission path is set only in consideration of the maximum speed for the first mass data transmission. Or the equipment therefor may be more expensive than before.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a node circuit which can monitor the traffic of an input packet and select the most appropriate output line, thereby optimizing the overall transmission efficiency.

It is another object of the present invention to provide a node circuit that reduces packet loss caused by switching buffer overflow by a simplified routing method.

It is a further object of the present invention to provide a node circuit that can absorb multimedia communications including burst data.

(Means for Solving the Problems) According to the present invention, a node circuit is connected to a plurality of input lines and output lines, and is used for packet communication. At least one of the input lines is connected to a packet terminal that handles burst data transmission.

The node circuit of the present invention is connected to an input line and an output line, can transfer an input packet to one output line based on the header information, and can store a combination of a specific input line and an output line. It comprises a switching network that can be used, an exchange control device for controlling the switching network, and a traffic control device connected to the exchange control device and having the following functions.

(A) upon setting up a call from the packet terminal, receiving traffic information of an input line to which the packet terminal is connected, for example, an average bit rate; (b) calculating a traffic amount required for the call from the traffic information; c) selecting an output line having a sufficient traffic capacity for the call from a plurality of output lines; and (d) controlling the switching controller so that the switching network selects the input line as the input line. Control is performed so that combinations of lines are stored as a link table.

Further, the present invention provides a traffic observation device to observe the actual call traffic of each input line. Further, the traffic control device refers to the observation result at a predetermined cycle and corrects the traffic information.

According to the present invention, each node circuit can select an optimum output line for accepting the call at the stage of setting up the call. Therefore, the throughput of the node circuit is improved, and the load of burst data transmission is spread. Thus, for example, the present invention can be applied to a packet switching center that accommodates moving image and audio data transmission.

(Operation) According to the present invention, when setting up a call for a packet from an input line, the traffic control means derives the bandwidth of the output line to be used for the call based on the used band information of the terminal that transmitted the packet. I do. The switching means transfers the packet to an output line corresponding to the band derived by the traffic control means when setting a packet call. Also, actual traffic is observed from the traffic observation means for the input line, and the residual band of the output line is corrected based on the observation result. Thereby, the overflow of the buffer in the switching means is minimized, and the communication quality and throughput of the communication processing node are improved.

Embodiment FIG. 1A is a schematic block diagram of a preferred embodiment of a node circuit according to the present invention.

FIG. 1B is a diagram showing an example of a control mode of the node circuit of the present embodiment. In the figure, node circuits 100 (transmission side) and 102 (reception side) to which the present invention is applied are packet terminals 101-1, 101-2,.
m or 102-1, 102-2, 102-n.
The node circuits are connected to each other by the communication line 103. However, another node circuit to which the present invention is applied may be installed between them.

Hereinafter, the present embodiment will be described with reference to FIG.

This node circuit 100 has a plurality of input lines 11-1, 11-2, 11
-3,..., 11-m are connected via traffic measuring devices 12-1,. These input lines are the first
As shown in FIG. 3B, the terminal is connected to a packet terminal for performing burst data transmission or another node circuit to which the present invention is applied.

These traffic measuring devices perform bit counting in the case of a fixed-length packet format. Further, those traffic measuring devices are connected to the switching network 150 (for example, 1
The communication lines 14-1, 14-2, and the like are described in Japanese Patent Application Laid-Open No. 63-169850 published on Jul. 13, 988.
14-3,..., 14-m. Here, hereinafter, the communication line includes a unidirectional or bidirectional communication line, a virtual transmission line in a time division manner, an abduction bus line, and the like.
Further, the traffic measuring device is connected to the traffic control device 240 via the communication line 130.

The communication lines 14-1, 14-2, 14-3,..., 14-m are provided for transferring an input packet to the switching network, and the other communication line 130 is used for measurement. It is provided to notify traffic information to the traffic control device 240.

The switching network 150 is connected to these communication lines 14-1 ... 14-m.
, And recognizes each piece of destination information, and transfers them to output lines 16-1, 16-2, 16-3,..., 16-n. Each combination of these communication lines and an output line to which a packet is to be sent is controlled by an exchange controller 251 (for example, a processing circuit described in the above-mentioned JP-A-63-169850, such as Mp, Ps). Directed through 255. However, the switching network itself can hold the combination in memory. Further, the exchange control device 251 is controlled by the traffic control device 240 via the communication line 245.

Further, the traffic control device 240 is connected via a communication line 230 to a terminal 220 which controls the entire node circuit. The terminal 220 is used to give various characteristics to the node circuit or to instruct the node circuit to perform functions not provided in the embedded software of the node circuit.

Although not shown in the drawings, all circuits except for the traffic control device 240 and the terminal 220 can be duplicated for backup.

Before describing the detailed configuration of the node circuit, the general functions of the node circuit of the present invention will be described with reference to FIGS. 2 (a), (b), FIGS. 4, 5 (a), and (a). This will be described using b).

According to the virtual call process used for general packet communication, a call request including a destination address (Call Raquest;
(Called CR) is sent to the destination packet terminal,
In the next stage, a call acceptance (Call Acknowledge, hereinafter referred to as CA) packet is returned from the receiving terminal to the calling terminal as acknowledgment of data transmission. During these packet transmissions, each node circuit (or exchange) on the transmission path stores in the link memory the combination of the input line and the output line used for the transmission.

According to the invention, instead of the conventional "CR" packet, at least (1) the destination address, for example as shown in CCITT Recommendation E.164, (2) the maximum transmission bit rate and (3) the average transmission bit rate, Is used when setting a transmission path.

Such information can be easily calculated, for example, in the packet terminal performing the burst data transmission by using the conventional technique. For example, the maximum throughput of the terminal is set to the maximum bit rate, and an appropriate coefficient Q (0 <Q <1)
May be multiplied to obtain the average bit rate.

FIGS. 2A and 2B are diagrams for explaining how the node circuit of the present invention handles the above information in the memory of the traffic control device 240. FIG. At the first stage, as shown in FIG. 2 (a), the node circuit on the calling side receives the maximum bit rate and looks up the maximum bit rate table to find the address of the next effective bit rate table. This address is stored at an address separated from the top address of the maximum bit rate table by an address distance corresponding to the maximum bit rate.

These maximum bit rate tables are provided for each input line. Of course, a common large-scale table can be provided for each line.

The effective bit rate to be determined is located at an address separated from the top address of the effective bit rate table by the same address distance as the value of the average bit rate received together with the maximum bit rate. In the initial setting, the average bit rate may be set as the effective bit rate.

Generally, the effective bit rate is set as a compromise between the maximum bit rate and the average bit rate. In other words, if the maximum bit rate is too large and the amount of data that follows is too small, if the average bit rate is too low, the buffer in the communication path may overflow during the first data transmission stage, and long low-speed intermittent transmissions may occur. In the case of data transmission, the average bit rate and the effective bit rate are substantially equal. In order to cover such a case, the effective bit rate is almost halfway between the maximum bit rate and the average bit rate.

Next, as shown in FIG. 2B, the node circuit analyzes the destination address and refers to the destination table. The destination table stores the top address of the characteristic table representing the characteristics of each valid output line.

In general, a plurality of output lines for the same destination are prepared for preventing congestion or for an emergency, and they are given a predetermined priority.

In the specific table, “A” represents the maximum bit rate of a certain output line (line number “C”), and both A and C are unique values. “B” is the transmission rate currently used by the output line “C”. This value is rewritten as follows.

First, the node circuit calculates the difference between "A" and "B" for each output line, and looks for the first output line whose difference is larger than the effective bit rate determined in the previous stage. If it is obtained, the node circuit stores in the link memory the combination of the input line number for sending the packet and the obtained output line number "C", and further stores "B" in "B". And the effective bit rate.

Note that all of the above tables can be loaded by appropriate software provided in the traffic control device, and can be input from the terminal 220 as needed.

In this way, the most appropriate output line having the most appropriate transmission rate is selected. Further, the subsequent data transmission is executed based on the contents of the link memory like a normal virtual call.

This situation will be described in detail after describing a packet format applied to the present invention.

The node circuit of the present invention can be applied to any of a variable length and a fixed length. FIGS. 3A and 3B are diagrams for explaining those applicable packets.

FIG. 3A shows an example of the structure of a fixed-length packet based on the CCITT recommendation. In the figure, GFC is an abbreviation of General Flow Control.
VPI stands for Virtual Path Identification (Vir
Abbreviation for tual Path Identification. YCI stands for Virtual Channel Identification.
Identification), PT stands for Payload Type
d Type), HEC stands for Header Error Control (Hea
der Error Control), and RES is reserved
rved). This packet structure consists of 5 octets (1 octet-8 bits) of header information and 48 octets of data information. VCI for empty packets
(12 bits)-"O", the information in the data field is discarded in the node circuit without any consideration.

Further, when a certain packet is used for the signal control of the present invention, the sequence of VCI in the packet takes a predetermined pattern.

FIG. 3 (b) shows a high-level data link (HighLevel
This is an example of a variable-length format packet based on the Data Link (hereinafter, HDLC) standard. In this structure, “F” is an 8-bit sequence “01111110”, which is used as a delimiter between packets. “A” is an 8-bit destination address, and “C” is an 8-bit control field. FCS stands for Frame Check Sequence
It is used for error correction of data to be transferred.

 Note that the maximum length of a packet is 4096 octets.

When there is no data to be transmitted, only the flag sequence is transmitted at predetermined intervals only for maintaining communication. Therefore, a packet having only the flag sequence is discarded without being measured.

Furthermore, in the present invention, when a certain packet is used for signal control, the LCGN in the "1" frame for information transfer is used.
(4 bits) and LCN (8 bits) (corresponding to the aforementioned VCI) have a predetermined pattern.

FIG. 4 and FIG. 5 (a) are diagrams for explaining the operation of the present invention.

FIG. 4 is an explanatory diagram when the packet terminal 260 on the calling side sets a call to the packet terminal 290 on the called side.

For generalization, in the following description, a signal packet is identified by its header information. of course,
The header information may be VCI information having the specific pattern described above in the case of the fixed-length packet format, or may be LCGH or LCN having a predetermined pattern in the case of the variable-length packet format.

First, when the packet terminal 260 on the calling side wants to originate a call with burst data transmission, the packet terminal 260 transmits a signal packet having header information SI indicating a call request from the packet terminal to the node circuit in the first packet. Node circuit 10
It is sent to the switching network 150 of 0 via the communication line 265.

In the switching network 150, the header information is translated, and it is recognized that the header information is a signal packet.
This signal packet contains the above-mentioned traffic information for selecting an output line, that is, (1) a destination address, and (2)
The switching network 150 sends the packet to the traffic control device 240 via the switching control device 251 including the maximum transmission bit rate information and (3) the average transmission bit rate information.

In the traffic control device 240, the above-described process of selecting the most appropriate output line is executed, and the traffic control device 240 controls the switching control device 251 so that the switching network 150 stores the combination of the input line and the output line in the memory. To control. For example, it is assumed that the link table is stored as L1: (11; 12). Further, the traffic control device 240 rewrites the header of the packet to S2 representing a signal packet between the node circuits,
The switching control device 251 is connected to the output line from which the packet is obtained.
And output via the switching network 150.

More specifically, the link table corresponding to each input line is a virtual table as shown in FIG. As described above, each call is managed by VCI (or LCGN, LCN for variable length packets). Further, this link table manages all calls of the corresponding input line, and is arranged at the address position of the corresponding input line in the switching network.

As shown in this figure, there is a case where the call is multiplexed to the output line number 4 together with the call of the other number 1 like the call of the VCI number 3, for example, but the call on the next output line, the number 10 and 8, respectively.
Is converted one-to-one.

Generally, the switching controller automatically assigns unused call numbers to calls on output lines.

The next node circuit 100A has a communication line 275 connected to the node circuit 100.
And receives a packet having the header information S2. In this node circuit 100A, processing similar to that performed in the previous node circuit 100 is performed by the switching network 150A, the switching control device 251A, and the traffic control device 240A.
And another link table L2: (12; 1
3) is set in the switching network 150A. Further, the traffic control device 240A rewrites the header information of the packet into S3 which means a signal packet from the node circuit to the packet terminal, and outputs the packet to the obtained output line via the switching control device 251A and the switching network 150A. Output.

The packet terminal 290 on the called side connects the communication line 2 to the node circuit 100A.
Connected via 85. The packet terminal 290 receives the packet having the header information S3, and
It has S3 and returns another packet with the traffic information of the packet terminal 290.

Thus, on the return trip, the link table L3: (11; 12)
And the link table L4: (12; 13) are set in the switching networks 150 and 150A, respectively.

FIG. 5 (a) is a diagram showing a state when data transmission is performed using these link tables. Once the link tables are set up on the communication path, they need not be rewritten until the next call occurs. Therefore,
These switching control devices 251, 251A and traffic control devices 240, 240A are released from processing in the subsequent data transmission.

During data transmission, the switching networks 150, 150A automatically rewrite the header information into address information from the calling side to the called side according to the obtained link tables L1, L2, L3, L4. For example, as shown in FIG.
The calling-side packet terminal 260 having the header information transmits a data packet having the header information I1. This header information is automatically rewritten on the communication path. Finally, this packet is sent to the packet terminal 290 on the called side, and the header information 13
Is received as a packet having

In this way, the optimum output line is selected by the above processing, and the switching control device and the traffic control device are released from the processing during the data transmission, thereby realizing high-speed data transmission corresponding to high-speed data burst data transmission in multimedia communication. it can.

Hereinafter, a more detailed configuration and functions of the embodiment of the present invention will be described with reference to FIGS. 6 (a), (b), 7 and 8.

If there is not much difference between the above-mentioned effective bit rate and the actual bit rate, it is possible to fix the effective bit rate at the average bit rate previously determined by the calling packet terminal.

However, when there is a large difference between the effective bit rate and the actual bit rate, it is necessary to adjust the effective bit rate to prevent overflow due to a calculation error and wasteful communication efficiency due to underestimation.

In order to solve such a situation, the node circuit of the present invention includes a traffic measuring circuit for measuring the actual input traffic, and adjusts the effective traffic of a subsequent call.

Also, since there are two different packet formats, two different traffic measurement devices are expected.

FIG. 7 is an example of a traffic measurement device corresponding to a fixed-length packet format, and FIG. 8 is an example of a traffic measurement device corresponding to a variable-length packet format.

In FIG. 7, a communication line 435 corresponds to one of the input lines 11-1 to 11-m shown in FIG. The shift register 430 holds the bit string input from the communication line 435 and transfers it to the comparator 440. Before that, some bucket patterns are written by the CPU 470 in the memory 450 (RAM).

In the comparator 440, the input bit string is assembled in a packet format and compared with the pattern described in the RAM 450 in advance. This is because there are not only empty packets but also other packets to be discarded without being counted in the node circuit.

When the comparator 440 detects a packet to be counted,
The comparator 440 controls the counter memory 460, and stores 1
Add. This measurement is performed for all calls identified by the VCI, and a plurality of measurement results are stored in the counter memory 460.

The control memory 480 is composed of a normal ROM and RAM combined with the CPU 470, and stores control software of the traffic measuring device and necessary data, for example, a pattern of a countable packet.

In this embodiment, the control memory 480 further stores unit time information for packet measurement.

The CPU 470 reads the contents of the counter memory 460 and the unit time information from the memory 480 at a predetermined cycle, and calculates the number of received packets per unit time for each call. CPU 470 sends this number of packets to traffic controller 240 for each call.

FIG. 6A is a block diagram of the traffic control device 240. The traffic control device 240 includes a central control device (hereinafter, CPU) 340, a read only memory (hereinafter, referred to as CPU).
ROM) 350, random access memory (hereinafter RAM) 360,
Inputs for traffic measurement devices 12-1 to 12-m /
It comprises output interfaces 370 and 380, input / output interfaces 390 and 400 for the exchange controller 251 and input / output interfaces 410 and 420 for the terminal 220. The above processing is based on the control software in the ROM 350 and the RAM 3
Executed within 60.

The traffic control device 240 receives the signal packet from the switching control device 251 via the input interface 390, processes it by the CPU 340, and stores the result in the RAM 360.
When the CPU 340 receives and processes the signal packet,
The U340 prepares an adjustment table as shown in FIG. 6 (b) in the memory, and stores the input line number and the obtained effective bit rate. Further, the CPU 340 receives the number of packets per unit time described above, and calculates an actual bit rate and an adjustment coefficient for matching the two. The adjustment coefficient may be, for example, AF or a numerical value such as (0 <AF <1) to decrease the effective bit rate.

The obtained adjustment factor is used in the setting stage of the next call.
For example, the traffic control device 240
May be configured to refer to the adjustment table corresponding to the input line number and to multiply the obtained effective bit rate in the effective bit rate table by the adjustment coefficient in the adjustment table.

Also, the actual bit rate of the packet having the VCI to be observed is observed, and the effective bit rate in Fig. 2 (a) is updated with the observed value. By doing so, it is possible to increase the accuracy of estimating the effective bit rate for traffic of data having the same characteristics (maximum value, average value).

FIG. 8 is a block diagram of a traffic measuring device, which corresponds to a variable length packet format. In FIG. 8, a communication line 485 corresponds to one of the input lines s11-1 to 11-m in FIG. Input buffer 490 is communication line 4
Take the bit string from 85 and transfer it to CPU500.

In this case, since the format of the discarded packet to be applied is only the flag sequence, a memory for storing a predetermined pattern and a comparator for comparing and detecting the same are unnecessary. Therefore, the CPU 500 can easily identify the flag sequence, and easily identify the packets to be counted.

On the other hand, since the packet length is variable up to 4096 octets,
In order to measure traffic accurately, we must count the packet length.

When the CPU 500 recognizes a packet to be counted, it adds “1” to the counter memory 510 and starts measuring the octet length of the packet. When the CPU 500 finishes measuring the octet length, the value is stored in the counter memory 510 in the same manner. The unit time information is stored in the control memory 520 similarly to the traffic measuring device shown in FIG. In addition, the CPU 500 uses LCG to identify each call.
It is assumed that N and LCN are extracted from the packet.

At a predetermined cycle, the CPU 500 reads the number of packets, the octet length, and the unit time information, and calculates the actual bit rate for each call. CPU 500 sends the result to traffic control device 240. Thereafter, the traffic control device 240 performs the same operation as that described in the traffic measurement device of FIG.

The present invention is not particularly limited to the data transmission of the virtual call system, but can be applied to a permanent virtual call system and the like. For example, the maximum bit rate information and average bit rate
The data can also be input from 220 to the memory of the traffic measurement device 240.

Further, it is possible to introduce another unit to represent traffic. For example, the actual number of packets per unit time can be used instead of the bit rate.

Furthermore, it is possible to use other types of packets, packets at different locations, for transmitting traffic information. For example, in the case of a fixed-length packet format, a CR packet may be provided with specific data to be used as a signal packet.
In the case of a variable-length packet format, a packet having a control field “C” having a specific value may be used as a signal packet.

Further, if necessary, the traffic control device of the present invention
240 is set away from the body of the node circuit 100,
Remote control may be performed via a communication line.

(Effect of the Invention) According to the present invention, the following effects are expected by observing actual communication and updating the effective band to be used. First, data loss due to buffer overflow in the communication processing node is minimized, and high quality communication is provided. In addition, it is possible to avoid setting a band having an excessive margin for actual communication, so that the use efficiency of the output line is increased. Therefore, the throughput in the communication processing node is improved.

[Brief description of the drawings]

FIG. 1 (a) is a schematic block diagram of a node circuit according to an embodiment of the present invention, FIG. 1 (b) is a diagram for explaining a connection mode of the node circuit according to the embodiment, and FIGS. (B) is a diagram for explaining how the routing method of the present invention handles traffic information in a node circuit, and FIGS. 3 (a) and (b) describe packet formats applicable to the present invention. FIG. 4 is a diagram for explaining how the present invention determines a communication path, and FIG.
FIG. 5B is a diagram for explaining how a node circuit of the present invention handles a packet during communication, FIG. 5B is a link table created in the present invention, and FIG. Block diagram of applied traffic control device, sixth
6B is an adjustment table stored in the traffic control device of FIG. 6A, FIG. 7 is a traffic measurement device of the present invention used for a fixed-length packet format, and FIG. 8 is a variable-length packet format 9 (a), 9 (b) and 9 (C) are diagrams for explaining burst data transmission. FIG. 12; traffic measurement device, 100, 102; node circuit, 150;
Switching network, 240; traffic controller, 251; switching controller.

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁶ , DB name) H04L 12 / 56,12 / 28

Claims (4)

(57) [Claims] A plurality of input lines and a plurality of output lines;
A node circuit used for packet communication, comprising: a switching network for storing a combination of the input line and the output line as a link table; and an input line and an output line of the combination based on header information of the packet. A switching network including a means for connecting the; and a packet observing means connected to the input line,
Means for observing and storing actual traffic of the input line; connected to the switching network and the traffic observing means; when a first call arrives at one input line, estimated traffic of the call And calculating a first value representing a traffic capacity required for the call, storing the first value in a memory, and utilizing at least a part of the first value to calculate the plurality of From the output lines, one of the output lines having a sufficient traffic capacity necessary for the call is selected based on the traffic information, and a combination of the input line and the selected output line is linked to a switching network. Instructs it to be stored as a table, followed by a call with the same estimated traffic on its input line based on the actual traffic stored. Node circuit, characterized in that said first value when having a means having a function of adjusting the second value. 2. A routing method for transmitting a packet from a calling terminal to a called terminal from an input line to an output line in a packet switching network, wherein data representing an estimate of call traffic to be transmitted is transmitted. Receiving the call through an input line having a node circuit, and calculating and storing a first value representing a traffic capacity required for the call from the estimate in the node circuit; An output line having a traffic capacity sufficient to transmit the call is selected from the plurality of output lines following the above based on the calculation, and a combination of the input line and the selected output line is stored as a link table. Connecting the combination of the input line and the output line based on the stored link table, and including traffic necessary for the call in the call. Sending a call to the node circuit, the call including a burst data transmission following a signaling packet, the estimated traffic being an estimated average bit rate, and a maximum bit rate; A routing method for observing actual traffic during data transmission and calculating the required traffic capacity for the next call based on the traffic observed in the node circuit for the next call. 3. The routing method according to claim 2, wherein a maximum bit rate table having a first address at an address corresponding to the value of the maximum bit rate is provided, and the estimated average bit rate is provided at the first address. And an effective bit rate table storing an effective bit rate corresponding thereto is provided. The maximum bit rate table is accessed from the estimated maximum bit rate in the signal packet to determine a first address. An effective bit rate table is accessed, a corresponding effective bit rate is obtained from an estimated average bit rate in a signal packet, and the obtained effective bit rate is used as a first value representing the required traffic capacity. And the routing method. 4. The routing method according to claim 3, wherein said selecting step comprises providing a characteristic table for each called terminal, and in said table, a maximum bit rate of an output line to each terminal and a used bit rate. , Store the line identifier and store it in the address of the corresponding memory, set up a destination table that can access the address of each characteristic table by destination, recognize the terminal to which the call is directed, and access the destination table To obtain the address of the characteristic table, thereby accessing the characteristic table to find an output line in which the difference between the stored maximum bit rate and the currently used bit rate is equal to or greater than the determined effective bit rate. A routing method, characterized in that: