JP3000545B2 - Congestion control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a congestion control method, and more particularly, to a congestion control method for changing a PDU size of a transport layer of data to be transmitted in response to a congestion occurrence notification from a network.

[0002]

2. Description of the Related Art In general, an unreliable protocol without error control is used for a physical layer (first layer), a data link layer (second layer), and a network layer (third layer) in an OSI reference model. When data communication is performed using a reliable protocol that performs error control in a transport layer (fourth layer), data retransmission control accompanying a transfer error is realized in the transport layer. When data is transferred from each layer to a lower layer, a PDU (Protocol Data Uni
As the size of (t: protocol data unit), a value set in advance according to the nature of the protocol of each layer and the line quality is used.

[0003] Normally, the maximum PDU size of the lower layer is smaller than the maximum PDU size of the upper layer. In most cases, the maximum PDU size of the lower layer is smaller than the maximum PDU size of the upper layer. . Therefore, at any time of non-congestion, congestion, and data retransmission, data transfer is performed in units of the maximum PDU size set by the protocol used in the transport layer. Is divided and stored in a plurality of PDUs and transferred.

[0004]

Therefore, in such a conventional congestion control method, since a fixed value set in advance is used as the maximum PDU size of the transport layer, data of relatively small PDU size is used. Even if data loss occurs in the lower layer that performs the transfer,
In the transport layer that performs error control, retransmission processing is performed in units of PDUs containing lost data, that is, PDUs having a relatively large size in the transport layer. There is a problem that a much larger amount of data must be retransmitted.

[0005] In an application in a layer higher than the transport layer, data is lost in units of a PDU size of the transport layer which is a fixed value, which makes it difficult to maintain data continuity. was there. The present invention has been made to solve such a problem, and provides a congestion control method that can reduce the amount of retransmitted data when data is lost and that can reduce disturbance of data continuity affecting an application. The purpose is.

[0006]

In order to achieve the above object, a congestion control method according to the present invention provides a method for transferring data from a transport layer to a lower layer in response to notification of occurrence of congestion from a network to a lower layer. A second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit to be used, and the first maximum size of the protocol data unit is set as the maximum size of the protocol data unit in response to the congestion release notification from the network to the lower layer. The maximum size is used. Also, the first maximum size is used as the maximum size of the protocol data unit after a lapse of a predetermined time from the occurrence of congestion, instead of the notification of congestion release from the network to the lower layer. Therefore, when the network is congested, a second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit.

[0007] When performing congestion control based on the congestion state of each layer lower than the transport layer,
As the second maximum size, a size corresponding to a protocol used in a corresponding lower layer is used. Therefore, the second maximum size corresponding to each lower layer that has become congested is used as the maximum size of the protocol data unit. Further, when a congestion state occurs in a plurality of lower layers, the smallest of the second maximum sizes corresponding to all the lower layers in the congestion state is used as the maximum size of the protocol data unit. It was done. Therefore, when a congestion state occurs in a plurality of lower layers, the smallest second maximum size among them is used as the maximum size of the protocol data unit.

When a lower layer receives an RM cell indicating the occurrence of congestion from the ATM network, the maximum size of a protocol data unit used for data transfer from the transport layer to the lower layer is set as the maximum size of the protocol data unit in non-congestion. When the second maximum size smaller than the maximum size of 1 is used and an RM cell indicating congestion release is received from the ATM network in the lower layer, the first maximum size is used as the maximum size of the protocol data unit. It was done. Therefore, a second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit in response to the reception of the RM cell indicating the occurrence of congestion from the ATM network, and the RM indicating the congestion release is used. A first maximum size is used in response to receiving a cell.

[0009] When a frame in which the BECN bit is set to "1" is received from the frame relay network in the lower layer, or a CLLM frame indicating the occurrence of congestion is received, the transport layer shifts from the transport layer to the lower layer. As the maximum size of the protocol data unit used for the data transfer of the second embodiment, a second maximum size smaller than the first maximum size during non-congestion is used, and a frame in which the BECN bit is not set to “1” in the lower layer is received. In this case, or when a predetermined period has elapsed after receiving the CLLM frame indicating the congestion state, the first maximum size is used as the maximum size of the protocol data unit. Therefore, as the maximum size of the protocol data unit, in response to the reception of a frame in which the BECN bit is set to “1” from the frame relay network or the CLLM frame indicating the occurrence of congestion, the first maximum size at the time of non-congestion is set. CL indicating the reception of a frame in which a small second maximum size is used and the BECN bit is not set to “1”, or a congestion state
The first maximum size is used according to the lapse of a predetermined period from the reception of the LM frame.

When the source quench packet is received from the IP network in the lower layer, the maximum size of the protocol data unit used for data transfer from the transport layer to the lower layer is set to the first maximum value in non-congestion. A second maximum size smaller than the size is used, and when a predetermined period has elapsed since the last source quench packet was received in the lower layer, the first maximum size is used as the maximum size of the protocol data unit. . Therefore, as the maximum size of the protocol data unit, a second maximum size smaller than the first maximum size during non-congestion is used according to the reception of the source quench packet from the IP network, The first maximum size is used as the predetermined period elapses.

[0011]

Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a terminal according to a congestion control method according to an embodiment of the present invention. In the figure, reference numeral 11 denotes an application for communication, 12 a transport layer protocol processing unit for controlling a protocol of a transport layer, 13 a lower layer protocol 2 processing unit for controlling a protocol of a layer lower than the transport layer, 14 Is a lower layer protocol 1 processing unit that performs lower layer protocol control.

Reference numeral 15 denotes a lower layer protocol 2 processing unit 13
And the maximum value of the PDU size at the time of data transfer in the transport layer, that is, the maximum PDU size (hereinafter, referred to as PDU size) according to the congestion state notified from each protocol of the lower layer protocol 1 processing unit 14 and managed. PDU size management unit. The PDU size management unit 15 stores the PDU size
And a DU size management table 17.

FIGS. 2A and 2B are explanatory diagrams showing these tables. FIG. 2A is a PDU size index table, and FIG.
It is a U size management table. In a PDU size index table (hereinafter referred to as an index table) 16, a PDU size at the time of non-congestion and an optimum PDU size at the time of congestion corresponding to each protocol type are set in advance for each line. The PDU size management table (hereinafter referred to as a management table) 16 stores, for each port number, the current PDU size currently used by the transport layer protocol processing unit 12 and the individual PDU size used in the lower layer. And a flag indicating a congestion state are set in accordance with the protocol type.

The port number referred to here is an identifier for identifying data to be transferred between the transport layer protocol processing unit 12 and each lower layer protocol processing unit. Out of the connections, it can be identified that the data is related to a predetermined connection. The protocol type is a type of various protocols used in the lower layer protocol 2 processing unit and the lower layer protocol 1 processing unit 14. Since different protocol types are used in the lower layer protocol 2 processing unit 13 and the lower layer protocol 1 processing unit 14, a plurality of protocol types are used in combination for one connection, that is, a port number.

Therefore, the index table 16 contains, for each of the physical lines 1 to n managed by the lower layer protocol 1 processing unit, the PDU size and PDU _def at the time of non-congestion determined by the line characteristics. 1 to PDU _def n, and the congestion PDU size of each protocol type having a value smaller than the non-congestion PDU size, PDU11 to PD
Umn is set in advance. Also, the management table 17
Includes a current PDU size currently set for each port number currently used by the transport layer protocol processing unit 12, flags 11 and 12 indicating the congestion state of each protocol type in the lower layer. Is updated. In the example of FIG. 2B, the protocol type 1 and the protocol type k are used in the lower layer for the port number 1, and the protocol type 1 and the protocol type 2 in the lower layer for the port number 2. Are used, and a flag indicating a congestion state is set in each column.

Next, the operation of the present invention will be described with reference to FIG. FIG. 3 is a flowchart showing the operation of the PDU size management unit 15, in which (a) is a connection setting notification reception process, (b) is a congestion notification reception process, (c) is a congestion release reception process, and (d) is a connection. This is a disconnection notification receiving process. First, at the start of communication, the transport layer protocol processing unit 12 notifies that a connection to which a predetermined port number has been assigned in the transport layer has been set, and the protocol type and line number used for this connection are: It is notified from the lower layer protocol 2 processing unit 13 and the lower layer protocol 1 processing unit 14.

In response, the PDU size management unit 15 executes the connection setting notification receiving process shown in FIG. First, the port number of the set connection,
The protocol type and the line number are obtained (steps 101 and 102), and the optimum maximum PDU size at the time of non-congestion is searched from the index table 16 based on the line number and the protocol type of the connection (step 10).
3).

Subsequently, a management area corresponding to the newly set connection is added to the management table 17,
The maximum PDU size searched from the index table 16 is registered corresponding to the port number indicating the connection (step 104). At the same time, a flag indicating the non-congestion state is registered as an initial value in the column of the protocol type used by the lower layer protocol 2 processing unit 13 and the lower layer protocol 1 processing unit 14 as the port number (step 104). Then, the series of connection setting notification reception processing ends.

Thereafter, the lower layer protocol 2 processing unit 1
When the congestion is detected and notified by the lower layer protocol 1 processing unit 14 or the congestion is notified, the PDU size management unit 15 executes the congestion notification reception process of FIG. First, a PDU size at the time of congestion indicating a value smaller than that at the time of non-congestion is searched from the index table 16 according to the line in which congestion has occurred and its protocol, and is set as a candidate PDU size (step 111). Next, management table 1
7 to check whether there is any protocol type corresponding to the same port number in which congestion has occurred and in which congestion has occurred (step 11).
2).

If congestion occurs in another protocol type (step 112: YES), the current PDU size is compared with the PDU size retrieved in step 111, and the smaller of the two is compared. The PDU size is set as a candidate PDU size (step 113). After this step 113, and when no other congestion has occurred in step 112 (step: NO), the candidate PDU size obtained in each case is registered as the current PDU size, and the port notified of the occurrence of congestion is registered. Register a flag indicating the occurrence of congestion in the column of the protocol type of the number,
The management table 17 is updated (Step 114).

Further, the new current PDU size is notified to the transport layer protocol processing unit 12, and the P is transferred from the transport layer to the lower layer.
An instruction to change the size of the DU is issued. As a result, the size of the PDU transferred from the transport layer to the lower layer is temporarily reduced as compared with the non-congestion state, and even if the congestion state is further deteriorated and data loss occurs, retransmission is performed in the transport layer. Since the number of PDUs to be transmitted is small, the amount of data actually retransmitted in the lower layer is reduced.

After step 115, the PDU size management unit 15 determines whether or not the congested notification protocol has a congestion release notification function (step 116). (Step 11
6: NO), activates the congestion release timer that times out after a predetermined period, and ends a series of congestion notification reception processing.
On the other hand, if it has the congestion release notification function (step 116: YES), the congestion notification reception processing is terminated as it is.

After the congestion notification receiving process has been executed in this way, if the lower layer protocol 2 processing unit 13 or the lower layer protocol 1 processing unit 14 detects the release of the congestion state and is notified of the congestion release, the PDU In the size management unit 15, the congestion release notification receiving process shown in FIG. Note that the congestion release notification receiving process is also performed when the above-described congestion release timer times out.

First, referring to each flag in the management table 17, whether any of the protocol types corresponding to the same port number from which congestion has been released is not informed of the release of congestion but is in a state of occurrence of congestion. Check if (Step 1
21). If there is another congestion-producing protocol type (step 121: YES), the smallest PDU size is searched from the index table 16 among the congestion-producing protocol types having the same port number. The size is set as a candidate PDU size (step 122).

On the other hand, when there is no other protocol type in which congestion is occurring (step 121: NO), the PDU size at the time of non-congestion corresponding to the port number is stored in the index table 1.
6 and set as a candidate PDU size (step 12
3). Thereafter, the candidate PDU size obtained in step 122 or step 123 is registered as the current PDU size, and a flag indicating the non-congestion state is registered in the protocol type column of the port number notified of the congestion release, and the management is performed. The table 17 is updated (Step 124).

Further, the new current PDU size is notified to the transport layer protocol processing unit 12, and the P is transferred from the transport layer to the lower layer.
An instruction to change the size of the DU is issued, and the series of congestion disclosure notification reception processing ends. As a result, when there is no other protocol type in which congestion has occurred, the size of the PDU transferred from the transport layer to the lower layer is returned from a small size during congestion to a relatively large original size during non-congestion. . If there is another protocol type in which congestion is occurring, the PDU size is changed to the optimum PDU size for that protocol type.

Thereafter, in response to the end of communication in the upper-level application 11, a notification of disconnection is sent from the transport layer protocol processing 12. In response to this, as shown in FIG. 3D, the PDU size management unit 15 deletes the management area of the port number corresponding to the notified disconnected connection from the management table 17 as the connection disconnection notification receiving process (step S3). 131),
The process ends. If the congestion release timer has been activated, the corresponding congestion release timer is cleared in step 131.

[0028]

Next, an embodiment of the present invention will be described with reference to FIG. FIG. 4 is a sequence diagram showing an embodiment according to the congestion control method of the present invention. As a transmission side, a transport layer protocol processing unit 4 for performing congestion control
1, a lower layer protocol processing unit 43 as a lower layer thereof, and a PDU size management unit 42 for managing the PDU size of the transport layer. As a receiving side, a lower layer protocol processor 45 and a transport layer protocol processor 46 for performing congestion control as an upper layer are provided, and these are connected via a network 44.

Now, it is assumed that the transport layer protocol processing unit 41 on the transmitting side and the transport protocol processing unit 46 on the receiving side are transferring data using the maximum PDU size “7” during non-congestion. I do. In this case, it is assumed that no congestion has occurred in any of the protocol processing units. Therefore, the transport layer protocol processing unit 41 divides the data from the upper application (not shown) using “7” as the maximum PDU size, generates the transfer data 51, and sends the transfer data 51 to the lower layer protocol processing unit 43. Hand over.

In response, the lower layer protocol processing unit 43 receives the transfer data 51, divides it, and
It is stored in a PDU of DU size “1” and transferred to the receiving side via the network 44. The PDU size here, for example, “7” does not indicate a value based on a physical unit, but is transmitted via the network 44 to the lowest lower layer protocol 43.
Are relative values where the size of the PDU exchanged at “1” is “1”.

The protocol used in each lower layer is such that a header composed of information for its own protocol processing is added to the data received from the upper layer and passed to the lower layer. Say p
The DU size ratio and the actual PDU size ratio are slightly different. Here, the transfer data (PDU size = “7”) from the transport layer protocol processing unit 41 is
It is assumed that the lower layer protocol processing unit 43 divides the data into seven pieces of transfer data 52 (PDU size = “1”) and transfers them to the receiving side.

The lower layer protocol processing unit 45 on the receiving side
Are the seven transfer data 52 from the transmission side via the network 44.
Are received, and are transferred to one transfer data 53 (PD
(U size = “7”) and passed to the upper protocol layer processing unit 45. Accordingly, the transport layer protocol processing unit 46 receives the transfer data 46 and sequentially transfers the data to a higher-level application (not shown). Thus, when a congestion state occurs in the network 44 while data is being transferred from the transmission side to the reception side, the network 44 sends the congestion information 55 to the lower layer protocol processing unit 43 on the transmission side. Then, the occurrence of the congestion state is notified.

Thus, the lower layer protocol processing unit 4
The PDU size management unit 42 uses the congestion occurrence notification 56 to indicate the line number in which the congestion state has occurred and its own protocol type.
Notify. In response to the congestion occurrence notification 56, the PDU size management unit 42 executes the congestion occurrence notification reception process shown in FIG. A new maximum PDU size “2” smaller than the PDU size “7” is newly obtained, and a PDU size change instruction 57 is notified to the transport layer protocol processing unit 41.

In response, the transport layer protocol processing unit 41 changes the current PDU size from “7” to “2” and changes the size of the transfer data 61 to be passed to the lower layer protocol processing unit 43 Make it smaller than during congestion. Therefore, the lower layer protocol processing unit 4
In step 3, the transfer data 61 is received and divided, and its own P
The data is stored in two transfer data 62 of DU size “1” and transferred to the receiving side via the network 44.

As a result, the lower layer protocol processing section 45 on the receiving side receives the two transfer data 62 from the transmitting side via the network 44 and transfers them to one transfer data 63 (PDU size = “2”). ) And passed to the upper protocol layer processing unit 45. Therefore, when there is no congestion, the transport layer protocol processing unit 4
In the lower layer protocol processing unit 43, the transfer data 51, which is a unit of retransmission processing in
In the case of congestion,
The transfer data 61 is divided into “two” transfer data 62 and transferred.

Here, it is assumed that the congestion state worsens and a part of the transfer data 65 constituting the transfer data 64 from the transport layer protocol processing unit 41 is lost. Transport layer protocol processing unit 4 on the receiving side
6, for example, upon receiving the next transfer data 66, detects the loss of the immediately preceding transfer data 65 and transmits a retransmission request 67 to the transmitting side. The transport layer processing unit 41 on the transmitting side detects the retransmission request 67 and
The lost data is transferred to the lower layer protocol processing unit 43 as the transfer data 71 by using the DU size, in this case, the PDU size “2” at the time of congestion.

In response to this, the lower layer protocol processing unit 43 divides the transfer data 71 into two pieces of transfer data 72 using the PDU size “1” and transfers them to the receiving side.
As a result, the transfer data 72 is received by the lower layer protocol processing unit 45 on the receiving side, passed to the transport layer protocol processing unit 46 as the transfer data 73, and the lost data is compensated for. Thereafter, when the congestion state is released, the network 44 sends the congestion release information 75 to notify the lower layer protocol processing unit 43 on the transmission side of the release of the congestion state.

Thus, the lower layer protocol processing unit 4
The PDU size management unit 4 uses the congestion release notification 76 to indicate the line number from which the congestion state has been released and its own protocol type.
Notify 2. In response to the congestion release notification, the PDU size management unit 42 executes the congestion release notification receiving process shown in FIG. 3C to determine the PDU size optimal for the state after the congestion release. In this case, since there is no other protocol in the congestion state, the PDU size “7” at the time of non-congestion is newly obtained, and the PDU size change instruction 77 is notified to the transport layer protocol processing unit 41.

In response, the transport layer protocol processing unit 41 changes the current PDU size from “2” to “7” and changes the size of the transfer data 61 to be passed to the lower layer protocol processing unit 43 to congestion. Return to the original size, which is sometimes large. Therefore, the lower layer protocol processing unit 43 receives the transfer data 81 and divides the transfer data 81 into seven transfer data 82 of its own PDU size “1”.
And transfer it to the receiving side via the network 44.

As described above, when congestion occurs, the transfer data length transferred from the transport layer protocol processing unit 41 on the transmission side to the lower layer, that is, the PDU size is “2” as compared with “7” during non-congestion. Even when the congestion state deteriorates and the lower layer transfer data is lost, the PDU size in the transport layer including the lost transfer data is "2". Retransmission data to be retransmitted is smaller than the size “7”.

Therefore, the amount of retransmission data at the time of data loss can be reduced and the time required for retransmission processing can be reduced as compared with the conventional case where a fixed PDU size is used both during non-congestion and when congestion occurs. It can be shortened. This makes it possible to reduce the disruption of data continuity affecting the application. For example, even when transferring image data or audio data in which data continuity is regarded as important as transfer data, the image by retransmission processing when congestion occurs And the occurrence of audio breaks can be reduced.

Next, an example in which a general protocol is used as a lower layer protocol will be described.
First, ATM (Asynch
The case where a (ronous Transmit Mode) protocol is used will be described. FIGS. 5A and 5B are explanatory diagrams showing congestion control when the ATM protocol is used. FIG. 5A shows a network configuration example, and FIG. 5B shows an ATM cell configuration.

In the ATM protocol, an RM cell is provided as a resource management cell, which is periodically transmitted from the transmitting terminal 1 to the receiving terminal 4 like other data cells, and further transmitted from the receiving terminal 4 to the transmitting terminal 1. By returning the information, a congestion state occurring in the network is notified. Actually, a cell in which “110” is stored in the PT (Payload Type) 157 provided in the header 155 of the ATM cell is an RM cell, and a CI (Congestion Indicatio) provided in the payload 156 of the RM cell.
n) An information bit indicating “non-congestion / congestion” is stored in 158 and notified.

For example, when congestion occurs in the link between the relay nodes 2 and 3, the relay node 2 sends the data cell 151 from the transmitting terminal 1 its EFCI (Ex
A plicit Forward Congestion Indication (not shown) bit is set and the data cell 152
Is transferred to the receiving terminal 4 side. The receiving terminal 4 receives the data cell 152, and sets the convergence to the CI 158 of the RM cell 153 and returns it to the transmitting terminal 1 because the EFCI bit is set. 2 via the transmission terminal 1.

Therefore, the lower layer protocol processing unit of the transmitting terminal 1 using the ATM protocol checks the PT157 of the received ATM cell and
Determine the M cell. Here, the CI 158 of the RM cell
Is checked to indicate "congestion", the PDU size management unit 15 (see FIG. 1) is notified of congestion. Thereby, during congestion, the PDU size in the transport layer of the transmitting terminal 1 temporarily decreases.

When the congestion is released, “non-congestion” is set in the CI 158 of the RM cell 151 at the relay node 2. Therefore, transmitting terminal 1 receives RM cell 152 returned from receiving terminal 4, and receives its CI 158
Indicates “non-congestion”, the PDU size management unit 15
To the congestion release. As a result, the PDU size in the transport layer of the transmitting terminal 1 is restored.

Next, a case where a frame relay (Frame Relay) is used as a lower layer protocol will be described. FIG. 5 is an explanatory diagram showing congestion control when using frame relay, where (a) is a network configuration example,
(B) shows a frame configuration.

In the frame relay, BECN (Back) is added to the address portion 165 of the frame used as a data transfer unit.
ward Explicit Congestion Notification) 166 is provided, and BECN 166 indicates “non-congestion /
An information bit indicating "congestion" is stored and transmitted to the transmitting terminal 1 from the frame relay network. In this case, not only the frames from the receiving terminal 4 communicating with the transmitting terminal 1 to the transmitting terminal 1 but also all the frames from other terminals toward the transmitting terminal 1 are used to be notified from the frame relay network.

For example, when congestion occurs in the link between the relay nodes 2 and 3, “congestion” is set in the BECN 166 of the frame 162 transmitted from the receiving terminal 4 by the relay node 2, and the frame 163 As sending terminal 1
Is forwarded to The lower layer protocol processing unit of the transmitting terminal 1 using the frame relay protocol checks the BECN 166 of the received frame and indicates "congestion". To notify congestion. As a result, the PDU size in the transport layer of the transmitting terminal 1 is temporarily reduced.

When the congestion is released, “non-congestion” is set in BECN 166 of frame 162 at relay node 2. Therefore, the transmitting terminal 1 notifies the PDU size management unit 15 of the congestion release because the BECN 166 of the received frame 162 indicates “non-congestion”. As a result, the PDU size in the transport layer of the transmitting terminal 1 is restored.

In the frame relay, besides the method using BECN as a method of notifying a transmitting terminal of congestion, a CLLM (Consolidated Link Ladder) as shown in FIG.
yerManagement: There is a method using an integrated link layer management frame. This message is used when the frame relay network notifies the terminal of various types of management information.
67 to notify the congestion state.

In this case, the relay node 2 generates a CLLM frame in which the reason indicating “congestion” is stored in the reason display 167 according to the occurrence of congestion, and transmits the frame to the transmitting terminal 1. The transmitting terminal 1 displays the reason indication 16 of the received CLLM frame.
7 is checked, and if "congestion" is indicated, the PDU size management unit 15 is notified. CLL indicating congestion release
In this case, since the M frame is not provided, the congestion release timer is started as shown in step 117 of FIG. 3B, and FIG.
Is executed.

Next, as a lower layer protocol, I
The case where P (Internet Protocol) is used will be described. FIGS. 5A and 5B are explanatory diagrams showing congestion control when IP is used. FIG. 5A shows an example of a network configuration, and FIG. 5B shows an SQ message configuration. The IP has a protocol for notifying a transmission terminal or the like of a failure that has occurred at the IP level as an Internetwork Control Message Protocol (ICMP).
tocol).

In this ICMP, an SQ (Source Quench) packet is provided as a kind of ICMP packet, and a congestion state is notified to a transmitting terminal.
Actually, Type 1 of the ICPM field 175
An ICMP packet in which 76 = “4” and Code 177 = “0” are set is recognized as an SQ packet.

For example, when congestion has occurred in the link between the relay nodes 2 and 3, the relay node 2 generates an SQ packet 172 and transmits it to the transmitting terminal 1. Sending terminal 1
Checks the Type 176 and the Code 177 among the received ICMP packets, and notifies the PDU size management unit 15 when the SQ packet is confirmed.
In this case, since no packet indicating congestion release is provided, in this case, the congestion release timer is started as shown in step 117 of FIG. Is executed.

[0056]

As described above, according to the present invention, the maximum size of a protocol data unit used for data transfer from the transport layer to the lower layer in response to the notification of occurrence of congestion from the network to the lower layer, Using a second maximum size smaller than the first maximum size of
The first maximum size is used as the maximum size of the protocol data unit in response to the congestion release notification from the network to the lower layer. Therefore, at the time of non-congestion and at the time of occurrence of congestion as in the related art, compared with the case of using a fixed size as the size of the protocol data unit, it is possible to reduce the retransmission data amount at the time of data loss, and to perform retransmission processing. The required time can be reduced. As a result, even if data loss occurs due to deterioration of the congestion state, since the amount of lost data is reduced, disruption of data continuity affecting applications can be reduced. For example, data continuity is regarded as important as transfer data. Even when image data, audio data, and the like to be transmitted are being transferred, it is possible to reduce the occurrence of breaks in images and audio due to retransmission processing when congestion occurs.

Also, instead of notifying the lower layer of the congestion release from the network, the first maximum size is used as the maximum size of the protocol data unit after a lapse of a predetermined time from the occurrence of congestion. Even if a given protocol does not have a means for notifying congestion release, the maximum size of the protocol data unit can be reliably returned to the size at the time of non-congestion.

When performing congestion control based on the congestion state of each layer lower than the transport layer,
As the second maximum size, a size corresponding to the protocol used in the corresponding lower layer is used, so even if congestion occurs in any of the lower layers among a plurality of lower layers using different protocols,
Data transfer can be performed with the maximum size of the protocol data unit optimal for various protocols used in the lower layer.

When a congestion state occurs in a plurality of lower layers, the smallest one of the second maximum sizes corresponding to all the lower layers in the congestion state is set as the maximum size of the protocol data unit. Since it is used, a maximum size smaller than that at the time of non-congestion is used for any lower layer, and in all lower layers in a congested state, the amount of retransmitted data at the time of data loss can be reduced and In addition, the time required for the retransmission processing can be shortened, and the disruption of data continuity affecting the application can be reduced.

As the maximum size of the protocol data unit, a second maximum size smaller than the first maximum size at the time of non-congestion is used in response to the reception of the RM cell indicating the occurrence of congestion from the ATM network. RM shown
The first maximum size is used in response to cell reception. In addition, the maximum size of the protocol data unit is the BE from the frame relay network.
In response to receiving a frame in which the CN bit is set to “1” or a CLLM frame indicating occurrence of congestion, a second maximum size smaller than the first maximum size in non-congestion is used, and the BECN bit is set to “1”. The first maximum size is used in accordance with a lapse of a predetermined period from the reception of a frame for which no is set or a CLLM frame indicating congestion.

As the maximum size of the protocol data unit, a second maximum size smaller than the first maximum size during non-congestion is used according to the reception of the source quench packet from the IP network. The first maximum size is used in accordance with a lapse of a predetermined period from reception. Therefore,
Even when ATM, frame relay, or IP is used as a protocol used in the lower layer, the amount of retransmission data when data is lost can be reliably reduced, and the time required for retransmission processing can be reduced.
Disruption of data continuity affecting applications can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of a communication terminal according to a congestion control method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an index table and a management table.

FIG. 3 is a flowchart illustrating a processing operation of a PDU size management unit.

FIG. 4 is a sequence diagram showing one embodiment according to the congestion control method of the present invention.

FIG. 5 is an explanatory diagram showing congestion control using the ATM protocol.

FIG. 6 is an explanatory diagram showing congestion control using a frame relay.

FIG. 7 is an explanatory diagram showing congestion control using IP.

[Explanation of symbols]

11 application, 12 transport layer protocol processing unit, 13 lower layer protocol 1 processing unit, 14 lower layer protocol 2 processing unit, 15 PD
U size management unit, 16 PDU size index table, 1
7. PDU size management table

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-7-283835 (JP, A) JP-A 9-284338 (JP, A) JP-A 8-251226 (JP, A) JP-A 5- 91144 (JP, A) IEICE Technical Report CQ 96-25 NTT R & D Vol. 45 No. 10 1996 p. 1041-1048 NTT Technical Journal 1996.4 p. 100-103 (58) Field surveyed (Int. Cl. ⁷ , DB name) H04L 12/56 H04L 12/28 INSPEC (DIALOG) JICST file (JOIS)

Claims (7)

(57) [Claims] 1. A congestion control method for performing congestion control in a transport layer of an OSI reference model based on a congestion state in a layer lower than the transport layer. Accordingly, a second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit used for data transfer from the transport layer to the lower layer, and a congestion release notification from the network to the lower layer. Congestion control method, wherein the first maximum size is used as the maximum size of the protocol data unit according to 2. The congestion control method according to claim 1, wherein instead of notification of congestion release from the network to a lower layer,
A congestion control method, wherein a first maximum size is used as a maximum size of a protocol data unit after a lapse of a predetermined time from occurrence of congestion. 3. The congestion control method according to claim 1, wherein when performing congestion control based on the congestion state of each layer lower than the transport layer, a corresponding lower layer is set as a second maximum size. Congestion control method, characterized by using a size corresponding to a protocol used in (1). 4. The congestion control method according to claim 3, wherein when a congestion state occurs in a plurality of lower layers, the most constrained one of the second maximum sizes corresponding to all lower layers in the congestion state. A congestion control method, wherein a smaller one is used as a maximum size of a protocol data unit. 5. A congestion control method for performing congestion control in a transport layer of an OSI reference model based on a congestion state in a layer lower than the transport layer, wherein congestion occurs from an ATM network in a lower layer. When an RM cell indicating the following is received, a second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit used for data transfer from the transport layer to the lower layer, A congestion control method characterized by using a first maximum size as a maximum size of a protocol data unit when an RM cell indicating congestion release is received from an ATM network in a lower layer. 6. A congestion control method for performing congestion control in a transport layer of an OSI reference model based on a congestion state in a layer lower than the transport layer.
When a frame in which N bits are set to “1” or a CLLM frame indicating the occurrence of congestion is received, the maximum size of a protocol data unit used for data transfer from the transport layer to a lower layer is defined as: The second smaller than the first maximum size during non-congestion
When the lower layer receives a frame in which the BECN bit is not set to “1”, or indicates a congestion state,
A congestion control method, wherein a first maximum size is used as a maximum size of a protocol data unit when a predetermined period has elapsed from reception of an LLM frame. 7. A congestion control method for performing congestion control in a transport layer of an OSI reference model based on a congestion state in a layer lower than the transport layer. When a packet is received, a second maximum size smaller than the first maximum size during non-congestion is used as the maximum size of the protocol data unit used for data transfer from the transport layer to the lower layer. Congestion control method, wherein a first maximum size is used as a maximum size of a protocol data unit when a predetermined period has elapsed since the last source quench packet was received.