JP3163479B2 - Bandwidth control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a band control method, and more particularly to a band control method for controlling a band of a protocol layer 4 in a node executing a communication protocol process.

[0002]

2. Description of the Related Art At present, the best-effort service is the main service form in the Internet / intranet, and each connection is provided with a band corresponding to a network condition that changes every moment. In TCP, which is a typical communication protocol, a change in network condition is detected based on detection of a packet loss event or the like, and the transmission amount is adjusted.

For this purpose, a reception notification window value (adwnd), which is a parameter indicating the reception capability of the partner node,
In addition, using a congestion window value (cwnd), which is a parameter indicating an estimated value of the transfer capability of the network,
Transmit in the range of in (adwnd, cwnd). However, min (A, B) indicates that the smaller value of A and B is used.

[0004] Bandwidth control for adjusting the transmission amount is cwnd.
And a slow start threshold (ssthresh) in addition to the above parameters. Although packet transmission is performed based on these parameters, there are two phases called a slow start phase and a congestion avoidance phase, and control methods are different.

The outline of the system is as follows.
Although the unit of each parameter of the bandwidth control uses the number of bytes in the implementation of TCP, it is expressed by the number of packets in the following to facilitate the description. First, after the connection is set, the bandwidth control parameter is set to ssthresh = ad
Initialize wnd and cwnd = 1.

Then, a slow start phase is entered.
It transmits the data packets (DT packets) and waits for the reception of the acknowledgment packet (ACK packet). If an ACK packet is received within a certain time, cwnd is incremented by +1.
Then, two packets are transmitted, and thereafter, each time an ACK packet is received, the number of DT packets confirmed is increased until the next transmittable amount, cwnd, reaches the threshold value ssthresh.

Accordingly, the transmittable amount cwnd is 1,
2,4,8, ... and so on. cwnd
When ssthresh reaches ssthresh, a congestion avoidance phase is entered. In the congestion avoidance phase, cwnd is increased by 1 / cwnd every time an ACK packet is received, so that the increase is much slower than in the slow start phase.

In a typical TCP implementation, a lost DT packet event may occur when an acknowledgment is not received and time monitoring times out, or when a certain number (usually three) or more
When K is received, it is determined that packet loss has occurred. Note that the duplicate ACK reception means that a plurality of ACK packets having the same reception sequence number are continuously received.

When it is determined that the DT packet is lost, the parameters of the bandwidth control are adjusted, and ssthresh is
(Cwnd, adwnd) / 2. For cwnd, in the case of duplicate ACK detection, new cwnd = old c
wnd / 2, and in case of timeout, new cwnd
= 1 (for example, WRStevens
Author, Chapter 21, TCP Illustrated Vol.1, Addison We
sley, 1994, etc.)

On the other hand, a selective response (SACK) function aimed at speeding up error recovery has been proposed as an Internet document in 1996 as RF as an option of TCP.
Specified in C2018. The SACK information is for notifying, in a concrete manner, a section in which the packet has been received discontinuously due to packet loss on the receiving side by adding a section in which the packet has been received to the ACK packet. The sender uses this information to retransmit only the lost packet.

Regarding the relationship with the band control (congestion control), since the existing function should be preserved in RFC2018, when the packet loss is detected by the SACK information, the same parameters as those at the time of detecting the duplicate ACK are used. Make adjustments. As described above, when a packet loss event is detected based on duplicate ACKs, response confirmation timeouts, and SACK information, control is performed to uniformly reduce the amount of transmission regardless of the user level or application type using the connection. Had become.

[0012]

However, in such a conventional bandwidth control method, particularly in the TCP currently implemented by various host / terminal apparatuses, when the network is congested such that a packet loss event occurs, the application is not controlled. And provides a best-effort service that uniformly reduces the communication bandwidth of each connection regardless of the communication requirements of the user. Therefore, when the Internet is congested, for example, high priority such as important / emergency is provided. There was a problem that a band could not be secured by a communication connection.

In other words, with the diversification and sophistication of information and communication services, priority is given to the use of communication bands as one of the so-called QoS (service quality) requirements in a best-effort service, and fine-grained. Service realization is becoming necessary. For example, on an intranet that is intended for internal use, it is conceivable to assign priority to the allocation of communication bandwidth according to the type of business and job position to secure throughput and shorten response time in important communications and emergency communications. Cannot be realized by TCP bandwidth control.

An object of the present invention is to solve such a problem, and an object of the present invention is to provide a bandwidth control method capable of realizing a fine-grained service based on prioritization of use of a communication band.

[0015]

In order to achieve the above object, a bandwidth control method according to the present invention comprises:
The present invention is a communication protocol that sets a communication band used for packet transmission based on a smaller value of a reception notification window value indicating a reception capability of a partner node and a congestion window value that estimates a transfer capability of a network, When a packet loss event is detected, the congestion window value is not reduced immediately but is held as it is, and only when the packet loss situation worsens thereafter, the congestion window value is reduced to reduce the communication band. Therefore, when a packet loss event is detected, the communication band is not reduced immediately but is kept as it is, and the communication band is reduced only when the subsequent packet loss situation worsens.

According to a second aspect of the present invention, in the first aspect of the present invention, it is determined whether or not priority communication for allocating a band preferentially according to a user level or an application type is necessary. Immediately reduces the congestion window value when a packet loss event is detected, and when priority communication is required, keeps the congestion window value as it is without immediately reducing it when a packet loss event is detected. Only when this is done, the congestion window value is reduced to reduce the communication band. Therefore, when a packet loss event is detected, the communication band of the connection that does not require priority communication is immediately reduced, and the communication band of the connection that requires priority communication is maintained using the released communication band as it is,
Only when the subsequent packet loss situation worsens, the communication band of the connection requiring priority communication is reduced.

According to a third aspect of the present invention, in the first or second aspect of the present invention, the packet loss amount is calculated for each predetermined monitoring section, and the first packet loss amount in an arbitrary monitoring section and the immediately following packet loss amount are calculated. And comparing the second packet loss amount in the monitoring section with the second packet loss amount, and if the second packet loss amount is larger than the first packet loss amount, it is determined that the packet loss situation is deteriorating. It is. According to a fourth aspect of the present invention, in the third aspect of the present invention, a section from a packet loss event detection point to a point in time when a delivery confirmation is notified from a receiving side for all transmitted packets is set as a monitoring section. is there.

According to a fifth aspect of the present invention, there is provided a communication protocol for measuring a throughput indicating a load state of a network for each measurement section having a predetermined length, and adjusting a packet transmission amount according to the change. However, the packet transmission amount is not immediately reduced but is held as it is, and the communication band is reduced by reducing the packet transmission amount only when the subsequent decrease in throughput is further deteriorated. Therefore, upon detecting a decrease in throughput, the communication band is not reduced immediately and is kept as it is, and the communication band is reduced only when the subsequent decrease in throughput is further deteriorated.

According to a sixth aspect of the present invention, in the fifth aspect of the present invention, it is determined whether or not priority communication for allocating a band preferentially according to a user level or an application type is necessary. If the decrease in throughput is detected immediately, the amount of packet transmission is immediately reduced.If priority communication is required, the amount of packet transmission is not immediately reduced when the decrease in throughput is detected, and is kept unchanged. The communication bandwidth is reduced by reducing the amount of packet transmission only in. Therefore, when a decrease in throughput is detected, the communication band of the connection that does not require priority communication is immediately reduced, and the communication band of the connection that requires priority communication is maintained as it is by using the communication band generated by this. Only when the deterioration is further worsened is the communication bandwidth of the connection requiring priority communication reduced.

The invention of claim 7 is the invention according to claim 5 or 6, wherein the difference between the throughput in an arbitrary first measurement section and the throughput in the second measurement section immediately after the first measurement section is the first Is calculated, and when the calculated first throughput change amount is equal to or more than a predetermined threshold value, it is determined that the throughput is reduced, and the throughput in the second measurement section and the immediately following throughput amount are determined. A second throughput change amount is calculated from a difference from the throughput in the third measurement section, and when the calculated first and second throughput change amounts are equal to or greater than a predetermined threshold, the decrease in throughput is further deteriorated. That is, it is determined that they are doing.

[0021]

Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a communication system according to a first embodiment of the present invention. In FIG. 1, an end node 1 is connected to an information communication network 2 and communicates with the other end node 1, and the information communication network 2 is a network including a communication line and a relay node.

The end node 1 includes a lower layer processing unit 1
1, layer 4 processing unit 12, application processing unit 13
It is composed of The lower layer processing unit 11 performs processing of the protocol layer 3 and below, that is, layer 1 for electrical matching with a communication line, layer 2 for frame assembly / disassembly, and layer 3 for routing (here, IP). ) Are performed.

The layer 4 processing section 12 performs data transmission / reception processing based on release of connection setting of layer 4 (here, TCP), flow control, and the like. In this case, the reception notification window value (adwnd), the bandwidth control parameter for the protocol layer 4 (TCP),
In addition to the congestion window value (cwnd) and the slow start threshold (ssthresh), a function for counting the number of lost packets, a variable loss1 for storing the number of lost packets
And loss2, a variable hi_c that holds the maximum transmitted sequence number transmitted and unconfirmed when a packet loss event is detected
hk is used.

The number of packets that can be transmitted is min (cwn
d, adwnd), and changes during communication according to the change of each parameter. Here, min (A, B) indicates that the smaller value of A and B is used. In addition, the upper layer has ON / OFF and has a priority communication flag for selecting a method of bandwidth control by TCP.

In the first embodiment of the present invention, when a packet loss event is detected in a connection whose priority communication flag is ON, the band control parameter ss is immediately set.
thresh and cwnd are kept without reduction, and the band control parameter ss only when the loss situation worsens thereafter.
This is to reduce thresh and cwnd. Also,
By setting a monitoring section until all packets transmitted and unacknowledged at the time when the packet loss event is detected are acknowledged, and comparing the total number of each lost packet lost in two adjacent monitoring sections, This is for determining whether or not the loss situation after detecting the packet loss event has deteriorated.

Next, referring to FIGS. 2 to 4, the first embodiment of the present invention will be described.
As an operation according to the embodiment, an operation of data transmission centering on band control will be described. 2 is a sequence explanatory diagram showing an outline of the entire communication sequence, FIG. 3 is a flowchart showing an outline of the entire band control, and FIG. 4 is a state transition table showing a band control process at the time of priority communication. Hereinafter, a case will be described as an example where a packet is transferred from a server (transmission side) in response to a request from a client side (reception side).

First, the layer 4 processing section 1 on the client side
2 (TCP) is the application processing unit 13 (upper A
Based on the communication start request from P), connection setting is performed by exchanging a connection setting request and a connection setting response with the server-side layer 4 processing unit 12 (TCP). At this time, if it is desired to use the SACK option in the TCP connection, a SACK indication is added to the connection setting request packet and negotiation is performed between the transmission side and the reception side.

The layer 4 processing section 1 on the client side
2 (TCP) is the application processing unit 13 (upper A
The connection is released by exchanging a connection release request and a connection release response with the layer 4 processing unit 12 (TCP) on the server side based on the communication termination request from P).

As shown in FIG. 2, after the connection setting is completed, the server enters a normal (N) state in which data transfer is possible (FIG. 3: step 31). The upper AP on the server side sets the TCP based on the user ID or the requested application ID notified from the upper AP on the client side.
It is determined whether or not priority band control is performed on the connection.

Here, when the notified ID indicates a connection requiring priority communication, the priority communication flag is set to O.
If N is set to indicate a connection that does not require priority communication, the priority communication flag is set to OFF. Therefore, in the normal (N) state of the subsequent data transfer, the layer 4 processing unit 12 of the server
When a loss event of a T packet is detected, it is determined whether priority communication is necessary for the connection based on ON / OFF of the priority communication flag.

As shown in FIG. 3, in the normal (N) state of data transfer (step 31), when the loss event of the TCP DT packet is detected by the layer 4 processing unit 12 on the server side, the priority communication is performed. The flag is checked (step 32). Here, the priority communication flag is set to OF
If F, it is determined that the connection is a connection for which the normal bandwidth control is designated as in the related art, and the bandwidth control parameters ssthresh and cwnd are immediately reduced (step 33).

On the other hand, when the priority communication flag is ON,
It is determined that the connection requires priority communication, and the following band control processing is executed. The packet loss event can be grasped from duplicate ACK, SACK (selective response) information, and delivery response confirmation timeout. Also, the number L of lost packets is 1 in the case of duplicate ACK and delivery confirmation timeout, but in the case of SACK information, it is a new number L of losses that is found from that information.

In the normal (N) state, when a duplicate ACK is received or a SACK is received, a monitoring section is started, and the maximum transmitted sequence number transmitted and unconfirmed at that time is set in a variable hi_chk, and a lost packet is stored in a variable loss1. The number L is set, and the state transits to the monitoring 1 (M1) state (step 34) without reducing the values of cwnd and ssthresh. Also, in the N state, the monitoring section is started even at the time of delivery confirmation timeout, the maximum transmitted sequence number that has been transmitted and not confirmed at that time is set in the variable hi_chk, the variable loss1 is set to 1, and cwnd and sst are set.
The state transits to the M1 state (step 34) without reducing the value of hresh.

In the M1 state, the processing at the time of ACK reception is as follows. When the reception sequence number (ACKseq) of the received ACK packet is hi_chk> ACKs
eq, ie, if all transmitted and unacknowledged packets have not yet been acknowledged, los
There is no change in s1, the state remains in the M1 state, and the monitoring section is continued.

On the other hand, if hi_chk ≦ ACKseq, that is, if transmission of all transmitted and unconfirmed packets has been acknowledged, the maximum transmitted sequence number that has been transmitted and unconfirmed at that time is set in hi_chk, and the variable loss2 is set. 0, and transits to the monitoring 2 (M2) state (step 35). Thus, the first monitoring section ends and the next monitoring section starts.

In the M1 state, the processing at the time of receiving the duplicate ACK or receiving the SACK is as follows. Receive sequence number of the ACK packet to be received (ACKse
If q) is hi_chk> ACKseq, that is, if all transmitted and unacknowledged packets have not yet been acknowledged, if a new packet loss is detected, the lost number L is added to loss1 and the M1 state is changed. Leave as is and continue the monitoring section.

On the other hand, if hi_chk ≦ ACKseq, that is, if transmission has been confirmed for all transmitted and unconfirmed packets, the maximum transmitted sequence number transmitted and unconfirmed at that time is set in hi_chk, and a new packet is transmitted. If a loss is detected, the loss number L is set in loss2, and the state transits to the M2 state. This allows
The first monitoring section ends, and the next monitoring section starts. Further, in the M1 state, at the time of delivery confirmation timeout,
Loss1 is incremented by 1, and the monitoring section is continued with the M1 state.

In the M2 state, the processing at the time of ACK reception is as follows. When the reception sequence number (ACKseq) of the ACK packet to be received is hi_chk> A
In the case of CKseq, that is, when the delivery of all transmitted and unconfirmed packets has not yet been confirmed,
There is no change in loss2, the state remains in the M2 state, and the monitoring section is continued.

On the other hand, when hi_chk ≦ ACKseq, that is, when the delivery of all transmitted and unconfirmed packets has been confirmed, it is determined that the two preceding and succeeding monitoring sections have been completed, and the values of loss1 and loss2 are compared ( Step 36). Here, if loss2 = 0,
Since no lost packet has occurred in the subsequent monitoring section, it is determined that the congestion situation has been resolved, and the band control parameters ssthresh and cwnd are not reduced, and the state transits to the normal (N) state (step 31).

Further, if loss1 ≧ loss2, the congestion situation is considered to be improving because the total number of lost packets in the subsequent monitoring section is reduced, so that the bandwidth control parameters ssthresh and cwnd are not reduced. Then, a new monitoring section is started to further see the situation. Therefore, the value of loss2 is set to loss1,
loss2 = 0, the maximum transmitted sequence number that has been transmitted and not confirmed at that time is set in hi_chk,
The state returns to the second state (step 35), and the monitoring section is continued.

If loss1 <loss2, the network congestion situation is considered to be deteriorating since the total number of lost packets in the subsequent monitoring section has increased, so that the bandwidth control parameters ssthresh and cw
After reducing nd (step 37), the state transits to the N state (step 31).

In the M2 state, the processing at the time of receiving duplicate ACK or receiving SACK is as follows. Receive sequence number of the ACK packet to be received (ACKse
If q) is hi_chk> ACKseq, that is, if transmission has not been confirmed for all transmitted and unconfirmed packets, the number of lost packets L is added to loss2 if a new packet is detected, and the number of lost packets is added to M2 state. Leave as is and continue the monitoring section.

If hi_chk ≦ ACKseq, that is, if transmission has been confirmed for all transmitted and unconfirmed packets, it is determined that the two preceding and succeeding monitoring sections have been completed, and the values of loss1 and loss2 are compared (step 38). ). Here, if loss1 ≧ loss2,
Since the total number of lost packets in the subsequent monitoring section has been reduced, the congestion situation is considered to be improving,
The bandwidth control parameters ssthresh and cwnd are not reduced.

Then, to see the situation further,
2 is set to loss1, and the maximum transmitted sequence number transmitted and unconfirmed at that time is hi_c.
hk, and if new packet loss is detected,
The lost number L is set to ss2, and the M2 state (step 3
Return to 5) and start a new monitoring section.

If loss1 <loss2, the network congestion situation is considered to be deteriorating because the total number of lost packets in the subsequent monitoring section has increased, and the bandwidth control parameters ssthresh and cw
After reducing nd (step 39), the state transits to the N state (step 31).

Further, in the M2 state, when the delivery confirmation times out, loss2 is incremented by 1, the M2 state is maintained, and the monitoring section is continued. The reduction of the band control parameters ssthresh and cwnd in steps 36 and 38 is as follows.
Here, as in the conventional method, ssthresh = min
(Cwnd, adwnd) / 2, new cwnd = old cwn
d / 2.

As described above, in the first embodiment of the present invention, when a packet loss event is detected, the bandwidth control parameters ssthresh and cwnd are immediately held without being reduced, and the subsequent loss situation becomes worse. Only when this is done, the bandwidth control parameters ssthresh and cwnd are reduced. Therefore, as compared with the conventional bandwidth control method shown in FIG. 5A, in which the bandwidth is reduced uniformly in response to the detection of the packet loss event, as shown in FIG. A bandwidth can be secured by a connection of a communication having a high priority such as urgency, and a detailed service based on the prioritization of use of the communication band can be realized.

A monitoring section is defined as a monitoring section until all packets transmitted and unacknowledged at the time of detecting a packet loss event are acknowledged, and each packet lost in an arbitrary monitoring section and a monitoring section immediately after it is determined. Since the total number of lost packets loss1 and loss2 are compared, it is possible to accurately determine whether the loss situation after detecting the packet loss event has deteriorated.

Further, based on the user ID or the requested application ID notified from the receiving side (client side), it is determined whether or not to perform preferential band control for the connection on the transmitting side (server side). Therefore, unrestricted priority communication to a large number of receivers can be easily suppressed, and the same connection as that in the related art without priority communication is immediately reduced in bandwidth in response to the detection of a packet loss event. Can be secured.

Next, a second embodiment of the present invention will be described. Here, a communication protocol (for example, TCP) that adjusts the transmission amount by grasping the network congestion status not from a packet loss event but from a change in throughput.
-TCP version called Vegas) will be described. Note that the concept is the same as described above, and will be briefly described here.

First, the throughput is calculated for each measurement section of a predetermined length. When calculating the throughput, first, the time required from the start of packet transmission in an arbitrary measurement section to the reception confirmation notification from the receiving side of all transmission packets transmitted in the measurement section is measured. Then, the throughput is calculated by dividing the amount of packets transmitted in the measurement section by the required time.

Subsequently, the throughput calculated in an arbitrary measurement section is set to P1, the throughput calculated in the immediately following measurement section is set to P2, and the throughput change amount D in both measurement sections is calculated as the difference between the two, that is, D = P1 −P
Find in 2. In this case, a predetermined threshold value a is determined and used to determine the situation as to whether or not the throughput is reduced.

The operation when the priority communication flag is ON is as follows. If D <a, it is determined that the throughput is good, and the packet transmission amount is increased. If D ≧ a, it is determined that the throughput has decreased, and the value of the current transmission amount is held and the state of the next period is observed.

Subsequently, the value of P2 is stored as P1, and the throughput in the next period is obtained as P2. Then, the throughput change amount D is calculated again and the threshold value a is calculated.
And if D ≧ a again, it is determined that the decrease in throughput is further deteriorated, and the value of the current packet transmission amount is reduced.

On the other hand, the operation when the priority communication flag is OFF is as follows. The throughput change amount D calculated in the same manner as described above is compared with the threshold value a. If D <a, it is determined that the throughput is good, and the transmission amount is increased. If D ≧ a, the throughput is reduced and the load on the network is increasing, and the value of the current transmission amount is immediately reduced.

Therefore, in a communication protocol that adjusts the transmission amount based on the change in the throughput, if a decrease in the throughput is detected, the packet transmission amount is immediately held without being reduced, and the subsequent decrease in the throughput is prevented. Since the amount of packet transmission is reduced only in the case of further deterioration, higher priority such as important / urgent etc. is compared with the conventional case where the bandwidth is reduced uniformly according to the detection of a packet loss event. A band can be secured by a communication connection, and a fine-grained service based on the prioritization of the use of the communication band can be realized.

[0057]

As described above, according to the present invention, when a network congestion situation is detected, the packet transmission amount is not immediately reduced without directly reducing the amount of packet transmission in the connection for priority communication. The communication bandwidth of the connection for performing the priority communication is reduced only when the communication speed is further deteriorated. Therefore, as compared with the conventional case in which the bandwidth is reduced uniformly in response to the detection of the packet loss event, the bandwidth can be secured in the connection of the communication of high priority such as important / urgent, and furthermore, Therefore, it is possible to cope with the diversification of QoS (service quality) in the best-effort service, and to realize a fine-grained service based on the prioritization of the use of the communication band. For example, in an intranet supposed to be used internally, it is possible to assign priorities to the allocation of communication bandwidth according to the type of business and the type of user, to secure throughput and reduce response time in important communications and emergency communications, which is aggressive. Best-effort service can be realized.

Also, it is determined whether or not priority communication for preferentially allocating a band according to a user level or an application type is necessary. For a connection that does not require priority communication, the amount of packet transmission is immediately reduced upon detecting a congestion situation. For connections requiring priority communication, the amount of packet transmission is not immediately reduced when the congestion state is detected, but is maintained as it is, and only when the congestion situation worsens, the amount of packet transmission is reduced to reduce the communication bandwidth. It was made. Therefore, even if congestion occurs in the network and the available communication bandwidth is reduced, the communication bandwidth of the connection that does not require priority communication that was immediately reduced in response to the occurrence of congestion is used for the connection that requires priority communication. The communication band is kept as it is. Further, the transmitting side can determine whether or not priority communication for allocating a band preferentially according to a user level or an application type is required, and it is possible to easily suppress unrestricted priority communication to a large number of receiving sides. For a connection similar to the conventional connection that does not perform communication, the bandwidth is immediately reduced in response to the detection of a packet loss event, and a bandwidth for performing priority communication can be secured.

Further, the packet loss amount is calculated for each predetermined monitoring section, and the first packet loss amount in an arbitrary monitoring section is compared with the second packet loss amount in the monitoring section immediately thereafter. If the second packet loss is larger than the first packet loss, it is determined that the packet loss situation has deteriorated. Since the section from the time until the notification of the delivery confirmation is set as the monitoring section, in the communication protocol that detects the network congestion state by the packet loss event and adjusts the packet transmission amount, the loss after the packet loss event is detected Accurately determine whether the situation is getting worse.

The first throughput change amount is calculated from the difference between the throughput in an arbitrary first measurement section and the throughput in the second measurement section immediately after the first measurement section, and the calculated first throughput change amount is calculated. If the amount is equal to or greater than the predetermined threshold, it is determined that the throughput has decreased, and the second throughput is determined from the difference between the throughput in the second measurement interval and the throughput in the third measurement interval immediately thereafter. The amount of change is calculated, and when the calculated first and second amounts of change in throughput are equal to or greater than a predetermined threshold value, it is determined that the decrease in throughput is further deteriorated. Detected by changes in throughput,
In the communication protocol for adjusting the packet transmission amount, it is possible to accurately determine a decrease in the throughput and a further deterioration thereafter.

[Brief description of the drawings]

FIG. 1 is a block diagram of a data communication system according to an embodiment of the present invention.

FIG. 2 is a sequence explanatory diagram showing an outline of an entire communication sequence.

FIG. 3 is a flowchart showing an outline of the entire band control.

FIG. 4 is a state transition table showing band control processing during priority communication.

FIG. 5 is an explanatory diagram showing an operation concept of the present invention.

[Explanation of symbols]

1 end node, 2 information communication network, 11 lower layer processing unit, 12 layer 4 processing unit, 13 application processing unit.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04L 12/56

Claims (7)

(57) [Claims] 1. A communication protocol for setting a communication band used for packet transmission based on a smaller value of a reception notification window value indicating a reception capability of a partner node and a congestion window value estimating a transfer capability of a network. When a packet loss event is detected, the congestion window value is maintained without being immediately reduced, and the communication band is reduced by reducing the congestion window value only when the packet loss situation worsens thereafter. Control method. 2. The bandwidth control method according to claim 1, wherein it is determined whether or not priority communication for preferentially allocating a bandwidth according to a user level or an application type is necessary. Immediately reduce the congestion window value at the time of event detection, and if priority communication is required, keep the congestion window value at the time of packet loss event detection without immediately reducing it, and only when the subsequent packet loss situation worsens. A bandwidth control method characterized in that a communication bandwidth is reduced by reducing a congestion window value. 3. The bandwidth control method according to claim 1, wherein a packet loss amount is calculated for each predetermined monitoring section, and a first packet loss amount in an arbitrary monitoring section and a monitoring section immediately thereafter are calculated. Comparing the second packet loss amount with the second packet loss amount, and if the second packet loss amount is larger than the first packet loss amount, it is determined that the packet loss situation is deteriorating. . 4. The bandwidth control method according to claim 3, wherein a section from a time point when the packet loss event is detected to a time point when a delivery confirmation is notified from the receiving side for all transmitted packets is set as a monitoring section. Control method. 5. A communication protocol for measuring a throughput indicating a load condition of a network for each measurement section of a predetermined length and adjusting a packet transmission amount in accordance with a change thereof. A bandwidth control method, wherein the communication bandwidth is reduced by holding the packet as it is without reducing it, and reducing the packet transmission amount only when the subsequent decrease in throughput is further deteriorated. 6. The bandwidth control method according to claim 5, wherein it is determined whether or not priority communication for preferentially allocating a bandwidth according to a user level or an application type is necessary. Immediately reduce the amount of packet transmission at the time of detection, and if priority communication is required, maintain the packet transmission amount at the time of detection of a decrease in throughput without immediately reducing it, and transmit the packet only when the subsequent decrease in throughput worsens A bandwidth control method characterized by reducing a communication bandwidth by reducing an amount. 7. The bandwidth control method according to claim 5, wherein a first throughput change amount is obtained from a difference between a throughput in an arbitrary first measurement section and a throughput in a second measurement section immediately after the first measurement section. When the calculated first throughput change amount is equal to or more than a predetermined threshold value, it is determined that the throughput has decreased, and the throughput in the second measurement section and the third measurement immediately thereafter are determined. A second throughput change amount is calculated from a difference from the throughput in the section, and if the calculated first and second throughput change amounts are equal to or greater than a predetermined threshold value, it is determined that the throughput decrease is further deteriorated. A bandwidth control method characterized by determining.