JP4977677B2 - Edge node and bandwidth control method - Google Patents

Description

  The present invention relates to flow control in a system that transmits and receives data between terminals via a network.

  Communication between terminals such as personal computers is performed via a computer network such as the Internet. Data such as a file is transferred by this communication. In this type of data transfer system, TCP (Transmission Control Protocol), which is a standard transport protocol, is often used as a transport layer protocol.

In data transfer, it may be required to transfer a large amount of data at high speed. On the other hand, studies are being made to improve throughput by controlling the TCP window size to be constant.
Yuta Hattori, 5 others, “Examination of TCP flow rate control method in edge router”, IEICE General Conference (B-6-54), March 2007, p. 54

  In TCP, bandwidth control that autonomously changes the window size according to packet discard is performed. However, in data transfer in a bandwidth guaranteed network that uses a token bucket type inflow restriction that defines a burst size at the entrance of the network to prevent traffic exceeding the guaranteed bandwidth from flowing into the network, the throughput is sawtooth. There was a problem that the guaranteed bandwidth could not be fully utilized.

  The technology for transmitting data at a constant throughput and the technology for controlling the advertisement window size consistently according to the bandwidth guaranteed by changing the TCP protocol described above have a countermeasure for this problem.

  However, in the technique described in Non-Patent Document 1 for controlling the advertisement window size to be constant, it has not been considered that flows are multiplexed. For this reason, the higher the number of flows to be subjected to bandwidth control is, the greater the difference between the packet transmission timing of the calculated flow and the packet transmission timing that can actually be transmitted. As a result, the flow becomes a bursty flow that exceeds the set burst size at the entrance of the network, so the packet discard occurs due to the inflow restriction, and the use efficiency of the guaranteed bandwidth for each flow decreases. was there.

  An object of the present invention is to provide a technique that enables efficient use of a guaranteed bandwidth even when bandwidth control target flows are multiplexed in data transfer in a bandwidth-guaranteed network.

In order to achieve the above object, the edge node of the present invention comprises:
An edge node disposed between the data transmitting terminal and the data receiving terminal in a bandwidth guaranteed network in which a bandwidth that can be transmitted from the data transmitting terminal to the data receiving terminal is guaranteed;
A bandwidth information holding unit for holding a guaranteed bandwidth allocated to a service target flow from the data transmitting terminal to the data receiving terminal in the bandwidth guaranteed network;
A propagation delay time calculating unit for measuring the propagation delay time of the service target flow;
From the guaranteed bandwidth of the service target flow held in the bandwidth information holding unit and the propagation delay time of the service target flow measured by the propagation delay time calculation unit, the data of the service target flow is the guaranteed bandwidth A window size calculation unit for calculating a window size that is within the bandwidth,
The difference between the arrived ACK number, which is the ACK number written in the ACK packet received from the data receiving terminal, and the transmitted ACK number, which is the ACK number written in the ACK packet previously transmitted to the data transmitting terminal An ACK scheduling unit that schedules the transmission timing of the ACK packet of the service target flow to be transmitted to the data transmission terminal in preference to the service target flow having a large ACK number difference;
An advertisement window size writing unit for writing the window size in the ACK packet;
have.

  According to the present invention, ACK packets of a plurality of service target flows are scheduled, and in the scheduling, priority is given to a service target flow having a large ACK number difference that is a difference between an arrived ACK number and a transmitted ACK number. It is possible to alleviate the bursting of the flow that sometimes occurs and to prioritize the service target flow in a state in which the data receiving terminal can receive more data, thereby efficiently transferring the data.

  Further, a token that is a data packet transmission right is managed for each of the service target flows, the token is added at a rate corresponding to the guaranteed bandwidth of the service target flow, and an ACK packet is transmitted to the data transmission terminal In this case, a token management unit that subtracts the token by a number difference between the ACK number assigned to the ACK packet and the transmitted ACK number of the data transmission terminal may be further included.

  In addition, when the ACK scheduling unit transmits an ACK packet of the service target flow, the ACK scheduling unit compares the ACK number difference of the service target flow with the amount of the token, and transmits data corresponding to the smaller one The ACK packet may be given an ACK number that permits the ACK packet.

  In addition, if the ACK number difference of the service target flow is larger than the token amount, the ACK scheduling unit adds a value obtained by adding the token to the transmitted ACK number of the service target flow. The ACK packet may be transmitted to the data transmission terminal.

  According to this, since the data transmitting terminal can transmit the maximum amount of data while complying with both the guaranteed bandwidth and the permission from the data receiving terminal, the throughput can be improved by efficiently using the guaranteed bandwidth. Can do.

  Further, it may further include a flow identification / distribution unit that performs flow distribution and sends the ACK packet of the service target flow received from the data receiving terminal to the ACK scheduling unit.

  In addition, the information processing apparatus may further include a flow information holding unit that holds the propagation delay time measured by the propagation delay time calculating unit in association with the connection information of the service target flow.

  According to the present invention, it is possible to alleviate flow bursting that occurs during flow multiplexing, prioritize a service target flow in a state in which a data receiving terminal can receive more data, and efficiently transfer data.

  In general TCP, two windows, an advertisement window and a congestion window, are defined. The size of the advertisement window is determined on the receiving side depending on the state of the reception buffer. The size of the congestion window is determined on the data transmission side according to the packet discard situation. As the window size used for flow control, the smaller one of the advertisement window size and the congestion window size is selected. This window size is updated when an ACK arrives.

  In the present embodiment, a band is guaranteed by a band guarantee network between a data transmitting terminal serving as a data transmitting side and a data receiving terminal serving as a data receiving side. Therefore, it is assumed that no packet is discarded as long as data is transmitted within the guaranteed bandwidth.

  Therefore, discarding packets can be prevented by adjusting the advertisement window size so that data is transmitted below the guaranteed bandwidth. If the packet discard does not occur, the congestion window size is not reduced, so the advertisement window size is always adopted as the TCP window size. As a result, the throughput does not fluctuate in a sawtooth shape, and the guaranteed bandwidth can be used efficiently.

  Therefore, in this embodiment, the edge node measures the propagation delay time between the data transmission side and the data reception side between the data transmission terminal and the data reception terminal, the propagation delay time, and the guaranteed bandwidth by the bandwidth guarantee network Then, the window size is calculated such that the data is transmitted within the guaranteed bandwidth, and the calculated window size is written as the advertisement window size in the ACK packet transmitted to the data transmitting terminal.

  As a result, the bandwidth of data to be transmitted from the data transmitting terminal to the data receiving terminal can be stably controlled within the guaranteed bandwidth, so that the guaranteed bandwidth can be efficiently used.

  Further, the edge node calculates an ACK number difference between the ACK number assigned to the ACK packet addressed to the data transmission terminal received from the data reception terminal and the ACK number assigned to the ACK packet transmitted to the data transmission terminal last time. The ACK packet of the service target flow having a large ACK number difference is preferentially transferred.

  Therefore, according to the present embodiment, since ACK packets of a plurality of service target flows are scheduled, even when the service target flows are multiplexed, it is possible to reduce the flow bursting. As a result, packet discard due to inflow restriction is reduced, and the guaranteed bandwidth can be efficiently used to maintain high throughput. For example, in the operation of a bandwidth guarantee service using TCP, there is no need to change the TCP protocol stack of the data transmission terminal.

  Furthermore, according to the present embodiment, in scheduling of ACK packets related to a plurality of service target flows, priority is given to a service target flow having a large ACK number difference that is a difference between an arrived ACK number and a transmitted ACK number. As a result, it is possible to prioritize the service target flow in a state where the data receiving terminal can receive more data, and efficiently transfer the data.

  When the edge node transmits an ACK packet of the service target flow, the edge node compares the ACK number difference of the service target flow with the amount of tokens, and the ACK number that permits sending data corresponding to the smaller one Is added to the ACK packet. Thereby, the data transmitting terminal can transmit the maximum data while complying with both the guaranteed bandwidth and the permission from the data receiving terminal. As a result, the guaranteed bandwidth can be efficiently used to improve the throughput.

  Hereinafter, the present embodiment will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing the configuration of the data transmission system according to the present embodiment. Referring to FIG. 1, the data transmission system of this embodiment includes edge nodes 11 and 12 and a bandwidth management server 13. The bandwidth management server 13 is expressed as RACS in the drawing. RACS is an abbreviation for Resource and Admission Control Sub-system.

  The data transmission terminal 14 is connected to the edge node 11, and the data reception terminal 15 is connected to the edge node 12. The edge node 11 and the edge node 12 can transmit and receive data on the bandwidth guarantee network 16. As a result, data from the data transmission terminal 14 is transferred to the data reception terminal 15 via the edge node 11 and the edge node 12.

  The bandwidth management server 13 is a device that centrally manages bandwidth information of all links of the bandwidth guarantee network 16. The bandwidth information includes information on the guaranteed bandwidth allocated to the link. The bandwidth management server 13 notifies the edge nodes 11 and 12 of bandwidth information of each link being managed. As a result of bandwidth management by the bandwidth management server 13, a guaranteed bandwidth in the bandwidth guaranteed network 16 is secured for the link from the data transmitting terminal 14 to the data receiving terminal 15. As a result, as long as the data transmission terminal 14 transmits data below the guaranteed bandwidth, packet discard does not occur in the bandwidth guaranteed network 16.

  FIG. 2 is a block diagram showing the configuration of the edge node of this embodiment. Referring to FIG. 2, the edge node 11 includes a data reception unit 21, a flow identification / distribution unit 22, a propagation delay time calculation unit 23, a data transmission unit 24, a band information holding unit 25, a flow information holding unit 26, and a window size calculation. Unit 27, advertisement window size writing unit 28, and ACK scheduling unit 29.

  The data reception unit 21 receives a packet such as a data packet or an ACK packet and sends the packet to the flow identification / distribution unit 22.

  The flow identification / distribution unit 22 guarantees the flow (service target flow) that is the target of the data transfer service that guarantees the bandwidth based on the user information that is held together with the bandwidth information in the bandwidth information holding unit 25. The flow that is not the target of the data transfer service (service non-target flow) is identified. The service non-target flow is, for example, a packet of a user who has not subscribed to the service.

  Further, the flow identification / distribution unit 22 distributes the ACK packet of the service target flow to the ACK scheduling unit 29 through the path 22a. The flow identification / distribution unit 22 distributes the packet of the non-service target flow and the data packet of the service target flow to the data transmission unit 24 through the path 22b.

  The propagation delay time calculation unit 23 measures a round trip propagation delay time (hereinafter simply referred to as “round trip propagation delay time”) 17 between the data transmission terminal 14 and the data reception terminal 15.

  The data transmission unit 24 transmits the packet from the flow identification / distribution unit 22 and the packet from the advertisement window size writing unit 29 to the edge node 12 via the bandwidth guarantee network 16.

  The bandwidth information holding unit 25 holds the bandwidth information of the service target user notified from the bandwidth management server 13 and user information (destination / source IP address, destination / source port number) in association with each other. The bandwidth information includes information on the guaranteed bandwidth allocated to the service target flow. The service target flow can be identified by user information.

  The flow information holding unit 26 holds the connection information of the service target flow identified by the flow identification / distribution unit 22 and the round trip propagation delay time 17 calculated by the propagation delay time calculation unit 23 in association with each other.

  The window size calculation unit 27 uses the round trip propagation delay time 17 held in the flow information holding unit 26 and the guaranteed band of band information held in the band information holding unit 25 to guarantee the data transmission band. Calculate the window size to be within the bandwidth.

  The advertisement window size writing unit 28 writes the window size value calculated by the window size calculation unit 27 into the window field of the ACK packet addressed to the data transmission terminal 14 created by the ACK scheduling unit 29, and the ACK packet is used as data. The data is sent to the transmission unit 24.

  The ACK scheduling unit 29 receives the ACK number difference, which is the difference between the ACK number of the ACK packet that has arrived from the data receiving terminal and the ACK number of the ACK packet that has been transmitted to the data transmitting terminal, and the service target flow in the ACK token management unit 30 Referring to the token amount held every time, an ACK number is calculated so that data corresponding to the smaller of the ACK number difference or the token amount is transmitted from the data transmission terminal 14, and the ACK number is calculated. The attached ACK packet is created and sent to the advertisement window size writing unit 28. The ACK number at which data corresponding to the token amount is transmitted is a value obtained by adding the token amount value to the ACK number in the ACK packet that has been transmitted to the data transmission terminal 14.

  The ACK token management unit 30 manages the token amount for each service target flow. Specifically, the ACK token management unit 30 adds a token according to the guaranteed bandwidth notified from the bandwidth information holding unit 25, and when an ACK packet is created by the ACK scheduling unit 29, the ACK number of the previous ACK packet And the token is subtracted by the difference between the ACK numbers of the current ACK packet.

  FIG. 3 is a diagram for explaining packet distribution by the flow identification / distribution unit 22. With reference to FIG. 3, packet distribution by the flow identification / distribution unit 22 will be described in detail.

  When a connection is established between the data transmission terminal 14 and the data reception terminal 15, data transfer is started. When the flow identification / distribution unit 22 of the edge node 11 receives the ACK packet from the data receiving terminal 15, the flow identification / distribution unit 22 determines whether the ACK packet is the service target flow. If it is an ACK packet for the service target flow, the flow identification / distribution unit 22 sends the ACK packet to the ACK scheduling unit 29 through the path 22a. For the non-service target flow packet and the service target flow data packet, the flow identification / distribution unit 22 sends the packet to the data transmission unit 24 through the path 22b.

  FIG. 4 is a sequence diagram showing exchange of signals between blocks in the edge node when the round-trip propagation delay time is measured. Referring to FIG. 4, when the flow identification / distribution unit 22 observes the SYN packet in the edge node 11 on the data transmission terminal 14 side (step 101), a round trip propagation delay time measurement start request is sent together with the user information of the service target flow. This is sent to the propagation delay time calculation unit 23 (step 102). Thereby, the propagation delay time calculation unit 23 starts measuring the round-trip propagation delay time 17. Thereafter, the flow identification / distribution unit 22 observes the SYN / ACK packet for the SYN (step 103). Thereafter, when the flow identification / distribution unit 22 observes the ACK packet for the SYN / ACK (step 104), the flow identification / distribution unit 22 sends a round trip propagation delay time measurement end request to the propagation delay time calculation unit 23 together with the user information (step 105). As a result, the propagation delay time calculation unit 23 ends the measurement of the round-trip propagation delay time 17.

  The propagation delay time calculation unit 23 that has finished measuring the round-trip propagation delay time 17 notifies the flow information holding unit 26 of the measured round-trip propagation delay time 17 together with user information as a round-trip propagation delay time report (step 106).

  FIG. 5 is a sequence diagram showing exchange of signals between blocks in the edge node when the window size is calculated. When the flow information holding unit 26 receives the round-trip propagation delay time report from the propagation delay time calculating unit 23 (step 201), the bandwidth guarantee window is sent to the window size calculating unit 27 together with the round-trip propagation delay time 17 and user information. A calculation request is sent (step 202).

  The window size calculation unit 27 that has received the bandwidth guarantee window calculation request requests the bandwidth information holding unit 25 for a guaranteed bandwidth corresponding to the user information based on the user information (step 203). The bandwidth information holding unit 25 returns the guaranteed bandwidth in response to the request from the window size calculation unit 27 (step 204).

  Based on the guaranteed bandwidth acquired from the bandwidth information holding unit 25 and the round-trip propagation delay time acquired from the flow information holding unit 26, the window size calculating unit 27 causes the data transmission terminal 14 to transmit data within the guaranteed bandwidth. A correct window size is calculated (step 205). At this time, a window size is preferable such that data is transmitted in a band that matches the guaranteed band.

  The window size can be obtained by, for example, equation (1).

  The header size includes Ethernet (registered trademark), IP, and TCP header sizes.

  FIG. 6 is a sequence diagram showing exchange of signals between blocks in the edge node when the advertisement window size is written in the ACK packet. Referring to FIG. 6, first, the window size calculation unit 27 calculates the window size of the service target flow (step 301). Subsequently, the window size calculation unit 27 notifies the calculated window size and the user information of the service target flow to the advertisement window size writing unit 28 (step 302). The advertising window size writing unit 28 writes the window size value notified from the window size calculating unit 27 in the window field of the ACK packet from the ACK scheduling unit 29 (step 303). The ACK packet in which the advertisement window size is written by the advertisement window size writing unit 28 is sent to the data transmission unit 24.

  FIG. 7 is a sequence diagram illustrating exchange of signals between functional blocks when a token rate is set for each user (service target flow). The token rate is a rate of a token added to a token managed by the ACK token management unit 30. The token rate is a value that depends on the guaranteed bandwidth.

  Referring to FIG. 7, the bandwidth information holding unit 25 notifies the ACK token management unit 30 of the guaranteed bandwidth and user information (step 401). At that time, the packet size may be further notified.

  The ACK token management unit 30 sets the token rate of the service target flow identified by the user information based on the guaranteed bandwidth (step 402).

  The token rate can be obtained by, for example, equation (2).

  Time granularity is a unit of time subdivision. The token addition timing may be, for example, when the ACK scheduling unit 29 refers to the token of the ACK token management unit 30. As the token amount to be added at that time, a value obtained by multiplying the difference between the time when the token is referred from the ACK scheduling unit 29 and the time when the token is previously referred to by the token rate may be used. .

  When the header sizes of Ethernet (registered trademark), IP, and TCP are taken into consideration, the ratio between the header and the data is obtained from the packet size, and the guaranteed bandwidth value is adjusted based on the ratio.

  FIG. 8 is an image diagram showing an example of the flow information table 33 used by the ACK scheduling unit 29 for scheduling. Referring to FIG. 8, the flow information table 33 includes a flow number 41, a transmitted ACK number 42, an arrived ACK number 43, and an ACK number difference 44.

  The flow number 41 is a number assigned to each service target flow. The transmitted ACK number 42 is an area for storing the ACK number given to the ACK packet that has been transmitted to the data transmitting terminal 14. The arrived ACK number 43 is an area for storing the ACK number assigned to the ACK packet arriving from the data receiving terminal 15. The ACK number difference 44 is an area for storing a value obtained by subtracting the value described in the transmitted ACK number 42 from the value described in the arrived ACK number 43.

  FIG. 9 is a block diagram illustrating a configuration of the ACK scheduling unit 29. FIG. 9 shows signal exchange between functional blocks when ACK packet transmission control is performed in cooperation with the ACK scheduling unit 29 and the ACK token management unit 30.

  Referring to FIG. 9, the ACK scheduling unit 29 includes an ACK packet creation unit 31, an ACK packet transmission control management unit 32, and ACK number storage units 34-1 to 34-N. The ACK packet transmission control management unit 32 includes a flow information table 33 for controlling the transmission timing of the ACK packet of each service target flow. The contents of the flow information table 33 are as shown in FIG.

  The ACK number storage units 34-1 to 34-N correspond to each service target flow and store the ACK number of the ACK packet 500 distributed from the flow identification / distribution unit 22. In addition, the ACK number storage unit 34 sends an ACK number notification 501 for notifying the ACK number of the arrived ACK packet to the ACK packet transmission control management unit 32.

  The ACK packet transmission control manager 32 updates the field of the arrived ACK number 43 of the service target flow in the flow information table 33 with the notified ACK number value. At the same time, the ACK packet transmission control manager 32 subtracts the transmitted ACK number from the arrived ACK number 43 and updates the field of the ACK number difference 44 as well.

  When the ACK number difference of any service target flow is updated, the ACK packet transmission control management unit 32 re-selects the service target flow having the largest ACK number difference among the plurality of service target flows.

  The ACK token management unit 30 manages the tokens 35-1 to 35-N of the plurality of service target flows. The ACK packet transmission control management unit 32 refers to the token value managed by the ACK token management unit 30 for the selected service target flow by executing the token reference 502 for the ACK token management unit 30. At this time, the ACK packet transmission control management unit 32 determines whether the token value managed by the ACK token management unit 30 is the difference between the time when the current token is referenced and the time when the previous token is referenced. Add the value multiplied by the rate. At that time, the ACK packet transmission control management unit 32 notifies the ACK token management unit 30 of the value of the ACK number difference held in the flow information table 33 within itself.

  The ACK token management unit 30 compares the managed token value with the ACK number difference notified from the ACK packet transmission control management unit 30.

  If the token value is larger than the value of the ACK number difference 44, the ACK token management unit 30 notifies the ACK packet transmission control management unit 32 of the value of the ACK number difference 44 by the token notification 503, and manages the token value. ACK number difference 44 is subtracted from. On the other hand, if the ACK number difference 44 is larger than the token value, the ACK token management unit 30 notifies the token value to the ACK packet transmission control management unit 32 by the token notification 503, and manages the managed token value. Set to zero.

  The ACK packet transmission control management unit 32 adds the value obtained by the token notification 503 from the ACK token management unit 30 to the transmitted ACK number 42 of the flow information table 33 and transmits the transmitted ACK number 42 obtained as the addition result. ACK transmission instruction 504 for instructing transmission of an ACK packet with ACK number as the ACK number is sent to the ACK number storage unit 34.

  The ACK number storage unit 34 that has received the ACK transmission instruction 504 sends an ACK packet creation request 505 for requesting creation of an ACK packet of the instructed ACK number to the ACK packet creation unit 31.

  The ACK packet creation unit 31 creates an ACK packet corresponding to the request in the ACK packet creation request 505 and sends it to the advertisement window size writing unit 28.

It is a block diagram which shows the structure of the data communication system by this embodiment. It is a block diagram which shows the structure of the edge node of this embodiment. It is a figure for demonstrating packet distribution by a flow identification and distribution part. It is a sequence diagram which shows the exchange of the signal between the blocks in an edge node when measuring a round-trip propagation delay time. It is a sequence diagram which shows exchange of the signal between the blocks in an edge node when calculating a window size. It is a sequence diagram which shows exchange of the signal between the blocks in the edge node at the time of window size calculation. FIG. 11 is a sequence diagram illustrating signal exchange between functional blocks when a token rate is set for each user in an ACK token management unit. It is an image figure which shows an example of the flow information table which an ACK scheduling part uses for scheduling. 3 is a block diagram showing a configuration of an ACK scheduling unit 29. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11, 12 Edge node 13 Band management server 14 Data transmission terminal 15 Data reception terminal 16 Band guarantee network 17 Round-trip propagation delay time 21 Data reception part 22 Flow identification and distribution part 23 Propagation delay time calculation part 24 Data transmission part 25 Band information holding | maintenance Unit 26 flow information holding unit 27 window size calculating unit 28 advertising window size writing unit 29 ACK scheduling unit 30 ACK token management unit 31 ACK packet creation unit 32 ACK packet transmission control management unit 33 flow information table 34 ACK number storage unit 41 flow number 42 Sent ACK number 43 Arrived ACK number 44 ACK number difference

Claims (9)

An edge node disposed between the data transmitting terminal and the data receiving terminal in a bandwidth guaranteed network in which a bandwidth that can be transmitted from the data transmitting terminal to the data receiving terminal is guaranteed;
A bandwidth information holding unit for holding a guaranteed bandwidth allocated to a service target flow from the data transmitting terminal to the data receiving terminal in the bandwidth guaranteed network;
A propagation delay time calculating unit for measuring the propagation delay time of the service target flow;
From the guaranteed bandwidth of the service target flow held in the bandwidth information holding unit and the propagation delay time of the service target flow measured by the propagation delay time calculation unit, the data of the service target flow is the guaranteed bandwidth A window size calculation unit for calculating a window size that is within the bandwidth,
The difference between the arrived ACK number, which is the ACK number written in the ACK packet received from the data receiving terminal, and the transmitted ACK number, which is the ACK number written in the ACK packet previously transmitted to the data transmitting terminal An ACK scheduling unit that schedules the transmission timing of the ACK packet of the service target flow to be transmitted to the data transmission terminal in preference to the service target flow having a large ACK number difference;
An advertisement window size writing unit for writing the window size in the ACK packet;
An edge node having   A token that is a data packet transmission right is managed for each of the service target flows, the token is added at a rate corresponding to the guaranteed bandwidth of the service target flow, and an ACK packet is transmitted to the data transmission terminal The edge node according to claim 1, further comprising: a token management unit that subtracts the token by a number difference between an ACK number given to the ACK packet and a transmitted ACK number of the data transmission terminal.   When the ACK scheduling unit transmits an ACK packet of the service target flow, the ACK scheduling unit compares the ACK number difference of the service target flow with the amount of the token, and permits transmission of data corresponding to the smaller one The edge node according to claim 2, wherein an ACK number is assigned to the ACK packet.   If the ACK number difference of the service target flow is larger than the token amount, the ACK scheduling unit adds a value obtained by adding the token to the transmitted ACK number of the service target flow as the ACK number of the next ACK packet. The edge node according to claim 3, wherein the ACK packet is transmitted to the data transmission terminal.   5. The edge according to claim 1, further comprising a flow identification / distribution unit that performs flow distribution and sends an ACK packet of the service target flow received from the data receiving terminal to the ACK scheduling unit. 6. node.   The edge according to any one of claims 1 to 5, further comprising a flow information holding unit that holds the propagation delay time measured by the propagation delay time calculating unit in association with connection information of a service target flow. node. A bandwidth control method for controlling a bandwidth of data transmission from the data transmission terminal to the data reception terminal in a bandwidth guarantee network in which a bandwidth that can be transmitted from the data transmission terminal to the data reception terminal is guaranteed,
In advance, the guaranteed bandwidth allocated to the service target flow from the data transmitting terminal to the data receiving terminal in the bandwidth guaranteed network,
Measure the propagation delay time of the service target flow,
From the guaranteed bandwidth of the service target flow and the propagation delay time, calculate a window size in which the data of the service target flow falls within the guaranteed bandwidth,
The difference between the arrived ACK number, which is the ACK number written in the ACK packet received from the data receiving terminal, and the transmitted ACK number, which is the ACK number written in the ACK packet previously transmitted to the data transmitting terminal Priority is given to the service target flow with a large ACK number difference, scheduling the transmission timing of the ACK packet of the service target flow to be transmitted to the data transmission terminal,
A bandwidth control method for writing an advertisement window size for writing the window size in the ACK packet. Managing a token that is a data packet transmission right for each of the service target flows;
Adding the token at a rate according to the guaranteed bandwidth of the service target flow;
When an ACK packet is transmitted to the data transmission terminal, the token is subtracted by the number difference between the ACK number assigned to the ACK packet and the transmitted ACK number of the data transmission terminal.
The bandwidth control method according to claim 7.   A value obtained by adding the token of the service target flow to the transmitted ACK number of the service target flow is set as an ACK number of the next ACK packet of the service target flow, and the ACK packet is transmitted to the data transmission terminal. The bandwidth control method according to claim 8.

Images