JPH1168845A - Data transfer system and transmission rate control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable a transmission system to transmit a packet at a transmission rate for enabling reception at a reception system even without receiving any confirm packet by transmitting a request data transfer rate to the transmission system at every prescribed timing after the reception of data and determining the transmission rate through the transmission system based on the received request data transfer rate. SOLUTION: When transmission data from an application 14 of a transmission system 10 are dispatched to a transport layer, a transmission buffer 11 divides these data into packets and performs queuing. The packet received from the low layer of a network 15 is queued in a reception buffer 21, the packet is extracted from the reception buffer 21 corresponding to the processing speed of an application 23, and it is dispatched to the application 23 as assembled data. The packet dispatched to the application 23 as data sends a confirm packet CP showing normal processing completion to the transmission system 10.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data transfer system and a transmission rate control method, and more particularly, to a transport protocol for performing high-speed data transfer between end systems in an information communication network and a data transfer system using the same. Applicable.

[0002]

2. Description of the Related Art First, functions targeted by the present invention in a transport protocol will be described.

[0003] The transport protocol has a flow control function for transmitting and receiving data between end systems (for example, terminals such as personal computers).
The flow control is to restrict the input on the transmitting side so that the input load on the receiving side does not exceed the processable range.

When the flow control is weak (when input control is weak), the receiving buffer on the receiving side overflows, the packet is discarded, and the packet must be retransmitted. As a result, the actual throughput ( Network use efficiency) will be reduced. On the other hand, when the flow control is strong (when the input regulation is strong), the input packet is not transmitted from the transmission side even though the reception side has an empty reception buffer or processing capacity, and as a result, the throughput is reduced. I can't go up.

[0005] Properly determining the flow control of the transport protocol in this way improves the throughput and realizes high-speed data transfer between end systems.

Next, a sliding window system which is a flow control technology as a general prior art will be described with reference to FIG. FIG. 2 shows a state where the transmission system transmits a data packet to the reception system, and the reception system transmits a confirmation packet for notifying the transmission system that the reception processing of the data packet has been completed. .

In FIG. 2, the code SN is Sequence Numbe
This is an abbreviation of r, which indicates the sequence number of the data packet of the transmission system. AN is an abbreviation of Acknowledgment Number, and indicates the number (reception processing confirmation number) of the data packet that the receiving system expects to receive next. That is,
In the receiving system, the receiving process of data packets up to the number of AN-1 has been completed. WS
Window Size is an abbreviation for the number of packets (window size) that can be received by the receiving side, starting from the value specified by AN. That is, AN + W
Data packets up to the number S-1 can be received.
ASN is an abbreviation of Available Sequence Number, and is a number of a data packet (a possible transmission number) that can be transmitted by the transmission system. That is, ASN = AN + WS-1.

Next, the operation will be described. For ease of explanation, the initial values of SN, AN, and WS are 1, 1, 3 respectively.
And These initial values are taught by both systems during the process of establishing a connection between the transmitting system and the receiving system.

[0009] The transmitting system transmits the data packet to which the SN is added to the receiving system. Since ASN = 3 at first, it is possible to transmit data packets up to SN = 3. When the reception processing of the data packet is completed, the receiving system transmits a confirmation packet to indicate that. AN and WS are added to the confirmation packet.
When the reception processing of the data packet of SN = 1 is completed,
A confirmation packet of AN = 2 and WS = 3 is transmitted. The transmission system that has received the confirmation packet calculates the ASN to be 4, and thus can transmit data packets up to SN = 4. This operation is continued.

Here, SN, AN, WS, and ASN are defined in units of packets, but are not limited thereto. For example, it may be specified in units of bytes. In addition, SN and AN are transmitted for each confirmation packet, but may be transmitted when reception processing for a certain number of packets is completed. Initial values of SN, AN, and WS are determined at the time of connection establishment by negotiation between the transmitting system and the receiving system. Further, the value of WS may change depending on the processing load state of the receiving system or the like.

[0011]

However, since the sliding window type flow control is designed on the assumption of a low-delay network, the throughput is reduced in a wide-area network having a large transmission delay and the data transmission speed is reduced. There is a disadvantage that transfer cannot be performed.

In a network having a large transmission delay, FIG.
As shown in (1), the transmission time of the packet between the transmission system and the reception system becomes longer. Therefore, the transmission system
Although the receiving system is in a state in which it can receive the data packet, the receiving system cannot transmit the data packet until receiving the transmitted confirmation packet, so that a so-called useless time elapses. This wasted time is a time called an idle time, and the idle time increases as the transmission delay increases. As a result, the throughput is reduced and high-speed data transfer cannot be performed.

Therefore, there is a need for a data transfer system and a transmission rate control method in which a transmission system can always transmit a packet at a transmission rate that can be received by a reception system without receiving an acknowledgment packet.

[0014]

According to a first aspect of the present invention, there is provided a transmission system which divides transmission data into a plurality of packets and transmits the packets at a predetermined transmission rate, a network connected to the transmission system, and transmission via the network. In a data transfer system including a receiving system for sequentially receiving packets transmitted from the system, the receiving system is in a state of waiting for processing, which is temporarily stored in a receiving buffer and held in the receiving buffer. From the amount of data and the maximum capacity of the reception buffer, a rate calculation means for generating a requested data transfer rate for the transmission system is provided,
After receiving the data, the requested data transfer rate is transmitted to the transmission system at a predetermined timing, and the transmission system determines the transmission rate based on the received requested data transfer rate.

Therefore, data transfer can be performed at an adaptive rate according to the state of the receiving buffer of the receiving system.

According to the present invention, packets are sequentially transmitted without waiting for reception confirmation from the receiving system. Therefore, even when the transmission rate does not match the state of the receiving buffer of the receiving system until data of the adaptive rate is transmitted. is there. The state of the network on the way is also dynamically changing. Therefore, the rate calculation means generates the required data transfer rate by assuming the maximum capacity of the reception buffer as a capacity smaller than the actual maximum capacity in advance so that the data in the reception buffer is not full of data waiting to be processed. It is desirable.

If the reception buffer has room,
Regardless of the amount of data held in the reception buffer and waiting for processing, transmission at the maximum transmission rate may increase overall throughput. In such a case, the rate calculation means generates a rate corresponding to the maximum transmission rate as the requested data transfer rate until the data waiting for processing in the reception buffer exceeds a predetermined safe capacity. Is desirable.

According to a second aspect of the present invention, there is provided a transmission rate control method for controlling a transmission rate from a transmission system when data is transmitted between a transmission system and a reception system connected to an information communication network. Is
A parameter is generated based on the ratio between the reception buffer length and the data queue length in the reception buffer and transmitted to the transmission system, and the transmission system controls its own transmission rate based on the received parameter. Features.

The first and second inventions described above have been made based on the following concept. The problem of the conventional example is that the transmission system cannot transmit a data packet unless a confirmation packet is received from the reception system.
Therefore, this problem is solved if the transmitting system can transmit data packets at a certain rate without receiving an acknowledgment packet. Therefore, the receiving system transmits an acknowledgment packet indicating that the data packet has been reliably received, and puts the rate at which the receiving system can receive (referred to as "adaptive rate") on the packet.
Accordingly, the transmission system can always transmit data packets at an adaptive rate that can be received by the reception system without receiving a confirmation packet in advance for each packet transmission.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described with reference to the drawings. FIG. 4 is a diagram for explaining the principle of this embodiment. The problem of the conventional example is that the transmission system cannot transmit a data packet unless a confirmation packet is received from the reception system. Therefore, this problem is solved if the transmission system can transmit a data packet at a certain rate without receiving an acknowledgment packet.

For this reason, the transmission system first sets the rate R
In step 1, a packet with sequence numbers SN1 to SN10... Is transmitted. Upon receiving the packet P1 with the sequence number SN1, the receiving system transmits an acknowledgment packet P2 carrying an adaptive rate R2, which is a rate that the receiving system can receive, and an acknowledgment number AN2. The transmission system transmits the packet at the adaptive rate R2 after transmitting the adaptive rate R2. The receiving system thereafter transmits a confirmation packet P2 on which the adaptive rates R3, R4,. The transmission system transmits the packet by changing the adaptation rate of the transmission packet to R3, R4 each time the confirmation packet P3, P4... Is received. Therefore, after receiving the acknowledgment packet, the transmission system can perform packet transmission according to the state of the transmission path and the reception system, and can perform data transfer with a short idle time.

(A) First Embodiment FIG. 1 shows an example of a transport protocol according to a first embodiment of the present invention using the above principle.

The transport layer of the transmission system 10 includes a transmission buffer 11, a rate control unit 12, and a rate calculation unit 13. The transmission buffer 11 has a function of dividing data received from the application layer 14 into packets and queuing the packets in the transmission buffer 11. The rate control unit 12 has a function of generating a packet transmission timing based on an adaptive rate that dynamically changes according to the confirmation packet P transmitted from the receiving system 20. The rate calculation unit 13 has a function of converting the adaptive rate sent from the reception system 20 into a parameter that can be recognized by the rate control unit 12.

On the other hand, the receiving system 20 includes a receiving buffer 21 and a rate calculator 22. The reception buffer 21 has a function of queuing a received packet once, assembling the packet according to the processing speed of the application 23, and delivering the packet to the application as data. The rate calculator 22 has a function of calculating an adaptation rate from the length of a packet queued in the reception buffer.

As a method of realizing the rate control unit 12 of the transmission system, there is a Leaky Bucket algorithm. An example of this algorithm will be described below.

The bucket length D, the current bucket length L, the additional bucket length W, and the update time T are assumed. Assuming that the length of a packet to be transmitted is X, if L−X> = 0, the packet can be transmitted, and
If −X <0, the packet cannot be transmitted. After transmitting the packet, L = L−X. Also, L = L + W is updated every T time. Therefore, a packet that could not be transmitted can be transmitted if LX> = 0 after updating. It should be noted that L <= D, and even if L> D as a result of updating every T time, L is suppressed by D and L = D is set. With such an algorithm, transmission rates can be averaged to achieve W / T. Conversely, if the transmission rate is R, then R = W / T, that is, W
By setting W such that = R × T, a desired transmission rate can be realized.

The Leaky Bucket algorithm shown here is an example, and there are various other rate control methods. In addition, Leaky is
The Bucket algorithm need not be used. The point is that it is only necessary to generate the packet transmission timing.

In the drawing, PR represents a Peak Rate (maximum transmission rate), and IR represents an Initial Rate (initial transmission rate).
CR is obtained by W / T, PR is the maximum transmission rate determined first between the transmission system and the reception system (for example, determined by the transmission rate of the physical layer), and IR is the transmission system and the reception first. The initial transmission rate determined between the systems (for example, determined by B / RTT, etc., where B:
Buffer size of the receiving system, RTT: round trip time.

Next, a communication operation using the protocol of FIG. 1 will be described. In addition, Leaky Buck
A description will be given of an example in which the et algorithm is used.

(1) Transmission of Data Packet in Transmission System When transmission data from the application 14 is passed to the transport layer, the transmission buffer 11 divides the data into packets and queues them. If the transmission is permitted from the rate control unit 12, the packet is transmitted. If there is no transmission permission, the packets are queued until the permission is lowered. An example of the algorithm of the rate control unit 12 is as described above. Transmission buffer 1 in FIG.
1 indicates a state in which three data packets of packet lengths X, Y, and Z are queued.

(2) Data packet reception and acknowledgment packet transmission in the receiving system Packets received from the lower layer of the network 15 are:
It is queued in the reception buffer 21. Packets are extracted from the reception buffer 21 according to the processing speed of the application 23, assembled, and passed to the application 23 as data. The packet passed as data to the application 23 sends a confirmation packet CP indicating normal process completion to the transmission system. At this time, the confirmation packet CP includes a confirmation number (AN) and an adaptation rate (AR).
It is included. AN indicates the packet number for which normal processing has been completed. In the description of the prior art, this AN is described as "the number of the data packet that the receiving system expects to receive next." However, since this is a definition problem, the description is simplified. Therefore, in the following, description will be made as "the packet number for which normal processing has been completed".

AR is a rate calculated by AR = PR × (1−H / B). B is the buffer length of the reception buffer, and H is the queuing length of the packet in the reception buffer 21. This calculation is performed in the rate calculator 22. The rate calculator 22 constantly monitors the queuing length of the reception buffer 21, and when transmitting the confirmation packet CP, calculates the adaptive rate AR according to the above equation and puts it on the confirmation packet. AR can be translated into a rate determined by the ratio R of the queuing length H of the packet in the reception buffer 21 to the buffer length B of the reception buffer. Since the buffer length B is fixed, the smaller the amount of queued data, that is, the smaller the queuing length H, the more data transfer is performed at a rate closer to the maximum transmission rate PR.

(3) Confirmation packet reception in the transmission system Upon receiving the confirmation packet CP, the transport layer
The acknowledgment number AN included in the acknowledgment packet CP confirms the delivery of the data packet transmitted by itself. However, unlike the prior art, the next packet is transmitted according to the rate determined by the rate control unit 12 even before the delivery is confirmed. Of course, if a missing packet is found by the confirmation packet CP, it goes without saying that the packet is retransmitted.

The rate calculator 13 calculates the adaptive rate AR from
The additional Bucket length W used by the rate control unit 12 is calculated. That is, W = AR × T is added from AR = W / T
Calculate Bucket length W. However, if the adaptation rate AR is equal to the previously received adaptation rate AR, it ignores it without doing anything. Thereby, the receiving buffer 21 of the receiving system
The transmission system will send data packets at a rate proportional to the queuing length of

Therefore, even in a wide area network having a large propagation delay, in order to control data transmission at a rate,
Throughput increases and high-speed data transfer becomes possible.

Further, since only the queuing length and the buffer length of the receiving system are used to determine the data transmission rate of the transmission system, the determination of the transmission rate is very simple.

Since the queuing length of the receiving system is reflected in determining the data transmission rate of the transmitting system, overflow of the receiving buffer can be suppressed.

(B) Second Embodiment Next, a data transfer system and a transmission rate control method according to a second embodiment of the present invention will be described.

The basic configuration of the transmission system and the reception system is the same as the system of the first embodiment. The difference from the system of the first embodiment is that an extra buffer is set in the reception buffer of the reception system. Even if the receiving system sets the adaptation rate to 0, the transmission system continues to transmit data packets until the adaptation rate (0) reaches the transmission system. Therefore, unless an extra buffer length is set in the reception buffer, the possibility of occurrence of reception buffer overflow may increase. In the second embodiment, an extra buffer S is introduced as shown in FIG. 5, and the reception buffer has a margin in advance.

Hereinafter, the protocol will be described. Note that, also in the second embodiment, the Leaky Bucket is similar to the first embodiment.
The algorithm is used.

(1) Transmission of Data Packet in Transmission System When transmission data from the application 14 is passed to the transport layer, the transmission buffer 11 divides the data into packets and queues them. If the transmission is permitted from the rate control unit 12, the packet is transmitted. If there is no transmission permission, the packet until the permission is lowered is queued and stays in the transmission buffer 11.

(2) Data packet reception and acknowledgment packet transmission in the reception system Packets received from the lower layer are received by the reception buffer 21 '.
Queued to. Packets are taken out of the reception buffer 21 'according to the processing speed of the application 23, assembled, and passed to the application 23 as data. The packet passed as data to the application 23 sends a confirmation packet CP indicating normal process completion to the transmission system. At this time, the confirmation packet CP includes a confirmation number (AN) and an adaptation rate (AR).
AN indicates the packet number for which normal processing has been completed.

AR is calculated by (a) AR = PR × (1−H / (BS)) when H <BS, and (b) AR = 0 when H> = BS. Is the rate calculated. B is the buffer length of the reception buffer, and H is the queuing length of the packet in the reception buffer 21 '. This calculation is performed in the rate calculator 22 '. The rate calculation unit 22 '
The queuing length of 1 'is constantly monitored, and when transmitting a confirmation packet, the adaptive rate AR is calculated according to the above equation and placed on the confirmation packet.

(3) Confirmation packet reception in the transmission system Upon receiving the confirmation packet CP, the transport layer
The acknowledgment number AN included in the acknowledgment packet CP confirms the delivery of the data packet transmitted by itself. The rate calculator 13 calculates an additional Bucket length W used in the rate controller 12 from the adaptive rate AR. That is, AR = W /
From T, the additional Bucket length W is calculated as W = AR × T.
However, the adaptive rate AR is the adaptive rate AR received last time.
If it is equal to, ignore it without doing anything. As a result, the transmission system sends data packets at a rate proportional to the queuing length of the reception buffer 21 of the reception system. As described above, since the AR is calculated by using (B−S) obtained by subtracting the extra buffer length S from the actual buffer length B, a margin is created in the actual buffer, and a slight change in the transfer rate occurs. Even so, the overflow of the receiving buffer is suppressed.

Suppression of packet discarding due to reception buffer overflow can be realized with a very simple parameter called a proportional parameter between the queuing length and the buffer length.

In the second embodiment, although the number of parameters is increased as compared with the protocol of the first embodiment, the effect of suppressing the overflow of the reception buffer is increased.
As a result, the overall data transfer throughput can be improved.

(C) Third Embodiment Next, a data transfer system and a transmission rate control method according to a third embodiment of the present invention will be described with reference to FIG.

The basic configuration of the transmission system and the reception system is the same as the system of the first embodiment. The difference from the system of the first embodiment is that secure queuing is set in the reception buffer of the reception system.
In the system according to the first embodiment, if at least one packet is queued in the reception buffer of the receiving system, the adaptation rate decreases in proportion to the queue. This can reduce throughput. While the packet queuing length is equal to or less than the safe queuing length, the throughput can be suppressed from decreasing by keeping the adaptation rate at the peak rate. Therefore, in the third embodiment, the concept of the safe queuing length P is introduced as shown in FIG. 6, and the adaptive rate is kept at the peak rate while the reception buffer has a margin.

Hereinafter, the protocol will be specifically described.
Note that, in the third embodiment as well, as in the first embodiment, Le
aky Bucket algorithm is used.

(1) Data Packet Transmission in Transmission System When transmission data from the application 14 is passed to the transport layer, the transmission buffer 11 divides the data into packets and queues them. If the transmission is permitted from the rate control unit 12, the packet is transmitted. If there is no transmission permission, the packet until the permission is lowered is queued and stays in the transmission buffer 11.

(2) Data packet reception and acknowledgment packet transmission in the receiving system Packets received from the lower layer are received by the reception buffer 21 ″.
Queued to. A packet is taken out from the reception buffer 21 "according to the processing speed of the application 23, assembled and passed as data to the application 23. The packet passed as data to the application 23 transmits a confirmation packet CP indicating completion of normal processing. At this time, the confirmation packet CP includes a confirmation number (AN) and an adaptation rate (AR).
AN indicates the packet number for which normal processing has been completed.

AR is (c) when H <= P (that is, when there is a margin in the reception buffer), AR = PR (d) When H> P, AR = PR × (1- (HP) / (BP)). B is the buffer length of the reception buffer, H is the queuing length of the packet in the reception buffer 21 ″,
P is the safety queuing length. This calculation is performed in the rate calculation unit 22 ″. The rate calculation unit 22 ″ constantly monitors the queuing length of the reception buffer 21 ″, and calculates the adaptive rate AR by the above equation when transmitting the confirmation packet CP. And put it in the confirmation packet CP.

(3) Confirmation Packet Reception in Transmission System Upon receiving the confirmation packet CP, the transport layer
It confirms the delivery of the data packet transmitted by itself using the confirmation number AN included in the confirmation packet. Rate calculator 1
3 calculates the additional Bucket length W used by the rate control unit 12 from the adaptive rate AR. That is, AR = W / T
Then, the additional Bucket length W is calculated as W = AR × T. However, if the adaptation rate AR is equal to the previously received adaptation rate AR, it ignores it without doing anything. As a result, the transmission system transmits data packets at a rate proportional to the queuing length of the reception buffer 21 of the reception system.

Suppression of packet discarding due to reception buffer overflow can be realized with a very simple parameter called a proportional parameter between the queuing length and the buffer length.

In the third embodiment, the number of parameters is increased compared to the protocol of the first embodiment, but the transmission rate of the transmission system is set to the peak rate while the packet queuing length is short. Therefore, the throughput is improved.

(D) Fourth Embodiment Next, a data transfer system and a transmission rate control method according to a fourth embodiment of the present invention will be described.

The basic configurations of the transmission system and the reception system are the same as the system of the first embodiment. The difference from the system of the first embodiment is that a correction parameter α is introduced when calculating the adaptive rate. With the introduction of this parameter, the relationship between the adaptation rate and the queuing length is no longer proportional, but becomes an α-order function.
As a result, while the queuing length is short, the adaptation rate does not decrease so much, and as the queuing length increases, the adaptation rate suddenly decreases. Through the introduction of the correction parameter α, the throughput can be improved and the buffer overflow can be suppressed.

The operation of the fourth embodiment will be described with reference to FIG. In the fourth embodiment, the Leaky Bucket algorithm is used as in the first embodiment.

(1) Data Packet Transmission in Transmission System When transmission data from the application 14 is passed to the transport layer, the transmission buffer 11 divides the data into packets and queues them. If the transmission is permitted from the rate control unit 12, the packet is transmitted. If there is no transmission permission, the packet until the permission is lowered is queued and stays in the transmission buffer 11.

(2) Reception of Data Packet and Transmission of Confirmation Packet in Receiving System The packet received from the lower layer is queued in the reception buffer 21. Packets are extracted from the reception buffer 21 according to the processing speed of the application 23,
It is assembled and passed to the application 23 as data. The packet passed as data to the application 23 sends a confirmation packet CP indicating normal process completion to the transmission system. At this time, the confirmation packet CP includes a confirmation number (AN) and an adaptation rate (AR). AN
Indicates the packet number for which normal processing has been completed.

AR is a rate calculated as AR = PR × {1− (H / B) ^α } as a function of (H / B) to the power of α. B is the buffer length of the receiving buffer, H is the queuing length of the packet in the receiving buffer 21 ″, and α is the correction parameter. This calculation is performed by the rate calculating unit 24. The rate calculating unit 24
Monitors the queuing length of the reception buffer 21 at all times, and calculates the adaptive rate AR by the above equation when transmitting the confirmation packet CP, and puts it on the confirmation packet.

(3) Confirmation Packet Reception in Transmission System Upon receiving the confirmation packet CP, the transport layer
The acknowledgment number AN included in the acknowledgment packet CP confirms the delivery of the data packet transmitted by itself. The rate calculator 13 calculates an additional Bucket length W used in the rate controller 12 from the adaptive rate AR. That is, AR = W /
From T, the additional Bucket length W is calculated as W = AR × T.
However, the adaptive rate AR is the adaptive rate AR received last time.
If it is equal to, ignore it without doing anything. As a result, the transmission system transmits data packets at a rate proportional to the queuing length of the reception buffer 21 of the reception system.

Suppression of packet discard due to reception buffer overflow can be realized by a very simple parameter called a proportional parameter between the queuing length and the buffer length.

In the fourth embodiment, the number of parameters is increased as compared with the protocol of the first embodiment. However, while the packet queuing length is short, the transmission rate of the transmission system is set high. Increased throughput,
If the queuing length becomes longer, the queuing length is rapidly set lower, so that the overflow of the reception buffer is suppressed and the overall data transfer throughput can be improved.

(E) Other Embodiments In the above description, the confirmation packet is transmitted for each data packet. However, the confirmation packet CP does not need to be transmitted for each data packet. May be collectively transmitted in one confirmation packet CP.

The receiving system does not need to generate a dedicated confirmation packet CP only for confirming reception, and transmits data to be transmitted from the system described as the receiving system to the system described as the transmitting system. Number (AN) or adaptation rate (A
R) can be transmitted.

The adaptive rate (AR) transmitted from the receiving system to the transmitting system does not need to be the rate itself, but may be any parameter indicating the relationship between the queuing length and the buffer length.

The above embodiments can be applied independently, but in combination, for example, the second embodiment and the third embodiment
The embodiments can be combined to further improve the throughput.

[0069]

As described above, according to the present invention, a packet can always be transmitted at a transmission rate that can be received by a receiving system without receiving a confirmation packet.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a system and a transport protocol according to a first embodiment.

FIG. 2 is a diagram showing a conventional transport protocol.

FIG. 3 is a diagram for explaining a problem of a conventional transport protocol.

FIG. 4 is a diagram illustrating the principle of each embodiment of the present invention.

FIG. 5 is a diagram illustrating a system and a transport protocol according to a second embodiment.

FIG. 6 is a diagram illustrating a system and a transport protocol according to a third embodiment;

FIG. 7 is a diagram illustrating a system and a transport protocol according to a fourth embodiment;

[Explanation of symbols]

 AN: Acknowledge number CP: Confirmation packet AR: Adaptive rate 10: Transmission system 11: Transmission buffer 12: Rate control unit 13: Rate calculation unit 15: Network 20 ... Reception system 21 ... Reception buffer 22 ... Rate calculation unit

Claims (8)

[Claims] 1. A transmission system that divides transmission data into a plurality of packets and transmits the packets at a predetermined transmission rate, a network connected to the transmission system, and a packet transmitted from the transmission system via the network. A data transfer system comprising a receiving system for sequentially receiving, a receiving buffer for temporarily holding received packets, an amount of data held in the receiving buffer and waiting for processing, and A rate calculating unit configured to generate a requested data transfer rate for the transmission system from a maximum capacity of a reception buffer, and after receiving data, transmitting the requested data transfer rate to the transmission system at predetermined timings; Determines the transmission rate based on the received request data transfer rate. Data transfer system, characterized by. 2. The data transfer system according to claim 1, wherein the receiving system returns confirmation data for confirming a packet received by the receiving system to the transmitting system. 3. The data transfer system according to claim 2, wherein the transmission system sequentially transmits packets at the predetermined transmission rate regardless of reception of the confirmation data. 4. The requested data transfer rate is generated based on a ratio R between a maximum capacity of the reception buffer and an amount of data waiting to be processed in the reception buffer. The data transfer system according to claim 1, wherein: 5. The data transfer system according to claim 4, wherein said rate calculation means generates the required data transfer rate based on the ratio R raised to the power of α. 6. The rate calculating means considers the maximum capacity of the reception buffer as a capacity smaller than an actual maximum capacity in advance so that the reception buffer is not filled with data waiting for processing. The data transfer system according to claim 1, wherein the requested data transfer rate is generated. 7. The rate calculating means generates the requested data transfer rate at a rate corresponding to a maximum transmission rate until data waiting to be processed in the reception buffer exceeds a predetermined safe capacity. The data transfer system according to claim 1, wherein: 8. A transmission rate control method for controlling a transmission rate from a transmission system when transmitting data between a transmission system and a reception system connected to an information communication network, the reception system comprising: A parameter is generated based on a ratio with a data queue length in a buffer and transmitted to the transmission system, and the transmission system controls its own transmission rate based on the received parameter. Transmission rate control method.