JPH11163936A - Flow control method and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a flow control method that can determine an optimal packet transmission speed in a communication terminal. SOLUTION: The device provides a flow control method in a packet communication network in which, when a reception terminal receives a packet, an ACK packet for informing a transmission terminal that the packet is received is returned to the transmission terminal which adjusts a transmission speed. At this time, when the ACK arrives, it is confirmed whether or not there is an RTT and a packet abandonment, and a feature amount for indicating a state of a packet network is obtained on the basis of whether or not there is the obtained RTT and the packet abandonment, then a parameter for indicating a relation between the feature amount for indicating the state of the packet network and the packet transmission speed is identified and the optimal packet transmission speed is determined on the basis of the identified parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a flow required by a communication terminal of a packet communication system or an ATM communication system for communicating and storing information using cells and packets when determining the transmission speed of cells and packets. The present invention relates to a control method and a flow control device.

[0002]

2. Description of the Related Art When packet discard is observed, the number of packets or the amount of information that can be transmitted at a time (referred to as a window size) is reduced, and when it is confirmed by ACK that a packet has reached a receiving terminal normally. In the TCP scheme for increasing the window size, Tah
oe and Reno (WR Stevens, TCP / I
P Illustrated, Vol. 1: The P
rotocols, Addison-Wesley, 1
994).

In addition, the minimum RTT (BaseRTT) and the current RTT (ActR
TT), window size (WindowSize),
From the number of packets or the amount of information (SendPacket) transmitted during the ActRTT, Dif is calculated by Expression (1).
f is calculated, and Expected = WindowSize / BaseRTT Actual = SendPacket / ActRTT Diff = Expected-Actual (1) Using appropriate thresholds α, β (α <β), Windas is determined as follows: L.
S. Blankmo, L .; L. Peterson, "T
CP Vegas: End to End Conge
STATION AVOIDANCE ON A GLOB
al Internet ", IEEE J. on Se
selected Areas in Comm. , Vo
l. 13, No. 8, pp. 1465-1480. Oc
t. 1995). 1) If Diff <α, during the next RTT, Wind
owSize is increased linearly. 2) If Diff> β, Wind during the next RTT
owSize is reduced linearly. 3) When α <Diff <β, WindowSize is not changed.

[0004]

SUMMARY OF THE INVENTION Conventional Taho
e, Reno, it is necessary to determine the speed of increase and decrease of the window size, and Vegas, it is necessary to determine the threshold values α and β. However, since the relationship between these parameters and the optimal window size control is not clear, it is necessary to determine the optimal parameters. Was difficult. Also, the network configuration used for parameter determination (for example, the minimum value of RTT (propagation delay)
And the buffer capacity of the packet switch), the throughput is greatly reduced.

The present invention has been made in consideration of the above circumstances, and has as its object to provide a flow control method and a flow control device capable of determining an optimum packet transmission speed in a communication terminal.

[0006]

According to the present invention, when a receiving terminal receives a packet, an ACK packet for notifying the transmitting terminal that the packet has been received is returned to the transmitting terminal.
A flow control method in a packet communication network in which a transmission terminal adjusts a packet transmission speed, wherein when an ACK arrives, an RTT (Round Trip Time; ACK)
The difference between the arrival time and the packet transmission time) and the presence / absence of packet discarding are obtained, and a feature quantity indicating the state of the packet network is obtained based on the obtained RTT and the presence / absence of packet discarding. A parameter indicating a relationship with the packet transmission speed is identified, and an optimum packet transmission speed is determined based on the identified parameter.

[0007] Preferably, the characteristic value indicating the state of the packet network includes an average value of an occurrence rate of a packet discard section and a packet discard section length (a section length at which packets are continuously discarded in a communication network or a receiving terminal). Alternatively, an average value of a product of at least two of the occurrence rate of the packet discard section, the packet discard section length, and the packet transmission speed may be obtained.

[0008] Preferably, the parameter may be identified on the basis of a past packet transmission speed and a characteristic quantity indicating a state of a packet network. The present invention relates to a flow used in a packet communication network in which, when a receiving terminal receives a packet, an ACK packet for notifying the transmitting terminal of the reception of the packet is returned to the transmitting terminal, and the transmitting terminal adjusts a packet sending speed. When the ACK arrives, the control device receives an RTT (Round Trip Tim).
e; means for determining the difference between the ACK arrival time and the packet transmission time) and whether or not to discard the packet;
Means for obtaining a characteristic amount indicating the state of the packet network based on T and whether or not the packet has been discarded; means for identifying a parameter indicating the relationship between the characteristic amount indicating the state of the packet network and the packet transmission speed; Means for determining an optimum packet transmission speed based on parameters.

Preferably, the characteristic value indicating the state of the packet network includes an average value of an occurrence rate of a packet discard section and a packet discard section length (a section length in which packets are continuously discarded in a communication network or a receiving terminal). Alternatively, an average value of a product of at least two of the occurrence rate of the packet discard section, the packet discard section length, and the packet transmission speed may be obtained.

[0010] Preferably, the identification means may identify the parameter based on a past packet transmission speed and a characteristic quantity indicating a state of a packet network. According to the present invention, it is possible to determine an optimum packet transmission speed in a communication terminal. The above-described invention is also realized as a machine-readable medium storing a program for causing a computer to execute the corresponding procedure or means.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration of a flow control device according to an embodiment of the present invention. FIG. 2 shows a processing procedure of flow control by the present flow control device.

The present flow control device can be realized as software. Further, the flow control device or software for realizing the flow control device is typically mounted on a communication terminal to be controlled, but is not limited thereto.

As shown in FIG. 1, the flow control device according to the present embodiment includes an observation unit 1, a feature amount calculation unit 2, a parameter identification unit 3, and a packet transmission speed calculation unit 4. When the ACK arrives at the transmitting terminal (step S1), the observation unit 1 sets an RTT (Round Trip Tim).
e; the difference between the ACK arrival time and the packet transmission time) and the presence or absence of packet discarding is detected (step S).
2, 3).

The feature value calculation unit 2 obtains a feature value indicating the state of the packet network based on the RTT obtained by the observation unit 1 and the presence or absence of packet discard (step S4). The parameter identification unit 3 indicates a relationship between the characteristic amount indicating the state of the packet network and the packet transmission speed based on the characteristic amount obtained by the characteristic amount calculation unit 2 and the past packet transmission speed at the transmitting terminal. The parameters are identified (Step S5).

The packet transmission speed calculation unit 4 determines an optimum packet transmission speed in the transmitting terminal based on the parameters identified by the parameter identification unit 3 (Step S6).

Hereinafter, an example of an algorithm for calculating the optimum packet transmission speed according to the present embodiment will be described in more detail. First, words and variables used in the following will be described.

As illustrated in FIG. 3, the packet includes an information segment obtained by dividing transmission information into an appropriate length, and a header in which destination information and the like are written. n indicates a sequence number for the information segment in the packet. ν [by
te] is the information segment length (hereinafter referred to as packet length)
Is shown.

Λ [byte / sec] indicates the transmission speed of packet n in the transmitting terminal. Note that the optimum packet transmission speed is represented by λ ^* . Although y will be described in detail later, y indicates an occurrence rate of a packet discard section caused by discarding a packet.

E [sec] indicates a packet discard section length, which will be described in detail later. z [byte / sec] is
It shows the information rate of information that the receiving terminal can receive without error by transmitting a packet (this is called throughput).

T [sec] is, as exemplified in FIG.
Indicates the packet transmission interval. Further, y _r represents an estimate of the occurrence rate of packet discard section, e _r [sec] represents an estimate of the packet discard section length when the packet is discarded, z _r [byte / sec] is a packet It indicates the expected value of the throughput when transmitting.

Next, identification of a parameter indicating a relationship between a feature quantity indicating a state of a packet network and a packet transmission speed will be described. Here, it is assumed that the product ye of the occurrence rate y of the packet discard section and the packet discard section length e is used as the feature quantity indicating the state of the packet network.

The mathematical model of the expected value z _r [byte / sec] of the throughput at the time of transmitting a packet is expressed by equation (2), where the packet length ν is fixed. _{z r = λ {1-y} r e r λ / ν} ... (2) where a λ = ν / T. Further, y and e are calculated by the feature amount calculation unit 2 from the presence or absence of packet discard and the RTT.

Since the product of y and e is represented by a function of λ in a steady state, K is a natural number of 1 or more, α _r0 , α _r1 ,
With α _r2 ,..., α _rK , β being real numbers, the product ye of y and e can be approximated by equation (3). Here, K and β are appropriately determined in advance. ye = α ₀ + α ₁ (λ−β) +... + α _K ((λ−β) ^K + δ (3) where δ is a residual.

[0024] When expressed in the formula the product y _r e _r of y _r and e _r (4), the parameter _{α r0, α r1, α r2} , ..., α rK it is theoretically optimally identified. _{y r e r = α r0 +} α r1 · (λ-β) + ... + α rK · (λ-β) K ... (4) For example, the method of least squares (Jiro Kondo, "mathematical model - the formula of the phenomenon", Maruzen ), Α _r0 , α _r1 , α _r2 , ...,
alpha _rK is a y _r e _r and ye error epsilon as _{ε = y r e r -ye ...} (5), is identified as equation (6) is minimized.

[0025]

(Equation 1)

Here, m is the total number of ACKs arriving at the transmitting terminal, and N is the length of the sliding window. y (i) and e (i) respectively represent the occurrence rate and the packet discard section length of the packet discard section observed when the i-th ACK arrives. The former is the i-th ACK and i
1 if there is packet discard during the first ACK,
Otherwise, 0, the latter is defined by the difference between the arrival times of the i-th ACK and the (i-1) -th ACK. y _r (i) is the estimate of the occurrence rate of packet discard section, e _r (i) is an estimate of the packet discard section length, when each packet checked delivered by i-th ACK is transmitted It is calculated. FIG. 6 illustrates the relationship between n, m, and e (i).

By the way, α which minimizes the equation (6)
_{In order} to identify _r0 , _αr1 , _αr2 ,..., _αrK , it is necessary to calculate Σ in equation (6) using a sliding window of length N. Become.

Then, for example, by _{performing the following} , α _r0 , α _r0 ,
α _r1 , α _r2 ,..., α _rK can be identified. That is, first, Σ in equation (6) is replaced with the weighted average W as in equation (7).
Replace with [•].

[0029]

(Equation 2) Then, by this replacement, the optimal α _r0 , α _r1 , α
_r2 ,..., α _rK can be obtained as a solution of equation (8) by the least squares method.

[0030]

(Equation 3)

Here, λ (m) is the packet transmission speed when the packet whose delivery is confirmed by the m-th ACK is transmitted. Note that the calculation here is performed by the parameter identification unit 3.

Next, the determination of the optimum packet transmission speed will be described. As described above, the optimal α _r0 , α _r1 , α
_{If r2} ,..., α _rK are identified, the optimum packet transmission speed λ ^* can be theoretically determined similarly using the optimization theory.

For example, optimization by the extreme value method (Jiro Kondo,
According to “Mathematical Model—Formulaization of Phenomenon”, Maruzen), the optimal packet transmission rate λ ^* is obtained by differentiating Equation (2) with λ to 0.
This is the solution of equation (9).

[0034]

(Equation 4) Therefore, the solution of equation (10) obtained by modifying equation (9) becomes the optimum packet transmission speed λ ^* .

[0035]

(Equation 5)

As described above, the optimum packet transmission speed λ ^* required for determining the packet transmission speed in the transmitting terminal can be obtained. Note that this calculation is performed by the packet transmission speed calculation unit 4.

Next, a description will be given of an example of obtaining the optimum packet transmission speed λ ^* by reducing the amount of calculation in the above algorithm. As described above, the parameters α _r0 , α _r1 ,
To calculate α _r2 ,..., α _rK and the optimal packet transmission speed λ ^* as solutions of equations (8) and (10), respectively, a relatively large amount of calculation or hardware is required.

Therefore, in order to reduce the amount of calculation, the optimum packet transmission speed λ ^* can be obtained, for example, as follows. First, α _r0 = α _r1 = α _r2 , ..., α _{r (K-1)}
= 0, β = 0. Then, the parameter to be obtained is only α _rK , and this parameter α _rK can be expressed by equation (11).

Α _rK = W [(y (m) e (m)) λ (m) ^K ] / W [λ (m) ^2K ] (11) Further, equation (11) is approximately expressed as α _rK = W [(y (m) e (m))] / W [λ (m) ^K ] (12) or α _rK = W [y (m)] W [e (m)] / W [ λ (m) ^K ] (13)

The optimum packet transmission speed λ ^* can be expressed by equation (14). λ ^* = (ν / ((2 + K) α _rK )) ^{1 / (K + 1)} (14) Therefore, first, any one of the equations (11), (12) and (13) is adopted. Then, in the feature value calculation unit 2, W is used as a feature value indicating the state of the packet network.
[(Y (m) e (m)) λ (m) ^K ] (based on equation (11)) or W [(y (m) e (m))] (based on equation (12) ) Or W [y (m)] W
[E (m)] (based on equation (13)) is calculated, and then the parameter identification unit 3 receives the calculation result (this value is Data) of the feature amount calculation unit 2 and receives Data.
/ W [λ (m) ^2K ] (in the case of equation (11)) or D
By calculating ata / W [λ (m) ^K ] (Equations (12) and (13)), the relationship between the feature quantity indicating the state of the packet network and the packet transmission speed as a result of the calculation is obtained. it can be identified parameter alpha _rK showing the.

Note that the feature value calculation unit 2 and the parameter identification unit 3 calculate the weighted average W [•] in the next parameter identification to obtain the value of each weighted average W [•] obtained in the current calculation. It is necessary to memorize.

When the parameter α _rK is identified by the parameter identification unit 3 as described above, the packet transmission speed calculation unit 4 calculates the above equation (14), thereby obtaining the optimum packet transmission speed in the device terminal. λ ^* can be determined.

As described above, based on the RTT and the presence / absence of packet discard, W [(y
(M) e (m))] or W [y (m)] W [e
(M)] or W [(y (m) e (m)) λ (m) ^K ]
Equation (11) or (12)
Or, from the equation (13), the parameters α _r0 , α _r1 , α _r2 ,
.., Α _rK are identified, and the optimal packet transmission speed is _calculated by the equation (1).
It can be determined from 4).

In the above, when the window size is used instead of the packet transmission speed, the optimum window size can be calculated by the product of λ ^* and RTT.

Next, the effect of the present embodiment will be described with comparison with the prior art. In the present embodiment, the presence / absence of packet discard and the RTT are converted into the average value of the packet discard section occurrence rate and the packet discard section length representing the average characteristics of the packet network, and these are used as the characteristic amounts of the packet network. Therefore, it is less susceptible to stochastic fluctuations in the network. In addition, since the packet transmission speed theoretically derived is calculated using the relational expression between the packet transmission speed and the characteristic amount so as to obtain the optimal control result, the packet transmission speed at which the optimal control result is obtained is calculated. Can decide. Further, since the parameters of the relational expression between the packet transmission speed and the characteristic amount are estimated from the past data, they have a feature that they do not depend on a change in the structure of the network state.

In particular, FIGS. 5 and 6 show the characteristics of this embodiment when the optimum control result is the maximum throughput. Here, since the unit time is normalized to 1, the RTT is represented by a magnification relative to the unit time, and the buffer length is represented by a magnification of the number of bytes that can be transmitted per unit time. FIG. 5 shows an average throughput (Throughput) and an average packet loss rate (PLR) when the minimum RTT (D _F ) of a packet network having an average service rate of 0.4 (assuming channel capacity is 1) changes. From FIG. 5, it can be seen that even if _DF changes, the average throughput (Z_O) of this embodiment is almost equal to the average channel capacity, and the average packet discard rate (PLR_O) of this embodiment is stable. FIG. 6 shows the average throughput (Throughput) when the buffer amount (Kc) of the bottleneck queue of a packet network with an average channel capacity of 0.4 changes.
And the average packet loss rate (PLR). According to FIG.
It can be seen that in this embodiment, an average throughput substantially equal to the average channel capacity is obtained even if c changes. For comparison, the performance of Vegas, which is a conventional method, is also shown. 5 and 6, the average throughput (Z_V) of Vegas and the average packet loss rate (PLR_
V) depends on the change of the minimum RTT and the buffer amount of the network.

Each of the above functions can be implemented as software. Further, the present invention can be embodied as a machine-readable medium storing a program for causing a computer to execute the above-described procedures or means. The present invention is not limited to the above-described embodiment, and can be implemented with various modifications within the technical scope.

[0048]

According to the present invention, it is possible to determine an optimum packet transmission speed in a communication terminal.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a flow control device according to an embodiment of the present invention.

FIG. 2 is an exemplary flowchart illustrating an example of a processing procedure of flow control in the embodiment.

FIG. 3 is a diagram for explaining a relationship between a packet and an information segment;

FIG. 4 shows a packet transmission sequence and received AC.
Diagram for explaining K

FIG. 5 is a diagram for comparing changes in the throughput and the packet discard rate between the embodiment and the conventional method with respect to the minimum value of RTT.

FIG. 6 is a diagram for comparing changes in the throughput and the packet discard rate between the embodiment and the conventional method with respect to the buffer amount of the packet switch;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 observation unit 2 feature calculation unit 3 parameter identification unit 4 packet transmission speed calculation unit

Claims (6)

[Claims] When a receiving terminal receives a packet, an ACK packet for notifying the transmitting terminal that the packet has been received is returned to the transmitting terminal, and the transmitting terminal adjusts a packet transmission speed in a packet communication network. When an ACK arrives, an RTT and the presence / absence of packet discarding are obtained; a feature quantity indicating a state of the packet network is obtained based on the obtained RTT and the presence / absence of packet discarding; A flow control method comprising: identifying a parameter indicating a relationship between an amount and a packet transmission speed; and determining an optimum packet transmission speed based on the identified parameter. 2. A feature quantity indicating the state of the packet network, the average value of the occurrence rate of packet discard sections and packet discard section lengths, or the occurrence rate of packet discard sections, packet discard section lengths, and packet transmission speeds. The flow control method according to claim 1, wherein an average value of at least two products is obtained. 3. The flow control method according to claim 1, wherein the parameter is identified based on a past packet transmission speed and a characteristic amount indicating a state of a packet network. 4. When a receiving terminal receives a packet, the receiving terminal returns an ACK packet for notifying the transmitting terminal of the reception of the packet to the transmitting terminal, and the transmitting terminal is used in a packet communication network for adjusting a packet sending speed. A flow control device, wherein when an ACK arrives, means for determining whether or not RTT and packet discarding are to be performed; means for determining a characteristic amount indicating a state of a packet network based on the determined RTT and whether or not packet is to be discarded; A flow comprising: means for identifying a parameter indicating a relationship between a characteristic quantity indicating a state of a network and a packet transmission speed; and means for determining an optimum packet transmission speed based on the identified parameter. Control device. 5. An average value of a packet discarding section occurrence rate and a packet discarding section length, or a packet discarding section occurrence rate, a packet discarding section length, and a packet transmission speed, as the characteristic quantity indicating the state of the packet network. The flow control device according to claim 4, wherein an average value of at least two products is obtained. 6. The apparatus according to claim 4, wherein said identifying means identifies the parameter based on a past packet transmission speed and a characteristic quantity indicating a state of a packet network.
3. The flow control device according to claim 1.