JPH11177618A - Congestion control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain impartiality of band reservation between a band by the Reno algorithm and a band by others intermingled with each other on a same network by increasing a congestion window size based on a just preceding congestion window size reduction rate and on a current congestion window size for each acknowledgement packet from a reception node in the case of avoiding congestion. SOLUTION: A layer 4 processing section of a transmission node 1 sets a connection based on a communication request from an application processing section 5 to start data transmission processing. In a congestion avoidance phase, a cwnd is increased for each reception of an ACK packet according to equation of new cwnd = old cwnd + 2×(1-R)/old cwnd, where (n) is number of packets whose reception is confirmed by the ACK packet and R is a just preceding reduction rate. Furthermore, the cwnd by the Reno algorithm is increased based on equation of new cwnd = old cwnd + 1/old cwnd.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a congestion control method, and more particularly, to a congestion control method in a transport layer protocol having a function of performing selective delivery confirmation.

[0002]

2. Description of the Related Art In general, as a typical example of a transport layer (layer 4) protocol, in TCP widely used in the Internet, window control in a transmitting / receiving node is used. In this window control, in addition to a reception notification window (adwnd) indicating the reception capability (buffer capacity) of the partner node, a congestion window (cwnd) in which the transfer capability of the network is estimated, That is, data is transmitted in the range of min (adwnd, cwnd).

[0003] Congestion control is performed using cwnd. The value of cwnd is determined based on a method called slow start, and starts from the transfer of one packet and increases the transmission amount every time a response is received until a predetermined threshold value (ssthresh) is reached. (That is, 1, 2, 4, 8).

Then, cwnd is set to a predetermined value ssthres
h, A is calculated using a method called congestion avoidance.
The transmission amount is gently increased by the following formula every time the CK is received. New cwnd = old cwnd + 1 / old cwnd

Conventionally, as another method of setting a cwnd value after detecting packet loss, a method of setting new cwnd = old cwnd / 2 (Reno algorithm) is widely known, and when a stable band is secured, A gradual increase in the amount of transmission and a reduction in the amount of transmission due to packet loss are repeated to secure a certain throughput.

Further, the amount of reduction of the Reno algorithm 1 /
With respect to No. 2, a method of setting the cwnd reduction amount according to the packet loss amount found from the SACK information is also considered (for example, see Japanese Patent Application No. 9-53215).
In this case, the rate of increase in the transmission amount is the same as in the Reno algorithm.

[0007]

However, in such a conventional congestion control method, the amount of cwnd reduction is constant particularly in the former TCP using the Reno algorithm, but the SACK information is not used in the latter method using SACK information.
Since the amount of cwnd reduction changes depending on information, when these congestion control methods coexist in the same network, the latter has a problem that a larger bandwidth is secured and fairness cannot be maintained. Was.

[0008] If the network congestion is not excessive,
Generally, packet loss occurs one by one.
Therefore, when the former congestion control method and the latter congestion control method are mixed in the same network, packets of both connections of the former method and the latter method may be discarded at the same time.

Here, since the cwnd reduction rates of both methods are not the same, when the RTT (RoundTripTime) is the same, the former method using SACK information is better than the former method using RACK information.
As compared with the method using the eno algorithm, cwnd is set to be large, and as a result, the latter method secures a large amount of bandwidth and cannot maintain fairness. The present invention is intended to solve such a problem, and when the amount of cwnd reduction is changed using SACK information, it is possible to obtain fairness in securing the bandwidth with that by the Reno algorithm mixed on the same network. It is intended to provide a congestion control method.

[0010]

In order to achieve the above object, a congestion control method according to the present invention provides a method for controlling a congestion of a data packet received by a receiving node from among a plurality of data packets transmitted discontinuously from a transmitting node. The selective response information indicating the individual acknowledgment for the packet is stored and the acknowledgment packet is transferred to the transmitting node. When the packet is lost, the selective response stored in the acknowledgment packet from the receiving node is transmitted by the transmitting node. Based on the lost data amount or the normal received data amount calculated from the information, reduce the subsequent congestion window size, at the time of subsequent congestion avoidance,
For each acknowledgment packet from the receiving node, the congestion window size is increased based on the immediately preceding congestion window size reduction rate and the current congestion window size. Further, by the transmitting node, at the time of avoiding congestion, for each acknowledgment packet from the receiving node,
From the immediately preceding congestion window size reduction rate R and the current congestion window size cwnd, cwnd + 2 × (1-
R) / cwnd to calculate a new congestion window size.

[0011]

Next, the present invention will be described with reference to the drawings. FIG. 1 is a flowchart illustrating a data transmission process according to a congestion control method according to an embodiment of the present invention;
FIG. 2 is a block diagram of a general communication system. FIG.
In FIG. 1, the transmitting node 1 and the receiving node 2 are accommodated in a communication network (network) 3, respectively, set a predetermined connection in the communication network 3, and perform data transfer via the connection.

In the transmitting node 1 and the receiving node 2, the application processing unit 5 is a processing unit that executes a higher-level application. The layer 4 processing unit 6
A processing unit that performs processing of data transmission and reception based on connection setting release, flow control, and the like; a lower layer processing unit 7
Is a processing unit that performs layer 2 processing such as frame assembly / disassembly and layer 3 processing such as routing.

The flow control of the protocol layer 4 executed by the layer 4 processing unit 6 is performed in a window system. The congestion control parameters include a reception window size (adwnd) declared to be receivable by a communication partner and a congestion window. It has a size (cwnd) and a slow start threshold (ssthresh). Further, it has start transmission sequence number information (SNO) for the purpose of preventing cwnd from being reduced more than necessary by detecting a packet loss event a plurality of times.

The number of packets that can be transmitted is min (cwn
d, adwnd), and changes during communication according to the change of each parameter. Hereinafter, a function for selecting the smaller one of A and B is represented as min (A, B). Further, when the packet loss is detected on the receiving side, the transmitting side and the receiving side are set at the time of connection setting so that selective response information (hereinafter, referred to as SACK information) is set in an acknowledgment packet (hereinafter, referred to as ACK packet). Negotiate with

Next, a congestion control process in the transmitting node will be described as an operation of the present invention with reference to FIG.
The layer 4 processing unit 6 of the transmission node 1 performs connection setting based on a communication request from the application processing unit 5, and starts the data transmission process shown in FIG.

First, ssthresh = adwnd, cwnd = 1 as initial values of each parameter for congestion control.
Packet, SNO = initial value of transmission sequence number (start transmission sequence number), and R = cwnd reduction rate (0
An initial value of <R <1) is set (step 10). Then, a slow start phase is entered, and min (cwn
d, adwnd), in this case, one data packet is transmitted (step 11), and the reception of an ACK packet is waited (step 12).

Here, an ACK packet from the receiving node 2 is received within a predetermined time (step 12: YES),
If the received ACK packet does not include the SACK information (step 15: NO), until cwnd reaches the threshold value ssthresh (step 16: YE
S), cwnd is increased by one packet (step 1)
7) If the transmission is not completed (step 19), the process returns to step 11 to transmit the subsequent packet.

Therefore, at the time of transmitting the next packet, n more packets are transmitted than at the time of the previous transmission. In this case, one packet transmitted first is confirmed to be received (n = 1), so that one packet is added and two packets are transmitted. Thereafter, every time an ACK packet is received from the receiving node indicating that the transmitted packet has been normally received by the receiving node, 1, 2, 4} and so on until the next transmittable amount cwnd reaches the threshold ssthresh. Exponentially increasing.

On the other hand, when cwnd has reached ssthresh (step 16: NO), a congestion avoidance phase is entered. In this congestion avoidance phase, cwnd is increased by the following equation based on the number n of packets for which cwnd has been acknowledged as an ACK packet each time an ACK packet is received, and the previous cwnd reduction rate R (step 18). New cwnd = old cwnd + 2 × (1-R) / old cwnd

Note that cwn based on the Reno algorithm
d is increased by the following equation. New cwnd = old cwnd + 1 / old cwnd Therefore, when all acknowledgments are received for cwnd, which is the immediately preceding transmission packet amount, cwnd increases by one packet, and gradually increases compared to the slow start phase.

In the present invention, cwnd is increased at the next congestion time, that is, at the time when cwnd reaches a peak value, using the above-described approximate expression that is close to cwnd by the Reno algorithm. It was done. Therefore, in the case of packet loss, the amount of increase in cwnd in the congestion avoidance phase becomes more moderate than that by the Reno algorithm.

In particular, when the network bandwidth is used by a plurality of connections, packet discarding occurs when the total bandwidth occupied by each connection exceeds the network bandwidth. Therefore, the method and the present invention are adopted by Reno. Even if both connections are mixed on the same network, a constant throughput can be obtained for each connection.

In the data packet loss event, no acknowledgment is received (step 12: NO), and when the time monitoring times out (step 13: YES), it is determined that a packet loss has occurred. In this case, each parameter of the congestion control is set as follows: ssthresh = min (cwnd, adwnd)
/ 2 cwnd = 1 packet (step 14).

On the other hand, when an ACK packet including SACK information is received (step 15: YES), the A
It is determined whether or not the reception sequence number indicated in the CK packet is larger than the start transmission sequence number SNO (step 20). If the reception sequence number is equal to or smaller than the start transmission sequence number SNO (step 20: NO), cwn
The reduction of d is not performed.

If the reception sequence number is larger than the start transmission sequence number SNO (step 20: YE
S), the start transmission sequence number is stored in the SNO (step 21), the cwnd reduction rate R is calculated, the value is stored and held (step 22), and cwnd reduction processing is performed (step 23). Note that the method proposed in Japanese Patent Application No. 9-53215 can be applied to the method of calculating the cwnd reduction rate R (reduction coefficient) and the process of reducing cwnd, and a description thereof will be omitted.

FIGS. 3A and 3B are explanatory diagrams showing changes in cwnd by various algorithms. FIG. 3A shows the results obtained by the conventional Reno algorithm, and FIG. 3B shows the results obtained by the conventional Reno algorithm and the result obtained by using the conventional SACK information. In the case of a mixture, (c) shows a case in which the result of the conventional Reno algorithm and the result of the present invention are mixed.

In FIG. 3A, at time T ₀ , cwn
d reaches ssthresh, and thereafter, the congestion avoidance phase occurs, and cwnd31A increases at a fixed increase rate based on the Reno algorithm described above, and the time T
Cwnd31A according to the congestion generated in _A1 and T _A2
Has been reduced.

In FIG. 3B, when the sum of the cwnd 31B by the Reno algorithm and the cwnd 32B by the conventional algorithm using the SACK information reaches the allowable transmission amount of the network, that is, congestion occurs at times T _B1 and T _B2. Occurs and the respective cwnd 31B, 32B are reduced. In this case, cw by the Reno algorithm
Compared with nd31B, the reduction rate of cwnd32B by the algorithm using the SACK information is low and increases at almost the same increase rate, so that the difference between the two gradually increases,
The fairness of the used bandwidth cannot be maintained.

In FIG. 3C, as in FIG. 3B, congestion occurs at times T _C1 and T _C2 , but cwn according to the present invention.
Since the increase rate of d32C is determined based on the immediately preceding reduction rate R, when the reduction rate R is small, the increase rate is also small, so that the cwnd immediately before the occurrence of congestion, that is, the peak value becomes almost the same value. A certain level of fairness is maintained for the bandwidth.

As described above, S notified from the receiving node
Since the receiving node calculates the amount of lost data that cannot be normally received by the receiving node based on the ACK information and reduces the amount of transmittable packets (cwnd) immediately after packet loss detection based on the amount of lost data. As compared with the case where the congestion window size cwnd is uniformly reduced irrespective of the packet loss status as described above, the number of transmission packets can be appropriately controlled according to the network load status, and network resources can be used effectively.

[0031]

As described above, the present invention relates to a method for controlling congestion of a transport layer using selective response (SACK) information from a receiving node, the method for avoiding congestion after reducing the congestion window size due to packet loss. For each acknowledgment packet from the receiving node, the congestion window size is increased based on the previous congestion window size reduction rate and the current congestion window size, so that the Reno algorithm using no selective response information Even when connections using the conventional congestion control method coexist on the same network, a certain level of fairness can be maintained in securing the bandwidth. In addition, new c
wnd = old cwnd + 2 × (1−R) / old cwnd, which is calculated using a new cwnd. Therefore, a congestion window size (peak value) that is almost the same as that by the Reno algorithm immediately before the occurrence of congestion can be obtained by a simple calculation ) Can be obtained.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a data transmission process according to a congestion control method according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a general communication system.

FIG. 3 is an explanatory diagram showing changes in cwnd due to various algorithms.

[Explanation of Symbols] 1 ... transmission node, 2 ... reception node, 3 ... communication network (network), 5 ... application processing unit, 6 ... layer 4
Processing unit 7, lower layer processing unit

Claims (2)

[Claims] 1. A transmission / reception node connected via a predetermined connection between a reception notification window size indicating a reception capability of a reception node and a congestion window size estimating a transfer capability of a network as a layer 4 protocol. In the congestion control method in which a data packet is transferred from a transmitting node to a receiving node using a smaller value as the number of transmittable packets, the receiving node receives the data packet discontinuously among a plurality of data packets transmitted from the transmitting node. The selective response information indicating the individual acknowledgment of the received data packet is stored and the acknowledgment packet is transmitted to the transmitting node. When the packet is lost, the transmitting node stores the acknowledgment packet from the acknowledgment packet from the receiving node. Lost data amount or normal received data amount calculated from selective response information Based on, reducing the subsequent congestion window size, the subsequent congestion avoidance when, every acknowledgment packet from the receiving node, and the previous congestion window size reduction rate,
A congestion control method characterized by increasing a congestion window size based on a current congestion window size. 2. The congestion control method according to claim 1, wherein the congestion window size reduction rate R and the current congestion window size cwnd for each acknowledgment packet from the receiving node when congestion is avoided. From, cwnd + 2 ×
A congestion control method, wherein a new congestion window size is calculated by (1-R) / cwnd.