JPS587954A - Control system for packet switching network - Google Patents

Abstract

PURPOSE:To increase the number of understood packets in a packet switching network by deciding an enquiry request whether packets can be transmitted or not in consideration of the number of packet repeaters or the number of packet repeating links. CONSTITUTION:Before transmitting a packet signal from a packet switching station PX1 to a packet switching station PX2, the propriety of transmission is enquired to the switching station PX2. After receiving the enquiry, a central control unit CC2 in the switching station PX2 discriminates whether the number of packet repeating stations from the switching station PX1 to the station PX2 or the number of packet repeating links exceeds a fixed value or not when the network is of a fixed congested status, decides the propriety of packet transmission from the discriminated result and returns the result to the switching station PX1. When the returned result shows the approval of transmission, packets can be transmitted from the switching station PX1.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a packet communication system, and particularly to a flow control system in a packet switching network.

In digital data exchange networks, including computer networks, the packet switching method, which is a type of store-and-forward method, is attracting attention as a communication method that can handle a variety of future communications, and is being used in some areas. . Packet switching means that communication messages are exchanged at
Divide into blocks of 00 bit size, add destination and other control information necessary for mesohege transfer to each block, and perform high-speed transfer in 1-block units via a storage exchanger or bucket alternator. It is a method.

However, packet switching network is a type of one-time switching network (Q
Because the network is a switching network, phenomena that cannot be seen in real-time switching networks may occur during sudden changes in traffic or heavy loads. One example of this is the so-called lock-up phenomenon, which occurs when a packet switch's loose memory becomes full, causing a network to become stuck. In order to prevent this phenomenon from occurring, it is necessary to control the entire flow of packets in the packet switching network, and this is called flow control.

There are several conventional flow control methods, but
One of these is a method in which the originating packet switch reserves a packet reception buffer in the destination packet switch before transmitting the packet.

Only 1. / In this conventionally known method, packets that can reach the destination station with one relay link and packets that can reach the destination station with, for example, 10 relay links are treated without distinction, but the latter packet is Therefore, the total amount of buffers used within the switching network and the load placed on the relay links become about 10 times larger. There is no problem when the communication network is empty, but when the network is in a busy state or near it, if short-distance packets are given priority over long-distance packets, the total number of packets that can be communicated will decrease even if the load on the network is the same. The number should be able to increase.

An object of the present invention is to provide a packet flow control method that can increase the number of transmitted packets in a packet switching network.

In a packet switching network, the terminating packet switching center responds to a request from the originating packet switching center to decide whether or not to send packets by determining the number of packet relay stations or packet relay links between the originating and terminating packet switching centers. It is characterized by taking the number into consideration to determine whether or not the request is possible, and returning the determination result to the sender.

As a result, when the network is congested, long-distance packets that place a large load on the network can be restricted from flowing into the switching network, which can quickly resolve network edge conditions and increase the communication rate of short-distance packets. It is possible to increase the number of communication packets in the network as a whole.

The present invention will be explained in more detail below using drawings and numerical formulas.

FIG. 1 is a system configuration diagram of the embodiment of the present invention, in which one packet switching station PX connected to the packet switching network, and the other stations (7) connected to the packet switching network.
- An example of transmitting a packet signal to the second packet switching center PX2 is shown. Packet switching center PX, other switching networks and other transmitting ends S1, S2...Sn are connected,
Similarly, the packet switching center PX2 has terminating ends R1, R2...
...Rn is connected. CC1 and 002 indicate the central control units of the respective packet switching centers px, , px2.

When another packet signal is to be transmitted from the packet switching center PXj to the packet switching center px2, before the transmission, the switching center PXj makes an inquiry to the switching center Px2 as to whether transmission is possible. The central control device aa2 of the switching center PX2 receives this, and when the network is in a congestion state above a certain level, the number of packet relay stations or the number of packet relay links from the switching center PX to the switching center Px2 exceeds a certain value. Based on this result, it is determined whether the packet can be transmitted or not, and this determination result is returned to the exchange PXI. Only when the return result is OK, the switching station PM transmits a packet.

FIG. 2 is a model of a packet switching network, and is an example in which N packet switching stations are connected by unidirectional relay links. Now, if we consider the case where any two packet switching centers perform communication, if we assume that there is no limit to the number of relay links and that the traffic exchange is uniform, then the communication between the two packet switching centers will be The average number of relay links is L = □ ・・・・・・・・・(1)
It is. Furthermore, if the number of packets sent from a specific packet switching center to another specific packet switching center within a unit time (1 second) is λ0, then the overall frequency of packet inflow into this network is λ=Nλo PKT/ sea---(
2). Since this arrives at the destination with an average of L links as described above, the load on each switching center and each relay link is .

As a conventional measure against congestion, if we include the quality condition that messages are not delayed for more than a certain time, the load on each relay link must be less than a certain value λθ,X, that is, λ8q ≦ λ8qX... ...
...(4) is required. If we restrict that anything exceeding this is blocked uniformly, the ratio of the number of packets transported to the number of generated packets is λ.喀λ0
At the time of ]. However, it is assumed to be N'.

On the other hand, in the method of the present invention, the number of relay exchanges is limited, and communication is performed only when the number of relay links is n or less (however, n - N). In the same way as above, the load at this time is 2 λo,, I---・λD ・・・・・(7)
Therefore, in the overload state, λ8q′ ≦λ8. It becomes X. Therefore, the ratio of the number of packets effectively transported to the number of packets generated is: That is, in the conventional method, from equation (5), Foc
−−−−−−− λ0, and in the method of the present invention, from equation (9), 1"~l −;
'-, which shows that the method of the present invention is more effective.

If the above-mentioned equations (5) and (9) are illustrated, it becomes as shown in FIG. 6. In other words, in FIG.
/λ and shows the ratio F of the number of packets carried to the number of packets generated, where the solid line is the method of the present invention and the broken line is the method of the conventional example.

The above example uses a ring network as a model and considers the case of unidirectional transmission, but the results are the same in the case of bidirectional transmission, and also consider the case of a multiple network model, for example, a chain connection of ring networks. However, the method of the present invention is still effective1. As mentioned above, according to the method of the present invention, when the network becomes congested, the number of effectively transported packets increases i~ Communication rate increases. This is advantageous when the degree of congestion is severe.

[Brief explanation of drawings]

FIG. 1 is a system configuration diagram according to an embodiment of the present invention, and FIG. 2 is a diagram showing a model of a packet switching network. Figure 3 shows an example of the effect of the present invention Figure 1゜Patent applicant
Naotaka Ide, patent attorney representing NEC Corporation

Claims (1)

[Claims] (1) In a control method for a packet switching network in which a packet signal is transmitted from a sending packet switching center connected to one packet switching network to a receiving packet switching network connected to that packet switching network, A first step in which the transmitting side packet switching center determines whether transmission is possible or not from the receiving side packet switching center, and when the switching network is in a state of width exceeding a predetermined level, the receiving side packet switching center Then, only when the number of relay links between the transmitting side packet switching center and the receiving side packet switching center is equal to or less than a predetermined number, a transmission permission signal is sent to the sending side packet switching center. A control method for a packet switching network, characterized in that it includes a second stage.