JPS61140253A - Packet transfer system - Google Patents

Abstract

PURPOSE:To attain transfer of a packet with a proper transfer delay according to the request by deciding an arrival parameter obtained from addition/ subtraction of a constant as a basic priority decided for each packet as a priority parameter at the arrival time of the packet. CONSTITUTION:The arrival parameter is calculated by adding/subtracting a predetermined constant to/from the arrival time of the packet in response to a delay request of the packet. For example, the delay request value itself is used as the constant added/subtracted to/from the arrival time, for example. In this case, the arrival parameter of the packet is calculated by (Ta+Td), where Ta is the arrival time of the packet and TD is the delay request value of the packet. When the packet transmission is finished by the packet transmission section 2, a parameter comparison section 4 compares the value of arrival parameter recorded in a parameter recording section3, selects the least arrival parameter and transmits the packet having the arrival parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a packet transfer method in a packet switching network.

Generally, the packet switching method is a store-and-forward method.

This is a type of system in which the data sent from the terminal device is stored in the exchange, and then transferred at high speed within the network to be sent to the other party's terminal device. At this time, the data transferred within the network is A header containing information necessary for intra-network transfer, such as destination optical information, error control information, and transfer order number, is attached to the beginning of each block. A data block with a header like this is written as a packet (
The present invention relates to a method for transferring such packets within a network.

[Conventional technology]

Conventionally, in packet communication via the switching network described above, the transfer method is usually designed so that the transfer time required for every packet to arrive at the destination (terminal device) from the starting point is the same. there were.

However, as seen in the Advanced Information and Communication System (INS), the telephone, telegraph,
With the advent of systems that integrate the five networks of telegraph, data, and facsimile into one digital network and thereby provide a wide variety of communication services, for example, the packets that make up one voice information The packet(s) may be required to be sent quickly, that is, with minimal transmission delay, and the packet(s) constituting the message information may tolerate transmission delay. It would also be permissible to send it out a little later.

In this way, in future packet communications, the requirements for transfer delay may differ depending on the packet, so
It is desirable to control the transfer rate of packet transfer according to the required delay value.

By the way, the conventional packet transfer method is explained in detail as follows.
A method was adopted in which packets waiting to be sent were simply sent out in the same order as they arrived (hereinafter referred to as the first-come, first-served method). In special limited fields such as military use, strict delay requirements are set for packets containing important information, and the transfer method applied in such cases is A method (hereinafter referred to as a priority method) was used in which the packet with the smallest delay request value is sent out first among the packets.

In these conventional packet transfer methods, in the case of a first-come, first-served method, packets 7) are sent out regardless of the delay request value, so the average packet transfer delay is the same for all packets, and the transfer delay varies depending on the packet. The drawback was that it could not meet the demands of the public.

In addition, in the case of the priority method, as long as at least one packet with a small delay request value is waiting to be sent, a packet with a large delay request value will not be sent, so the difference between a packet with a small delay request value and a packet with a large delay request value is necessary for the packet transfer delay. This results in a large discrepancy as described above, and there is a drawback that it cannot meet the demand for appropriately varying transfer delays depending on packets.

[Problem that the invention seeks to solve]

The problem to be solved by the present invention is to enable, in a packet transfer system, to transfer each packet with an appropriate transfer delay according to the request.

Therefore, it is an object of the present invention to provide a packet transfer method that makes the above possible.

[Means and effects for solving the problem] Therefore, in the present invention, in order to achieve the above object, the basic priority determined for each packet is obtained by adding or subtracting a constant, which is determined for each packet, to the arrival time of the packet. The arrival parameter is determined as a priority parameter, and the order in which packets are sent is determined based on the magnitude of this parameter.

〔Example〕

FIG. 1 is a block diagram showing a first embodiment of the present invention. In the figure, 1 is a sending buffer, 2 is a packet sending section, 3 is a parameter recording section, and 4 is a parameter comparing section. The parameter recording section 3 constitutes means for creating and recording the packet arrival time or the above-mentioned arrival parameters, and the parameter comparison section 4 mutually compares the packet arrival times or arrival parameters recorded in the parameter recording section 3. constitute means to do so.

Note that in the following explanation, it is assumed that the smaller the time value, the more past the time.

The operation of the first embodiment will be described by taking as an example the case where arriving parameters are created and recorded in the parameter recording section 3. The arrived packet is transferred to the sending buffer 1, and at the same time, the arrival parameter of the packet is created and recorded in the parameter recording section 3. The arrival parameters are
It is calculated by adding or subtracting a predetermined constant to the packet arrival time in accordance with the delay request value of the packet (this is written in the header and can be known by reading it).

For example, the delay request value itself can be used as a constant to be added or subtracted from the arrival time. In this case, if the packet arrival time is T1 and the packet delay request value is T4, the packet arrival parameter is (Ta +'l
', ) to calculate. When the packet sending unit 2 finishes sending a packet, the parameter comparison unit 4 compares the values of the arrival parameters recorded in the parameter recording unit 3, selects the arrival parameter with the smallest value, and selects the arrival parameter having the smallest value. Instructs the packet sending unit 2 to send the packet. According to the instruction, the packet sending unit 2 takes out the packet from the sending buffer 1 and sends it out.

The arrival parameters of the sent packet are erased from the parameter recording section 3.

Through the above operations, the information on both the packet arrival order and the delay request value is appropriately combined, and packet transfer is performed with an appropriate difference in delay time depending on the delay request value of each packet.

In other words, even for packets whose delay requirement is not strict,
Since the transmission priority is determined by taking into account the arrival time (in other words, the waiting time), there is no possibility that the transmission will not be transmitted forever.

In addition, when recording the arrival time of a packet in the parameter recording unit 3, when comparing the values of arrival parameters in the parameter comparison unit 4, convert the arrival time of the packet node into an arrival parameter in the same manner as described above. For comparison purposes, exactly the same results can be obtained.

However, in this case, what is deleted from the parameter recording section 3 is:
Naturally, this is the arrival time of the sent packet.

Next, the case where packets are divided into multiple classes will be explained. Packets are classified according to their delay requirements (in other words, their basic priorities).

A second embodiment will be described below, taking as an example a case where there are packets belonging to two classes with different delay request values (basic priorities).

FIG. 2 is a block diagram showing a second embodiment of the invention. In the figure, 5 is a class identification section, 6 is a class 1 transmission waiting buffer, 7 is a class 2 transmission waiting buffer, and 8 is a class identification section.
is a packet sending unit, 9 is a class 1 parameter recording unit,
10 is a parameter recording section for class 2, and 11 is a parameter comparison section.

The parameter recording units 9 and 10 constitute means for creating and recording the arrival time or arrival parameters of a packet under the function of the class identification unit 5, and the parameter comparison unit 11
constitutes means for comparing the arrival times or arrival parameters of the packets recorded in the parameter recording units 9 and 10.

Regarding the second embodiment, parameter recording units 9 and 10
The operation will be explained by taking as an example the case where arrival parameters are recorded.

The class of the arrived packet is identified by the class identification unit 5, and the packet of class 1 is sent to the transmission waiting buffer 6.
, class 2 packets are transferred to the sending buffer 7. At the same time, the arrival parameters of that packet are
A packet of class 1 is created and recorded in the parameter recording unit 9, and a packet of class 2 is created and recorded in the parameter recording unit 10. The arrival parameter is calculated by adding or subtracting a constant predetermined for each class to the packet arrival time according to the packet delay request value (basic priority order).

For example, the delay requirement value for class 1 packets is T, +1)
, if the delay request value of the class 2 packet is Ta (21, then the class 1 constant and class 2
The constants are Ta (1) and Ta (2), respectively.

In this case, if the arrival time of the packet is T, the arrival parameter (priority parameter) of the packet is [:'r' a + Ta (1
) ], for class 2 packets [T−+Ta (2
)]. Since the arrival parameter values are relative to each other, for example, the arrival parameter of a class 1 packet is [T:, a +Ta (11Ta (2)
The arrival parameters of class 2 packets are calculated by T
The operation of packet transmission control is the same for case a.

When the packet sending unit 8 finishes sending a packet, the parameter comparison unit 11 compares the smallest arrival parameter (this value will be expressed as a) recorded in the class I parameter recording unit 9 with the class 2 parameter record. Part 10
If a<b, the oldest packet from the class 1 sending buffer 6 is selected, and if a>b, the oldest packet is For example, it instructs the packet sending unit 8 to extract the +-1 oldest packet from the class 2 sending buffer 7 and send it.

According to the instruction, the packet sending unit 8 takes out the packet from the sending buffer 6 or 7 and sends it out. The arrival parameters of the sent packet node are deleted from the parameter recording section 9 or 10. Here, the oldest packet means the packet with the smallest arrival time among the packets waiting to be sent out in the sending waiting buffer of each class.

Therefore, it is clear that the arrival parameter of the packet has the smallest value among the arrival parameters recorded in the parameter recording section of the class to which the packet belongs.

Note that the sending buffers 6 and 7 are stored in first-in, first-out format (
If a FIFO (FIFO) memory buffer is used, the oldest packet in the same class can be easily retrieved from the beginning of the memory buffer. Similarly, parameter recording sections 9 and 10
If a FIFO memory buffer is used, the smallest arriving parameter in the same class can be easily referenced and deleted at the beginning of the memory buffer.

Through the above operations, the information on both the packet arrival order and the delay request value is appropriately combined, and packet transfer is performed with an appropriate difference in delay time for each class depending on the packet delay request value.

Note that when recording the packet arrival time in the parameter recording section 9 and IO, when comparing the arrival parameter values in the parameter comparison section 11, the packet arrival time is converted into an arrival parameter in the same manner as described above, and then the comparison is performed. Exactly the same result can be obtained by doing so.

However, in this case, what is erased from the parameter recording unit 9 or 10 is, of course, the arrival time of the sent packet.

As an example, in the second embodiment of the present invention, the results of a simulation evaluation of the time during which a packet waits for transmission in the transmission waiting buffer (hereinafter referred to as waiting time) are shown below.

The simulation conditions are as follows. The required delay value for class 1 packets is 100 milliseconds.

The required delay value for class 2 packets is 200 milliseconds.

Packets arrive randomly, and the arrival rate (average number of packets arriving per second) is equal to 4 packets/second for class 1 and class 2. The time required for sending a packet in the packet sending unit 8 is 10 per bucket/node.
0 milliseconds.

The resulting average waiting times are 194 milliseconds for both class 1 and class 2 packets in the first-come, first-served method, and 67 milliseconds for class 1 packets and 67 milliseconds for class 2 packets in the priority method. 32 for packets
While the difference is large at 2 milliseconds, in the second embodiment of the present invention, it is 166 milliseconds for a class 1 packet and 223 milliseconds for a class 2 packet, so the difference is moderate.

As can be seen from this result, the present invention makes it possible to transfer packets with an appropriate difference in delay time depending on the packet delay request value and waiting time, compared to the conventional method.

〔Effect of the invention〕

As explained above, according to the present invention, an arrival parameter (priority parameter) is created taking into account the delay request value (basic priority) and packet arrival time (waiting time), and the size of the arrival parameter is By controlling the order of packet transmission based on the delay request value and waiting time, it is possible to transfer packets with an appropriate difference in delay time according to the delay request value and waiting time, eliminating the drawbacks seen in the conventional technology. It has the advantage of being possible.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention;
The figure is a block diagram showing a second embodiment of the present invention. Code explanation...Transmission waiting buffer, 2...Packet sending section, 3...Parameter recording section, 4...Parameter comparison section, 5...Class identification section, 6...Class 1 Sending waiting buffer, 7... Class 2 sending waiting buffer, 7 buffer,
8... Packet sending section, 9... Class 1 parameter recording section, 10... Class 2 parameter recording section, 1
1...Parameter comparison section. Agent Patent Attorney Akio Namiki Agent Patent Attorney Kiyoshi Matsuzaki Shin 2! It

Claims (1)

[Claims] 1) In a packet transfer method in which arriving packets are stored once and then sent to others, for each arriving packet, the priority level when transferring the packet to another is expressed. Means for determining and storing a priority parameter that takes into account a basic priority determined for each packet and its arrival time as a priority parameter, and comparing the priority parameters stored in the storage means with each other. 1. A packet transfer method, comprising means for determining a priority order by selecting a packet having a parameter determined to have a high priority and transmitting the packet in the order of the priority. 2) In a packet transfer method in which arriving packets are stored once and then sent to others, the class to which each arriving packet belongs is identified according to the basic priority determined for each packet. means and
Means for classifying and storing the arriving packets into classes according to the identification results, and a priority parameter that takes into account the arrival time as a priority parameter indicating the priority when transferring packets belonging to the same class to another. A means for determining and storing the parameters, and a means for selecting the highest parameter for each class among the parameters determined as described above, and then further comparing the priorities of the highest parameters with each other. A packet transfer method, characterized in that the packets are extracted from the storage means and sent out in order of priority according to the comparison result.