JPS63305643A - Packet network configuration system - Google Patents

Abstract

PURPOSE:To obtain a system facilitating network expansion or configuration changes by connecting a packet data terminal, a packet exchange and a voice communication equipment such as a telephone set to a basic replacement module called the constitution unit and interconnecting the constitution units so as to be as the hierarchical tree structure selecting the number of stages of the hierarchy in response to the scale of the network. CONSTITUTION:A subscriber terminal 1 comprising a packet data terminal a packet exchange or a voice communication equipment such as telephone set is connected to a constitution unit 4 comprising a basic exchange module. The constitution units 4 are interconnected so as to be of hierarchy tree structure and the number of stages of the hierarchy is selected in response to the scale of the network. A subscriber address 3 is given to be arranged for each constituting unit 4 connecting to the subscriber terminal 1 and its length depends on the number of stages of the hierarchy of the constitution units 4. When it is required to accommodate a new subscriber terminal to the network, for example, the constitution unit 6 is connected newly to the subscriber terminal allocated line of the constitution unit 4 of the lowest hierarchy order and the new subscriber terminal, is accommodated in the constitution unit 4.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to data communications, and particularly to a packet network configuration system suitable for packet switching.

[Conventional technology]

The switching capacity of current packet switching equipment, for example, NTT's D51 type packet switching equipment, has achieved a switching capacity of 104 packets/second.

In the future, as the amount of information in data communications increases, service functions expand, and the application of packet switching to new media (images, audio, etc.) increases, it will be necessary to develop a packet switch with a switching capacity of about 10B packets/second. Become.

However, the basic configuration of conventional packet switching equipment is a multi-processor configuration with a uni-bus, loop bus, and memory sharing system, and moreover, complex software processing is performed for inter-processor communication and protocol processing. . therefore.

An increase in packet switching traffic immediately causes bus and memory contention, so an increase in the number of processors has the disadvantage that switching capacity decreases.

On the other hand, if we adopt a crossbar switch, which is seen in line switching systems, or a fully coupled system, as a coupled system aiming for high-speed transfer with the highest priority on performance, when the number of processors is N, the crossbar switch can use N2 switch elements. In a fully connected network, each processor must have N-1 input/output ports. This means that as the number of processors N increases, it becomes uneconomical.

Against this background, we have developed a processor coupling method for packet switching equipment that supports high-speed transfer and an increase in the number of processors.
Nikkei Electronics 12-21゜1981, No. 97
As stated on pages 1 to 100, it is advantageous to place the switching modules of the crossbar switch in multiple stages and adopt a coupling system in which the number of switching elements does not increase on the order of N2 as in the case of a single stage. It has been stated.

This multi-stage crossbar switch configuration can be applied to the combination of 1000 processors or more, and the Omega Network, which was adopted by the supercomputer of Palose Inc. in the United States, combines shuffle exchanges in multiple stages. It exists as an example of a multi-stage switch configuration.

[Problem that the invention seeks to solve]

In a multi-stage switch network configuration method such as the Omega network mentioned above, when it becomes necessary to expand the network due to the addition of subscribers, etc., it is necessary to connect the additional subscribers and all existing subscribers. In particular, as the scale of the network increases, the number of connections increases rapidly, making it impossible to easily expand or change the configuration of the network.

An object of the present invention is to realize a Kerato network configuration method that allows network expansion and configuration changes while employing a multi-stage switch network configuration method.

[Means for solving problems]

The above purposes are for packet data terminals, packet switches,
By connecting voice communication devices such as telephones to basic exchange modules called component units, and configuring a 1W4 in which the component units are interconnected in a hierarchical tree structure in which the number of layers can be selected depending on the scale of the network, achieved.

The packet data exchange process is performed by providing a subscriber address for subscriber terminal identification, adding the packet transfer destination subscriber address to the header part of the packet data, and having the above-mentioned component unit determine the subscriber address. Execute.

The switching function of the constituent units is realized by a switching element called a bit switch, which determines the output switching path as a result of a 0.1 determination of one particular bit of input packet data.

The configuration unit operates by sequentially exchanging bit switches, but when the destination subscriber of the input packet data does not exist among the subscribers accommodated below it, the configuration unit operates by sequentially exchanging the bit switches before forwarding the packet to the next higher configuration unit. A determination section;
When the destination subscriber exists in the subscribers accommodated below itself, the destination subscriber or the subordinate constituent unit accommodating the destination subscriber is determined, and after finally determining the destination subscriber and forwarding the packet. It has a determination section.

[Effect]

In the bit switch exchange process of the component unit, the status of the internal FIFO (first-in first-out) buffer of the bit switch at the next stage of the transfer destination is monitored using a control line, and packet data is transferred to the next stage only when there is free space. bit switch.

When there is no free space, packet data is held in the internal FIFO buffer. This makes it possible to prevent packet data from colliding or being lost due to buffer overflow, thereby ensuring reliable exchange and relay operations. Furthermore, since the determination result of the bit switch always creates a unique exchange path once the destination subscriber is determined, there is no reversal of the order of packet data, and highly reliable packet J data transfer can be performed.

〔Example〕

Hereinafter, one embodiment of the present invention will be explained in order starting from FIG. FIG. 1 is an overall configuration diagram of a packet network according to the present invention. 1. In the present invention, there is a packet data terminal.

A subscriber terminal 1 consisting of either a CClTlClTl recommendation type packet switch or a voice communication device such as a telephone is connected to a constituent unit 4 consisting of a basic switching module.

The constituent units 4 are interconnected in a hierarchical tree structure as shown in FIG. 1, and the number of layers can be selected depending on the scale of the network.

The connection line 2 between the subscriber terminal and the network is assigned a subscriber address 3 that is uniquely determined within the network to identify the subscriber terminal, and the destination of packet data is this subscriber address 3. The configuration unit 4 determines the exchange route by making a determination.

The subscriber address 3 is attached to each component unit 4 to which the subscriber terminal 1 connects, and its length is determined by the number of layers in the component unit 4. For example, the two-digit number ( 2 bits) indicates that this configuration unit accommodates subscriber terminals whose upper two bits of the subscriber address are equal to these two digits in the lower hierarchy. That is, among the subscriber terminals, subscriber terminals whose subscriber addresses have the same upper two bits are collectively connected to this configuration unit.

In FIG. 1, packets are exchanged between subscriber terminals by determining the destination subscriber address added to the header part of the packet data by the constituent units. This is done by sequentially determining at each layer stage whether or not the packet data exists, that is, the packet data exchange route is determined as shown in 5a or 5b.

In FIG. 1, when it is necessary to accommodate a new subscriber terminal in the network, for example, a new component unit 6 is connected to the subscriber terminal accommodating line of the conventional component unit 4 at the lowest level of the hierarchy, This configuration unit 6 accommodates new subscriber terminals. therefore.

The subscriber address held by the new subscriber terminal is the subscriber address (OO-...
...) (... represents an arbitrary bit string of arbitrary length), two digits are further added for identifying the subscriber terminal accommodated in the new configuration unit 6 (00NeeL...).・・・
), in this way, there is no need to change the connection relationships and subscriber addresses in the existing configuration range when adding new subscriber terminals or changing connections.

Unlimited expansion and changes are possible for each constituent unit.

Note that there is no restriction that the connection line for a configuration unit added to accommodate new subscriber terminals must be that of the lowest-level configuration unit in the hierarchy, and that the connection line that can be connected to each configuration unit is There are no restrictions on the number of user terminals. For example, each component unit in FIG. 1 accommodates 22 = 4 subscriber terminals, and by using 3 of these
It can accommodate 8 subscriber terminals. Generally, 1+2+3+
・・・・・・+n = n (n + 1) / 2
can be used to form a construction unit accommodating 2n subscriber terminals. This is because one level of hierarchy can be constructed using this configuration unit as a unit.

FIG. 2 is a schematic diagram of the constituent unit. Constituent unit 4
has multiple subscriber terminal connection lines 10 (108 to 1.O
b) Connection line 11 to the upper stage component unit. There is a connection line 12 to the lower stage constituent units. Note that the topmost component in the hierarchy does not have a connection line to the uppermost component.

FIG. 3 shows a switching element called a bit switch that realizes the function of exchanging component units.

This bit switch 13 is also known as the switching element of the omega network, which is a multistage extension of the conventional shuffle exchange, and one person manually determines whether a specific bit of data is 0.1, and performs a switching operation based on the determination result. sort the input data. For example, the third
A specific bit of the subscriber address in the header of the input data packet 14 in the figure is determined, and if it is 0, the output line 15 is
When the bit a is 1, a data packet is output to the output line 15b.

FIG. 4(a) shows the connection configuration of the bit switches in the configuration unit 4. The component unit 4 has a pre-determination section 20 that extracts packet data to be transferred via the training unit in the upper layer, and a packet data to the component units in the lower layer and subscriber terminals connected to the main component unit 4.
It consists of a post-judgment section 21 for sorting data.

A configuration unit 4 in FIG. 4(a) shows the internal connections of the bit switches in the second stage from the top of the hierarchy in FIG. The reason why the internal connections of the constituent units differ depending on the hierarchy level is that the lower the constituent units are in the hierarchy, the more times the subscriber address is judged in the pre-judgment part. This is due to the different connections.

The configuration unit in FIG. 4(b) is an internal connection of the pitch switch at the top of the hierarchy in FIG.・Only sort data.

Next, an actual method of exchanging packet data will be explained using FIG. 4(a). Assume that packet data is input from a lower stage constituent unit (or a subscriber terminal) connected to the connection line 12 among the four routes extending downward in FIG. 4(a).

The packet data first enters the bit switch 22 of the predetermining section 2o, where it is determined whether the packet data is addressed to a higher-order constituent unit. That is,
A bit switch is used to determine whether the first bit (most significant bit) of the subscriber address attached to the header of the packet data is 0.1, and if it is 1, the packet data is transferred to a higher-order component unit. 23 to the connection line 11. On the other hand, if the first bit is 0, then the second bit is determined to be 0.1 by the bit switch 24, and if it is 1, the bit switch 23 similarly determines that the second bit is 0.1. It is output to the connection line 11 via.

When the second bit is O, the packet data is not a packet to be transferred to an upper level, so it is inputted from the bit switch 24 to the rear determination unit 21 of the constituent unit, where it is sent to the second bit of the lower constituent unit of this constituent unit. It determines which of the four lines the packet data is addressed to and allocates the transfer destination.

Specifically, among the four routes extending downward in FIG. 4(a), the destination subscriber address [0100...
...] is input, the pre-determination section determines the underlined two bits of (0100...-) from left to right, and the packet data is transferred from the bit switch 23 to the upper component unit. Also, the destination subscriber address (0010...
...], the previous judgment section oo O10...---
-) are judged from left to right and input to the bit switch 25 of the post-determination section. Next, in the post-judgment section (0
010...] The underlined 2 bits are determined from left to right, and the connection line 10b to the lower-level component unit accommodating the subscriber terminal set having the subscriber address [0010...] Transfer to.

Similarly, the method for exchanging the top component unit shown in FIG. 4(b) is to determine the first and second bits of the subscriber address in the header part of the packet data transferred from the four lines connecting to the lower component units. death.

When the destination subscriber address is (O0-------), the connection line 26; when it is (01...-), the connection line 27;
When it is (10...---), it is the connection times &? I28. (11
°------), the packet data is transferred to the connection line 29, respectively.

Next, FIG. 5(a) shows a control method and hardware schematic diagram of the bit switch that operates as described above.

This will be explained using (b).

Conventional Omega network etc. have each bit switch (
While the load (traffic amount) on the switching element (switching element) is average, the packet network configuration method of the present invention uses bits and
Load concentration may occur on the switch 23, such as a concentration of transfer packets to higher-level constituent units. This load concentration is more likely to occur in higher-level configuration units. Therefore, bit switches are used to prevent deterioration in data reliability due to load concentration, such as blocking of packets and data and loss or error of packets and data. control is necessary.

FIG. 5(a) shows a flowchart of a bit switch control procedure that can deal with the above-mentioned blocking and deterioration of data reliability.
This shows a hardware control procedure in which a first-in, first-out buffer is provided and transfer is not started until a transfer permission signal is received from the next-stage bit switch at the packet transfer destination. Specifically, when a packet is received at block 30, the FIF is applied to the bit switch on the input side.
After turning on the O-in-use control signal to prevent packet collisions, it is determined in block 31 whether the determined bit of the destination address is O or 1.

In block 32, it is checked whether the result and the FIFO buffer empty control line of the next stage bit switch determined based on the result are off, and if both are true, that is, the AND condition is true, then block 33 forwards the packet to the bit switch. If the FIFO of the bit switch at the next stage is not empty, the packet is held in the FIFO in this state in loop 34 and a wait state is entered.

One multiplexed bit switch (e.g., Figure 4(a)
), if multiple bit switches (e.g. 22 and 24 for 23) enter the I"I l"O queue in 23), configure them so that they are serviced fairly by hardware switching. put.

A hardware block diagram of the bit switch that performs the above control is shown in FIG. 5(b). Bit judgment/route selection circuit 41. Vacuum check circuit 42, switch/switch control section 43. Packet transmission circuit 44. Packet receiving circuit 4
Consists of 5. The time requirement for switch control depends on the transmission speed, but considering a packet of about 32 bytes on a transmission line of about 3 Mbps, the packet transit time is 86μ
It becomes s. Assuming that a 10% reduction in efficiency of the transmission line is allowed, the time is approximately 9 μS. The hardware that analyzes the subscriber address of the packet and performs the routing process and switch control within this time becomes possible by incorporating the circuit shown in FIG. 5(b) into an LSI.

FIG. 6 explains the connection between the constituent unit 4 and existing packet data terminals 10, packet switching equipment 20a, and subscriber terminals such as voice communication devices 30 such as telephones in the packet network formation system of the present invention.

That is, the packet data terminal 10 is directly connected to the configuration unit 4 because it has internal functions for assembling input data into the packet format according to the present invention and disassembling the packets into data. A conventional packet switch 1, for example, a switch 2a based on CCITT Recommendation By exchanging.

It is designed to perform packet exchange between x and 25 subscriber terminals and connections between networks. This is possible because the address system X, 25 is set outside the subscriber address system of the packet network configuration method of the present invention. A voice communication device 130 such as a telephone is an A/D.

It is connected to the component unit 4 via means 7 including a D/A converter, decoder, encoder, bucket assembly functions, etc.

FIG. 7 shows the overall structure shown in FIG. 1 in relation to each other including the connections of the bit switches inside the constituent units. The address assigned to each constituent unit indicates a set of subscriber addresses of subscriber terminals accommodated in the lower hierarchy of that constituent unit. In other words, the constituent unit (0010) has subscribers (OO10...) below it.
...) (the ... part is an address with an arbitrary bit length), and the constituent unit [φ] is at the top of the hierarchy, and all subscribers of the network It shows that it accommodates. However, the names of the above constituent units do not give $1 @ to the address of the subscriber they accommodate (because the subscriber address can be of any length). Explain the overall operation of packet switching using Figure 7.0 For example, subscriber 71 with subscriber address (000000) is subscriber 72 (
Consider the operation of sending a packet at 000011). Since the most significant bit of the destination subscriber address is (000011), the bit switch LO stage of the component unit 4 connected to the subscriber address coooooo) outputs the packet to the bit switch L1 stage with a 1 determination of O. niL
The second bit from the highest bit is set to (0).
00011), so bit switch 5 with determination 0
Outputs the packet to two stages. Similarly, the 3rd and 4th bits are sequentially determined, and the packet is output to the next stage bit switch according to the determination result. - Through the switch, the destination subscriber can be forwarded.

Next, the same subscriber 71 (000000) becomes subscriber 73
Consider the operation of sending a packet at (001110).

In this case, configuration unit 40 (00
The packet is transferred up to the 52nd stage of the bit switch 00). Here, the third bit of the destination subscriber address (001
110), and the destination of the packet is the configuration unit 40.
It is determined that (oooo) is not the subscriber terminal that it accommodates. As a result, the upper component unit 44 (
00), and the pre-determination unit again determines whether or not there is a destination subscriber in the subscriber terminal (00...) that it accommodates. That is, the most significant 4 bits of the packet header section constitute the constituent unit 40 (
0000).

41 (0001), 42 (0010), 43 (0011
) is sequentially determined whether the destination subscriber is accommodated in
When not accommodated, a higher-level component unit 4
Outputs the packet to UU (φ).In this way, the pre-judgment section of each component unit determines whether the subscriber it accommodates is the destination subscriber, and if there is no destination subscriber, it sends the packet to the higher-level component unit in the hierarchy. The packet is transferred to the pre-judgment unit of the unit, and the presence or absence of the destination subscriber is determined again.When the destination subscriber is found in a certain configuration unit, the post-judgment unit of the configuration unit transfers the packet to the lower hierarchy that accommodates the destination subscriber. The constituent units are determined, and the destination subscriber is determined through the post-judgment section of the lower constituent units in order.

Returning to the example of the destination subscriber 73 (001110), the pre-judgment section of the configuration unit 40 (00) transfers the packet to the configuration unit 43 (0011) based on the determination result of the upper 4 bits, and then the determination section transfers the packet to the upper 5 bits. , 6 bits (001
110) and transfers the packet to destination subscriber 73 (001110).

According to the above-described embodiment of the packet network configuration method according to the present invention, when a destination subscriber in a constituent unit does not exist among the subscribers accommodated by itself, the packet is transferred to a constituent unit one layer higher than the constituent unit, and subsequent destination subscriber determination is performed. When the destination subscriber exists in the subscriber that the destination subscriber accommodates,
By providing a post-determination unit that determines the destination subscriber or the lower-level component unit that accommodates the destination subscriber, finally determines the destination subscriber, and transfers the packet there, for example, traffic to a nearby area can be Compared to the conventional Omega network, which performs the same number of determinations and switching for each destination subscriber and forwards the data, the number of constituent units that go through the network is reduced, and the number of determinations and switching is reduced. In addition, there are no restrictions on the number of digits or definition method of the subscriber address that identifies the destination subscriber, and the addition of configuration units or configuration changes to the subscriber address This is easily possible simply by increasing the number of digits and changing the definition.

Regarding packet switching processing speed, compared to the conventional software processing method that determines the packet type from the packet header and performs call control, the switching route is faster by using hardware to determine the bits in the packet header. The call control software processing such as order control required when reversing the arrival of packet data can be uniquely determined.

〔Effect of the invention〕

According to the present invention, it is possible to easily increase the number of hierarchical stages of packet switching modules or the number of stages of bit switches in packet switching modules, thereby making it possible to configure a highly expandable network. Also, bit
Since network call control is omitted from switch control and packet transfer is performed by hardware and firmware, it is possible to have a processing capacity of 10B packets/sec compared to the conventional 10' packet/sec processing capacity. Become.

[Brief explanation of drawings]

FIG. 1 is an overall configuration diagram of the *a construction method according to an embodiment of the present invention.
Figure 2 is a schematic diagram of the configuration unit in Figure 1, Figure 3 is a configuration diagram of a bit switch that realizes the exchange function of the configuration unit, and Figure 4(a) is a diagram of the bit switch inside the configuration unit.
FIG. 4(b) is a connection configuration diagram of the bit switch inside the highest layer configuration unit; FIG. 5(a) is a flowchart of the packet data transfer control procedure in the bit switch; FIG. 5(I)) is a hardware schematic diagram of a bit switch, FIG. 6 is a connection relation diagram between constituent units and subscriber terminals, and FIG. be. ?゛・、・

Claims (1)

[Scope of Claims] 1. In a packet network formed by connecting a plurality of subscriber terminals corresponding to any of packet data terminals, packet switches, and voice communication devices, the constituent units consisting of basic switching modules are arranged in a hierarchical tree. A packet network configuration method characterized by interconnecting each other to form a structure. 2. The packet network configuration method according to item 1, wherein the configuration unit is formed by combining bit switches having a switching function for each bit in multiple stages. 3. The subscriber terminal has a forwarding destination subscriber address uniquely determined within the network, and the forwarding destination of the packet data is indicated by adding the subscriber address to the header part of the packet. Packet network configuration method in Section 1. 4. The packet network configuration method according to item 2, wherein the bit switch sequentially judges the bits of the transfer destination subscriber address and determines an exchange route according to the judgment result. 5. The above-mentioned structural unit is connected to the first structural unit directly connected to the structural unit or accommodated in a hierarchically lower structural unit.
In order to identify the subscriber terminal and the other second subscriber terminal, a pre-determination unit performs a bit determination on the bit positions of the common part of the subscriber address of the first subscriber terminal, and a bit position of the non-common part. 1. The packet network configuration method according to item 1, further comprising a post-determination unit that performs bit determination. 6. Putting the packet input from the packet data terminal into the pre-judgment section, looking at the forwarding destination subscriber address, and determining whether the forwarding destination is the first subscriber terminal or the second subscriber terminal; When the subscriber terminal is the second subscriber terminal, the packet is forwarded to the constituent unit one level higher, and the subsequent determination of the subscriber terminal is left to it; when the subscriber terminal is the first subscriber terminal, or when the constituent unit When a packet is transferred by entrusting the judgment of the first subscriber terminal, the post-judgment section determines the transfer destination subscriber terminal or the constituent unit one step lower that accommodates the transfer destination subscriber, and makes the final decision. 5. The packet network configuration method according to item 5, wherein the packet is transferred to a destination subscriber.