JPH0370331A - Path selecting system in atm communication - Google Patents

Abstract

PURPOSE:To improve the throughput by distributing a traffic unit with every the same service class uniformly in each path (link). CONSTITUTION:The subject system is constituted of a communication control section 10 having a connection request application parameter reception means 11, a connection enable path identification means 12, a traffic unit comparison means 13 for a relevant service class, and a service class traffic unit storage section 14 and of a switching section 15. When a path is selected, the traffic unit of the same service class is compared as to each to select a fewer traffic unit, then the path is selected so that the traffic unit in each service class among paths is always uniformized. Thus, the load to cells of different properties is balanced to improve the throughput.

Description

[Detailed Description of the Invention] [Summary] This invention relates to a bus selection method in asynchronous transfer mode (ATM) communication, which improves network throughput without degrading the quality of service required for ATM communication, and which also has a simple configuration and high speed. Optimal bus selection is possible [Industrial Application Field] The present invention relates to a bus selection method in asynchronous transfer mode (ATM) communication.

ATM (Asynchronous
US Transfer Mode) exchange technology is CCI
TT, and wideband ISDN (Inte
rated 5 services Digital I N
Research is actively being conducted at various institutions as a technology to realize the network. Bus selection (or routing) is used to control traffic in ATM communications.

Here, "bus" refers to each transmission path of the ATM communication network,
Refers to a path through which calls are connected, including links between switches in a TM exchange (synonymous with route).

There are two types of bus selection: one is used to decide which bus to connect to when setting up a call, and the other is used to change the bus during prostitution as part of the prostitute control, but bus selection is already set for both purposes. Rerouting, which changes the bus in use, causes an inversion of the cell order, which is undesirable. Therefore, bus selection (routing) at the time of call setup, which eliminates the need for cell order preservation processing, is important.

On the other hand, in ATM communication, various types of traffic such as voice, data, video, etc. with different information bearer speeds and burstiness are handled uniformly, so if traffic with strong burstiness is mixed, fluctuations in the traffic added to the communication path may occur. As the number of cells increases, there are problems such as cell discards, delays, and deterioration of service quality, and a solution to these problems is desired.

[Prior Art] In ATM communication, transfer is performed by inserting fixed-length (for example, 53 bytes) packets called cells into vacant time slots, and the effective transfer speed can be increased by increasing or decreasing the number of cells. It can be changed. A cell is composed of a header and data, and the header includes information such as the destination.

The network looks only at the contents of the header, exchanges each cell, and delivers it to the other party. In this case, continuous signals such as audio and moving images are also converted into cells and transferred, and then converted back into continuous signals on the receiving side.

In ATM communication, as mentioned above, information with various characteristics is transmitted, and information with extremely high communication speeds (several tens of MHz, such as video) and information with low communication speeds, such as voice, are transmitted on the same bus. It is transmitted in a mixed manner. In this case, the nature of the call that occurs (a variety of terminal devices can be installed, each with different characteristics) may be one that generates information that occurs continuously (data communication information) or one that generates intermittent information. Due to the existence of devices that generate bursty information (such as audio and video information), it is necessary to control the bus selection of the transmission line that multiplexes and transfers cells.

In conventional ATM communication, since ATM communication itself is a new technology, it is not possible to give specific examples of routing control methods, but there are traffic control methods used in conventional circuit switching and synchronous packet switching. One possibility is to use technology to control prostitutes. In other words, in the case of circuit switching, the outgoing route is selected by determining whether or not the line is occupied, and in the case of packet switching, the method is to determine whether the outgoing route has free capacity. I was able to make a decision and choose a bus that had space.

When this packet switching technology is applied to ATM communication,
■Calculate the sum of the peak bandwidths declared by calls for each bus, and select randomly from among the buses that can pass.

- A possible method is to calculate the sum of peak bandwidths declared by calls for each bus, and then select the bus with the smallest used bandwidth from among the buses that can pass.

[Problem B to be solved by the invention] However, in the methods described above and (ii), other call parameters (average bandwidth, burst duration, burst occurrence interval, burst rate, etc.) are not considered. For example, when a high-speed, bursty connection request call is assigned to a bus, it is difficult to reduce cell discards and improve throughput.
Although it differs depending on burstiness and transfer speed, if the generation points of many cells overlap (peak), the transmission capacity will be exceeded, and in that case, cells will be discarded (cells are discarded without being transferred). arise. Such cell discard can be avoided by increasing the capacity of the buffer provided within the network, but this increases the delay time and is undesirable in the case of information such as voice that requires real-time performance.

On the other hand, if routing is controlled so that only calls with a capacity smaller than the transfer capacity of the transmission path are transferred, it is not possible to maximize network throughput from the viewpoint of efficiently using common resources within the network.

An object of the present invention is to provide a bus selection method for ATM communication that improves network throughput without degrading the service quality required for ATM communication, has a simple configuration, and can perform optimal bus selection at high speed. .

[Means for solving L1 problems] FIG. 1 is a basic configuration diagram of the present invention.

In FIG. 1, reference numeral 10 indicates a communication control unit, and 11 indicates a connection request/request call unit that receives declaration information including the service class of the connection request call.
12 is a connectable bus identification means; 13 is a corresponding service class call number comparison means; 14 is a service class-specific call number storage unit; and 15 is a switching unit.

The present invention provides information on a connection request call such that the attribute parameters included in the call are provided as declaration information, and when the ATM communication control device obtains the attribute parameters of the declaration information,
Calculates the number of calls for each service class currently connected to each outbound route, calculates the number of calls that would occur when the call is newly added, selects the route with the lowest number, and transfers the call. It is controlled to perform the following.

[Operation] Before explaining the operation shown in FIG. 1, the present invention will be explained in detail.

The burst traffic model is shown in A of 1 (l). This is shown in order to consider the maximization of network throughput using the multiplexing characteristics of burst traffic. Burst duration, β−'(
=1/β) represents the average burst generation interval time, T represents the cell formation time, and Nc(-α-1/T) represents the average number of cells generated during the burst duration time.

FIG. 10B shows how the traffic load on the transmission path changes over time when such bursty cells and continuous cells are multiplexed and transmitted.

In this way, load fluctuations occur due to burst data, and if the amplitude of the fluctuations is large, transmission capacity may be exceeded (cell discard occurs), and network throughput decreases.

Therefore, it can be seen that in order to maximize network throughput, it is necessary to at least balance the cell loads (number of cells generated) on each bus.

On the other hand, when such a plurality of bursty data is multiplexed, each bus is considered to contain a mixture of calls having different characteristics (burst rate, peak band, average band, etc.).

Therefore, in the present invention, calls with similar characteristics are considered to be in the same class (type), and each call is divided into classes according to its characteristics to manage each bus, and when a call occurs, it is connected. When selecting one from multiple possible buses,
The principle is to select buses so that the number of calls of the same service class on each bus is balanced.

For this purpose, parameters including the service class are declared in the connection request call, and bus selection control is performed based on the service class.

Next, the operation of FIG. 1 will be explained. The basic configuration shown in FIG. This configuration is also applied to the link routing control method for each link (including cases where the link is configured in multiple stages).

When a connection request call is generated from a terminal or a transmission line, parameters including destination information and information representing the service class and band of the call are first reported to the exchange or network (when the request is generated).

This service class is information that classifies device characteristics such as the average bandwidth of the call that generated the transfer request and the coefficient of variation into multiple types, and indicates which service class it belongs to. It is determined according to the characteristics of

The communication control unit 10 shown in FIG. 1 receives and holds the contents of the connection request call using the connection request/declaration parameter receiving means 11.

Connectable bus identification means 12 identifies a bus (usually, a plurality of routes are prepared) that can be connected to the destination indicated by the destination information included in the connection request call information.

Next, when the corresponding service class call number comparison means 13 receives the output from the connectable bus identification means 12 and further receives the declared parameters from the connection request/declaration parameter receiving means 11, the number of calls by service class storage section 14 The number of calls of the same service class as the declared parameter in the selected bus is extracted and compared.

As shown in FIG. 1, the service class call number storage unit 14 stores classes A and B for each bus 1 to n.

C...Another number of calls C1l, Cl2=, Cnl, Cn2...
is stored. The call number comparison means 13 compares the number of calls of the declared service class from among the plurality of buses selected by the connectable bus identification means 12, and selects the bus with the least number of calls.

When one bus is selected by the call number comparing means 13, the information is supplied to the switching section 15, which switches the cell of the connection request call to be connected to the selected bus.

In this way, when selecting a bus, by comparing the number of calls in the same service class for each bus and selecting the smaller one, the bus is created so that the number of calls in each service class is always equalized between each bus. By selecting , the loads on cells with different characteristics are balanced, and throughput can be improved.

[Embodiment] FIG. 2 is a configuration diagram of an ATM communication network according to the first embodiment of the present invention, FIG. 3 is a control flow diagram of the first embodiment, and FIG.
The configuration diagram of the management table of the embodiment, FIG. 5 is the second embodiment of the present invention.
FIG. 6 is a control flow diagram of the second embodiment, and FIG. 7 is a diagram of the ATM switch according to the third embodiment of the present invention.
FIG. 8 is a block diagram of the TM switch, FIG. 8 is a control flow diagram of the third embodiment, and FIG. 9 is a block diagram of the management table of the third embodiment.

In the present invention, various types of "service class" can be considered as declared by a connection request call, but the following two are specific examples.

Service class classification method (classification at higher layer) A, video service B, voice service C, data service D.

Service class classification method ■ (Classification by bearer) A, peak band α, average band β6, dispersion coefficient 7'+B, peak band α2
．． Average band β21 dispersion coefficient T1C, peak band α8. Average band β11 variance coefficient T1D,... In the case of the above classification method ■, the subscriber declares the service class when making a call, and in the case of the classification method ■, the subscriber declares the above service class when making a call, or Various parameters (α, β, r
) and the exchange determines the service class.

In the following embodiment, the operation of declaring a service class will be simply explained, but it goes without saying that any classification method among the above-mentioned types can be adopted.

[Configuration of First Embodiment] The configuration of the ATM communication network according to the first embodiment of the present invention shown in FIG. 2 will be described.

In FIG. 2, 20 is a subscriber (indicated by SL, S2);
21 is an ATM switch (indicated by SW), 22 is a transmission line (J
23 represents a transmission control unit (represented by DBC), and 24 represents a transmission path information (represented by DDB) storage unit.

Although only Sl and 32 are shown as subscribers 20, a large number of subscribers (not shown) are connected to each ATM exchange [21], and ATM communication is constituted by these multiple ATM exchanges and the transmission line 22. This is done via the network. The transmission control unit 23 performs transmission control in the ATM communication network, that is, how calls are connected to transmission lines connecting a plurality of exchanges.

The control flow in the transmission control unit 23 is shown in FIG.
Various types of information used in transmission control are stored in the transmission path information storage section 24 as a transmission path management table. The contents of the transmission path management table are shown in FIG. 4 and are composed of four tables.

In Figure 4, (1) is the transmission path usage table (D
F>, and is updated periodically depending on whether the transmission path from the monitor device is empty or blocked. (In this case, "empty/busy" indicates whether the transmission capacity of the transmission path is not exceeded by all the calls currently connected to the transmission path (there is room) or not (there is no room).

(2) is a transmission path usage band table (expressed as DAc), which represents the usage band of each transmission path based on the average band declared at the time of call origination, and is updated every time a call is generated or terminated. (3)
is a transmission path limited band table (expressed in DSc),
It represents the limited band of each transmission path and is set when initializing the transmission path. Further, (4) is a table of connected calls by service class (represented by DN), which holds the number of connected calls for each service class of each transmission path, and is updated every time a call occurs or ends.

The control flow in FIG. 3 shows the flow executed in the transmission control unit 23, which receives a connection request call from each ATM switch 21 in FIG. Performs control to instruct which transmission path to connect to.

When the subscriber (20 in Figure 2) first makes a call, A
It is received by the TM exchange 21 and transferred to the transmission control section 23. The transmission control unit 23 then starts transmission route selection. When making a call, the subscriber declares the service class (assumed to be X) and the average bandwidth (assumed to be Zc) as parameter information, and a transmission route is selected in the sequence shown below.

First, the transmission path usage table DF ((1) in Figure 4)
, a search is made for an empty transmission route that can be used for the desired call (step 31). If there is an empty transmission route (step 32), then a search for an empty transmission route using the declared peak band is performed (step 33). In this case, from among the available transmission routes, the transmission path used bandwidth table DAc
((2) in Figure 4), transmission line limit band table DSc
The average bandwidth Zc of the call declared as ((3) in FIG. 4) is used.

In other words, for each transmission path, the transmission path usage band table DA
The average band Zc is added to c, and if the added value does not exceed the limit band of the corresponding transmission path limit band table DSc, it is determined as an empty transfer route and the search is performed sequentially.

Next, it is determined whether there is an empty transmission route (step 34), and if there is, the transmission route with the least number of connected calls of the relevant service class is selected (step 35). In this case, the service class connection call loss table DN (fourth
From (4) in the figure, the number of calls of the service class declared from each transmission route selected in the above step is extracted, compared, and the transmission route with the smallest number of calls is selected.

This selected transmission route information is notified to the calling ATM switch.

In this way, the table related to the transmission path (DAc and DSc) selected for the connection request call is updated (Zc is added to the usage band of the transmission path number used by DAc, and the corresponding service class of the transmission path number used by DSc is updated. 1 is added to the number of calls (step 36), and the control is ended.

(Configuration of Second Embodiment) FIG. 5 shows the countries in which the ATM exchange according to the second embodiment of the present invention is constructed.

In FIG. 5, 50 is a subscriber (indicated by S), 51 is a network terminal device (indicated by NT), 52 is a multiplexing device (indicated by MUX), 53 is an ATM switch, and 54 is a signal processing device (indicated by MUX).
55 is a communication path control device (indicated by SWC) that controls cell connection; 56 is a call processing device that determines which outgoing line (bus) a connection request call should be connected to and performs other controls. 57 represents an ATM switch connection information (represented as DB) storage device, and 58 represents an outgoing line (bus).

FIG. 6 shows the control flow executed in the call processing device of this ATM switch.

In the control shown in FIG. 6, a table (indicated by F) stores the current usage status of each link, a table (indicated by Ac) stores the current usage band of each link according to the declared value at the time of call origination, A table (S
(indicated by c), and a table (indicated by N) storing the number of connected calls for each service class of each link. Although specific examples of each of these tables are not shown, they correspond to the tables with the same names as the management tables ((1) to (4) in FIG. 4) of the first embodiment, and have the same contents. Illustrations are omitted. However, the "bus (outgoing line) number" is used instead of the "transmission line number."

The control flow in FIG. 6 is similar to that in FIG. 3 above, and its contents will be summarized. First, in response to a connection request call from a subscriber (including declared parameters such as service class), the current usage status (empty/busy) of each bus is checked using Table F and an empty bus is selected (Ste knob 60). If there is an empty bus, step 61) selects an empty bus based on the declared average bandwidth (step 62). This search is similar to step 33 in Figure 3 above, and as a result of the search, an empty bus that satisfies the conditions is detected. Then, via step 63, the bus with the least number of connected calls for the corresponding service is selected (step 64). After this, the data is updated (Ac, N) and the process ends (step 65). If there is no free bus in step 61 or 63, it is assumed that bus selection is not possible, and the process waits or searches for another bus. It is possible to perform controls such as

[Configuration of Third Embodiment] FIG. 7 shows the countries in which the ATM switch of the third embodiment of the present invention is constructed.

This ATM switch is composed of multi-stage self-routing switches in which a route (bus) is selected in each switch, and can be used as a switch in the exchange of the second embodiment.

In FIG. 7, 70 is the first stage switch (S+ to S+3
), 71 is the primary link (Lz~Lss), 72 is the intermediate link (7)
te display), 73 is a secondary link (M + + ~ M33),
74 is a final stage switch (indicated by S, ~S0), 75 is a communication channel control device (SWC), and 76 is a signal processing device (SI
G), 77 represents a call processing device (CNP), and 78 represents a connection information storage unit (database) of the ATM switch.

The control flow executed in the call processing device of this third embodiment is shown in FIG. is shown in FIG. 9 as the structure of the management table.

A in Figure 9 is a link usage status table (indicated by F)
represents the blockage status of the primary link (71 in FIG. 7) and the secondary link, and the status is regularly updated by the monitor device. B.

is a link usage bandwidth table, which represents the bandwidth currently used by each link based on the average bandwidth declared at the time of call origination for each link, the primary link and the secondary link,
Updated each time a call occurs or ends. C1 is a link limit band table, which is the limit band of each primary link and secondary link, and is set at the time of channel vinylization.

The control flow of the third embodiment shown in FIG. 8 will be explained with reference to the tables shown in FIG. 9 above.

In advance, the service class (Zc) is sent from the subscriber's terminal as the declaration information of the connection request call.
) and the average bandwidth (representing one of the (selection possible).

In this state, control of bus selection by the switch is started, and first a search is performed for a bus that can be connected to the outgoing line. This search is based on link usage table F (A in Figure 9,
) and select an empty bus connectable to the selected outgoing line (step 81).Next, in the above selection, it is determined whether there is an empty bus or not.step 82). If so, proceed to the next step 83, retrieve the currently used bandwidth of each selected link from the used bandwidth table Ac (B in Figure 9), and calculate the average bandwidth Xi declared for each. to add. This addition is performed for each selected link of the primary link and the secondary link. The calculation results are stored in a memory work area (not shown).

Next, in step 84, it is determined whether the above added result does not exceed the link limit bandwidth of each link by referring to the corresponding contents of the link control bandwidth table Sc (C0 in FIG. 9). , if it exceeds the limited band, it will be removed from the candidates. At this time, a check is also made to see if there is a match between the primary and secondary links remaining as candidates (whether there is a bus that connects them), and if this is not possible, they are removed from the candidates.

After this, step 8 to determine whether there are any candidates remaining.
5), if any remain, match the primary link and 2
For the next combination of links, the number of connected calls table N (
The number of calls for the declared service class is read from () in Figure 9 and compared. In this case, the comparison is made by comparing the number of calls for each of the combined I-order link and secondary link. , the larger number of calls is determined, and the determined number of calls is the number of calls of the link combining the primary and secondary links. Similarly, when the number of calls for each link that is a combination of a plurality of primary links and secondary links is determined, the link with the smallest number of calls is selected from among the numbers of calls of those combined links (step 86 ). Note that the switches 70, 72 and 74 in FIG. 7 are switched according to each selected link information.

As a result, the contents of the link used bandwidth table Ac are updated with the information on the selected link (primary link, secondary link) (step 87). Furthermore, the contents of the connected call number table N are updated (step 88), and 1 is added to the number of calls of the service class declared for the corresponding link.

[Effect of the invention] As described above, according to the present invention, the ATM communication network and the
F for bus selection method in TM switch and ATM switch
The following effects can be achieved.

i. It is possible to uniformly distribute the number of calls for the same service class on each bus (link), ■ Increase in cell discard due to uneven burstiness ■ Decrease in throughput due to uneven usage band ■ Due to uneven holding time The problem of reduced throughput in conventional ATM communication can be solved.

il, the algorithm for bus selection only requires two items: calculation of the used bandwidth (sum-difference calculation) and calculation of the number of calls for each service class (sum-difference calculation), so it does not require complex calculations and can perform high-speed processing. be.

The configuration diagram of the ATM switch of the second embodiment, FIG.
FIG. 7 is a configuration diagram of an ATM switch according to the third embodiment of the present invention, FIG. 8 is a control flow diagram of the third embodiment, and FIG. 9 is a control flow diagram of the third embodiment of the present invention. FIG. 10 is a diagram showing a burst traffic model and bus load fluctuations.

In FIG. 1, 10: Communication control section 11: Connection request/declaration parameter receiving means 12: Connectable bus identification means 13: Number of calls of corresponding service class comparison means 14: Number of calls by service class storage section 15: Switching section

[Brief explanation of drawings]

Claims (3)

[Claims] (1) In a bus selection method in asynchronous transfer mode (ATM) communication, a call connection request includes parameters including the service class of the call as declaration information, and the ATM communication control unit (10) selects each of all buses. A call number storage unit (14) that stores the number of calls for each service class currently connected to the bus, a connection request/declaration parameter receiving means (11), and a call number storage unit (14) that stores the number of calls for each service class currently connected to the bus; A call number comparison means (13) which extracts and compares the calls from the call number storage section and identifies the bus with the least number of calls, and a switching section (15) which switches the bus based on the output of the call number comparison means (13). A bus selection method in ATM communication characterized by comprising: (2) A where multiple ATM exchanges are connected by transmission lines
In the bus selection method in the TM communication network, when a call connection is requested, the exchange notifies the ATM communication network of parameter information including the service class of the call, and the network control device of the ATM communication network controls all of the ATM exchanges that make up the communication network. comprises a database storing the number of calls for each service class connected to each of the transmission paths; upon receiving the connection request call from the exchange, the network control device identifies connectable transmission paths; In an ATM communication network, the number of calls in the database corresponding to the declared service class for each transmission path is compared, the transmission path with the smallest number of calls is selected, and the selected transmission path is notified to the ATM switch concerned. Bus selection method. (3) In the bus selection method in the switch of an ATM switch, when a call connection is requested, the subscriber terminal notifies parameter information including the service class of the call, and the switch control device of the ATM switch selects each link of each ATM switch. The control device includes a database storing the number of calls for each service class participating in the connection request call, and the control device identifies a connectable link when the connection request call is received, and the control device identifies the link that can be connected to the link that corresponds to the service class declared for the identified link. A bus selection method in a switch of an ATM exchange, characterized in that the number of calls in a database is compared, a link with the smallest number of calls is selected, and the switch is controlled to connect a connection request call to the selected link.