JPH07303089A - Selective delay type multistage multiplexing method - Google Patents

Abstract

PURPOSE:To provide a selective delay type multistage multiplexing method for easily making the degree of freedom for the resource allocation of connection or priority control compatible with efficient connection acception control by efficiently performing the multistage multiplexing of traffic at a connection orientated network for performing connection acception control and virtual path control for collecting similar traffic. CONSTITUTION:It is reported to a rear-stage multiplexing part (rear MUX) 15 by a request signal that a front-stage multiplexing part (front MUXi) 11 holds transmittable frames, the rear MUX 15 selects any set channel provided with minimum set channel time out of all the set channels which receive the request signal, and it is reported to the front MUXi 11 to be multiplexed to the set channel by a select signal. The front MUX 11i, that receives the select signal, selects any channel provided with minimum channel time out of the channels holding the frames and transmits the frame at the head of the channel to the selected set channel, and the rear MUX 15 transmits the frame to an access line.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a selective delay type multistage system in which a multiplexer, which receives transmission frames from a plurality of applications using a network, from a channel, finally sends out one multiplexed output to an access line. It relates to a multiplexing method.

[0002]

2. Description of the Related Art In a connection-oriented packet communication network, when an application using the network provides a service, various communication qualities are required of the network depending on the nature of the service provided by the application. In order for the network side to provide services that meet various communication quality requirements, it is necessary to reserve and secure the resources used by the connection even within the network at the start of the connection that connects applications. become.

Usually, each application declares to the network a parameter that describes the required communication quality when a connection start request is made, and the parameter is taken into consideration in consideration of the resources currently available on the network side. This is realized by determining whether or not to provide the indicated communication quality, determining routing, actually reserving resources, and performing connection acceptance control for establishing or rejecting a connection.

As long as the network secures the resources required for the connection at the time of establishing the connection and carries out the communication within the range of the communication quality parameter declared in advance, the stable communication is not affected by the load level of the network. Be able to guarantee quality.

However, relying only on application declarations for network resource allocation means that
In order to ensure stable operation of the entire network, it is necessary to adjust the parameters of all applications that use the network, which causes a difficult problem in network management. Also, the connection acceptance control itself must perform a series of processes such as connection approval / disapproval, routing, and resource reservation based on a large number of parameters, which makes it difficult to perform it at high speed and at low cost.

In order to solve such a problem, a virtual pipe (usually called a virtual path) for collecting and accommodating connections having similar traffic characteristics is set, and a virtual path is set for this virtual path. A method of performing routing and resource allocation has been proposed.

Consider a multi-stage multiplexing in which a plurality of connections (hereinafter referred to as channels) are multiplexed into one virtual path (hereinafter referred to as an aggregate channel) and a plurality of aggregate channels are multiplexed into one access line. As shown in FIG. 3, a transmission packet (hereinafter, referred to as a frame) from an application is queued in the queue of the pre-multiplexing units 31 to 33. These are first multiplexed by the former-stage multiplexer (previous MUX _i (i = 1, ... M)) 31-33. Then, the outputs of these m (preceding MUXs) 31 to 33 (corresponding to the aggregated channels) become the inputs of the post-stage multiplexing unit (post-MUX) 35, and finally one multiplexed output is obtained and accessed. Sent to the line.

FIG. 2 shows a conventional multistage multiplexing system. For simplification, regarding the pre-stage multiplexer, only the pre-stage multiplexer (pre-MUX _i ) 21, which is the i-th multiplexer, is illustrated. The pre-stage multiplexer (pre-MUX _i ) 21 has an input queue for each channel, and a frame to be transmitted is queued here (procedure i). Previous MU
In X _i 21, one channel is selected from the channels whose input queues are not empty and the head frame is transmitted to the collective channel i according to an appropriate measure described later (procedure ii). The frame sent to the collective channel i is queued in the intermediate queue (procedure iii). In the latter-stage multiplexing unit (rear-MUX) 25, one aggregate channel is selected from the aggregate channels whose intermediate queues are not empty and the head packet is transmitted to the access line in accordance with an appropriate measure as in the former stage (procedure). iv).

In steps ii and iv described above, the time-dependent variable C _i assigned to each input line i is used as a policy for selecting each multiplexer.
Priority control is used to select the first frame of the line having the smallest C _i . The following is a method of changing C _i when the j-th packet arrives at the _i- th queue. However, P _i and R _i represent the priority and the average communication band of the i-th input line, respectively. Further, T _ij and L _ij respectively represent the arrival time and the packet length of the j-th packet in the i-th queue.

C _i : = max (T _ij , C _i −P _i ) C _i : = C _i + L _ij / R _i

[0011]

In the prior art, the multiplexing in the former stage multiplexing section and the multiplexing in the latter stage multiplexing section were performed independently. Therefore, there are the following problems.

(1) An intermediate queue is required between the former-stage multiplexer and the latter-stage multiplexer (that is, the aggregate channel), and it is difficult to prevent overflow. Moreover, the hardware for that is needed.

(2) It is difficult to reflect an event that occurs after the output is actually selected by the post-multiplexing unit after being multiplexed by the pre-multiplexing unit. To prevent this, a complicated operation of rearranging the frames in the intermediate queue is required.

(3) Since the priority as a criterion for frame selection depends on time, the post-multiplexing unit needs to recalculate the priority of the frames in the intermediate queue.

These problems become more serious when the access line usage rate is high and the frames selected by the pre-multiplexing unit tend to accumulate in the intermediate queue.

In addition to priority control, work-conser
ving / non-work-conserving, time-window, packet-spa
There are various control methods such as cing, but they are implemented separately and it is difficult to control them in combination.

The present invention has been made in view of the above,
Its purpose is to efficiently perform multi-stage multiplexing of traffic in a connection-oriented network that performs connection admission control and virtual path control that collects similar traffic, and provides flexibility in connection resource allocation and priority control. It is an object of the present invention to provide a selective delay type multistage multiplexing method that facilitates compatibility with efficient connection acceptance control.

[0018]

In order to achieve the above object, in the selective delay type multi-stage multiplexing method of the present invention, a multiplexing unit which receives from a channel transmission frames from a plurality of applications using a network is finally provided. In a multiplexing method for sending one multiplexed output to an access line, a pre-multiplexing unit that collects channels having similar traffic characteristics and multiplexes each into one aggregated channel,
Further, the multiplexer is connected to the preceding multiplexer, and the latter multiplexer for multiplexing a group of the aggregated channels into one access line and the second multiplexer is provided in two stages, and increases at regular intervals with the passage of real time. Before the access line time is longer than the access line time to be transmitted to the access line by the post-multiplexing unit, if the access line time is larger, it is replaced with the value of the access line time, and the post-stage multiplexing unit replaces the frame. After being transmitted to the access line, the aggregate channel time is increased by the width having the smaller one of the frame transmission time or the predetermined maximum increment value with respect to the access line time, and the normal channel in the preceding stage multiplexing unit. When the value becomes smaller than the sum of the predetermined minimum value for the aggregate channel time and the aggregate channel time, it is replaced with the sum,
After the frame arrives on the channel, if the queue of the channel is empty, if the corresponding aggregated channel time is larger, it is replaced with the value of the aggregated channel time, and the pre-multiplexing unit transfers the frame to the aggregated channel. After transmitting, the channel time increasing by the width of the smaller one of the transmission time of the frame or the predetermined maximum increment value for the aggregate channel time is defined, and the preceding stage multiplexing unit
A request signal notifies the latter-stage multiplexing unit that a frame that can be transmitted is held, and the latter-stage multiplexing unit determines the minimum aggregate channel time from all aggregate channels that have received the request signal. Select the aggregate channel to have, notify the preceding multiplexing unit that multiplexes to the relevant aggregate channel by a selection signal, and the preceding multiplexing unit that has received the selection signal selects the smallest channel among channels holding frames. The gist of the present invention is to select a channel having the channel time, send the head frame of the channel to the selected aggregate channel, and send the frame to the access line by the latter-stage multiplexing unit.

In addition, the selective delay type multi-stage multiplexing method of the present invention is, in the above, a first offset and a first offset for making it appear that a frame has arrived earlier than the actual time with respect to the channel time and the aggregate channel time. 2 offset value is determined in advance, and after the frame has arrived at a channel whose queue is empty, the channel time of the channel is the first offset from the aggregate channel time of the corresponding aggregate channel. If the subtraction value is less than the subtraction value, the channel time is replaced with the subtraction value, and the post-multiplexing unit sets the aggregate channel before transmitting the frame from the selected aggregate channel to the access line. If the channel time is smaller than a subtraction value obtained by subtracting the second offset value from the access line time, then The case channel time and gist be replaced by the subtraction value.

Further, according to the selective delay type multi-stage multiplexing method of the present invention, when a frame arrives in the aggregate channel in a burst state, a burst amount that can be transmitted by the aggregate channel is determined in advance, and the latter-stage multiplexing unit is configured to When selecting an aggregate channel having the minimum aggregate channel time from all aggregate channels that have received a request signal, the aggregate channel time does not have a subtraction value obtained by subtracting the access line time from the aggregate channel time. The gist is to select a collective channel with time.

[0021]

In the selective delay type multi-stage multiplexing method of the present invention, the pre-stage multiplexing unit notifies the post-stage multiplexing unit of the fact that it holds a frame that can be transmitted, and the post-stage multiplexing unit sends the request signal. The aggregate channel having the minimum aggregate channel time is selected from all the aggregate channels received, and is notified to the pre-multiplexing unit that multiplexes on the aggregate channel by a selection signal, and the pre-multiplexing unit that receives the selection signal is The channel having the smallest channel time is selected from the channels holding the frame, the head frame of the channel is transmitted to the selected collective channel, and the latter-stage multiplexing unit transmits the frame to the access line.

Further, in the selective delay type multi-stage multiplexing method of the present invention, in the above, after the frame arrives at the empty channel of the queue, the pre-stage multiplexing unit sets the set channel corresponding to the channel time of the channel. If it is smaller than the subtraction value obtained by subtracting the first offset value from the channel time,
The channel time is replaced by a subtraction value, and the aggregate channel time of the aggregate channel is subtracted by subtracting the second offset value from the access channel time before sending the frame from the aggregate channel selected by the latter-stage multiplexing unit to the access channel. If less than the value, replace the aggregate channel time with the subtracted value.

Further, in the selective delay type multi-stage multiplexing method of the present invention, in the above, when a frame arrives in the aggregate channel in a burst state, the burst amount that can be transmitted by the aggregate channel is determined in advance, and the post-stage multiplexing unit minimizes the burst amount. When selecting an aggregated channel having an aggregated channel time of, the aggregated channel time having an aggregated channel time whose subtraction value obtained by subtracting the access line time is not larger than the burst amount is selected.

[0024]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a multiplexer for implementing a selective delay type multi-stage multiplexing method according to an embodiment of the present invention. The multiplexer of the present embodiment shown in the same figure multiplexes a plurality of channels into one aggregated channel and a former stage multiplexer (pre-MUX _i ) 11 which multiplexes a plurality of aggregated channels into one access line. Post-stage multiplexing unit (post-MUX) 1
It consists of 5.

Further, the following parameters are held for determining the order of frame selection in each of the multiplexing units 11 and 15. The black circles (●) are system parameters that change during the operation of the multiplexer, and their values cannot be changed externally. The white circles (◯) have a direct effect on the actual operation of the multiplexer, and it is necessary to define the value before the operation.

[0027] Next, the operation of the multiplexing device using the above parameters will be described. This is an operation that needs to be performed at a certain fixed time interval, an operation when a frame arrives at the j-th channel of the former stage multiplexing unit (previous MUX _i ) 11, and the latter stage multiplexing unit (rear MUX) 15 has an access line. It is divided into three operations when a frame is transmitted.

(A) Processing at Fixed Time Intervals First, processing that needs to be performed at fixed time intervals will be described. See the flowchart shown in FIG. 4 for this process.

<< Post-MUX >> 1. ochtime: = ochtime + 1 << all previous MUX _i >> For all channels ij

## EQU1 ## ichtime _ij : = max (ichtim
e _ij , gchtime _i + ichmin _ij ) (B) Processing when a frame arrives at the j-th channel of the previous MUX _i The processing when the frame arrives at the j-th channel of the pre-multiplexing unit (previous MUX _i ) 11 will be described. To do. For this processing, refer to the flowchart shown in FIG.

<< Previous MUX _i >> 1. Check if the queue for channel j is empty, and if empty,

## EQU00002 ## ichtime _ij : = max (ichtim
e _ij , gchtime _i- offset _ij ) 2. The frame is put at the end of the queue of the j-th channel (procedure ii shown in FIG. 1).

3. The request signal REQ _i is notified to the post-multiplexing unit (post-MUX) 15 (procedure ii shown in FIG. 1).

(C) Processing when transmitting frame to access line The processing when the post-stage multiplexing unit (rear MUX) 15 tries to transmit a frame to the access line will be described. For this processing, refer to the flowchart shown in FIG. For the procedure of selecting the collective channel p in the subsequent MUX and the procedure of selecting the channel u in the previous MUX _p in the processing of FIG. 6, refer to the flowcharts shown in FIGS. 7 and 8, respectively.

<< Post-MUX >> 1. From all the aggregate channels receiving the request signal REQ _q ,

[Equation 3] Then, the collective channel p is selected.

2. The selection signal SEL _p is notified to the previous MUX _p (procedure iii shown in FIG. 1).

<< Front MUX _p >> 3. From all the channels whose queues are not empty, select the channel u for which ichtime _pu = min (ichtime _pu ).

4. The head frame of the channel u is sent to the collective channel p (procedure iv shown in FIG. 1).

[0037]

## EQU00004 ## 5. ichtime _pu : = min (ichtime _pu + L
_pu / rate _pu , gchtime _p + ichmax _pu ) where L _pu is the transmission frame length.

<< After MUX >>

## EQU00005 ## 6. gchtime _p : = max (gchtime _p , o
chtime-goffset _p ) 7. Send the frame to the access line.

[0039]

## EQU00006 ## 8. gchtime _p : = min (gchtime _p + L
_pu / grate _p , ochtime + gchmax _p ) where L _pu is the transmission frame length.

As described above, in the selective delay type multi-stage multiplexing method of the present invention, the multiplexing in the pre-multiplexing unit and the multiplexing in the post-multiplexing unit are interlocked with each other, and the selection is performed in the post-multiplexing unit. It is characterized by performing delay selection type multiplexing in which the selection is performed in the preceding stage multiplexing unit, and for this reason, (1)
A signal notifying the post-multiplexing unit that the pre-multiplexing unit holds a frame that can be transmitted, (2) the post-multiplexing unit selects the pre-multiplexing unit as the highest priority, and the next frame Signal for notifying the former stage that it can be sent to the output of the latter stage multiplexer, (3) one of the input queues is selected in the former stage multiplexer, and the frame of the queue is selected by the latter stage multiplexer. It has the function of sending to the collective channel as an input or to the access line as an output of the subsequent stage.

Each multiplexing section has a variable for determining the selection order for each input channel, and the variable represents the ratio of the value of the bandwidth actually communicated by the traffic and the value of the average communication bandwidth. Further, in order to maintain the characteristics of each traffic on the input side and the output side of the multiplexer without being affected by other traffic volume, a value for suppressing transmission when the variable exceeds a predetermined value, A value is used to make the arrival time of the frame appear earlier than it looks. Furthermore, the maximum and minimum values that the variable can take are used.

Then, the latter-stage multiplexer receives a request signal that the former-stage multiplexer holds a frame that can be transmitted, instead of checking the intermediate queue for queuing the output frame of the former-stage multiplexer. ing. The post-stage multiplexing unit determines the pre-stage multiplexing unit to be transmitted next, from among the pre-stage multiplexing units receiving the request signal, in consideration of the time when the pre-stage multiplexing unit issues the request signal. Then, the selection is notified to the preceding-stage multiplexing unit. Upon receiving the signal from the latter-stage multiplexer, the former-stage multiplexer selects one from the non-empty input queues according to the variable indicating the priority of each channel, and The first frame is sent to the access line which is the input or output of the latter-stage multiplexer.

Each multiplexer has a variable serving as a selection criterion for each input channel, and when a frame is transmitted, it is increased by the time actually required for communication (that is, frame length / average communication band). The values of the variables are constrained to be predetermined maximum and minimum values. Further, by setting a value that makes the frame appear to arrive earlier than the actual arrival time, a finer priority control than the conventional priority control method is performed. Specifically, it is possible to preferentially transmit a frame of an input line assigned to an application in which the frame arrives sparsely at a low rate. Furthermore, in the post-stage multiplexing unit,
By defining a value that inhibits transmission when the variable is exceeded, communication is inhibited when burst communication is performed over the average communication band.

[0044]

As described above, according to the present invention,
Paying attention to the fact that the frames to be multiplexed by the pre-multiplexing unit have similar properties, when the frame arrives at the pre-multiplexing unit, the pre-multiplexing unit sends the frame itself to the post-multiplexing unit. Instead, only a request signal indicating that there is a packet to be transmitted is transmitted. This eliminates the need for intermediate queues for aggregate channels. Further, since the selection of the frame in the former-stage multiplexing unit is delayed until the selection signal from the latter-stage multiplexing unit is received, it is not the time when the request signal is sent to the latter-stage multiplexing unit.
A frame can be selected by reflecting the latest queue state at the time point selected by the latter-stage multiplexing unit. Further, even if the priority depends on the time, no frames are accumulated in the intermediate queue, so that the recalculation of the priority is unnecessary.

Further, in the present invention, various control methods can be combined and implemented by changing the setting of the parameter value. Furthermore, the number is 9
Since it is very small, it is easy to implement software and hardware.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a multiplexer for implementing a selective delay type multi-stage multiplexing method according to an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a conventional two-stage multiplexing system.

FIG. 3 is a configuration diagram showing multiplexing when a virtual path is used.

FIG. 4 is a flowchart showing a processing procedure at regular time intervals.

FIG. 5 is a flowchart showing a procedure when a frame arrives at the j-th channel of the previous MUX _i .

FIG. 6 is a flowchart showing a processing procedure when transmitting a frame to an access line.

7 is a flowchart showing a procedure for selecting an aggregate channel p by the MUX after the processing shown in FIG.

FIG. 8 shows the channel u in the previous MUX _p in the process shown in FIG.
It is a flow chart which shows the procedure of selecting.

[Explanation of symbols]

11 Pre-Multiplexing Unit (Pre-MUX _i ) 15 Post-Multiplexing Unit (Post-MUX)

Claims (3)

[Claims] 1. A multiplexing method in which a multiplexing unit, which inputs transmission frames from a plurality of applications using a network from a channel, finally outputs one multiplexed output to an access line, has similar traffic characteristics. Collect channels with one each
The former-stage multiplexer for multiplexing to one aggregated channel and the latter-stage multiplexer that is connected to the former-stage multiplexer and further multiplexes the aggregate of the aggregated channels into one access line. And the access line time that increases at regular intervals with the passage of real time, and if the access line time is larger than the access line time before the latter-stage multiplexing unit sends the frame to the access line, Value, and after the latter-stage multiplexing unit sends the frame to the access line, the latter is increased by the width having the smaller one of the sending time of the frame or a predetermined maximum increment value for the access line time. The aggregate channel time and the normal value in the preceding-stage multiplexing unit are smaller than the sum of the predetermined minimum value for the aggregate channel time and the aggregate channel time. If the queue of the channel is empty after the frame arrives at the channel, if the corresponding aggregate channel time is larger, then the aggregate channel time is replaced by the value of the aggregate channel time. After the pre-multiplexing unit sends the frame to the collective channel, the channel time is defined to increase by the smaller of the sending time of the frame or a predetermined maximum increment value with respect to the collective channel time. The former-stage multiplexing unit notifies the latter-stage multiplexing unit that it holds a frame that can be transmitted, by a request signal, and the latter-stage multiplexing unit determines the minimum among all aggregate channels that have received the request signal. Select the aggregate channel having the aggregate channel time of, notify the preceding multiplexing unit that multiplexes on the aggregate channel by a selection signal, and receive the selection signal. The pre-stage multiplexing unit selects a channel having the smallest channel time among channels holding frames, sends the head frame of the channel to the selected collective channel, and performs the post-stage multiplexing. A selective delay multistage multiplexing method, wherein the unit sends the frame to an access line. 2. A first offset and a second offset value for pretending that a frame has arrived earlier than an actual frame are predetermined with respect to the channel time and the aggregate channel time, and the pre-multiplexing unit After a frame arrives on a channel whose queue is empty, if the channel time of the channel is less than a subtraction value of the aggregate channel time of the corresponding aggregate channel minus the first offset value, the channel time is Substituting with the subtraction value, and before the latter-stage multiplexing unit sends the frame from the selected aggregation channel to the access line, the aggregation channel time of the aggregation channel is changed to the second offset value from the access channel time. If the subtracted value is smaller than the subtracted value, the aggregate channel time is replaced with the subtracted value. Select delayed multistage multiplexing method Motomeko 1 wherein. 3. When a frame arrives at a collective channel in a burst state, a burst amount that can be transmitted by the collective channel is predetermined, and the latter-stage multiplexing unit selects from among all the collective channels that have received the request signal. When selecting an aggregate channel having the smallest aggregate channel time, a aggregate channel having an aggregate channel time in which a subtracted value obtained by subtracting the access line time from the aggregate channel time is not larger than the burst amount is selected. 2. The selective delay type multistage multiplexing method according to claim 1.