JP2002135269A - Priority controller and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a priority controller which does not affect the priority of a data matrix group which is out of compensation. SOLUTION: When an output signal obtained by multiplexing (k), (m) and (n) input matrices (k, m and n are positive integers) inputted to respective multiplexers (1 to 3) is transmitted to input routes (4 to 6) as data matrices, the output signal of the data matrices is selected, weighted by a weighted round Robin(WRR) control unit (10), and outputted as a single output matrix. Consequently, the average of total bands of data matrices is less than the bands of a data matrix group transmitted to an output route (7). When a specific input data matrix out of input data matrices inputted to a certain input data matrix group is inputted in a state exceeding a weighting state, the weighting of input data matrices which keep the other weightings existing in the same data matrix group is prevented from being reduced relatively.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a priority control apparatus and a priority control method, and more particularly to, for example, a specific input data matrix of a malicious input data matrix input to a certain input data matrix group exceeding a weight. The present invention relates to a priority control device and a priority control method for preferentially processing an input data matrix that has been subjected to a well-intentioned input data matrix (which obeys other weights existing in the same data matrix group).

[0002]

2. Description of the Related Art Conventionally, a priority control device and a priority control method generally use an input data matrix and a malicious input data matrix when a malicious input data matrix and a well-intentioned input data matrix are mixed in a certain data matrix group. Classify into a data matrix and
The present invention is applied as a control device and a control method for assigning a processing priority to the classified data matrix.

[0003] Prior application invention example 1 similar to the present invention in the technical field
Japanese Patent Application Laid-Open No. 06-090255 discloses a “data network congestion control method”. In the invention example 1 of the prior application, congestion control is performed on a resource that has been subjected to congestion, and congestion control is performed within the network and at the access point. In particular, congestion control is continuously performed for each resource. As a result, the problem of fairness in the conventional congestion control method, the problem that the active channel cannot actually use all the contracted bandwidth, and the data elements from all the channels that exceeded the contracted bandwidth during congestion. , And the problem of exacerbating congestion by causing such a channel to retransmit all the discarded data elements, delay problems, etc., is said to be overcome.

[0004] Further, Japanese Patent Application No. 11-170 of the prior invention invention example 2
No. 641 uses a method of sequentially increasing weight values based on statistics.

[0005]

However, the above-mentioned conventional WRR (Weighted Round Robi)
n) With the control method alone, when a malicious input data matrix and a well-intentioned input data matrix are mixed in a specific data matrix group, no compensation is performed for the reduction of the weight for the good-intentional input data matrix, and as a result, The disadvantage was that priority was no longer guaranteed.

[0006] Further, as disclosed in Japanese Patent Application Laid-Open No. 06-090255, a method of separating a data matrix storage unit for each input data and managing the priority of a well-intentioned input data matrix is disclosed in
When the number of input data increases, the management method becomes complicated.

In the method of increasing the weight value sequentially based on statistics as disclosed in Japanese Patent Application No. 11-170641, if the weight value for the data matrix group to be compensated becomes too large, the priority for other data matrix groups is increased. Is no longer guaranteed.

An object of the present invention is to provide a priority control device and a priority control method which do not affect the priority of a data matrix group not to be compensated.

[0009]

In order to achieve the above object, a priority control apparatus according to the first aspect of the present invention includes a multiplexing apparatus in at least two multiplexing apparatuses to which at least two input matrices are input and at least two multiplexing apparatuses. And at least two input paths through which the respective output signals are transmitted as data matrices, and weighted rotation priorities in which output signals of the respective data matrices are selected and transmitted to the output path as a single output matrix by weighting ( (WRR) control unit, and the average of the total bandwidth of each data matrix is smaller than the bandwidth of the data matrix group transmitted on the output path.

In the priority control apparatus according to the present invention, a first multiplexing device (1) to which k (k is a positive integer) input matrices are input, and m (m is a positive integer) input matrices. A second multiplexer (2) to which a matrix is input, a third multiplexer (3) to which n (n is a positive integer) input matrices are input, and a first multiplexer (1) A first input path (4) in which the multiplexed output signal is transmitted as a first data matrix, and a second input path in which the output signal multiplexed in the second multiplexer (2) is transmitted as a second data matrix. Two input paths (5),
A third input path (6) through which the output signal multiplexed in the third multiplexer (3) is transmitted as a third data matrix
And the output signals of the first to third data matrices are selected,
Output path (7) as a single output matrix by weighting
Weighted rotation priority (WRR) control unit (1
0), and the average of the total bandwidth of the first to third data matrices is smaller than the bandwidth of the data matrix group transmitted to the output path (7).

According to a third aspect of the present invention, in the priority control apparatus according to the second aspect, the WRR control unit (10) stores the data matrix transmitted from the first input path (4) in series. One data matrix storage unit (11), a second data matrix storage unit (12) that stores data matrices transmitted from the second input path (5) in series, and a third input path (6) A third data matrix storage unit (13) for serially storing data matrices transmitted from
And a condition control unit (14) for selecting whether to read out data from any one of 1 to 13) and send it to the output path (7).

The data matrix storage section (11 to 13) is constituted by a first-in first-out storage element (FIFO memory), and a data matrix group to be transmitted to the output path (7) is:
The first to third data matrix storage units (11 to 13) selected by the WRR method are used by the condition control unit (14) to determine whether the first to third data matrix storage units are empty. The indicated signal and the information on the band of the data matrix group input to the first to third data matrix storage units (11 to 13) within a unit time may be output.

According to a sixth aspect of the present invention, in the priority control method, at least two input matrices are input to two multiplexing units, and each of the at least two output signals multiplexed in the at least two multiplexing units is converted into a data matrix. And output signals of at least two data matrices are selected by weighting of a weighted rotation priority (WRR) control processing method, and are transmitted and output as a single output matrix;
It is characterized in that the average of the total bandwidth of at least two data matrices is smaller than the bandwidth of the data matrix group transmitted and output.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment of a priority control device and a priority control method according to the present invention will be described in detail with reference to the accompanying drawings. 1 to 8 show an embodiment of a priority control device and a priority control method according to the present invention.

FIG. 1 is a block diagram showing a processing unit of the priority control apparatus according to the embodiment, and includes a weighted rotation priority (WR).
R) shows a priority control device of a method of inputting a data matrix group to the control unit 10.

K input matrices are input to the first multiplexer 1, m input matrices are input to the second multiplexer 2, and n input matrices are input to the third multiplexer 3. (K,
m and n are positive integers), and the multiplexed outputs are subsequently transmitted as a data matrix group through the first to third input paths 4 to 6, and are internally processed by the WRR control unit 10 according to a predetermined weight. A matrix group is selected, and on output path 7,
It is transmitted as a single output matrix. It is assumed that the average of the total bandwidth of the data matrix group transmitted through the first to third input paths 4 to 6 is smaller than the bandwidth of the data matrix group transmitted through the output path 7.

FIG. 2 is a block diagram showing the internal configuration of the WRR control unit 10 used in this embodiment. This FIG. 2 corresponds to FIG.
2 shows a more detailed configuration example of a weighted rotation priority control unit (WRR control unit) 10 in FIG.

The WRR control unit 10 has first to third data matrix storage units 1 for the first to third input paths 4 to 6.
Input matrix groups are stored in series, and the condition control unit 14 selects from which data matrix storage unit the data is read and transmitted to the output path 7.

The data matrix storages 11 to 13 are constituted by, for example, first-in first-out storage elements (FIFO memories). From the first to third data matrix storage units 11 to 13, a signal indicating to the condition control unit 14 whether the data matrix storage units 11 to 13 are empty, and the first to third data matrix units within a unit time. The information of the band of the data matrix group input to the data matrix storage units 11 to 13 is output. FIG.
In the example, the number of input paths is three, but any positive integer path may be used.

The data matrix group transmitted to the output path 7 is W
It is selected by the RR method. In other words, after the condition control unit 14 rotationally selects an input path for which the remaining weight counter 15 is not "0" and outputs data, the corresponding counter of the remaining weight counter 15 is set to "-1", and the next time selection is performed. To be the final candidate. If all the counters are "0" or if the data matrix storage units 11 to 13 are not empty and there is no input path in which the weight counter remains, a request is issued to the weight setting unit 17 and the initial setting is performed in advance. At the time of setting, the weight value set by main controller 18 is reassigned to remaining weight counter 15. At this time, the condition control unit 14 has determined that the bandwidth of each of the data matrix groups input to the first to third data matrix storage units 11 to 13 is equal to or greater than the weight assigned within a unit time. In this case, the timer 16 is started according to the excess degree, and the weight compensation time is added. The weight compensation time is, for example, a time directly proportional to the excess band.

While the timer 16 is running, each time the remaining weight counter 15 is reassigned, the weight value is set to the normal value "+1", and the priority of outputting the corresponding input data matrix group is increased. When 16 becomes "0", the compensation of the weight value is completed. Here, the timer 16
Are common to each input matrix group, so it is not necessary to prepare a plurality of them.

(Operation Example) FIG. 3 shows that in this operation example, the number of inputs in FIG. 1 is k = 1, m = 1, n = 1, and the initial values of the weight setting section 17 are first to third. The case where "5, 4, 8" is set for the input path is shown. FIG.
Shows an example of the bandwidth of the input matrix in the operation example. FIG. 5 is a diagram illustrating a state example of a main part in an operation example. FIG. 6 is a flowchart illustrating an operation example. FIG. 7 shows an example of a band of data input to the third data matrix storage unit 13 in the operation example. FIG.
In the case of the example, the bandwidth of the data matrix output from the third input path is shown.

In the following operation example, k = 1, m = 1, n = 1, the number of inputs in FIG.
Is set to "5, 4,
Consider the operation shown in Fig. 3 where 8 "is set. Here, the schedules of the bands input to the input matrices k1 and m1 are" 5 and 4 ", respectively, which are as planned. , N2, and n3 have respective scheduled bandwidths of “3, 2,
3 "(total 8).

In FIG. 4 showing the bandwidth of the actual input data matrix, the input matrix n3 is a malicious input data matrix and the input matrix group (n1 + n2) during time t1 to t4.
+ N3) exceeds the scheduled band 8. WR at this time
The operation of the R control unit 10 will be described with reference to the state values of each unit in FIG. 5 and the control flowchart activated for each unit time in FIG. In FIG. 4, the input data matrix n
Although the bandwidths of 1, n2, and n3 are shown, the WRR control unit 10 does not individually monitor the bandwidths.

In FIGS. 4 and 5, until time t1, each of the first to third input paths 4 to 6 inputs a data matrix group with a predetermined band value. For this reason, the flowchart of FIG. 6 passes "No" for both step A1 and step A5, and ends.

From time t1 to time t4, the condition control unit 14 detects that the third data matrix accumulation unit 13 has received an excess bandwidth exceeding the predetermined bandwidth (step A1 / Ye).
s), the operation state of the timer 16 is checked (step A2),
If the timer 16 has not been started, it is started (step A3), and a value corresponding to the excess band is set in the timer 16 (step A4). In this example, one unit time is added to the excess band 1. When re-assigning the value of the weight counter, the weight for the input data matrix group whose timer has been started is set to be one more than the normal value (= initial setting value) (steps A5 and A6).

From time t4 to time t8, no timer value is added because the bandwidth input to the third data matrix storage unit is within the expected bandwidth, but since the timer 16 continues to operate, Steps A5 and A6 are repeated. The priority compensation for the input data matrix n1 and the input data matrix n2 is performed at t1 to t8. After time t9, since the timer of the timer 16 is "0", the addition of the wait value is not performed.

The above operation suppresses an increase in the amount of data staying in the third data matrix storage unit 13 due to the input data matrix n3 being a malicious input data matrix.
With respect to the input matrices n1 and n2 in a good faith, it is possible to guarantee the priority based on a preset scheduled bandwidth. FIG. 7 shows the bandwidth of the input data matrices n1 to n3 in the third data matrix storage unit during a series of operations, and FIG. 8 shows the bandwidth of the data matrix output from the third input path.

According to the above embodiment, the weighted rotation priority (WR) for determining the priority of a plurality of queued data queues
R) In the control device and method, the priority is determined for each data matrix group for traffic in which a plurality of input data matrix groups exist and a plurality of input data matrices are further multiplexed in the input data matrix group.

According to this configuration, a specific input data matrix among the input data matrices input to a certain input data matrix group is input exceeding the weight (such an input data matrix is referred to as "malicious input data"). )
It is possible to prevent the weight of an input data matrix that complies with other weights existing in the same data matrix group (such an input data matrix is referred to as “good input data”) from being relatively reduced.

For this reason, the weight for the data matrix group is increased by the time set in the timer, and the input band management is not performed for each input data matrix unit, and complicated control such as input control is not performed. Of the input data matrix is compensated, and the priority is guaranteed.

According to the present invention, it is possible to compensate for damage to a well-intentioned input data matrix only by adding a timer-controlled constant weight value to a data matrix group. Further, since the increase in the weight value is suppressed to a constant value, it does not affect the priority of the data matrix group not to be compensated. Further, it is possible to guarantee the priority for a well-intentioned input data matrix by a simple management method without separating the data matrix storage unit for each input data matrix.

According to each of the above-mentioned inventions, when a malicious input data matrix and a well-intentioned input data matrix are mixed in a certain data matrix group, it is possible to guarantee a data output band for the well-intentioned input data matrix. This means that the priority of the well-being input data matrix reduced by the malicious input data matrix is converted into a timer value by converting the excess bandwidth of the data matrix group into a timer value. This is for adding a certain constant weight value.

Further, the function of compensating the data output band for a well-intentioned input data matrix can be simply configured. This is because priority control is not performed for each input data matrix, but weight control for each input path is performed based on bandwidth information for each data matrix group.

The above embodiment is an example of a preferred embodiment of the present invention. However, it is not limited to this.
Various modifications can be made without departing from the spirit of the present invention.

[0036]

As is apparent from the above description, in the priority control device and the priority control method of the present invention, at least two input matrices are input to two multiplex processing units and multiplexed in these multiplex processing units. The respective output signals are transmitted and output as a data matrix, and the output signals are selected by weighting in a weighted rotation priority (WRR) control processing method, transmitted and output as a single output matrix, and output in a total bandwidth of the data matrix. The average is smaller than the band of the data matrix group. Therefore, since the increase in the weight value is suppressed to a constant value, it does not affect the priority of the data matrix group not to be compensated.

[Brief description of the drawings]

FIG. 1 is a block diagram of a processing unit showing an embodiment of a priority control device of the present invention.

FIG. 2 is a block diagram showing a more detailed configuration example of a weighted rotation priority control unit (WRR control unit) 10 in FIG. 1;

FIG. 3 shows the number of inputs in FIG. 1 as k = 1, m = 1, n
= 1, and the initial value of the weight setting unit 17 is set to "5, 4, 8" for the first to third input paths.

FIG. 4 shows an example of a bandwidth of an input matrix in an operation example.

FIG. 5 is a diagram illustrating an example of a state of a main part in an operation example.

FIG. 6 is a flowchart illustrating an operation example.

FIG. 7 shows a third data matrix storage unit 13 in an operation example.
3 shows an example of the band of data input to the.

8 shows the bandwidth of the data matrix output from the third input path in the case of the example of FIG.

[Explanation of symbols]

1, 2, 3 Multiplexer 4, 5, 6 Input path 7 Output path 10 Weighted rotation priority (WRR: Weighted Round Rob)
in) control unit 11, 12, 13 data matrix accumulation unit 14 condition control unit 15 remaining weight counter 16 timer 17 weight setting unit 18 main controller

Claims (6)

[Claims] 1. At least two multiplexers to which at least two input matrices are input, and at least two input paths through which respective output signals multiplexed in the at least two multiplexers are transmitted as a data matrix. A weighted rotation priority (WRR) controller for selecting an output signal of each of the data matrices and transmitting the output signal as a single output matrix to the output path by weighting; Is smaller than the band of the data matrix group transmitted through the output path. 2. A first multiplexer (1) to which k (k is a positive integer) input matrices are inputted, and a second multiplexer (m) (m is a positive integer) input matrix inputted thereto. A multiplexing device (2); a third multiplexing device (3) to which n (n is a positive integer) input matrices are input; and an output signal multiplexed by the first multiplexing device (1). A first input path (4) transmitted as one data matrix, and a second input path (5) transmitted as the second data matrix in the second multiplexer (2). ), A third input path (6) through which an output signal multiplexed in the third multiplexer (3) is transmitted as a third data matrix, and an output signal of the first to third data matrices. Is selected,
Output path (7) as a single output matrix by weighting
Weighted rotation priority (WRR) control unit (1
0), and the average of the total bandwidth of the first to third data matrices is smaller than the bandwidth of the data matrix group transmitted to the output path (7). Priority control unit. 3. The WRR control section (10) includes: a first data matrix storage section (11) for serially storing a data matrix transmitted from the first input path (4); A second data matrix storage unit (12) for serially storing the data matrix transmitted from the input path (5); and a third data matrix storage unit for serially storing the data matrix transmitted from the third input path (6). And a condition control unit (14) for selecting whether to read data from any of the data matrix storage units (11 to 13) and send the data to the output path (7). 3. The priority control device according to claim 2, wherein: 4. The data matrix storage section (11 to 13).
Is a first-in first-out storage element (FIFO memory), and a data matrix group to be sent to the output path (7) is W
The priority control device according to claim 3, wherein the priority control device is selected by an RR method. 5. The first to third data matrix storage units (11 to 13) determine whether the first to third data matrix storage units are empty to the condition control unit (14). 5. A signal indicating whether or not, and information on the band of the data matrix group input to the first to third data matrix storage units (11 to 13) within a unit time are output. 3. The priority control device according to 1. 6. The at least two input matrices are input to two multiplexing units, and the at least two output signals multiplexed in the at least two multiplexing units are transmitted and output as a data matrix; The output signals of the two data matrices are selected by weighting in a weighted rotation priority (WRR) control processing scheme, transmitted and output as a single output matrix, and the average of the total bandwidth of the at least two data matrices is:
A priority control method, wherein the bandwidth is smaller than the band of the data matrix group to be transmitted and output.