JP5290216B2 - Network delay distribution monitoring apparatus, method and program - Google Patents

Description

  The present invention relates to a technique for monitoring a delay distribution of a network, and more particularly to a technique for estimating a percentile value of a delay distribution without keeping a large number of measurement data.

  Although the delay amount of packets passing through the network varies depending on the traffic amount, the delay amount is one of the important parameters of signal quality, so the network operator monitors the distribution of the delay amount and the delay amount When a state that does not satisfy a predetermined standard occurs, it is required to take some measures for eliminating the state. The predetermined criterion is defined by a percentile value, for example, “95% of packets have a delay within X seconds”.

  For example, Patent Document 1 discloses a configuration in which a simultaneous coupling distribution of measured values such as a delay amount is estimated, and a percentile value is estimated based on the estimated simultaneous coupling distribution. However, there is a problem that the processing amount is large, such as obtaining a histogram from the collected measurement values, and then creating a discrete probability distribution approximating the histogram and performing maximum likelihood estimation.

JP 2006-33715 A

  Therefore, an object of the present invention is to provide a monitoring device, method, and program for estimating the percentile value of a delay distribution by simple processing.

According to the monitoring device according to the invention,
Means for collecting and accumulating n (n is a natural number) measurement values of delay measurements performed between two node devices of the network, and between the two node devices based on the collected n measurement values Means for updating the 100p percentile value (0 <p <1) of the delay distribution, and the means for updating is based on the pre-update value of the 100p percentile value and the n measured values. And a second means for determining an updated value of the 100p percentile value using the weighting coefficient. To do.

According to another embodiment of the monitoring device of the present invention,
When the number of measurement values that are equal to or less than the value before the update of the 100p percentile value among the n measurement values is k, the weighting factor indicates that the delay distribution between the two node devices is the 100p percentile. In the case of following the value before updating the value, it is also preferable that the value is based on a probability that k measurement values among the n measurement values are equal to or less than the value before the 100p percentile value is updated.

According to another embodiment of the monitoring device of the present invention,
The weighting factor is when k is equal to 0 and n:

であり、それ以外の場合には、

Otherwise,

であることも好ましい。

It is also preferable.

Furthermore, according to another embodiment of the monitoring device of the present invention,
The updating means further comprises third means for determining a 100p percentile value in the n measurement values, and the second means comprises the 100p percentile value in the n measurement values, It is also preferable to use the updated value of the 100p percentile value for determination.

Furthermore, according to another embodiment of the monitoring device of the present invention,
It is also preferable to delete the n measurement values after updating the 100p percentile value.

According to the program according to the present invention,
A computer is caused to function as the monitoring device.

According to the monitoring method of the present invention,
A delay information collecting means for collecting n (n is a natural number) measurement values of delay measurements performed between two node devices; a weighting factor calculating means is for the collected n measurement values; Based on the 100p percentile value (0 <p <1) of the delay distribution between the two node devices, obtaining a weighting factor indicating the probability of the 100p percentile value of the delay distribution between the two node devices, and updating the estimated value And means for updating the 100p percentile value of the delay distribution between the two node devices based on the weighting factor.

  A process of creating a histogram and estimating the distribution is not necessary, and the delay distribution can be monitored with a simple process. In addition, each measurement result can be deleted after being used for updating, which has the advantage of requiring less storage capacity.

1 is a system configuration diagram according to the present invention. It is a schematic block diagram of the monitoring apparatus by this invention. It is a schematic block diagram of an estimation part.

  EMBODIMENT OF THE INVENTION The form for implementing this invention is demonstrated in detail below using drawing. FIG. 1 is a diagram showing a system configuration according to the present invention. The system according to the present invention includes a monitoring apparatus 100 and a network 200 according to the present invention. The network 200 includes a plurality of node devices indicated by squares in the figure, and the monitoring device 100 according to the present invention is configured to be able to communicate with the plurality of node devices of the network 200. Each node device has a function of transmitting a signal for delay measurement to any other node device and measuring a transmission delay with the other node device.

  FIG. 2 is a schematic configuration diagram of a monitoring apparatus according to the present invention. As shown in FIG. 2, the monitoring apparatus according to the present invention includes a delay information collection unit 1, a temporary storage unit 2, an estimation unit 3, a history storage unit 4, and an information output unit 5.

  The delay information collection unit 1 instructs the node device to execute delay measurement, collects the measurement results, and stores the collected measurement values in the temporary storage unit 2. Note that when instructing execution of delay measurement, a node device to be measured is also designated. Hereinafter, a packet path specified between any two node devices is referred to as a section.

  The estimation unit 3 updates the 100p percentile value of the section based on the 100p percentile value (0 <p <1) of each section and the measured delay value of the section stored in the temporary storage unit 2, and updates the section. The result is output to the history storage unit 4 and the information output unit 5.

  The history accumulating unit 4 holds the history of the 100p percentile value of each section. For example, the history storage unit 4 vibrates with a section in which the 100p percentile value fluctuates by a predetermined threshold or an amplitude of the 100 p percentile value by a predetermined value or more. And a section exceeding a predetermined reference, such as a section where the 100p percentile value exceeds a threshold, is output to the information output unit 5. The information output unit 5 is an interface unit with the user, and has a function of displaying the latest 100p percentile value updated by the estimation unit 3 and a section exceeding a predetermined standard from the history storage unit 4 to the user. Have.

Below, the estimation process of the 100p percentile value by the estimation part 3 is demonstrated. First, the estimation unit 3 obtains the initial value ε ₁ of the 100p percentile value of the section from the n measurement values obtained by the first delay measurement of the section stored in the temporary storage unit 2. Specifically, an n = 100, when the p = 0.95 (95 percentile values), a small value 95th, the initial value epsilon ₁ of 100p percentile value of the interval. Note that the n measurement values in the first delay measurement used for obtaining the initial value ε ₁ of the 100p percentile value are deleted from the temporary storage unit 2.

Subsequently, the estimation unit 3 updates the 100p percentile value of the section from the n measurement values newly obtained in the temporary storage unit 2 and obtained by the second delay measurement of the section (after the update). Is ε ₂ ). Note that the n measurement values in the second delay measurement used to update the 100p percentile value are deleted from the temporary storage unit 2 after the 100p percentile value is updated. Thereafter, similarly, the estimation unit 3 obtains the updated value ε _{t + 1} of the 100p percentile value of the section from the ₁₀₀ p percentile value ε _{t of the} section and n measurement values in the t-th delay measurement, Repeatedly deleting the n measurement values in the t-th delay measurement from the temporary storage unit 2.

Then, the update process of the 100p percentile value in the estimation part 3 is demonstrated. FIG. 3 is a schematic configuration diagram of the estimation unit 3, and the estimation unit 3 includes a percentile value calculation unit 31, a weight coefficient calculation unit 32, and an estimated value update unit 33. First, percentile value calculator 31, of n measured values in the t-th delay measurement, to determine the 100p percentile value x _t in the t-th delay measurements, 100p percentile x _t in the t-th delay measurement Is output to the estimated value update unit 33. Specifically, when n = 100 and p = 0.95 (95th percentile value), the 95th smallest value in the t-th delay measurement is expressed as the 100p percentile value x _t in the t-th delay measurement. To do.

The weighting factor calculation unit 32 acquires the current 100p percentile value ε _t from the history storage unit 4, and among the n measurement values in the t-th delay measurement, the value is equal to or less than the 100 p percentile value ε _t. The number k of measured values is obtained, and the weighting coefficient _Wt is calculated by the following equation.

The estimated value update unit 33 calculates the updated 100p percentile value ε _{t + 1} of the interval from the ₁₀₀ p percentile value x _t in the t-th delay measurement, the weighting factor W _t, and the current 100 p percentile value ε _t of the interval. Obtained by the following formula.
ε _{t + 1} = W _t · ε _t + (1−W _t ) · x _t (2)

The percentile value update processing according to the above equations (1) and (2) will be described below. First, if the delay distribution of a certain section follows the current 100p percentile value ε _t of the section, the probability that the delay amount is equal to or less than ε _{t in the} section is p. Therefore, among the n measurement values in the t-th delay measurement, the probability P that the value is less than or equal to the current 100p percentile value ε _t is k.
P = _n C _k p ^k (1-p) ^n−k (3)
It becomes.

The maximum value P _Max of Expression (3) is when k = np, and is expressed by the following expression.
P _Max = _n C _np p ^np (1-p) ^n-np (4)

  Α (p, n, k), which is a function of p, n, and k shown below, can be obtained by normalizing equation (3) with equation (4) and approximating it using the Stirling formula. Here, for simplicity, it is assumed that n and p are selected so that np becomes an integer value.

The present invention uses Equation (5) as a weighting factor. As is apparent from the above equation, the weighting factor is np when the result of the t-th delay measurement completely follows the current 100p percentile value ε _t , that is, the number of measured values with a delay of ε _t or less is np. The value is sometimes 1 and becomes smaller as the value of k becomes farther from np.

Equation (2) is the weighting factor calculated by equation (1), that is, the value indicating the likelihood of the current 100p percentile value epsilon _t, a value obtained by adjusting the current 100p percentile value epsilon _t, t-th delay the 100p percentile value x _t in the measurement, which also adds the adjusted value in the likelihood, therefore, 100p percentile value, in response to a change in state of the network, and gradually will be updated. In the present invention, the measurement result of each time can be deleted after being used for updating. In other words, there is an advantage that it is not necessary to hold past measurement results, and the required storage capacity can be reduced. Further, it is not necessary to create a histogram and estimate the distribution, and the delay distribution can be monitored with a simple process.

The average value of ε _t and x _t is compared with the value of equation (2), and the larger one is used as the updated 100p percentile value, or when the weighting factor W _t is larger than a predetermined threshold value, ε _t Otherwise, x _{t is set} to the updated 100p percentile value, or the value obtained by the following equation is set to the updated 100p percentile value ε _{t + 1} instead of the equation (2). good.
ε _{t + 1} = (W _t · ε _t ² + (1−W _t ) · x _t ² ) ^0.5 (6)

  In the above-described embodiment, the monitoring device is different from the node device. In the monitoring device according to the present invention, since the necessary storage capacity and the processing load are light, the function illustrated in FIG. The block may be incorporated in each node device, and each node device may monitor the 100p percentile value and notify the higher management device. In this case, the delay information collection unit 1 collects only measured values measured by the incorporated node device. Furthermore, in the embodiment described above, the number of measurement values of delay measurement at each time is n, but n may be a different value at each time.

  The monitoring apparatus according to the present invention can be realized by a program that causes a computer to function as each unit shown in FIGS. These computer programs can be stored in a computer-readable storage medium or distributed via a network. Furthermore, the present invention can be realized by a combination of hardware and software.

DESCRIPTION OF SYMBOLS 1 Delay information collection part 2 Temporary storage part 3 Estimation part 4 History storage part 5 Information output part 31 Percentile value calculation part 32 Weight coefficient calculation part 33 Estimated value update part 100 Monitoring apparatus 200 Network

Claims (7)

Means for collecting and accumulating n (n is a natural number) measurement values of delay measurements performed between two node devices of the network;
Means for updating the 100p percentile value (0 <p <1) of the delay distribution between the two node devices based on the collected n measured values;
With
The means for updating is
First means for obtaining a weighting factor indicating a probability of the value before update based on the value before update of the 100p percentile value and the n measurement values;
Second means for determining an updated value of the 100p percentile value using the weighting factor;
Monitoring device. When the number of measurement values that are equal to or less than the value before the update of the 100p percentile value among the n measurement values is k, the weighting factor indicates that the delay distribution between the two node devices is the 100p percentile. In the case where the value before the update of the value is followed, k measurement values among the n measurement values are values based on a probability that the value is not more than the value before the update of the 100p percentile value.
The monitoring device according to claim 1. The weighting factor is when k is equal to 0 and n:

Otherwise,

Is,
The monitoring device according to claim 2. The means for updating is
A third means for determining a 100p percentile value of the n measured values;
The second means uses the 100p percentile value in the n measurement values and the updated value of the 100p percentile value for determination.
The monitoring apparatus according to any one of claims 1 to 3. Deleting the n measured values after updating the 100p percentile value;
The monitoring device according to any one of claims 1 to 4.   A program that causes a computer to function as the monitoring device according to any one of claims 1 to 5. A step in which delay information collection means collects n measurement values (n is a natural number) of delay measurements performed between two node devices;
Based on the collected n measurement values and the 100p percentile value (0 <p <1) of the delay distribution between the two node devices, the weighting factor calculation means calculates 100p of the delay distribution between the two node devices. Obtaining a weighting factor indicating the probability of the percentile value;
An estimated value updating means updating a 100p percentile value of a delay distribution between the two node devices based on the weighting factor;
A delay distribution monitoring method comprising:

Images