JP2007158742A - Node device and communication network system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable the communication path control to be improved in responsive performance, even if the bursts of traffic are made smooth. <P>SOLUTION: A traffic data collecting unit 21 collects traffic data, and the traffic data are stored in a traffic data storage unit 31. An accumulated distribution data calculation unit 22 calculates the accumulated distribution data from the traffic data, and the accumulated distribution data are stored in an accumulated distribution data storage unit 32. A feature parameter extraction unit 23 extracts feature parameters, which can be made correspond to the magnitude of the bursts from the accumulated distribution data. A moving average parameter acquisition unit 24 acquires moving average parameters, corresponding to the extracted feature parameters that refer to the moving average parameter storage unit 33. A moving average calculation unit 25 calculates the moving average of the traffic data by using the moving average parameters. A communication path control unit 26 controls the communication paths based on the moving average of the traffic data. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a node device that acquires traffic data of communication data to be transmitted and received and controls a communication path based on the acquired traffic data, and a communication network system using the node device.

  In recent communication networks, the introduction of GMPLS (Generalized Multi-Protocol Label Switching) technology, etc. has made it easier to manage resources and control paths across multiple layers. For this reason, it is possible to set and release communication paths at shorter time intervals. For example, it is possible to set and release communication paths according to increase or decrease in traffic during the day and night.

Incidentally, Patent Document 1 discloses an example of a technique for setting a communication path according to traffic in an optical communication network using GMPLS. According to this technique, it is possible to set a communication path capable of efficiently optimizing the cost of the entire network in consideration of traffic flowing through the network.
E. Oki, K. Shiomoto, S. Okamoto, W. Imajuku, and N. Yamanaka, “A Heuristic Multi-Layer Optimum Topology Design Scheme Based on Traffic Measurement for IP + Photonic Networks”, Optical Fiber Communication Conference and Exhibit, pp .17-22, Mar. 2002.

  In general, traffic flowing through a communication network such as the Internet is said to have self-similarity and has a large burstiness. Therefore, when the traffic is monitored and used, the monitored traffic data is not directly used, but the data obtained by smoothing the burstiness by using a moving average is used. However, when such a smoothing process is performed, it takes time until the traffic data currently monitored is reflected in the control of the communication path. That is, in the conventional technology, there is a problem that the response performance of the control is lowered when trying to solve the problem of traffic burstiness.

  Accordingly, the present invention provides a node device and a communication that can improve the time until the monitored traffic data is reflected in the control of the communication path, that is, the response performance of the control, even if the traffic burstiness is smoothed. An object is to provide a network system.

  The invention according to claim 1 is a node device that is connected to a communication line constituting a communication network and controls a communication path of communication data communicated via the communication line. For each communication path set in the communication data, traffic data collecting means for collecting traffic data of the communication data at a predetermined time; (2) traffic data accumulating means for accumulating the collected traffic data; (3 ) Feature parameter extraction for obtaining traffic data statistical data for traffic data stored in the traffic data storage means for a time sufficiently longer than the predetermined time, and extracting feature parameters representing the characteristics of the statistical data And (4) a low-level smoother that appropriately smoothes the burstiness of the traffic data. A smoothness parameter accumulating means for accumulating the smoothness parameter value of the filter corresponding to the characteristic parameter value; and (5) referring to the smoothness parameter accumulating means based on the extracted characteristic parameter value; (6) a low-pass filter having the acquired smoothness parameter value is configured, and the collected data is obtained by the low-pass filter. Traffic data smoothing means for smoothing traffic data and calculating smoothed traffic data; and (7) a communication path control means for controlling a communication path in the communication network based on the calculated smoothed traffic data. It is characterized by that.

  According to the first aspect of the present invention, the characteristic parameter extracting means obtains statistical data of the traffic data for the traffic data collected by the traffic data collecting means and accumulated in the traffic data accumulating means. The characteristic parameter which has is extracted. This characteristic parameter corresponds to an amount indicating the degree of the burstiness of the traffic data obtained from the statistical data of the traffic data.

  On the other hand, the traffic data smoothing means smoothes the collected traffic data with a low-pass filter. At this time, the traffic data smoothing means appropriately determines the smoothness parameter of the low-pass filter according to the feature parameter obtained by the feature parameter extraction means, that is, the degree of the burstiness of the traffic data. That is, the smoothness parameter acquisition means refers to the smoothness parameter storage means based on the characteristic parameter, obtains the smoothness parameter value of the low-pass filter that appropriately smoothes the burstiness of the traffic data, and smoothes the traffic data. The means constitutes a low-pass filter based on the value of the smoothness parameter.

  Therefore, the smoothed traffic data calculated by the traffic data smoothing means can be appropriately smoothed according to the degree of burstiness of the traffic data. In addition, the smoothness parameter acquisition means performs smoothing as weak as possible among the smoothness parameters of the low-pass filter that can appropriately smooth the traffic data, that is, the response performance of the low-pass filter is good. You can choose. Therefore, even if the traffic burstiness is smoothed, it is possible to realize a node device with good response performance capable of reflecting the traffic data to the communication path control as quickly as possible.

  According to a second aspect of the present invention, in the node device according to the first aspect, the low-pass filter is configured by a moving average means for moving average the collected traffic data.

  According to the second aspect of the present invention, the low-pass filter can be realized by a simple numerical calculation process.

  According to a third aspect of the present invention, in the node device according to the first or second aspect, the feature parameter extracting means fits the statistical data of the traffic data to a Pareto cumulative distribution function, thereby the feature parameter. Is extracted.

  According to the third aspect of the present invention, the feature parameter extraction means fits the statistical data of the traffic data to the Pareto cumulative distribution function, so that it is possible to extract a feature parameter suitable for the case where the traffic is congested. .

  According to a fourth aspect of the present invention, in the node device according to the first or second aspect, the feature parameter extracting means fits the statistical data of the traffic data to an exponential cumulative distribution function, thereby the feature parameter. Is extracted.

  According to the fourth aspect of the present invention, the feature parameter extraction means fits the statistical data of the traffic data to the exponential cumulative distribution function, so that it is possible to extract a feature parameter suitable for a case where the traffic is not congested. .

  According to a fifth aspect of the present invention, in the node device according to the first or second aspect, the feature parameter extraction unit (1) fits the statistical data of the traffic data to a Pareto cumulative distribution function, A first feature parameter is extracted as a feature parameter of the statistical data, and first correlation data indicating the strength of correlation between the statistical data and the Pareto cumulative distribution function is calculated, and (2) the traffic data A second feature parameter is extracted as a feature parameter of the statistical data by fitting the statistical data to the exponential cumulative distribution function, and a second correlation indicating the strength of the correlation between the statistical data and the exponential cumulative distribution function Data is calculated, (3) the first correlation data and the second correlation data are compared, and (4) the previous When the first correlation data indicates that the correlation is stronger than the second correlation data, the first feature parameter is selected as the feature parameter of the statistical data, and (5) the second correlation data is When the correlation is stronger than the first correlation data, the second feature parameter is selected as the feature parameter of the statistical data.

  According to the fifth aspect of the present invention, the feature parameter extraction means fits the statistical data of the traffic data to both the Pareto cumulative distribution function and the exponential cumulative distribution function, and further, the statistical data of the traffic data and the respective cumulative distributions Calculate correlation data with the function. Then, based on the correlation data, a feature parameter fitted to the cumulative distribution function having the stronger correlation is selected. Therefore, the feature parameter extraction means automatically selects the cumulative distribution function having the stronger correlation without acquiring whether or not the traffic is congested, and acquires the feature parameter based on the cumulative distribution function. Can do.

  According to a sixth aspect of the present invention, in the node device according to any one of the third to fifth aspects, the feature parameter extracting unit performs the fitting by a least square method.

  According to the sixth aspect of the present invention, the feature parameter extracting means can perform the fitting by a simple numerical calculation process.

  According to a seventh aspect of the present invention, in the node device according to any one of the first to sixth aspects, the smoothed traffic data calculated by the traffic data smoothing means is once or a predetermined upper limit. The communication path control means controls the predetermined communication path when the threshold value is continuously exceeded a predetermined number of times, or when the predetermined lower limit threshold value is continuously decreased once or a predetermined number of times. It is characterized by that.

  According to the seventh aspect of the present invention, the communication path control means controls the predetermined communication path when the smoothed traffic data exceeds or falls below the predetermined upper limit or lower limit threshold. Stable communication path control can be performed even for large traffic data. Further, the stability of control of the communication path can be further improved by setting the number of times that exceeds or falls below the predetermined upper limit or lower limit threshold to be a plurality of times.

  A communication network system according to an eighth aspect of the invention includes the node device according to any one of the first to seventh aspects.

  According to the eighth aspect of the present invention, the node device used in the communication network can reflect the traffic data in the control of the communication path as quickly as possible even if the traffic burstiness is smoothed. Therefore, it is possible to realize a communication network that performs control of a communication path that is stable and has good response performance.

  According to the first to seventh aspects of the present invention, it is possible to realize a node device capable of reflecting traffic data as quickly as possible in communication path control even if the traffic burstiness is smoothed. it can. According to the eighth aspect of the present invention, it is possible to realize a communication network that performs control of a communication path that is stable and has good response performance.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram illustrating an example of a configuration of a node device according to an embodiment of the present invention. As shown in FIG. 1, the node device 1 is communicated via a communication line 5 by controlling the link connection unit 10 to which a plurality of wired or wireless communication lines 5 are connected and the link connection unit 10. And a control unit 20 that controls a communication path of communication data. Specifically, a so-called router or the like corresponds to the node device 1.

  The link connection unit 10 includes an input link IF (Interface) 11, a link switch 12, an output link IF 13, and the like. The input link IF 11 demodulates the signal input from the communication line 5, extracts path information included in the packet of the communication data, and inputs it to the control unit 20 as appropriate. Further, the output link IF 13 appropriately adds path information or the like to the packet of communication data to be output based on an instruction from the control unit 20, modulates the communication data, and outputs it to the communication line 5. The link switch 12 is a switch group that switches connection between the communication line 5 that is input for each packet of communication data and the communication line 5 that is output in accordance with an instruction from the control unit 20.

  The control unit 20 is configured by a so-called computer including a CPU (Central Processing Unit), a storage device, and the like (not shown), a traffic data collection unit 21, a cumulative distribution data calculation unit 22, a feature parameter extraction unit 23, and a moving average parameter acquisition unit 24. And a functional block such as a moving average calculation unit 25 and a communication path control unit 26. The functions of these functional blocks are realized by the CPU executing programs for the respective functional blocks stored in the storage device. The control unit 20 includes a traffic data storage unit 31, a cumulative distribution data storage unit 32, a moving average parameter storage unit 33, a communication path information storage unit 34, and the like in its storage unit.

  Here, the storage device is normally configured by a semiconductor random access memory (RAM), a flash memory, or the like, but may be configured by using a hard disk storage device or the like as necessary.

  The traffic data collection unit 21 in FIG. 1 is the traffic data collection unit in the claims, the cumulative distribution data calculation unit 22 and the feature parameter extraction unit 23 are the feature parameter extraction unit in the claims, and the moving average parameter acquisition unit 24. Are the smoothing parameter acquisition means of the claims, the moving average calculation unit 25 is the traffic data smoothing means of the claims, the communication path control unit 26 is the communication path control means of the claims, and the traffic data storage unit 31. Corresponds to the traffic data storage means, and the moving average parameter storage section 33 corresponds to the smoothness parameter storage means.

  FIG. 2 is a diagram showing a procedure from collecting traffic data to controlling a communication path based on the traffic data in the embodiment of the present invention. Hereinafter, detailed functions of the functional blocks of the control unit 20 will be described in accordance with the procedure of FIG.

(Step S01)
First, the control unit 20, the traffic data collecting unit 21, the amount of communication data arriving at the input link IF11 during which time T _M for each time T _M of the monitor a predetermined cycle, that is, the traffic data, The communication data is collected separately for each communication path, and the collected traffic data is stored in the traffic data storage unit 31.

That accumulation, the traffic data is traffic data storage unit 31 collected as described above, each communication path, and, as the data for each time chopped in a monitor interval time T _M (e.g., in increments of 1 minute time) Is done. The traffic data accumulated at this time is the amount of data communicated per unit time, that is, a value converted into the traffic amount per unit time, and the unit is represented by, for example, bps (bit per second). It is.

(Step S02)
Next, the control unit 20 uses the cumulative distribution data calculation unit 22 to analyze every analysis unit time T _A (for example, 1 hour) that is a second predetermined time sufficiently longer than the predetermined time T _M. for traffic data accumulated in the traffic data storage unit 31 during the analysis unit time T _a, and calculates the cumulative distribution data, and accumulates the calculated cumulative distribution data on the cumulative distribution data storage unit 32.

FIG. 3 is a diagram showing an example of cumulative distribution data of traffic data according to the present embodiment. In FIG. 3, the horizontal axis of the graph represents traffic data, that is, the traffic amount x per unit time, and the vertical axis of the graph represents cumulative distribution data p with respect to the traffic amount x. Here, the cumulative distribution data p in traffic volume x, the collected during the analysis unit time T _A, among the accumulated traffic data, the value means the ratio of traffic data is less traffic x. That is, it can be said that the cumulative distribution data p is the cumulative probability density of the traffic amount x. In FIG. 3, the hatched circle indicates an example in which the cumulative distribution data p calculated in step S02 is plotted.

  As described above, in step S02, the control unit 20 creates table data in which the traffic amount x capable of representing the graph as shown in FIG. 3 and the cumulative distribution data p are associated with each other. Data is stored in the cumulative distribution data storage unit 32.

(Step S03)
Next, the control unit 20 analyzes the cumulative distribution data for each communication path accumulated in the cumulative distribution data accumulation unit 32 by the feature parameter extraction unit 23, and based on the analysis result, accumulates the communication path for each communication path. The feature parameters α and γ of the distribution data are calculated. Hereinafter, the process in step S03 will be described in detail.

In general, in a communication network such as the Internet, the graph of the cumulative distribution of such traffic data follows the Pareto cumulative distribution function when the traffic is congested, and the exponential cumulative distribution when the traffic is not congested. It is said to follow a function. Here, the Pareto cumulative distribution function F (x) and the exponential cumulative distribution function G (x) are represented by the following expressions (1) and (2), respectively.
F (x) = 1− (β / x) ^α (1)
G (x) = 1−exp (−x / γ) (2)

Therefore, the feature parameter extraction unit 23, the parameter α of the cumulative distribution function, beta, gamma, and the traffic data for each communication path that is accumulated in the cumulative distribution data storage unit 32 during the analyzing unit time T _A Fit to fit. For fitting the parameters α, β, and γ, a least square method that is often used in the past is used. That is, the parameters α, β, and γ of the cumulative distribution function are calculated by the least square method.

FIG. 4 is a diagram showing how the cumulative distribution of traffic data in FIG. 3 is fitted with the Pareto cumulative distribution function F (x). In FIG. 4, the hatched circles plot the cumulative distribution data calculated in step S02, and the solid line shows the curve represented by the Pareto cumulative distribution function F (x). The control unit 20 sets the parameters α and β of the Pareto cumulative distribution function F (x) to the square of the difference between the cumulative distribution data calculated in step S02 and the value of the Pareto cumulative distribution function F (x) by the least square method. Decide to minimize the sum. That is, the parameters α and β are determined so that the following expression (3) is minimized.
ΔF ² = Σ (p _i −F (x _i )) ² (3)

Similarly, the control unit 20 determines the parameter γ for the exponential cumulative distribution function G (x) so that the following expression (4) is minimized.
ΔG ² = Σ (p _i −G (x _i )) ² (4)

  Here, the parameters α and β of the Pareto cumulative distribution function F (x) are called a shape parameter and a scale parameter, respectively. Of these, the shape parameter α is strongly related to the burstiness of traffic data, and therefore, here, the shape parameter α is defined as a characteristic parameter characterizing the Pareto cumulative distribution function F (x). Similarly, a parameter γ of the exponential cumulative distribution function G (x) is defined as a characteristic parameter of the exponential cumulative distribution function G (x).

Thus, the control unit 20 calculates the characteristic parameter α of the Pareto cumulative distribution function F (x) and the characteristic parameter γ of the exponential cumulative distribution function G (x). Therefore, the control unit 20 determines whether the cumulative distribution data calculated in step S02 has a strong correlation with the Pareto cumulative distribution function F (x) or the exponential cumulative distribution function G (x). Then, the minimum value of ΔF ² calculated by Expression (3) and Expression (4) is compared with the minimum value of ΔG ² , and it is determined that the cumulative distribution function having the smaller value has a strong correlation.

(Step S04)
Next, the control unit 20 refers to the moving average parameter storage unit 33 by the moving average parameter acquisition unit 24 and acquires the moving average parameter k that matches the feature parameter α or γ. Hereinafter, regarding the process in step S04, the case where the characteristic parameter is the shape parameter α of the Pareto cumulative distribution function F (x) will be described in detail. The moving average parameter k corresponds to the smoothness parameter in the claims.

  The shape parameter α of the Pareto cumulative distribution function F (x) usually takes a value satisfying 1 <α <2. However, the more the burstiness and the unstable traffic, the smaller the shape parameter α and the burstiness. It is known that the smaller and more stable traffic, the larger the shape parameter α. That is, in the present embodiment, the shape parameter α is used as a parameter indicating the magnitude of the burst property.

On the other hand, although details will be described later, in the present embodiment, when the communication path control unit 26 controls the communication path, the traffic data collected by the traffic data collection unit 21 is not used as it is, but collected before that. A moving average value obtained by taking a moving average with the traffic data thus obtained is used. That is, the moving average value R _n of the traffic data for the traffic data X _n collected by the traffic data collection unit 21 at the n-th time is calculated by the following equation (5).
_{R n = k * X n +} (1-k) * R n-1 (5)
Here, k is a moving average parameter, and is a numerical value satisfying 0 <k ≦ 1. According to this equation (5), the moving average value R _n of the traffic data is such that the burst property is suppressed as k is smaller.

  Therefore, in this embodiment, an appropriate moving average parameter k is determined according to the degree of burstiness of the traffic data represented by the characteristic parameter α (that is, the shape parameter α) calculated in step S03. However, since there is no specifically known relational expression between the shape parameter α and the appropriate moving average parameter k, the experimentally obtained relation is used here. Note that the appropriate moving average parameter k means that appropriate smoothing is possible for traffic data with a large burstiness, and the time delay of control is minimized for traffic data with a low burstiness. That is, the moving average parameter k that can improve the response performance of the control.

  FIG. 5 is a diagram showing a profile of an appropriate moving average parameter k obtained experimentally with respect to the shape parameter α of the Pareto cumulative distribution function F (x). In the graph of FIG. 5, the horizontal axis represents the shape parameter α, and the vertical axis represents the moving average parameter k. The area of the shaded display portion sandwiched between the lower limit value and the upper limit value indicates that the moving average parameter k is applicable to the moving average calculation.

  The appropriate profile of the moving average parameter k shown in FIG. 5 is obtained by conducting an experiment in which traffic corresponding to various shape parameters α is passed through a communication network that has been experimentally configured in advance, and further moving for various moving average parameters k. By obtaining average data, the smoothness and response performance of the moving average data are evaluated, and an appropriate range of the moving average parameter k is determined.

  According to the profile of the moving average parameter k shown in FIG. 5, when the shape parameter α is small, the traffic data has a large burstiness. Therefore, the moving average parameter k has a smaller value as the moving average of the traffic data. It is an appropriate value for calculation. In this case, in the moving average calculation of the traffic data, the traffic data at the current traffic data acquisition time is evaluated smaller than the traffic data acquired before that, so even traffic data having a large burstiness is appropriate. To be smoothed.

  In addition, when the shape parameter α is large, the burstiness of the traffic data is small. Therefore, a larger value of the moving average parameter k is a more appropriate value for calculating the moving average of the traffic data. In this case, in the moving average calculation of the traffic data, the traffic data at the acquisition current time point is evaluated to be larger than the traffic data before that, so that the control response performance is improved.

  In FIG. 5, when the moving average parameter k exceeding the upper limit value is applied to the moving average, the traffic burstiness is not sufficiently smoothed, so that the communication path that responds to the traffic burstiness is controlled. there is a possibility. Further, when the moving average parameter k that is lower than the lower limit value is applied to the moving average, smoothing is excessively performed, and there is a possibility that control of a communication path that cannot respond to increase / decrease in traffic may be performed.

  The profile data of the moving average parameter k obtained by experiments as described above is stored in advance in the moving average parameter storage unit 33 of the control unit 20. Therefore, when the shape parameter α is calculated in step S03, the control unit 20 refers to the moving average parameter accumulating unit 33 based on the calculated shape parameter α, and the moving average parameter suitable for the shape parameter α. Get k.

  FIG. 6 is a diagram showing how the moving average parameter k matching the shape parameter α is acquired by referring to the moving average parameter accumulating unit based on the shape parameter α in the present embodiment. Since the appropriate moving average parameter k range is stored as the upper limit value and lower limit value data in the moving average parameter storage unit 33 in association with the value of the shape parameter α, the control unit 20 determines in step S03. The upper limit value and the lower limit value of the moving average parameter k are acquired according to the calculated shape parameter α. At this time, when there is no upper limit value and lower limit value of the shape parameter α and the moving average parameter k corresponding to the shape parameter α stored in the moving average parameter storage unit 33, for example, linear interpolation is performed using data before and after that. Ask by. In the example of FIG. 6, when the moving average parameter k for α = 1.37 is obtained, the upper limit value is 0.67 and the lower limit value is 0.3. Therefore, k satisfying 0.3 <k <0.67 is an appropriate moving average parameter.

  When the control unit 20 obtains the upper limit value and the lower limit value of the appropriate moving average parameter k as described above, it is as large as possible out of k included between the upper limit value and the lower limit value, usually the upper limit value. Is selected as a moving average parameter k that is actually applied to the moving average. Here, the reason why the moving average parameter k as large as possible is selected is that the larger the moving average parameter k is, the better the response performance of the control using it.

  In the above description, the profile information of the moving average parameter k for the shape parameter α of the Pareto cumulative distribution function F (x) is accumulated in the moving average parameter accumulating unit 33, so that an appropriate moving average parameter is obtained from the shape parameter α. The procedure for obtaining k has been described. This procedure can be similarly applied to the relationship between the characteristic parameter γ of the exponential cumulative distribution function G (x) and the moving average parameter k. That is, if the profile information of the moving average parameter k with respect to the characteristic parameter γ is experimentally acquired in advance and the information is accumulated in the moving average parameter accumulating unit 33, the control unit 20 calculates the characteristic parameter γ calculated in step S03. By referring to the moving average parameter accumulating unit 33 based on the above, an appropriate moving average parameter k suitable for the feature parameter γ can be acquired.

(Step S05)
Next, the control unit 20 uses the moving average calculation unit 25 to calculate the moving average value of the traffic data using the moving average parameter k calculated in step S04. The moving average value is calculated according to the equation (5) described above. Note that taking the moving average of the traffic data means smoothing the traffic data, and therefore the moving average calculating unit 25 corresponds to a low-pass filter in the claims.

(Step S06)
Next, the control unit 20 controls the communication path by the communication path control unit 26 according to the moving average value of the traffic data calculated in step S05. Hereinafter, control of the communication path will be described.

  The moving average value of traffic data calculated in the process up to step S05 is applicable to control of various forms of communication paths. Here, as one example, the moving average value of traffic data is applied to control of a communication path in a hierarchical communication network composed of an L1 (Layer 1) path and a packet layer path. In this case, it is assumed that the L1 path in this communication network is a wavelength path, and the packet layer path is an MPLS path.

  As an example of communication path control in the communication network as described above, there is control for newly establishing or deleting a communication path by increasing or decreasing the traffic volume. Normally, a plurality of MPLS paths are accommodated in one wavelength path, but the maximum amount of traffic that can be accommodated in the wavelength path is determined in advance as a physical limit amount.

  Therefore, the control unit 20 sets the upper limit threshold of the traffic amount that can be accommodated in the wavelength path, for example, 50% of the physical limit amount. The control unit 20 can acquire the traffic amount for each MPLS path as the moving average value by the processing up to step S05. Therefore, the control unit 20 calculates, for each wavelength path, the total traffic amount of the MPLS path accommodated in the wavelength path. When the total traffic volume increases beyond a preset threshold value, a part of the MPLS path is transferred to another wavelength path.

  Similarly, the control unit 20 sets a lower limit threshold of the traffic amount accommodated in the wavelength path, for example, 10% of the physical limit amount. If the total traffic amount of the MPLS paths accommodated in a certain wavelength path is below the lower threshold, it is useless to occupy one wavelength path with that much traffic amount. 20 transfers the MPLS path accommodated in the wavelength path to another wavelength path. Then, the empty wavelength path is deleted.

  By controlling the communication path as described above, it is possible to improve the communication efficiency of the entire communication network. In the present embodiment, since the communication path is controlled based on the moving average value of the traffic amount considering the degree of burstiness, the traffic data with burstiness can be appropriately smoothed. As a result, control response performance can be improved.

  A part of the embodiment described above can be variously modified.

(Modification 1 of embodiment)
First, in the embodiment described above, the moving average process, which is one of the low-pass filters, is used to smooth the traffic data. However, instead of the moving average process, other digital signal processing (analog signal processing is also used). May be used. In this case, the moving average parameter k is associated with the reciprocal of the time constant of the low-pass filter. Therefore, in this case, profile information of the low-pass filter time constant for the characteristic parameter α or γ of the Pareto cumulative distribution function F (x) or the exponential cumulative distribution function G (x) is prepared in advance, and the traffic data Traffic data is smoothed by a low-pass filter having a time constant determined according to the magnitude of burstiness.

(Modification 2 of embodiment)
In the embodiment described above, the cumulative distribution data is calculated for each analysis unit time T _A (for example, 1 hour) sufficiently longer than the time T _M (for example, 1 minute) of the monitoring period of the traffic data (step S02) and subsequent processes are executed, but the calculation of the cumulative distribution data may be executed at every monitoring period time T _M or every time shorter than the analysis unit time T _A. Good. In this case, since the feature parameter α or γ Pareto cumulative distribution function F (x) or exponential cumulative distribution function G (x) is calculated in a shorter time than the analysis unit time T _A, the traffic data burst Changes can be reflected in communication path control more quickly.

(Modification 3 of embodiment)
In the embodiment described above, the control of the communication path is performed every time the moving average value of traffic data exceeds or falls below the upper limit threshold or the lower limit threshold once. The communication path may be controlled at this time. By doing so, the response performance of the control is reduced, but stable communication path control can be performed even when the degree of traffic burst itself fluctuates in a short cycle. In particular, in the embodiment of the modified example 2, the moving average value of traffic data is calculated in a short cycle, so that the effect of performing stable communication path control is considered to be significant.

  At this time, the communication network administrator determines how many times the moving average value of traffic data (one time or a plurality of designated times) exceeds or falls below the upper or lower threshold. Or it is good also as what an operator can set suitably.

It is the figure which showed the example of the structure of the node apparatus which concerns on embodiment of this invention. It is the figure which showed the procedure until it arrives at the embodiment of this invention until traffic data is collected and a communication path is controlled based on the traffic data. It is the figure which showed the example of the cumulative distribution data of the traffic data which concern on embodiment of this invention. FIG. 4 is a diagram showing a state of fitting the cumulative distribution of traffic data in FIG. 3 with a Pareto cumulative distribution function F (x). It is the figure which showed the profile of the suitable moving average parameter k acquired experimentally with respect to the shape parameter (alpha) of the Pareto cumulative distribution function F (x). In the embodiment of the present invention, referring to the moving average parameter accumulating unit based on the shape parameter α, it is a diagram showing how to obtain a moving average parameter k that matches the shape parameter α.

Explanation of symbols

1 Node device 5 Communication line 10 Link connection 11 Input link IF
12 Link switch 13 Output link IF
20 control unit 21 traffic data collecting unit (traffic data collecting means)
22 Cumulative distribution data calculation unit (feature parameter extraction means)
23 feature parameter extraction unit (feature parameter extraction means)
24 Moving average parameter acquisition unit (smoothness parameter acquisition means)
25 Moving average calculator (traffic data smoothing means)
26 Communication path control unit (communication path control means)
31 Traffic data storage unit (Traffic data storage means)
32 Cumulative distribution data storage unit 33 Moving average parameter storage unit (smoothness parameter storage means)
34 Communication path information storage unit

Claims (8)

A node device connected to a communication line constituting a communication network and controlling a communication path of communication data communicated via the communication line,
For each communication path set in the communication data to be communicated, traffic data collection means for collecting traffic data of the communication data at predetermined time intervals;
Traffic data storage means for storing the collected traffic data;
Feature parameter extraction means for obtaining statistical data of traffic data for the traffic data stored in the traffic data storage means for a time sufficiently longer than the predetermined time, and extracting characteristic parameters representing the characteristics of the statistical data When,
Smoothness parameter accumulating means for accumulating a smoothness parameter value of a low-pass filter that appropriately smoothes the burstiness of the traffic data in correspondence with the value of the feature parameter;
Smoothness parameter acquisition means for referring to the smoothness parameter storage means based on the extracted feature parameter value and acquiring a smoothness parameter value of a low-pass filter suitable for the smoothing;
Configuring a low-pass filter having the value of the obtained smoothness parameter, smoothing the collected traffic data by the low-pass filter, and calculating traffic data smoothing traffic data;
A node device comprising: communication path control means for controlling a communication path in the communication network based on the calculated smoothed traffic data. The node apparatus according to claim 1, wherein the low-pass filter is configured by moving average means for moving average of the collected traffic data. The feature parameter extraction means includes
The node device according to claim 1, wherein the feature parameter is extracted by fitting statistical data of the traffic data to a Pareto cumulative distribution function. The feature parameter extraction means includes
The node device according to claim 1, wherein the feature parameter is extracted by fitting statistical data of the traffic data to an exponential cumulative distribution function. The feature parameter extraction means includes
A first feature parameter is extracted as a feature parameter of the statistical data by fitting the statistical data of the traffic data to the Pareto cumulative distribution function, and indicates the strength of the correlation between the statistical data and the Pareto cumulative distribution function Calculating first correlation data;
A second feature parameter is extracted as a feature parameter of the statistical data by fitting the statistical data of the traffic data to an exponential cumulative distribution function, and indicates the strength of correlation between the statistical data and the exponential cumulative distribution function Calculating second correlation data;
Comparing the first correlation data and the second correlation data;
When the first correlation data indicates that the correlation is stronger than the second correlation data, the first feature parameter is selected as a feature parameter of the statistical data;
The second feature parameter is selected as the feature parameter of the statistical data when the second correlation data indicates that the correlation is stronger than that of the first correlation data. Item 3. The node device according to Item 2. The feature parameter extraction means includes
The node device according to claim 3, wherein the fitting is performed by a least square method. The communication path control means includes
When the smoothed traffic data calculated by the traffic data smoothing means exceeds a predetermined upper limit threshold value once or a predetermined plural times continuously, or once or exceeds a predetermined lower limit threshold value The node device according to any one of claims 1 to 6, wherein a predetermined communication path is controlled when continuously falling. A communication network system comprising the node device according to any one of claims 1 to 7.

Images