SYSTEM AND METHOD OF FORECASTING USAGE OF NETWORK LINKS

• US 20090245107A1
• Filed: 03/25/2008
• Published: 10/01/2009
• Est. Priority Date: 03/25/2008
• Status: Active Grant

• Current Assignee: Atlassian US, Inc. (Atlassian Corporation)
• Original Assignee: Verizon Data Services Incorporated (Verizon Communications Inc.), Verizon Services Corporation (Verizon Communications Inc.)
• Inventors: Krishnaswamy, Murali, Bhatta, Sambasiva R., Ritzer, Edward W.
• Granted Patent: US 7,808,903 B2
• Time in Patent Office: Days
• US Class Current: 370/232
• CPC Class Codes: H04L 41/142 using statistical or mathematical methods
• CPC Class Codes: H04L 41/147 for predicting network behaviour

Abstract

An approach is provided for forecasting usage of network links. The method includes determining traffic data values for traffic over a network link that is used to provide one or more services to one or more customers, where the traffic data values are taken at intervals for a predetermined range of time. A historical data set of peak traffic data values is compiled from the traffic data values. A mathematical confidence factor is applied to the peak traffic data values to achieve modified data values, and a time-based weighting factor is applied to the modified data values to achieve further modified data values. And, a forecast trend curve of traffic over the network is calculated using the further modified data values.

Claims

1. A method comprising:

determining traffic data values for traffic over a network link that is used to provide one or more services to one or more customers, wherein the traffic data values are taken at intervals for a predetermined range of time;

compiling a historical data set of peak traffic data values from the traffic data values;

applying a mathematical confidence factor to the peak traffic data values in the historical data set to achieve modified data values;

applying a time-based weighting factor to the modified data values to achieve further modified data values; and

outputting a forecast trend curve of traffic over the network calculated using the further modified data values.

2. The method according to claim 1, further comprising modifying a capacity of the network link based on the forecast trend curve.

3. The method according to claim 1, further comprising determining an arithmetic mean of the peak traffic data values in the historical data set,

wherein the mathematical confidence factor assigns a weight value to an individual peak traffic data value based upon a proximity in value of the individual peak traffic data value to the arithmetic mean.

4. The method according to claim 3, wherein the weight value assigned to the individual peak traffic data value is determined by:

determining a standard deviation of the peak traffic data values in the historical data set;

subtracting the arithmetic mean from the individual peak traffic data value to find a difference value; and

determining the weight value from a predetermined list of weight values that each correspond to a value of the difference value.

5. The method according to claim 3, wherein the weight value assigned to the individual peak traffic data value is determined by:

determining a standard deviation of the peak traffic data values in the historical data set;

subtracting the arithmetic mean from the individual peak traffic data value to find a difference value; and

determining the weight value from a predetermined list of weight values that each correspond to a value of the difference value as compared to the standard deviation.

6. The method according to claim 1, wherein the time-based weighting factor assigns a higher weight value to modified data values that correspond to peak traffic data values that are more recent than other peak traffic values in the historical data set, and a lower weight value to modified data values that correspond to peak traffic data values that are less recent than other peak traffic values in the historical data set.

7. The method according to claim 1, wherein the time-based weighting factor assigns a weight value to an individual peak traffic data value based upon a proximity in time of the individual peak traffic data value to a first-in-time peak traffic data value in the historical data set.

8. The method according to claim 1, further comprising:

determining updated traffic data values for traffic over the network;

recompiling the historical data set of updated peak traffic data values from the updated traffic data values;

applying the mathematical confidence factor to each of the updated peak traffic data values in the historical data set to achieve modified updated data values;

applying the time-based weighting factor to the modified updated data values to achieve further modified updated data values; and

outputting the forecast trend curve of traffic over the network recalculated using the further modified updated data values.

9. The method according to claim 1, further comprising:

adjusting the time-based weighting factor based on a comparison between the forecast trend curve and actual peak traffic data values for traffic over the network link during a period of time corresponding to the forecast trend curve;

applying the adjusted time-based weighting factor to the modified data values to achieve further modified adjusted data values; and

outputting the forecast trend curve of traffic over the network recalculated using the further modified adjusted data values.

10. The method according to claim 1, further comprising:

determining updated peak traffic data values for traffic over the network;

adjusting the time-based weighting factor based on a comparison between the forecast trend curve and the updated peak traffic data values;

recompiling the historical data set of peak traffic data values from the updated peak traffic data values;

applying the mathematical confidence factor to each of the updated peak traffic data values in the historical data set to achieve modified updated data values;

applying the adjusted time-based weighting factor to the modified updated data values to achieve further modified updated data values; and

outputting the forecast trend curve of traffic over the network recalculated using the further modified updated data values.

11. The method according to claim 1, wherein the traffic data values for traffic over the network link are segmented into separate historical data sets for different types of services and/or different customers, and wherein forecast trend curves are separately calculated for the separate historical data sets.

12. The method according to claim 1, further comprising applying a correlation coefficient to the forecast trend curve, wherein the correlation coefficient is representative of a number of customers allocated to the network link, and wherein the correlation coefficient is removed from the forecast trend curve when the number of customers allocated to the network link reaches a predetermined maximum number.

13. A system comprising:

a monitoring device configured to audit a network link to determining traffic data values for traffic over the network link that is used to provide one or more services to one or more customers, wherein the monitoring device is configured to audit the traffic data values at intervals for a predetermined range of time; and

an analysis module configured to compile a historical data set of peak traffic data values from the traffic data values, apply a mathematical confidence factor to the peak traffic data values in the historical data set to achieve modified data values, apply a time-based weighting factor to the modified data values to achieve further modified data values, and output a forecast trend curve of traffic over the network calculated using the further modified data values.

14. The system according to claim 13, wherein the analysis module is further configured to determine an arithmetic mean of the peak traffic data values in the historical data set, and wherein the mathematical confidence factor applied by the analysis module assigns a weight value to an individual peak traffic data value based upon a proximity in value of the individual peak traffic data value to the arithmetic mean.

15. The system according to claim 14, wherein the weight value assigned by the analysis module to the individual peak traffic data value is determined by:

determining a standard deviation of the peak traffic data values in the historical data set;

subtracting the arithmetic mean from the individual peak traffic data value to find a difference value; and

determining the weight value from a predetermined list of weight values that each correspond to a value of the difference value.

16. The system according to claim 14, wherein the weight value assigned by the analysis module to the individual peak traffic data value is determined by:

determining a standard deviation of the peak traffic data values in the historical data set;

subtracting the arithmetic mean from the individual peak traffic data value to find a difference value; and

determining the weight value from a predetermined list of weight values that each correspond to a value of the difference value as compared to the standard deviation.

17. The system according to claim 13, wherein the analysis module assigns a time-based weighting factor having a higher weight value to modified data values that correspond to peak traffic data values that are more recent than other peak traffic values in the historical data set, and a time-based weighting factor having a lower weight value to modified data values that correspond to peak traffic data values that are less recent than other peak traffic values in the historical data set.

18. The system according to claim 13, wherein the analysis module assigns a time-based weighting factor to an individual peak traffic data value, wherein the time-based weighting factor has a weight value based upon a proximity in time of the individual peak traffic data value to a first-in-time peak traffic data value in the historical data set.

19. The system according to claim 13, wherein:

the monitoring device is further configured to determine updated traffic data values for traffic over the network; and

the analysis module is further configured to recompile the historical data set of updated peak traffic data values from the updated traffic data values, apply the mathematical confidence factor to each of the updated peak traffic data values in the historical data set to achieve modified updated data values, apply the time-based weighting factor to the modified updated data values to achieve further modified updated data values, and output the forecast trend curve of traffic over the network recalculated using the further modified updated data values.

20. The system according to claim 13, wherein the analysis module is configured to apply an adjusted time-based weighting factor to the modified data values to achieve further modified adjusted data values, and output the forecast trend curve of traffic over the network recalculated using the further modified adjusted data values, wherein the adjusted time-based weighting factor is based on a comparison between the forecast trend curve and actual peak traffic data values for traffic over the network link during a period of time corresponding to the forecast trend curve.

21. The system according to claim 13, wherein:

the monitoring device is further configured to determine updated peak traffic data values for traffic over the network; and

the analysis module is further configured to recompile the historical data set of peak traffic data values from the updated peak traffic data values, apply the mathematical confidence factor to each of the updated peak traffic data values in the historical data set to achieve modified updated data values, apply an adjusted time-based weighting factor to the modified updated data values to achieve further modified updated data values, and output the forecast trend curve of traffic over the network recalculated using the further modified updated data values, wherein the adjusted time-based weighting factor is based on a comparison between the forecast trend curve and the updated peak traffic data values.

22. The system according to claim 13, wherein the analysis module is further configured to apply a correlation coefficient to the forecast trend curve, wherein the correlation coefficient is representative of a number of customers allocated to the network link, and wherein the analysis module is configured to remove correlation coefficient from the forecast trend curve when the number of customers allocated to the network link reaches a predetermined maximum number.