MXPA99011161A - Processes and systems for dynamically measuring switch traffic - Google Patents

Abstract

A process and system is disclosed for determining automatically and dynamically the correct time segment during which a switch or components of a switch receive peak traffic from a communications network. The process collects traffic data regarding the usage of a selected switch or selected switch components. The traffic data is collected over a selected journalling period (e.g., 30 days), which is periodically updated with new traffic data to maintain a journal of the most recent 30 days of collected traffic data. Collected traffic data is filtered to remove aberrant or corrupting data. An average peak usage segment for all of the segments of the journalling period is selected from the remaining filtered traffic data. By regularly adding new data to and removing old data from the journal, the resulting moving window will reflect recent changes in peak usage. Regularly (e.g., daily) selecting the average peak usage segment detects such changes quickly, with the result that the process of the present invention dynamically reacts to and determines changes in the peak usage segment for particular switch components. Peak usage segments and associated traffic data determined and collected by the present invention result in improved network traffic engineering. A system is disclosed for collecting, processing and evaluating traffic data in order to engineer network elements for optimal traffic flow. Additional methods are disclosed for use by the system in dynamically filtering aberrant data with improved accuracy and in determining average traffic usage limits for particular switch components. The methods of the present invention allow more effective use of expensive network switching resources by providing more accurate determination and more dynamic tracking of switch usage at the component level.

Description

PROCESSES AND SYSTEMS TO MEASURE DYNAMICALLY THE SWITCHING TRAFFIC This invention relates to processes and systems for measuring, dynamically and automatically, traffic through a communications network, in order to ensure the efficient allocation of the resources of this network.

BACKGROUND OF THE INVENTION The control of network traffic is aimed at effectively minimizing the cost of the number of unsuccessful communications attempts, caused by congestion or network failure, while also ensuring that the most expensive network equipment does not it is used neither excessively nor minimally. The ultimate goal is to provide a given grade of service with the least amount of equipment. To do this one must determine the amount of traffic handled by the network, particularly by the network switches. Traffic data describes the amount and characteristics of communications traffic (voice, video or data) through the network. They are collected to help operators of communications networks determine how efficiently those networks are operating and, if necessary, plan reductions, repairs or improvements to the network. Data traffic is also useful to large clients who rent or rent network establishments. The owner of this invention also owns a patent that describes a "Telephone System Adapted to Inform Customers of Traffic Data of the Telephone Establishment." U.S. Patent No. 5, 410, 589, which describes the advantages of, and procedures for, collecting and reporting traffic data, is hereby incorporated by reference in its entirety. Two typical ways to design switching capacity include the design of the extreme value, in which engineers work to accommodate maximum traffic, or the design of the hour in time consistently occupied, which engineers change to accommodate peak traffic. during a period that, on average, is mostly in a busy state. However, there is a tension between the provision of quality versus the cost efficient service. The design of the extreme value provides maximum quality at high cost, the design of busy time, consistent over time, provides a chosen level of quality at a lower cost. - - For example, a particular switch component can service 256 lines, but will be provided with only 64 slots or time connections to which those lines serve. (Line units allow analog subscriber lines to communicate with the digital network). The time slots are smaller than the number of lines, because most of the 64 lines are rarely used (if used) at the same time. When, however, 65 lines seek access to communications services from the switch at the same time, the number 65 caller is blocked by the switch. Typically, networks are designed to maintain blocking below a certain absolute percentage, such as% of attempted calls, and within a certain average percentage, such as 1.5% of attempted calls. As an example, the determination of "fast busy" signals that generate congestion for more than 2% of communications attempted through a particular switch at selected time periods predicts network designers in (a) expect what from the customers about the bad service and (b) that the network may need a repair or an improvement, if the problem persists. Typically, the huge volume and types of traffic through network switches allows the collection and analysis of only statistically significant samples of traffic data, rather than the collection and analysis of traffic in real time. Thus, the above methods of analyzing switching traffic use the "average peak usage time" for the entire switch. The use of the peak or busy time refers to the time of day in which the traffic through the switch reaches a peak point. (Although the phrase uses the term "hour", the crest can be in any period of time). The "average" refers to the fact that the average peak usage time is selected by averaging the traffic over many days and then determining the time of day during which the use of the network switching peak occurs. Generally, the average peak usage time has been determined once a year, manually analyzing the use of switching traffic. The "mean peak usage time" is then used to design network traffic for the rest of the year, with switching designed to handle the volume and type of traffic that occurs during its average peak usage time, for example, systems commercial as the system COER, supplied by Lucent Technologies, Inc. (formerly part of AT & T), takes traffic data from a particular switch and then organizes and reports that data to allow manual design to determine if the limits of switching have been reached. This requires that the traffic analysis process involves only a small subset of available data. Also, the COER system determines the busy time once a year and only for the total switch. Other efforts have been made to monitor traffic. For example, U.S. Patent No. 4,456,788 to Kline, et al., Describes a "Telecommunication Trunk Circuit Informer and Consultant", a system and method that analyzes trunk circuit data. The Kline et al. Patent mentions determining the busy state hours, but does not describe doing so in a switch component or a continuous base. U.S. Patent No. 5,359,649 describes systems for "Tuning Congestion of Telecommunication Networks", which monitors the elements and_ routes of the network to identify congested routes and repair them or redirect traffic. These previous processes and systems, however, do not solve several problems. First, a certain number of switching components are designed beyond (or below) their capacity. This is because the traffic data has been collected only for the average peak usage time of the entire switch, which means that the traffic data for this time is the only data analyzed in the network design. But many switching components will have different hours of average peak usage; The same components may also have different traffic levels during their time of actual crest usage. This results in an overload or minor use of the switching components, which may fail or be more expensive to operate. ~ Similarly, the average peak usage time is usually determined once a year. That was fine in the past, when the relative stability of the use of traffic through switches required the determination of the time of average peak usage only annually, in order to aid the design of the traffic. The network equipment and client allocation procedures used when the processes of traffic use and previous systems were developed result in a relatively homogeneous traffic usage through the switch components, which also allow the infrequent selection of one hour of. use average peak. ~ The statistical analysis of the traffic data predicts the amount of blocking expected for a given level of use of the switch. Two key measurements have an impact on the amount of blockage; the volume of traffic handled by the switch and the volatility of the traffic. The volume is typically measured in hundreds of seconds occupied by the call or "CCS" handled by a switch during an average busy hour, determined according to the methods described above. These methods effectively decrease the average usage capacity of all switching components to match the worst performance component of the switch. Volatility refers to the degree of variance of traffic from "a calculated average." Volatility is typically measured by simply discarding data collected for days that are thought to be unrepresentative, for example, in many systems, traffic data for holidays. Saturdays, Sundays, and even Fridays are ignored or ignored when switching charges are channeled As volatility and capacity increase, so does the blockade. ~ 1 Recently, however, - numerous changes in technology and industry have occurred. These changes have had an impact, drastically and negatively, on the effectiveness of current processes in analyzing switching capabilities, for example, two separate causes have resulted in non-homogeneous allocation in the switching components., the subscriber carrier delivery systems have reduced the random extension of the client allocation process. Most subscriber carrier systems typically handle around 96 lines each and they serve a very small geographical area. Because of this, systems often primarily serve only residential customers or only business customers. For example, new subdivisions can have all residences assigned to a new switching component. Thus, customers that handle a switching component, located in a residential area with many second lines and computer modems, can handle significantly more and different traffic than components of the rest of the switching. Clients with different usage characteristics focus on different switching components, which lead to widely different business hours between the components. A second reason for the inhomogeneous allocation is increasingly the practice of reusing the previously assigned facilities, when the service to a new client in an old location is introduced (for example, when an old client moves, the connection to the residence remains and the service is restored simply when someone else reoccupies the residence). That significantly reduces labor ~ associated with the establishment of the new service in * previously occupied locations, but results in less homogeneous traffic, because new customers end up with a specific service of a new switching component. Finally, there has been a constant and growing proliferation of high usage lines for Internet service providers, telecommunication host computer connections and the like. These new and numerous high-use lines also have increases in the differences in traffic load between various switching components. In general, these changes have reduced the homogeneous nature of the traffic load through the switching components. Increasing numbers of switching components, therefore, have average peak usage hours that are different from the average peak usage time for the entire switch and other components. Also, not only are the hours of average peak usage between the components different, but the "peak" traffic handled by a particular component may be significantly different from the "peak" traffic handled by the other components or all of the switching. This is especially aggravated by groups of new subscribers, who can use recent technologies in a simple switching component, as described above. Also, many of the weekend and holiday traffic data that has been considered as representative river are actually or become more representative of general switching traffic. Many components are not, therefore, designed for the appropriate traffic load. In addition, ~ rapid growth in telecommunications has resulted in average peak usage hours that change dynamically. For example, as businesses increasingly provide Internet access for their employees, the commute traffic during the lunch hour, previously little, has greatly increased as employees make use of a "free" Internet connection during lunch breaks. . Better methods of filtering or eliminating non-representative traffic measurements are also necessary. _ In short, there is a need for traffic analysis systems that take into account the actual loads through the switching components when analyzing traffic usage. Finally, the improved determination of business hours and filtering results in a more accurate determination of the use of real traffic on a basis of the individual switching component, which current processes and systems do not allow. That information is then used to (a) adjust the load on a particular switch or its components or (b) redesign the network to the optimal configuration for its actual load.

SUMMARY OF THE INVENTION The present invention aims to determine, accurately and dynamically, the traffic capacity of the individual switching components, in order to provide an improved service and to use more effectively the switching resources of the network. The invention includes an automatic process and system that determines the correct peak time and average usage at that time (or other period of time) for selected components of network elements, such as a switch. The method involves the steps of periodically collecting and storing traffic data segments in each switching component over a selected period of time, averaging the traffic data of each switching component for each segment over a selected period of time; and select the peak usage segment. It also reveals methods for filtering aberrant or statistically corrupt traffic data, from the gathered traffic data. Such methods dynamically determine if the particular traffic data is not statistically adequate. The methods for determining the average correct usage and the peak time can be repeated continuously in order to take into account the changes in the network load and consider those changes in the selection of the peak segment for a particular component. This, in turn, allows the method of the present invention to determine, accurately and dynamically, the capacity limit for each switching component. The traffic data, which describes the traffic through the switching component in the selected peak usage segment, is collected and analyzed. Depending on the results, the load on the selected component can be adjusted or reconfigured. the network in another way. A system that performs these processes automatically is revealed. The dynamic and automatic process and system of the present invention readily recognize the volatile traffic usage changes that take place in the communications network in order to increase the accuracy of the traffic analysis. The use of this process when service increases results in increasing money savings and lower labor requirements. For example, the more accurate use of data informs traffic designers to take more active measures to prevent new conditions from impacting the service. The "equipment can be deployed in more correct amounts, in order to save capitalization costs." The replacement of a manual process with an automatic one will reduce labor costs, and associated operating costs, required to perform the analysis. In the embodiment of the invention, the selected switching components generate and store the traffic data A collector periodically retrieves that traffic data, which describes the use of the component in a selected segment or time period, such as 30 minutes. collector supplies traffic data to a database and processor.Since the "collector and processor can monitor numerous (eg hundreds of) switches with multiple components, the database is useful in maintaining large volume of data. traffic data necessary to select each average peak segment of the component. The database contains at least 'enough traffic data to create a daily relationship that describes a representative period of use of the component. Generally, the relationship should be large enough to minimize the impact of a grouping of daily average peak-use peak segments, while also short enough to respond to a fundamental change in the use of a particular switch or component. The relationship acts as a mobile window that analyzes the adjustable number, most recent, of average daily peak segments, to select the average peak usage segment for the period of the relationship. If a new datum of the day is added to the relationship, any day prior to the previous 30 days (or any other time frame that includes the relationship period) can be ignored. Selecting an average peak usage segment of a given operating ratio period prevents an acute business hour impulse of the particular component from affecting the selection of the average peak usage time. In any case, the meeting of a full day of traffic data ensures sufficient data to allow the determination of an average peak usage segment that appropriately reflects the average peak traffic through several "commutations and switching components. filtered traffic data collected in each switching component for each segment through a given complete period of the relationship.This process is repeated until successive segments in a single day have been averaged with the other 29 (for a ratio of 30 days) corresponding segments.

This determines the average use of the component during each time segment. The result is a full day comprising multiple segments with each showing average usage. The segment in each peak traffic occurred in the component, can then be selected. A selection method involves choosing from the traffic data the average peak usage of two consecutive segments; this gives the busy time of the switching component (or other period of time) if the segments are time periods of 30 minutes. The traffic data measurements for the selected peak time are stored for each day in the database. Traffic data can be automatically filtered to eliminate aberrant data, that is, data that incorporate errors or that should be excluded from the analysis of the design of traffic usage because they are outside the statistically acceptable limits. Then, the processor averages the traffic data without flag, in order to determine the average use of a particular component in its busy time. _ This selection of a peak segment and the determination of the average use during the peak segment can be done daily in the segments recorded in the daily relationship, which is updated routinely to maintain only the segments that describe the traffic over the selected number of days more recent. Because the process selects the average peak usage segment over an updated operating relationship, changes in traffic patterns for particular switching components or commutations during that "period of time will be detected, dynamically and automatically. For the time that the shift in peak usage has repeated for half the period of the daily ratio, the data indicating a new average peak usage segment exceed the data that supports the old average peak usage segment. If particular switching components have handled traffic from a prison, which allows only calls during the period from 5:00 p.m. to 6:00 p.m., that period will probably be the average peak segment. Prison permanently change the call time at 7:00 p.m. until 8:00 p.m., the average crest segment will be displaced to this last Thymus period Assuming a 30-day daily relationship period, the present process will detect displacement within about 15 days. Therefore, a temporary change of less than 15 days will not change the average peak segment. Using the data collected as described above, the methods of the present invention can be used to more accurately determine the capacity of the individual switching components. First, non-representative traffic measurements for a component's average peak use segment are identified and removed from the data from which the capacity is determined. Next, the method uses the remaining traffic data to determine an average traffic usage limit for the switching component. In order to determine exactly which measurements are truly unrepresentative, the method of the invention selects a first average value for all measurements of traffic data in a selected average peak usage segment for a particular daily relationship period. Thus, when a component has an average peak usage segment of 2:30 to 3:30 p.m., the average traffic in this time is determined for the daily relationship period of the previous 30 days. Then a lower limit is established. The present invention selects the lower limit as a percentage of the first average value. Preferably, 95% of the first average value is used as the lower limit. This value has been found to exclude traffic measurements from days of low usage not representative of the normal predominant traffic patterns. Relevant measurements remain to allow proper design of the component., an upper limit is established. This is done by determining a second average value, based on traffic data for the daily ratio (for example before 30 days) that remains after traffic below the lower limit (for example, 95% of the first average value), is excluded . The method then determines the standard deviation between the second average value and the traffic data remaining above the lower limit. -An upper limit can then be established based on the second average value plus 2 standard deviations. All measurements of the daily peak usage segment, which exceed the upper limit, are excluded.

By making non-representative measurements, according to this method, the present invention more accurately eliminates the actual unrepresentative traffic data, in contrast to the previous methods, which simply make approximate assumptions about which data will probably not be representative. Also, the method of the present invention detects changes in traffic patterns that create non-representative data. For example, while the above methods will not detect a snow day or the like, during which traffic is particularly high, the present method detects and excludes such unrepresentative data. With the data limited by the lower and upper limits, a third average value representing the average monthly traffic usage for the computing component can be determined. This third average value is used as the average measurement of actual use for design purposes. This value is compared against the capacity of use calculated to "" determine the proximity to the limits of the service. It is also used in calculations of average use per line assigned in the component, which can be used to estimate the number of individual subscribers that can be supported by the component. Usage measurements that remain after the above exclusion process can be used to calculate a measurement of volatility, displayed by the data. This measured volatility is calculated by the standard deviation of the usage measure divided by the average. The volatility measurement can be used to then calculate the average usage capacity of the component. A database stores the average usage for a busy hour as well as the traffic usage limit calculated for the components of the switch. Such traffic data may include the use of the busy time, call attempts or call blocks. Average-use traffic data can then be computed to the threshold capabilities of the component, in order to handle the load on the component. Components that are near or above capacity can have their loads properly adjusted. A management network can be provided and can access the stored traffic data and generate reports in order to determine if the target service levels are met by the communications network.

For example, several network administrators can connect to the network information store that holds the collected traffic data for the selected average peak segments of various computing components. Network administrators can then operate reports to obtain measurements of the level of services, such as determinations of a particular component (or switch): percent of the dial tone delay; percent of capacity; occupation percent; Percent blocking or any other measurements desired by the network design. The process of the present invention can be performed on a computer platform in communication with a particular switch or numerous switches. For example, in one embodiment, a personal computer can be coupled to selected switch components. In another embodiment, a hold computer platform can be coupled to a manifold that examines multiple components of multiple switches in order to gather traffic data describing the traffic handled by those switch components. For example, the hold platform can be coupled to the collector that monitors traffic through switches located within a regional communications network, such as a Local Access and Transport Area ("LATA"), or multiple switches in regional networks . The warehouse platform stores the collected traffic data and processes it, as described above, in order to select the average peak usage segment for a particular switch component. The present invention, therefore, is aimed at achieving at least one or more or combinations of the following objectives: Accurately calculate the operating capacity of individual commutator components, in order to assign the correct number of lines to particular components. - Automatically select an average peak usage segment, useful for allocating resources within a communications network and thus make more efficient use of network resources and decrease network operation costs; Collect and analyze the traffic data from the "switch components, in order to select the average peak usage segment of at least selected" switch components, - dynamically determine the offsets in an average peak usage segment of a particular component; Monitor traffic through selected switch components, during the selected average peak usage segment of that component, and compare traffic at that time to the capacity of the component, in order to adjust the loads through the switch components selected; Support the traffic design, analyzing the traffic data collected during the average peak usage segments of the switching components, in order to generate several reports on the service levels of the switching components; Provide methods to eliminate non-representative traffic measurements for the design of the switching component; - more accurately determine the average limit traffic for each switch component; and Providing a system for carrying out the methods of the present invention, which accomplish the above objects. Other objects, features and advantages of this invention will become apparent from the remainder of this document, which includes the Figures.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1A shows a system, according to the present invention, for the processing of traffic data collected from a representative communications network; Figure IB shows the components of a switch from which the traffic data is collected and processed by means of the system shown in Figure 1A; Figure 2A shows the traffic data collected for several segments over several days and averaged, - Figure 2B shows a collection of the averaged segments of Figure 2A, each reflecting the average peak traffic for each sement of a full day; Figure 2C shows the process of updating a daily relationship, to create a mobile window whose contents reflect the value of the most recent 30 days of averaged traffic data; Figure 3 is a flow diagram illustrating the steps for collecting and processing traffic data according to the method of the present invention; Figure 4 shows a traffic design system for monitoring the load on the switching components during their selected average peak segments, which compares the actual load to the thresholds '' reconfiguring the network to optimize the use of the switch component; Figure 5 shows traffic data for a particular component and a method for signaling "aberrant" data; - Figure 6 shows a load service curve, illustrating the difference in operating the point capacity of a particular switch component for a particular constant; Figure 7 is a bar graph, comparing "(a) the operation of the point capacity, determined according to the above processes and only based on the general switching capacity and (b) the operation of the point capacity determined in accordance with the processes of the present invention and based on the component per commutator DETAILED DESCRIPTION OF THE DRAWINGS Review of a Communications Network Figure 1A shows a communication network 10, which, in the mode shown, is a telecommunications network provided with network capabilities Intelligent Advanced ("AIN") and a Signaling System 7 (SS7). The network 10 illustrates a type of network with which the present invention operates. This network 10 has service switching points ("SSP") 20, 22 and a signal transfer point ("STP") 26, which also is coupled to a mobile switching center ("MSC") 24. The SSP 20, 22"can be any telecommunication switch, such as 1AESS, 5ESS, DMS-100/200, E S2, DC0, DMS10 or comparable switches (which include the packet switches) .The service control point (" SCP ") 28 is coupled to STP 26 (as are other STPs, not shown in Figure 1A.) In" example form, any of the telephones 30, 32 can originate a call through SSP 20 by dialing a number termination to, for example, the cordless telephone 34. The originating SSP issues a question containing the dialed destination digits; the question is_ switched through the SS7 network 10 via STP 26 ~ to the SCP 28, which comprises the guide and the characteristic databases of the client. The SCP 28 provides a route by which the call from the telephone 30, 32"can travel to the wireless telephone 34. A route can be from the originating SSP 20 through the tandem SSP 22, which, in turn, the communication to the MSC 24 ends. This MSC 24 communicates with the wireless telephone 34 through the well-known base stations, located in various cell sites.

Collection of Traffic Data from the Switching Components _ -_ _ The network 10 also includes a traffic data collector 40, a traffic data processor 42, and a network information store ("N") 44, which has a database 41. The collector 40 can be a TDMS ("Traffic Data Surveillance System"), sold by Lucen Technologies (formerly AT &; T) or a DCOS 200, sold by Bell Communications Research (also named "BellCore"), deployed, for example, on a UNXX 3B2 / 600 platform. The collector 40 is coupled to the SSPs 20, 22, which periodically examine the stored traffic data. Generally, switch manufacturers have standardized SSPs 20, 22 to measure and then store traffic data every 30 minutes, although SSPs 20, 22 can be modified to measure traffic data more or less frequently. These measurements of unprocessed traffic data include measurements at a component level of total use, maintenance usage, termination calls and termination "" concentrator calls. The measurements of total use, usually in hundreds of seconds of calls ("CCS"), the time that the component is in use during the segment. The total use includes regular use (for example, per customer) and the use of maintenance, which measures the time that the component is not available to handle the traffic of the regular customer. Termination calls refer to the number of calls terminated to the line unit ("LU") (or if it is a 3-way LU, the number of 3-way calls handled by that component). Termination hub calls are the number of failed events that occur because all equipment is in use (for example, blocked calls). The link data can also be measured. (In other words, the traffic data includes the measurements of the use of the switching processors). The collector 40 supplies a stream of traffic data to the processor 42 and NIW 44. (Alternatively, for the old networks 10, the collector 40 can supply its traffic data stream first to a network data system, which supports the processor 42 of traffic design and the NIW 44 can then obtain traffic data from the network data system). The collector 40 and the processor 42 can be displayed on several separate or simple computer platforms. For example, a "workstation or personal computer" platform can accommodate both processor 42 and NIW 44, in order to collect and process traffic data from the selected components of multiple SSPs, 20 or 22. Alternatively, the processor 42 and the NI 44 can be deployed on the warehouse platform, such as an NCR 5100 Unix base server, which operates the Teradata Relational DBMS and sold by NCR.The database 41 can include data registers of traffic collected by collector 40 and include: (1) the detailed traffic data for each monitored switch component for the most recent 90 days (or more) and (2) the historical summary data in the form of a daily ratio of average occupied hours for each of the 30 most recent days, used to determine the 30-day average of operation and historical data showing the dynamic copy of the daily summary data for the first of each month for a selected period of time (for example, 3 years) "" . The database 41 also maintains several description and reference files. Additionally, several auxiliary data sources 45 can be coupled to the NIW 44 and provide various types of information to the database 41 in order to assist in the traffic design functions. For example, cross-reference switching tables (to help correlate switching location keys that differ between administrative systems), traffic tables (Poisson distribution tables, used in determining component capabilities), ODF (File of Office Data, with data of manual configuration or data of line of work), SCM (Address of Capacity of Switching, that is the designer of the traffic and the glider of the network), LSD &F (Demand and Establishment of Line Switching, which is a planning system), MR7 (Management Report 7, which produces monthly statistical reports that count the work quantities of clients that use certain switching characteristics), LNA (Line and Number Management, which assigns lines and carrier systems to switches) or COSMOS (a system that mechanically assigns lines to line units), can be provided. These auxiliary sources 45 generally describe features of the SSP, 20 and 22 or their components. Figure IB shows a detailed view of a particular SSP, 20, 22. Generally, independently of the manufacturer or type of SSP, 20, 22, each will have at least the following components, from which collector 40"obtains traffic data: (1) line units "ending in copper lines (analog LUs); (2) line units terminating in integrated" "digital" "loop bearer lines (DLs, LUs); (3) line units terminating in the ISDN ("Integrated Services Digital Network"), lines (ISDN LUs); "(4) 3-way conference call or bridge line units (3-way LUs); and (5) switching processors, which operate several software (switching) programs and which the present invention similarly monitors to determine if the processor is overloaded and in need of improvement (Switching Processors) Figure IB shows that each of these components interferes with the collector 40, which periodically retrieves the stored traffic data.Several switch vendors can supply the operator service or other switching components from which the traffic data can be collected, if desired. - Also, as new technologies are deployed within network 10, SSPs 20, 22 are enhanced to handle the physical line connections required by those technologies. if ADSL becomes a viable technology option, SSPs 20, 22 will be adapted as components to handle clients that have those particular types of lines and traffic data can be collected and analyzed from those ADSL line units, according to this invention. Generally, traffic data is collected periodically through a full 24 hours (or other period of time). A "segment" refers to a measurement of simple traffic data from a single switch or switch component, taken over a selected period of time, typically every 30 minutes. The collector 42 takes enough segments to describe the traffic handled by the particular switching component for a work day - which may be a full 24 hour period or a shorter period (eg, 12 or 18 hours more active) of the day). Figure 3 is a flow diagram showing the steps involved in collecting and processing traffic data from one or multiple switching components. Step 100 begins the process having the information _ of memory usage and of generation of the switching component, during periods of time of 30 minutes (or other) in the day. The collector 40 retrieves this data from the buffer. This may involve the collector 40 determining in step 102 from which the components of the SSP 20, 22, the traffic data will be collected. Alternatively, the collector 40 can be configured to collect traffic data from a previously selected set of switch components. For example, the collector 40 may periodically check the SSPs 20, 22 to obtain data from each of the components, shown in Figure IB.Filtering and Processing of Traffic Data Stage 102 averages the last 30 days of the data collected in the first time period. Step 104 repeats step 102 for each successive time period through the full 30 days until all traffic data for each time period has been averaged. Figure 2A shows representative sets of segment traffic data collected from a switch component selected over several days and averaged. The present invention can be configured so that the collector 42 collects similar traffic data from each component of each SSP 20, 22 within the network 10. The step 104 results in a number of average component uses for each time slot, as shown in Figure 2B.The step 106 then selects from the averaged traffic data the 2 average measurements of 30 consecutive higher minutes (e.g., time 2000.) This is the selected crest usage segment or time "busy", as indicated by step 108. Step 109 waits for an additional day of measurements and then automatically repeats steps 100 to 108, using the last 30 days of data collected.Thus, a daily ratio of 30 days plus Recent data is continuously maintained, as shown in Figure 2C.After the selection of ~ a particular busy hour, step 110 stores the traffic data for that busy hour, such data, stored by the user. 40 in the database 41, can be used to determine usage limitations.

Step 112 effectively filters the stored usage data by signaling the days with abnormally high or low and non-repetitive uses. These data are excluded from the average process. For example, segments collected in 24 hours can be filtered to remove data describing the highest single instantaneous peak segment for a particular component or commutation. Thus, the segment for day 5/8/97, time 2300, shown in Figure 2A, will be deleted from the traffic data. Alternatively, the improved processes described below and in Figures 4-6 can be used to filter data. All this filtering removes ridges or repeats, so they do not adversely affect the determination of an average peak segment for a selected daily ratio period. Figure 3 shows that step 114 that averages the measurements of occupied hours without signal, recorded during the last 30 days, to determine the average use of the component during the busy hour. Step 116 repeats the process to determine the average use of the occupied time of the switch component using the last 30 days of the busy hour data without signaling. Step 118 uses the average use of the component in the busy hour to compare with the thresholds, such as the component's calculated usage capacity, which describes the component's capacity. Such comparisons are used for the administration of the load.

Network Traffic Design Figure 4 shows a traffic design system 60 for monitoring and adjusting loads in the components of the SSP 20, 22, within a network 10 and based on the selection of the average peak segment for each component of the network. SSP 20, 22. Generally, the traffic design system 60 is directed to: (1) monitor the load on each component during the selected average peak segment, according to this invention; (2) compare the actual determined load of such monitoring with the threshold that is the previously determined or selected capacity of that component; and (3) alert the network designers of the smaller or greater use of particular components, so that the network 10 can be configured more optimally. The processor 42 (or the separate surveillance processor) may be coupled to the database 41 and the NI 44 to perform these functions. The processor 42 generally performs the following steps, in order to determine whether the particular switching components are under- or over-utilized. First, the average peak usage segment or the average peak occupied time, is selected for the component, as described above. The processor 42 ensures that the traffic data describing the use of the component during that time period is collected and stored in the database 41. The processor 42 determines the average use of the switching components in that average peak segment, according to the methods described above. Next, the processor 42 identifies the theoretical load of the particular component, using tables and Poisson calculations. (Phone traffic "is often assumed to have Poisson's exponential retention and arrival times, which allow the theory load to be calculated using the cumulative Poisson distribution function.) The theoretical load must be adjusted to a load of operation that ensures that the component does not block an acceptable percentage of calls.The Network Information Warehouse ("N") 44 stores and manages the enormous amount of data necessary to determine the average peak and the theoretical load of the many components of the multiple SSP, 20, 22, within the network 10. The percentage of capacity of the component can then be determined by the processor 42, comparing the average peak load in the occupied peak peak time with the operating load. then compared to the "thresholds within the database 41. If the capacity is in, near or above, the threshold, the processor 42 may signal the component as potentially overblowing. reloaded. The processor 42 may also point to the sub-loaded components. The merchandiser / terminal 64 and the demand server 66 allow network traffic administrators 10"to operate queries and reports against the NIW 44 data. Questions may be through the management network 62, which may be a wide area network that provides Internet connections within of a carrier region A simple or multiple merchandiser / terminals 64 allow for the sending and viewing of reports, ad hoc questions, and the maintenance of description and reference files. printing of these reports.

A. Non-Representative Data Removal An additional benefit of the present invention "involves the ability of dynamic signaling and removal of the above analysis traffic data, which does not represent the typical traffic handled by a particular switching component. non-representative traffic data while also accounting for the overall traffic volatility that drives a particular switching component Figure 5 illustrates a sampling of the traffic data representing the CCSs handled by a particular Line Unit in the SSP 20 during the Use segment "average peak on a particular day. The method of the present invention establishes an upper limit 50 and a lower limit 52, within these limits 50, 52 are the representative traffic data and out of which are the non-representative traffic data. The traffic data is collected for a period "of selected daily ratio, usually 30 days, but Figure 5 shows the data collected in approximately a period of 5 months.A first average value 54 is determined based on all the data of collected traffic for a daily relationship period A lower limit 52 is selected, which is less than, but based on, the first average value 54. For example, a lower limit 52 may be set as 95% of the first average value 54. Daily traffic data below the first mean value 54 are removed from the daily ratio, a second average value is calculated based on the remaining first filtered daily data, and the method then determines the standard deviation between the two. filtered daily ratio traffic data, and the second average value 56. An upper limit 50 can then be adjusted based on the second average value 56 and the data exceeding the upper limit are discarded to create a doubly filtered daily relationship. For example, an upper limit 50 that is equal to the second average value 56 plus two standard deviations, has been found useful in removing excessively volatile traffic data, while maintaining representative data. By selecting the lower limit 52 to be just below the first average value 54, while selecting the upper limit 50 to be substantially greater than the second average value 56, the traffic data for the particularly low traffic days are removed from the calculations of the average peak loads of the switching components, while many of the days of high traffic data, but not unusually high days, are maintained.This method ensures the elimination of low traffic data that would otherwise suggest to the network designers that the switching component could handle more traffic At the same time, except for high, exceptionally volatile traffic data, most data is maintained and analyzed in order to properly design the switching component to handle high loads .

B. Calculation of the Use of the Operation Point As described in Practice No .23"5-070-100 of AT &T, expedition 10 (November 1995), analog LUs end up switched to analog subscribers and transform their signals into digital signals. comprised by SSP 20 and the rest of the network. Multiple service lines of analog LUs, vary from 250 to 640 lines; however, only 64 slots or time connections are available for the LU. Thus, it is said to have line concentration ratios of 4: 1 (256/64), 6: 1 (384/64), 8: 1 (512/64) or 10: 1 (640/64). The above processes do not calculate the capacity of each component (for example of each analog LU). Instead, the operating capabilities for a full central office, SSP 20, are calculated, based on the average busy time for the full SSP 20, and the approximate assumptions about which traffic data to the flag or connection, are aberrant The present invention accurately determines the actual use capacity or the use of the Point of Operation ("OP") of each component of SSP 20, with the use of the following formula: OP = 111 + 4 * PL * K * CV2} - i 0 72 *? * cv2 The following table describes the values for the variables in the equation.

OP is the Use of the Operating Point of the LUs (ie the capacity of the LU) PL is the capacity of CCS (hundreds of seconds of the call) of the appropriate service load curve (there is no load variation, CV = 0.0) based on the blocking criteria (B0) chosen or calculated, as described in the Pile AC No. 2 .5-0U / 'U loo ae AT &T, Edition 10.00"(November 1995) In other words, it is the load capacity theory that is derived from the Poisson tables, such as that indicated in the Practice No. .235-060-110 of AT & T, Edition 7.00 (February 1993).

K is a constant, empirically derived from the appropriate load service curve, as shown in Figure 6, which projects the Operation Point (in CCS) versus the CV or coefficient variation.

CV is the variation of the coefficient for day data] per day. The sample size for which the coefficient of variation is calculated is one defined by the user In order to calculate exactly the capacity of the operating point of each LU or other switching component, the system must determine a coefficient of variation ("CV") that characterizes the variations in the traffic load that a particular LU or _ot. ro-switching component perform during the selected average peak segment. Basically, the CV measures the volatility of the data around a selected average, the formula for " CV is: CV = standard deviation / average or = s / average Where: = Square root 1. { ? (xt - average) "'/ (n-l).} average = S X, / n, n = the total number of average LUs in a month X is the total use of the busy time for the LU during the 1st day Figure 7 shows a comparison between the determinations of use, according to the above processes and the determinations of use made using the methods of the invention. The above processes simply determine a general CV for the entire SSP_ 3, 0 and then use it to determine the average usage capacity (in CCS) for each component in the SSP29. In contrast, the present invention calculates a CV for each component of SSP 20 and then determines each limit of actual use of the component. As an example, the SSP 20 may be a 5ESS switch provided by Lucent Technologies with at least 2, Line Units (LU). Figure 6 compares the actual use capacity of each LU that the method of the present invention determines against the ability to use the previous processes. Figure 6 illustrates that if the above processes are used, LU 1-11 and 13-23 each will be over-designed; in other words, each LU 1-11, 13-23, will be able to handle more calls. However, the LU 12 has been sub-designed and can not keep up with the traffic assigned to it. This sub-design will create quality service problems that will be costly or impossible to diagnose using the above methods. Note that although these methods are described for use in determining traffic usage through analog LUs, they can be easily used for other types of computer components. As an example, suppose that the present invention monitors component 12 of SSP 20. If this component 12 is designed to handle mostly residential traffic, the method illustrated in Figure 3 will determine that an average peak usage segment for the component 12 occurs at 1900 hours, as shown in Figure 2B Figure 4 can represent the process of all traffic data for component 12 collected during peak usage of 1900 hours over a selected period. previously used to indicate unrepresentative measurements and determine the actual average usage capacity for component 12. If this component 12 is designed for the traffic load determined in accordance with the above methods (as also shown in Figure 6), the system 60 will instead have to be determined as a sub-designed component 12 for its given traffic flow. Those skilled in the art will recognize that the present invention can be used with various types of communication networks, in addition to the network 10. Such persons will also recognize that the configuration and method of operation of the particular illustrative network 10, shown in FIG. Figure 1. Similarly, the skilled person will recognize that the methods of the present invention know "can apply to other elements of the network, such as STPs" or packet switches, for which similar traffic data can be collected. Additionally, the type of traffic data collected from such network elements may include the traffic data discussed herein or other data such as the traffic data described in the tables set forth in the US Patents, Nos. 5,359,649 and 5,333,183, each of which is hereby incorporated by reference in its entirety, The foregoing is provided for purposes of illustration, explanation and description of the various embodiments of the present invention, and will be apparent to ordinary experts in the art. and - adaptations that can be made without departing from the scope or spirit of the invention and the following claims For example, the illustrative time periods, switches or networks, described herein, in no way limit the scope of the following claims, since that persons skilled in the art will recognize that the present invention can easily be modified for use with "periods of" time, switches or interchangeable networks.

Claims (25)

1. A process to monitor and improve the control of communications network traffic; miañejadas by at least one switch, which has a plurality of components, the process comprises the steps of: a.) generate, in selected components of the switch, traffic data, which describe the communications traffic of the network, handled by each of the components of the switch; b-) periodically collect traffic data from the selected switch components, "to form segments describing traffic data for different periods of time; c.) automatically select at least one segment for each component of the switch, which describes the peak usage average of each component of the switch; and d.) analyze the traffic data, which describes the traffic of communications handled by a particular switching component, during the selected segment of the particular switch component.

2. The process of claim 1, further comprising the step of filtering the traffic data to eliminate the aberrant traffic data.

3. The process of claim 1, wherein the selection step further comprises the steps of: a.) Collecting the segments describing the traffic during a single day, -b.) Storing a set of segments to form a daily relationship that comprises a preselected number of traffic data days; c. ) averaging the traffic data of common segments through the daily relationship that share common time periods on different days; e d.) identify the segment during which the highest average peak use occurs.

4. The process of claim 1, further comprising the step of dynamically detecting a displacement in the segment, which describes the average peak usage.

5. The process of claim 4, wherein the detection step further comprises the steps of: a.) Updating, automatically and periodically, the daily relationship to include segments for the most recent preselected number of days; and b.) repeat the automatic selection stage.

6. The process of claim 1, wherein the analysis step further comprises the steps of: a.) Determining a real load, handled by the particular component during the selected segment; and b.) compare the actual load to an operating load of the component, to determine if the component is being underused or overused.

7. A method for monitoring selected components of multiple telecommunications switches, in order to better control communications traffic, within a communications network, having a number of network elements, this method comprises the steps of: a.) Generating and temporarily storing segments of selected switching components of traffic data describing telecommunications traffic, handled by the component over a discrete "" time period; b.) interface with the component, to periodically recover the segments, -c.) store the segments in order to create a daily relationship comprising a mobile window d communications traffic, handled by the component; d. )) select from the segments a period of time of peak usage, during which the peak use of the component occurs e) dynamically detect changes in the peak usage of a component, updating, periodically and automatically, the daily relationship with new segments, and repeat the selection stage; and f) compare an average load of the component during the peak usage time period with an operating load of the component, in order to handle the load on the component.

8. The method of claim 7, further comprising the step of filtering the segments to remove the aberrant traffic data, before performing the selection step.

9. The method of claim 7, wherein each of steps a.) Through f.) Are performed for each of the selected components of multiple switches.

10. The method of claim 7, wherein the comparison step includes the step of determining the average load of the component, averaging the traffic data of segments within a daily relationship, which describes the telecommunications handled by the component over the period of Peak usage time.

11. A system for mechanically controlling communications traffic, selecting, automatically and dynamically, a period of time, during which a network element for a telecommunications network handles peak communications traffic and analyzes the amount of traffic handled by the element network, this system comprises: a.) a collector, to periodically record traffic data comprising information about communications traffic, which passes through the network element; b.) a database, to receive and store the registered traffic data; c.) elements, coupled to the database, for selecting from the traffic data a first busy state hour, which describes the average peak usage of the network element; d.) elements for updating the traffic data in the database, in order to automatically detect _a displacement in the average peak usage of the network element at a second occupied state hour.

12. The system of claim 11, further comprising resources for filtering the traffic data to remove that traffic data that does not conform to the preselected acceptable parameters.

13. The system of claim 11, wherein the network element comprises a component of a switch, selected from the group consisting of analog line units, digital circuit line units, 3-way call-line units, processors switching and ISDN line units.

14. The system of claim 11, further comprising an administration network, for determining an operating capacity of the switching component and comparing this operating capacity with the traffic handled by the switching component, during the ridge use time period. average.

15. A communication network, which comprises: a.) A plurality of network elements, each with multiple components, adapted to generate traffic data describing the traffic handled by each of the selected components for a previously established segment of time; b.) elements, coupled to the switches, to periodically collect traffic data of the selected components of each network element; c.) elements to receive and store the collection elements, enough traffic data to create a daily relationship, which describes the traffic handled by the selected components over a selected period of days; and d.) elements to select, automatically and dynamically, a segment that describes the average peak usage of each of the selected components.

16. The system of claim 14, further comprising elements for determining the actual load on each of the selected components, during their respective average peak-usage segments.

17. The system of claim 15, further comprising elements, coupled to the selection elements, for determining an operating capacity of each switching component and comparing the operating capacity of each switching component with the actual load.

18. The system of claim 15, further comprising a plurality of data sources, coupled to the receiving and storage elements, to provide data describing characteristics of the network elements and components.

19. A method for determining the capacity of a selected component of a switch, used in a communications network, the component has a selected average peak time usage segment for which the traffic data is collected and stored in a database, to create a daily multi-day relationship of traffic measurements, this method comprises the steps of: a.) identifying traffic data that is representative of the communications traffic handled by the component; and b.) determine an average traffic usage limit for the particular component of the representative traffic data.

20. The method, according to claim 19, wherein the identification step comprises the "stages of: a.) Establishing a lower limit; b) setting an upper limit; c.) Determining the standard deviation of the traffic data that is within the upper and lower limits, and d) create a subset of representative traffic data that includes data that (i) _ falls above the lower limit and (ii) falls under at least one deviation standard above the upper limit.

21. The method, according to claim 20, wherein the lower limit is based on a percentage of the first average value, determined from all "traffic related to component usage" data during the average peak usage segment.

22. The method, according to claim 20, wherein the upper limit is based on a second average value, determined from the remaining traffic data, after excluding any traffic data that falls below the lower limit.

23. The method, according to claim 19, wherein the limit of so of average traffic is the average value of the remaining traffic data.

24. The method, according to claim 19, wherein the identification step comprises the steps of removing from the database (i) traffic data describing particularly low traffic periods and (ii) eliminating traffic data. which describe unusually high traffic periods.

25. The method, according to claim 19, wherein the determination step comprises the step of calculating the operating capacity for the selected component in the average peak usage segment. SUMMARY OF THE INVENTION A process and system is disclosed to automatically and dynamically determine the correct time segment, during which a switch, or components of a switch, receives peak traffic from a communication network. The process collects traffic data regarding the use of a selected switch, or selected switch components. Traffic data is collected over a selected period of a daily ratio (for example 30 days), which is periodically updated with new traffic data to maintain a daily relationship of the most recent 30 days of traffic data collected. These collected traffic data is filtered to remove the aberrant or corrupted catos. An average peak usage segment for all segments of the daily relationship period is selected from the remaining filtered traffic data. By regularly adding new data to, and removing old data from, the daily relationship, the resulting mobile window will reflect recent changes in peak usage. Regularly (for example, daily) select the average peak usage segment that detects such changes rapidly, with the result that the process of the present invention dynamically reacts to, and determines the changes in, the peak usage segment for particular switching components. . The peak usage segments and associated traffic data, determined and collected by the present invention, result in improved network traffic designs. A system for collecting, processing and evaluating traffic data is disclosed in order to design network elements for optimal traffic flow. Additional methods are disclosed for "the use by the system in dynamically filtering aberrant data with improved accuracy and in determining average traffic usage limits for particular switching components." The methods of the present invention allow for the most effective use. of costly network switching resources, by the provision of a more accurate determination and a more dynamic tracking of the use of the switch at the component level.