Method for determining network switch capacity

• US 5,978,358 A
• Filed: 04/04/1997
• Issued: 11/02/1999
• Est. Priority Date: 10/01/1996
• Status: Expired due to Term

• Current Assignee: Apple Inc., Bell Northern Research, LLC (Hilco Inc. (d/b/a Hilco Global))
• Original Assignee: Nortel Networks Corporation
• Inventors: Madhavapeddy, Seshagiri R., Yang, Ping, Trink, Eddy H., Wang, Jing
• Primary Examiner(s): Kizou, Hassan
• Assistant Examiner(s): Pezzlo, John
• Application Number:
  US08/833,362
• Time in Patent Office: 942 Days
• Field of Search: 370/230, 370/232, 370/233, 370/234, 370/235, 370/237, 370/250, 370/252, 370/253, 370/377, 370/378, 370/379, 370/410, 455/560, 455/507, 455/405, 455/422, 379/133, 379/134, 379/137, 379/138, 379/139, 379/113, 379/111
• US Class Current: 370/234
• CPC Class Codes: H04M 3/365 Load metering of control unit
• CPC Class Codes: H04W 16/22 Traffic simulation tools or models
• CPC Class Codes: H04W 24/02 Arrangements for optimising operational condition

Abstract

The switch capacity prediction process of the present invention collects operational measurements from the cellular switch. These operational measurements indicate the frequency of the call events processed by the switch. The other input/output (I/O) time used by the call processor to process messages not related to call processing is separated from the call processing I/O time. A relationship between the other I/O time and the sum of the call processing time and call processing I/O time is determined. A weighted average is determined from the summation of the products of: the frequency of events and the call processing time, the frequency of events and the call processing I/O time, and the other I/O time. The weighted average is then divided into the call processor occupancy time, in hours, to determine the switch capacity.

Claims

1. A method for determining a capacity level of a network switch, the switch having a processor to process a plurality of call events, the processor'"'"'s time comprising message processing time and maintenance time, the message processing time comprising call processing time, call processing input/output time, and non-call processing related input/output time, the method comprising the steps of:

collecting operational measurement data from the network switch;

separating the non-call processing related input/output time from the message processing time;

determining a frequency of the call events;

determining a weighted average for the real-time cost of the plurality of call events in response to the frequency of call events, the call processing time, the call processing input/output time, and the non-call processing related input/output time; and

computing the capacity level in response to the weighted average and the message processing time.

2. The method of claim 1 wherein the capacity level is computed by determining a result of the division of the call message process time by the weighted average.

3. The method of claim 2 and further including transforming the call event process time into milliseconds before dividing by the weighted average.

4. A method for predicting busy hour call attempts for a cellular switch, the cellular switch having a processor to process a plurality of call events that is comprised of different sets of call events, the plurality of call events being processed within a message processing time that comprises call processing time and input/output time, the input/output time comprising call processing input/output time and remaining input/output time, the method comprising the steps of:

collecting operational measurement data from the cellular switch;

retrieving the call events from the operational measurement data;

separating the input/output time from the call event process time;

defining an input/output relationship between the call event process time and the remaining input/output time;

determining a call event frequency for each set of call events;

generating a weighted average by summing a product of each call event frequency and the call event'"'"'s respective message processing time, a product of each call event frequency and the call event'"'"'s respective call processing input/output time, and the input/output relationship; and

determining the busy hour call attempt prediction in response to the weighted average and the call event process time.

5. The method of claim 4 wherein the input/output relationship is a constant times each set of different call events.