Dismiss
Surety is performing system maintenance this weekend. Electronic date stamps on new Prior Art Database disclosures may be delayed.
Browse Prior Art Database

# Dynamic Smoothing for FDDI Link Error Rate Calculations

IP.com Disclosure Number: IPCOM000110961D
Original Publication Date: 1994-Jan-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 76K

IBM

## Related People

Mieczkowski, DJ: AUTHOR

## Abstract

Disclosed is a method for calculating the Link Error Rate (LER) for FDDI Station Management (SMT) using a dynamic smoothing factor algorithm that reduces transitory fluctuations in the reported LER, yet provides adequate response to substantial LER variations.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Dynamic Smoothing for FDDI Link Error Rate Calculations

Disclosed is a method for calculating the Link Error Rate (LER)
for FDDI Station Management (SMT) using a dynamic smoothing factor
algorithm that reduces transitory fluctuations in the reported LER,
yet provides adequate response to substantial LER variations.

The FDDI (Fiber Distributed Data Interface) SMT (Station
Management) ANSI standard 7.2 specifies that link errors on the fiber
must be detected and measured over time.  This measurement is the

When the LER exceeds a predefined level, an alarm message is
sent on the ring to notify FDDI management agents of the LER
condition.  The LER may fluctuate under certain circumstances.  These
fluctuations may result in the LERexceeding the alarm threshold for a
short period of time.  It may not be desirable to report these
short-lived anomalies as an alarm condition to the management agents.
A way to avoid reporting these fluctuations is to use a smoothing
algorithm that tempers the LER.

A smoothing algorithm typically is implemented as follows:

LER = (Weight_Factor * Current_LER) + ((1-Weight_Factor) *
Previous_LER)

The weight factor value must be greater than 0 but less than or
equal to 1.  A weight factor of 1 is equivalent to no smoothing.  A
weight factor close to 0 represents a high degree of smoothing.

This smoothing algorithm provides adequate results when each
LER calculation is based on an identical time interval.  However,
problems arise when the time interval between LER calculations
varies.  These difficulties can be resolved by using a weight factor
that varies.

When link errors occur in bursts, the time interval between
calculations will be small and significant smoothing is desired.  A
small weight factor will provide this degree of smoothing.  However,
a small weight factor will provide very slow reponse to adjusting the
LER rate in low error rate conditions.

When link errors occur only over a period of hours or days, the
time interval between calculations is obviously large.  Little or no
smoothing is required in this case.  A large weight factor close to
or equal to 1 will provid...