Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Mean Time Between Errors for Error Counters Management

IP.com Disclosure Number: IPCOM000101206D
Original Publication Date: 1990-Jul-01
Included in the Prior Art Database: 2005-Mar-16
Document File: 3 page(s) / 110K

Publishing Venue

IBM

Related People

Owen, SA: AUTHOR [+2]

Abstract

There are various counters defined in today's communication network. These counters, by monitoring the errors as well as the traffic, provide valuable information to the management of the network.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 49% of the total text.

Mean Time Between Errors for Error Counters Management

       There are various counters defined in today's
communication network.  These counters, by monitoring the errors as
well as the traffic, provide valuable information to the management
of the network.

      Normally, there is a threshold associated with an error counter
such that when the threshold is reached, some predefined action
(e.g., Alert) will be initiated.  However, this threshold mechanism
has no concept of time.  All errors are treated equally regardless of
the time interval.  The result is that an Alert may be generated even
though the time it takes the error counter to get to the threshold is
very long.  One simple solution is to define the age of the counter
as the time it takes the counter to reach the threshold.  If the time
is too long, then an Alert will not be generated.  However, this
solution also has problems, because it includes all the errors during
the measurement period even though some may have a long time interval
between them.  The ideal thing is to understand the error arrival
rate and discard those errors that are considered aged, such that
they will not be included in any of the analysis.  The result will be
much more accurate and realistic.

      This article describes a technique for accomplishing this goal
by determining the average arrival of the errors or the Mean Time
Between Errors (MTBE).  If MTBE is relatively long, then the error
count can be discarded; otherwise, errors are counted.

      Definition:
      n         The current value of the error counter, n ranges from
0 to N.
      t0        Relative time value, it is the time when the error
counter starts counting.
      tn        Relative time value, it is the time when the error
counter has reached the value of n.
      tn-t0     Total time for the error counter to go from 0 to n
      MTBE      Mean time between errors, the formula for MTBE is
(tn- t0)/n, where n>0.
      TM        Threshold for the MTBE.  If the calculated MTBE is
greater than Tm, the average mean time between errors is higher
(better) than the threshold and the error counter is reset. If the
MTBE is not greater than Tm, the average time between errors is less
than the accepted tolerance, and the error counter continues to be
incremented.
      N         Threshold set for the error counter.  When n=N, a
predefined action, such as Alert or Event Report will be initiated.

      Algorithm:
      Initialization
           n=0;
           t0=tn=Current time; (This is the time when counter starts
counting)
      Error detected
           tn=Current Time; (Time stamp the latest error)
           n=n+1; (increment the error counter)
           IF MTBE > Tm  (MTBE is  greater than the threshold limit)
           THEN n=0;  (all errors are d...