Browse Prior Art Database

Multiple Phase Pseudo-Error Rate Monitor

IP.com Disclosure Number: IPCOM000121050D
Original Publication Date: 1991-Jul-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 3 page(s) / 146K

Publishing Venue

IBM

Related People

Bates, RJS: AUTHOR [+4]

Abstract

This article describes a technique for over-sampling a single bit of data in order to provide diagnostic information about the status of a fiber-optic link, either for error detection and fault isolation or as a means to improve the error rate of the link.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 51% of the total text.

Multiple Phase Pseudo-Error Rate Monitor

      This article describes a technique for over-sampling a
single bit of data in order to provide diagnostic information about
the status of a fiber-optic link, either for error detection and
fault isolation or as a means to improve the error rate of the link.

      Fiber-optic communication links play an increasingly important
role in computer systems - in channel communications and for
multiprocessor interconnection.  In these applications it is
important to provide a robust link design in terms of insuring low
error rates with a system that is tolerant of manufacturing
variations.  This article presents a technique by which a fiber-optic
link can monitor its own bit error rate performance (a pseudo-error
rate) (see [1,2]) by multiple sampling of the data signal within a
single bit interval.  This information can be used either as a link
status signal to a supervisory processor or as a feedback control
signal to optimize the data link performance.

      The basic idea is illustrated in Fig. 1, which shows the eye
diagram for a typical fiber-optic data communication link.  The data
is sampled at the optimum point, that at which the signal amplitude
is a maximum.  However, by sampling at other times within the bit
interval, it is possible to monitor the error rate performance of the
link. Because the error rate is an exponential function of the
signal-to- noise ratio, and because of the finite rise/fall times of
the signal, it is possible to compare two samples within the same bit
interval and estimate the error rate. In general, the optimum sample
has a much greater probability of being the correct data than a
sample which is offset from the optimum sampling point.  By comparing
the two sample results, it is possible to estimate the error rate,
and so monitor the performance of the link.  As the frequency of the
disagreements between the two samples increases, one can infer that
the link performance is degrading, either due to decreasing amplitude
of the eye diagram or degradation in the horizontal eye opening
(i.e., timing window).  This error rate signal can be used to flag a
service action by a customer engineer or be used as an electrical
feedback signal to improve the link performance. Alternatively, one
could sample at three different times within the bit interval, and
based on a voting scheme, optimize the error performance.

      Implicit in the above discussion is the assumption that the
different samples within a bit interval are independent of each
other.  If the noise is strongly correlated, then if one sample is in
error, it is...