Photonic Network Communications, 1:3, 235±255 (1999)
# 1999 Kluwer Academic Publishers, Boston. Manufactured in The Netherlands.
Fault Localization at the WDM Layer
Carmen Mas, Patrick Thiran, Jean-Yves Le Boudec
Institute for Computer Communications and Applications (ICA), EPFL (Swiss Federal Institute of Technology),
CH-1015 Lausanne, Switzerland
Received April 1, 1999; Revised July 19, 1999
Abstract. A single failure in a communication network may trigger many alarms. When the communication network uses optical ®bers as
transmission medium and increases its capacity by using Wavelength Division Multiplexing (WDM), the number of alarms and the dif®culty to
locate the failure are considerably higher. In this case, a single failure may interrupt several channels which causes a large information loss.
We propose an Alarm Filtering Algorithm (AFA) for the fault management of an optical network that supports multiple failures and works in
the presence of non-alarming elements, that is, network components which may fail but never generate an alarm (e.g. optical ®bers). The
algorithm provides a list of components whose failure explain the observed alarms. It avoids the use of failure probabilities, which are dif®cult
to estimate, and does not need a global knowledge of the network topology. Moreover it also tolerates alarm losses and false alarms.
The algorithm is tailored to the speci®c behavior of the hardware components of an optical network when a failure occurs. The classi®cation
of the network components according to the alarm signals they generate enables a formalization of the alarm-®ltering problem and results in an
ef®cient algorithm for localizing the failure(s). This algorithm is applied to the WDM rings of the COBNET network  (COBNET is a
European ACTS project) and to a meshed optical network with the ARPA2 topology.
Keywords: alarm ®ltering, optical communication networks, fault diagnosis
Communication networks have been developing
rapidly during the last few years. The evolution of
the transmission technology and the rapid increase in
user needs have in¯uenced major changes in net-
works. One new technology is the use of optical ®ber
as transmission medium, which allows high transmis-
sion rates because of their low attenuation and high
capacity. The capabilities of optical ®ber are increased
by Wavelength Division Multiplexing (WDM) that
makes possible the transmission of several informa-
tion channels through a single link so that the capacity
of the link can become considerably higher.
Nowadays, it is possible to ®nd market equipment
that supports from 80 to 128 wavelengths on a ®ber,
each ®ber carrying up to 10 Gbit/s of information .
In a near future, this equipment may support up to one
thousand wavelengths. But the capacity advantage
comes however with some drawbacks if a failure
occurs. This ®rst one is that a failure can now result in
the loss of a large amount of information. The second
one is that the network management application is
overwhelmed by a large number of alarms due to a
single failure involving several channels. For these
reasons a fast fault recognition process is needed and
is critical in ensuring network availability. It is
composed of two phases: alarm ®ltering and testing.
The output of the ®rst phase is the list of network
components that might have failed and triggered the
alarms received by the manager. The second phase
con®rms or discards the previous result by checking
these components. This testing part is outside the
scope of this paper, which is devoted to the alarm
processing phase only.
The problem of identifying failures is closely
related to the physical layer. Upper layers react on
physical failures in different ways in order to re-
establish the communication (for example, by
rerouting or changing switch positions). But even if
the connection is successfully re-established, the
physical layer management must localize the failure.
This should be based on the alarms emitted by the
hardware components and the knowledge of the
components' behavior. As the physical layer is
E-mails: (carmen.mas@ep¯.ch), ( patrick.thiran@ep¯.ch)