On the Design of Observers for Fault Detection in Communication Networks

Abstract

This paper provides a framework for approaching the problem of fault detection in large Communication Networks. Communication process are modeled as Discrete Event Systems (D.E.S.), so any fault will appear as a change of a D.E.S. Here we concentrate on the simplest case of Discrete Event systems, namely Finite State Machines (FSM).

The model of computation consists of two FSMs. One, called the observer, observes part of the behavior of the other. Two fundamental questions are formulated.

(a) Given a specific fault that could happen in one FSM, can the observer detect this fault even though it observes only part of the behavior of the FSM?

(b) Given a specific fault that may happen in a FSM, can we design the Minimum Alphabet Observer, i.e., an observer which observes the minimum number of event transitions and is still able to detect this specific fault?

After these questions are introduced we answer them for one specific class of faults. This is the class of faults which can be described as strings. In this case we prove that the first question can be answered in polynomial time while the second is NP-complete. For the second question a heuristic algorithm is given.

This work was carried out under the support of NSF grant #CDR-88-11111, as well as the New York Science and Technology Foundation, through its grant to the Columbia University Center for Advanced Technology.