Fault diagnosis in distributed computer networks

Abstract

This thesis introduces the concept of a diagnosis algorithm in the contest of the Preparata-Metze-Chien (PMC) model. It represents a Computer-Aided-Design (CAD) tool for use in analyzing such algorithms. That is, with the tool, the user can establish a multiprocessor system, a set of test outcomes, and then analyze the properties of specified distributed diagnosis algorithm. Examples in this thesis include a system in which: 1. Correct diagnosis is achieved in a small number of iterations. 2. Correct diagnosis is never achieved. 3. An oscillating situation exists in which, faulty processors become alternately enabled and disabled.