Operative diagnosis algorithms for single-fault in graph-based systems

A number of real-life systems can be modeled, to a certain level of abstraction, as directed graphs for the purpose of fault diagnosis. In these graphs, system components are represented as nodes and the fault propagation between any two nodes is represented by an edge between them. Alarms are attached to some components to detect faults. The fault diagnosis problem deals with computing the set of all potential faulty components, P-S, corresponding to a set of ringing alarms A(R). Exactly one component can become a fault source at any time, and the fault can subsequently propagate to others. In this paper, we present two algorithms for the single fault diagnosis problem which perform better than existing algorithms under different conditions.