A Unified Theory for f1/f2-Diagnosable Communication Networks

We propose a new diagnosis model that allows a continuum between the two classical invalidation models in system-level diagnosis, namely, the PMC and BGM models. This new invalidation model assumes that there is no maximal set of connected faulty units of cardinality greater than f2 (1≤f2) in which all the faulty units can declare other faulty unit in this set to be fault-free out of a total of up to f1 faulty units in a network with n units (f1≤n). This f1/f2-invalidation model provides a realistic representation of behavior of faulty units in the diagnosis of heterogeneous communication networks. A complete characterization theorem provides the necessary and sufficient conditions for a network to be one-step f1/f2-diagnosable. We propose an adaptive algorithm for the diagnosis of communication networks of arbitrary topologies using the f1/f2-invalidation model.