Distributed Causal Model-Based Diagnosis Based on Interacting Behavioral Petri Nets

This paper deals with the problem of causal model-based diagnosis of distributed systems. The setting we consider is a collection of interacting behavioral Petri nets (BPNs). Each BPN model represents the causal behavioral model of one subsystem and its interactions with neighboring subsystems. Interactions among subsystems are modeled by tokens that pass from one model to another via common places. Diagnosis reasoning scheme exploits, in a first step a backward reachability analysis on each net model to obtain local diagnoses; and in a second step, it exploits a forward reachability analysis for ensuring that local diagnoses are consistent and form global ones.