Exploiting Petri nets to support fault tree based dependability analysis

Abstract

This paper explores the possibility of converting Fault Trees (FT) into the Generalized Stochastic Petri Net (GSPN) formalism. Starting from a slightly modified version of a conversion algorithm already appeared in the literature, the aim of the paper is to exploit the modeling and decision power of GSPN for both the qualitative and the quantitative analysis of the modeled system. The qualitative analysis resorts to structural properties and is based on a T-invariant analysis. In order to alleviate the state space explosion problem deriving from the quantitative analysis, the paper proposes a new formalism for FT, that is referred to as High Level FT (HLFT), in which replicated redundant units are folded and indexed. Starting from the HLFT formalism, a new conversion algorithm is provided that translates a HLFT into a Stochastic Well-formed Net (SWN). The computational saving of using SWN with respect to GSPN is carefully examined considering an example of a fault-tolerant multiprocessor system.