A translation of State Machines to temporal fault trees

Abstract

State Machines (SMs) are increasingly being used to gain a better understanding of the failure behaviour of safety-critical systems. In dependability analysis, SMs are translated to other models, such as Generalized Stochastic Petri Nets (GSPNs) or combinatorial fault trees. The former does not enable qualitative analysis, whereas the second allows it but can lead to inaccurate or erroneous results, because combinatorial fault trees do not capture the temporal semantics expressed by SMs. In this paper, we discuss the problem and propose a translation of SMs to temporal fault trees using Pandora, a recent technique for introducing temporal logic to fault trees, thus preserving the significance of the temporal sequencing of faults and allowing full qualitative analysis. Since dependability models inform the design of condition monitoring and failure prevention measures, improving the representation and analysis of dynamic effects in such models can have a positive impact on proactive failure avoidance.