Using both formal methods and well-adapted Monte Carlo rare event simulation based on GSPN to process dependability assessments on highly critical systems

摘要

Dependability assessments of highly critical systems by means of behavioural Monte Carlo simulation often lead to the problem of rare event simulation, whose solving is dramatically time-consuming. Considering this difficulty, a specific Monte Carlo simulation technique has been developed, allowing to save processing time and to get accurate estimates. This technique is based on the determination of particular events, which occur during the evolution of a system, and are necessary to the occurrence of the rare feared studied event. This position paper proposes a general scheme to perform Dependability assessments on highly critical systems in the case described above (when the studied feared event is rare). This propounded scheme uses both formal methods and well-adapted Monte Carlo behavioural simulation.