Assessing dependability of safety critical systems using diverse evidence

Abstract

A primary objective of the DATUM (Dependability Assessment of safety critical systems Through the Unification of Measurable evidence) project was to improve the way dependability of software intensive safety-critical systems was assessed. The authors' hypothesis was that improvements were possible if multiple types of evidence could be incorporated. To achieve the objective, the authors had to investigate how to obtain improved dependability predictions given certain specific information over and above failure data alone. A framework for modelling uncertainty and combining diverse evidence was provided in such a way that it could be used to represent an entire argument about a system's dependability. The various methods and technologies for modelling uncertainty were examined in depth and a Bayesian approach was selected as the most appropriate method. To implement this approach for combining evidence, Bayesian Belief Networks (BBNs) were used. With the help of a BBN tool, a framework for dependability assessment was provided that met the original objective and which was subsequently proved to be practical and highly popular. A major benefit of this approach was that otherwise hidden assumptions used in an assessment become visible and auditable.