Applying Bayesian Belief Networks to System Dependability Assessment

The dependability of technological systems is a growing social concern. Increasingly computer based systems are developed that carry the potential of increasing catastrophic consequences from single accidents. There have been significant research advances in assessment methods. However dependability assessment of computer systems in practice is still a very uncertain and often adhoc procedure. Decision making about system dependability is an uncertain affair and must account of failures in expertise and be capable of integrating different sources of evidence. A more meaningful way of reasoning about systems dependability can be achieved by rejecting current ad-hoc dependability assessment methods and replacing them with the idea of dependability argumentation. Bayesian Belief Networks (BBN’s) is proposed as the most promising technology to support this kind of dependability argumentation.