On Dependability Assessment of Fault Tolerant Systems by Means of Statistical Model Checking

The problem of dependability assessment can be solved analytically just under predefined conditions. If they do not hold, alternative approaches must apply. Widely, they rely on the Monte Carlo simulation, suffering by the high computational complexity. Some rest on further instruments such as probabilistic timed automata that have been shown to be efficient to solve problems in various areas. However, more general as well as precise and faster instruments such as stochastic timed automata (STA) and statistical model checking (SMC) are available for the same purpose the moment. In the paper, basic terms and principles behind the construction of reliability models and dependability assessment on the STA/SMC basis are summarized, followed by a demonstration of their practical applicability in the area of non-repairable systems. Our main goal is to show that, instruments of STA/SMC can facilitate the dependability assessment process even in adverse conditions such as presence of multiple faults of various parameters.