Combination of component fault trees and Markov chains to analyze complex, software-controlled systems
This work addresses the problem of analyzing safety-critical systems for industries like automotive, though it appears incremental as it combines existing methodologies.
The paper tackles the limitations of Fault Tree analysis and Markov Chains in analyzing complex, software-controlled systems by integrating Markov Chains into Component Fault Tree models, enabling modular and compositional safety or reliability analysis as demonstrated in an automotive case study.
Fault Tree analysis is a widely used failure analysis methodology to assess a system in terms of safety or reliability in many industrial application domains. However, with Fault Tree methodology there is no possibility to express a temporal sequence of events or state-dependent behavior of software-controlled systems. In contrast to this, Markov Chains are a state-based analysis technique based on a stochastic model. But the use of Markov Chains for failure analysis of complex safety-critical systems is limited due to exponential explosion of the size of the model. In this paper, we present a concept to integrate Markov Chains in Component Fault Tree models. Based on a component concept for Markov Chains, which enables the association of Markov Chains to system development elements such as components, complex or software-controlled systems can be analyzed w.r.t. safety or reliability in a modular and compositional way. We illustrate this approach using a case study from the automotive domain.