Mariëlle Stoelinga, UT-EEMCS-FMT
In this talk, I will discuss my work on models for risk analysis, both fundamentals algorithms and applications: Fault trees are a prominent formalism to model and analyse the propagation of component failures. It is used by a range of institutions, like NASA, ESA, Rijkswaterstaat, Airbus to compute system a number of system metrics, such as the reliability (ie, the probability that the system is up during its mission time) and availability (the average percentage of up time).
In this talk, I will show how the deployment of stochastic model techniques leads to more powerful and flexible analysis methods: I will discus several techniques to minimize the underlying stochastic models, via bisimulation minimization (aka lumping) and confluence reduction. Furthermore, I will briefly go into attack trees, which are similar to fault trees, but model malicious system attacks.