Exploiting Bayesian Networks for the analysis of combined Attack Trees

The growing need to find proper countermeasures able to protect critical infrastructures from threats has addressed the definition of quantitative methodologies for risk assessment. One of the most difficult aspects in this topic is the evaluation of the effects of attacks. Attacks Trees represent one of the most used formalisms in the modeling of attack scenarios: notwithstanding some extensions have been proposed to enrich the expressiveness of the original formalism, some effort should be spent on their analyzability. This paper defines a transformational approach that translates Attack Trees into Bayesian Networks. The proposed approach can cope with different Attack Trees extensions; moreover, it allows the quantitative evaluation of combined attacks modelled as a set of Attack Trees.