Model-based Confidentiality Analysis under Uncertainty

In our connected world, ensuring the confidentiality of the software systems we build becomes increasingly difficult. Model-based design time confidentiality analyses have been proposed to cope with this complexity early. However, the usefulness of such analyses is limited due to uncertainty about the software architecture itself and the software’s execution environment. This leads to conclusions about confidentiality violations that lack both precision and comprehensiveness. Although there exist approaches to deal with design time uncertainty, existing research lacks precise statements about the impact of uncertainty on confidentiality. To address this, we include uncertainty as part of our software architectural model. We extend a data flow-based analysis to include the impact of uncertainty on confidentiality violations. The results of the case study-based evaluation show high accuracy with typical design time uncertainty. Also, our analysis yields more precise statements about the impact of uncertainty on confidentiality than the state of the art.