Uncertain Analytic Hierarchy Process for Risk Assessment in Cyber-Physical Systems

This paper introduces an innovative holistic risk assessment framework that integrates Bayesian Networks (BN) with Multi-Criteria Decision Making (MCDM) to calculate and consolidate heterogeneous risk values into a single, comprehensive risk metric. Specifically, the framework employs an array of risk-specific BNs to derive a set of heterogeneous risk metrics, which are then integrated using the Uncertain Incomplete Analytic Hierarchy Process (UIAHP) technique. This approach involves soliciting pairwise comparisons of risks from a panel of experts, whose assessments are used to compute weights associated with each risk metric; further to that, the methodology incorporates information related to the degree of certainty in the experts’ evaluations to enhance the robustness of the derived weights. The effectiveness of the proposed risk assessment methodology is demonstrated through a real hardware-in-the-loop case study conducted in a laboratory environment, which simulates a scaled-down critical infrastructure for water distribution.