An Approach To The Analysis Of Security Measure Robustness Considering Epistemic Uncertainty In Scenario Likelihood

Abstract

The current dynamic development of the security situation is pushing the risk of attacks on critical infrastructures further into the focus of both, operators and the authorities. Legislation requires critical infrastructure operators to take appropriate physical security and resilience measures. In this context, there are increased efforts to develop concepts for securing critical infrastructures against possible attacks. However, the lack of knowledge regarding the likelihood of threat scenarios causes epistemic uncertainty that impacts risk analysis. In previous work, we proposed a combination of models to make the influence of uncertainties visible: a threat model that describes a wide range of potential scenarios, a threat likelihood model, in which a probability distribution over these scenarios represents scenario likelihood, and a vulnerability model that enables the assessment of security measure effectiveness in these scenarios. Here, we extend that approach, thereby enabling the analysis of security measure robustness against vulnerability from scenarios with uncertain likelihood. For this purpose, we represent uncertain knowledge regarding that likelihood via prior distributions. Based on a notional example, we calculate vulnerability for three alternative configurations of security measures in a set of scenarios and weight that vulnerability by uncertain scenario likelihood. The resulting probability distributions show that the degree of variance in overall security measures effectiveness depends on the configuration of the security measures. Introducing simple robustness indicators, we compare the probability distributions and discuss their relevance for the robustness of security measures.