A Subjective Network Approach for Cybersecurity Risk Assessment

We propose a subjective Bayesian network approach for cybersecurity risk assessment to address the limitations of traditional risk assessment models, which use precise values for the likelihoods of cyber-attacks. In many situations, it is often difficult to elicit accurate probabilities due to lack of knowledge, or insufficient historical data, making the evaluation of risk in existing approaches unreliable. With this approach, we seek to better reflect the reality underpinning the model and offer a better approach to decision-making via the modelling of uncertainty about the probability distributions in the form of subjective opinions, resulting in a model taking second-order uncertainty into account. We develop a subjective Bayesian network for cybersecurity risk, and then discuss the risk evaluation and decision analysis problem under the proposed model. Finally, our approach is evaluated against classical Bayesian networks using the scenario of wiper malware in an industrial control system. Our results show that taking uncertainty about the probabilities into account during security risk analysis can lead to different outcomes, and therefore different security decisions.