A Bayesian-node fusion approach for optimization of cyber attack modeling at the virtual layer of the cloud

Networks, the cloud included, are susceptible to cyber-attacks owing to the myriad of vulnerabilities exhibited in networked components. Attackers chain together these vulnerabilities to generate optimized attack paths with varying degrees of success. The challenge in cloud networks has been to capture the dependencies amongst the vulnerabilities in the attack paths of the exploited components. We in this paper partition the cloud into three discrete layers with concentration on the virtual layer where, via applied node fusion in the resultant Bayesian attack network, endeavor to capture the aforementioned relationships by grouping like nodes together for conjunction probabilities of intersection and disjunction probabilities of union of two or more events. We further explore these dependencies through the connectivity matrix and employ CVEs and edge weighting for effective path determination. We likewise demonstrate how failure nodes can be induced and utilized for attack mitigation and prioritization in security plan formulation.