Network modelling in analysing cyber-related graphs

To improve the resilience of the computer network infrastructure against cyber attacks or causal influences and find ways to mitigate their impact, we need to understand their structure and dynamics. Here, we propose a novel network-based influence-spreading modelling approach to investigate event trajectories or paths in attack and causal graphs with directed, weighted, cyclic and/or acyclic paths. In our model, we can perform probabilistic analyses that extend beyond traditional methods to visualise cyber-related graphs. The model uses a probabilistic method to combine paths that join within the graph. This analysis includes vulnerabilities, services, and exploitabilities. To demonstrate the applicability of our model, we present three cyber-related use cases: two attack graphs and one causal graph. This model can serve cyber analysts as a tool to produce quantitative metrics for prioritising tasks, summarising statistics, or analysing large-scale graphs.