Difficulty-level metric for cyber security training

摘要: Cyber security training systems work as a suitable learning environment for educating cyber analysts on how to detect and defense before real attacks happen. As is an iterative process, the assessment component not only assesses knowledge gained by analysts, but also adjusts difficulty of lessons accordingly based analysts’ performance. In this paper, we present attack graphbased probabilistic metric measure lesson scenarios’ levels. Based causal relationships between vulnerabilities in graph, apply Bayesian Reasoning aggregate individual into value representing attackers success likelihood achieve goal. However, one major complication using that it does allow cycles, which exists graphs. We identify different types cycles graphs propose efficient algorithm remove while preserving cyclic influence probability calculation.