Assessing the likelihood of cyber network infiltration using rare-event simulation

Abstract

Network infiltration is one of many types of cyber-based attacks that may be of interest to a cyber security analyst. Sufficient observation of particular events that may be uncommon during network infiltration requires special simulation techniques. This paper presents an application of the importance sampling method to estimate the likelihood of a successful network infiltration, given that sufficiently many network alerts have not been generated to achieve said success. The benefits of utilizing importance sampling within this context are assessed against the use of standard simulation.