Survivability prediction of ad hoc networks under attack

Survivability analysis focuses on the ability of network entities to function during incidents such as attacks. Currently, testing survivability of ad hoc networks consists of running scenarios with several configurations, often thousands, to obtain an understanding of the impacts of an attack. This process is very latent, choice of configurations are subjective or random, and results do not generalize to different scenarios. Focusing on these problems, in this paper, we introduce a novel method for efficient survivability analysis that uses machine learning and an attacker-focused network representation. We have collected a dataset and use it to build a classifier that accurately (above 97% true positive rate) predicts flow loss due to spoofing and data forwarding attacks.