Evading Stepping-Stone Detection with Enough Chaff.

Stepping stones are used extensively by attackers to hide their identity and access restricted targets. Many methods have been proposed to detect stepping-stones and resist evasive behaviour, but so far no benchmark dataset exists to provide a fair comparison of detection rates. We propose a comprehensive framework to simulate realistic stepping-stone behaviour that includes effective evasion tools, and release a large dataset, which we use to evaluate detection rates for eight state-of-the-art methods. Our results show that detection results for several methods fall behind the claimed detection rates, even without the presence of evasion tactics. Furthermore, currently no method is capable to reliably detect stepping-stone when the attacker inserts suitable chaff perturbations, disproving several robustness claims and indicating that further improvements of existing detection models are necessary.