Swarm Attacks against Network-Level Emulation/Analysis

It is always assumed that if the attackers can achieve their goal by exploiting a vulnerability once, they won’t exploit it twice. This assumption shapes our view of what attacks look like, and affects the design of many security systems. In this work, we propose the swarm attack, in which the attacker deliberately exploits the same vulnerability multiple times, each intended to carry out only a small part of the attack goal. We have studied eight systems that detect attacks using networklevel emulation/analysis, and find them surprisingly vulnerable to attacks based on this strategy.