Effective Detection of Active Worms with Varying Scan Rate

Abstract

Active worms have been posing a major security threat to today's Internet. It is widely believed that active worms continue their evolutions. In this paper, we model a new form of active worms called varying scan rate worm (the VSR worm in short). The VSR worm deliberately varies its scan rate and is able to avoid being effectively detected by existing worm detection schemes. The emerging "Atak" worm belongs to this category of worms. To countermeasure the VSR worm, we design a new worm detection scheme called attack target distribution entropy based dynamic detection scheme (DEC detection in short). DEC detection utilizes the attack target distribution and its statistical entropy in conjunction with dynamic decision rules to distinguish worm scan traffic from non-worm scan traffic. We conduct extensive performance evaluations on the DEC detection scheme, using real-world traces as background scan traffic. Our data clearly demonstrates the effectiveness of the DEC detection scheme in detecting VSR worms as well as traditional worms