Popularity-based Detection of Domain Generation Algorithms

Abstract

In order to stay undetected and keep their operations alive, cyber criminals are continuously evolving their methods to stay ahead of current best defense practices. Over the past decade, botnets have developed from using statically hardcoded IP addresses and domain names to randomly-generated ones, so-called domain generation algorithms (DGA). Malicious software coordinated via DGAs leaves however a distinctive signature in network traces of high entropy domain names, and a variety of algorithms have been introduced to detect certain aspects about currently used DGAs.