TrumanBox: Improving Dynamic Malware Analysis by Emulating the Internet

Abstract

Dynamic analysis of malicious software (malware) is a powerful tool in countering modern threats on the Internet. In dynamic analysis, a malware sample is executed in a controlled environment and its actions are logged. Through dynamic analysis, an analyst can quickly obtain an overview of malware behavior and can decide whether or not to indulge into tedious manual analysis of the sample. However, usual dynamic analysis exposes the Internet to the threats of an executed malware (like portscans) because advanced concealment techniques of malware often require full Internet access. For example, a missing link to the Internet or the unavailability of a specific server often causes the malware to not trigger its malicious behavior. In this paper, we present TrumanBox, a technique to emulate relevant parts of the Internet to enhance dynamic malware analysis. We show that TrumanBox not only prevents many threats but also enlarges the scope of the types of malware that can be analyzed dynamically.