A Framework for Understanding Dynamic Anti-Analysis Defenses

Abstract

Malicious code often use a variety of anti-analysis and anti-tampering defenses to hinder analysis. Researchers trying to understand the internal logic of the malware have to penetrate these defenses. Existing research on such anti-analysis defenses tend to study them in isolation, thereby failing to see underlying conceptual similarities between different kinds of anti-analysis defenses. This paper proposes an information-flow-based framework that encompasses a wide variety of anti-analysis defenses. We illustrate the utility of our approach using two different instances of this framework: self-checksumming-based anti-tampering defenses and timing-based emulator detection. Our approach can provide insights into the underlying structure of various anti-analysis defenses and thereby help devise techniques for neutralizing them.