Presentation
SCAGuard: Detection and Classification of Cache Side-Channel Attacks via Attack Behavior Modeling and Similarity Comparison
DescriptionExisting approaches for detecting cache side-channel attacks (CSCAs) and mitigating information leaks either fail to detect new variants or have to impractically update deployed systems. In this work, we propose SCAGUARD, to detect CSCAs via attack behavior modeling and similarity comparison. Specifically, to capture both attack-relevant syntactic code and semantic cache information, we introduce the notion of cache state transition enhanced basic block sequences (CST-BBSes) and propose an approach to automatically construct such CST-BBS models. We evaluate SCAGUARD using real-world attacks and diverse benign programs. The results confirm our approach outperforms, in particular, on new attack variants, existing detection approaches.
Event Type
Research Manuscript
TimeThursday, July 13th3:55pm - 4:10pm PDT
Location3003, 3rd Floor
RISC-V
Security
Hardware Security: Attack and Defense