MAD-EN: Microarchitectural Attack Detection Through System-Wide Energy Consumption
Published in IEEE Transactions on Information Forensics and Security (IEEE TIFS), 2023
Existing microarchitectural attack detectors rely heavily on hardware performance counters (HPCs), which are limited in number and may be unavailable in virtualized environments. MAD-EN proposes a new detection approach using system-wide energy consumption data collected via Intel RAPL (Running Average Power Limit). By training deep learning models on energy traces, MAD-EN achieves high detection accuracy across a wide range of microarchitectural attack types, providing a complementary and deployable alternative to HPC-based detectors.
Recommended citation: D. R. Dipta and B. Gulmezoglu, "MAD-EN: Microarchitectural Attack Detection Through System-Wide Energy Consumption," IEEE Transactions on Information Forensics and Security (IEEE TIFS), vol. 18, pp. 3006-3017, 2023.
Download Paper