Presentation
Smart Detection of Obfuscated Thermal Covert Channel Attacks in Many-core Processors
DescriptionThis paper introduces the first machine learning-based detection method for short-duration thermal covert channel (TCC) attacks. Our smart solution employs short windows of time-domain measurements, instead of the traditional Discrete Fourier Transform (DFT), to detect TCCs. To evaluate our solution, we introduce a new obfuscated short-duration attack. Our experiments show that the new attack remains undetected under advanced DFT-based state-of-the-art approaches, reducing their accuracy to 18%. In contrast, our smart technique detects state-of-the-art and obfuscated attacks with an accuracy of 99%. Moreover, our solution outperforms the overhead of the DFT-based state-of-the-art solution by more than 14x.
Event Type
Research Manuscript
TimeWednesday, July 12th5:12pm - 5:29pm PDT
Location3012, 3rd Floor
Security
Embedded and Cross-Layer Security
