Research Manuscript: Security Techniques across the Board (STAB)
Event TypeResearch Manuscript
Hosted in Virtual Platform
Embedded and Cross-Layer Security
DescriptionThis session presents security defenses and attacks across a range of abstraction layers. The first two papers present machine learning techniques to hardware-based malware detection. The first paper's approach estimates and reduces uncertainty in the detection process and the second paper presents a collaborative federated learning approach. The next two papers investigate side-channel attacks, proposing an architecture to obfuscate timing side-channels and presenting a new interruption-based side-channel attack against a secure enclave. The next paper presents a Network-on-Chip architecture to defend cloud services. The final paper presents a method to detect confidentiality vulnerabilities in mixed hardware/firmware systems using formal verification.