Princeton University

School of Engineering & Applied Science

Deriving Abstractions to Address Hardware Platform Security Challenges

Pramod Subramanyan
Engineering Quadrangle B327
Friday, December 16, 2016 - 2:00pm to 3:30pm

Today's computing devices store and process an enormous amount of security-critical assets. These assets are a lucrative target for cybercriminals and protecting them from malicious actors remains a key challenge in computer security. Hardware is especially important in this context: security protections implemented in software may be invalidated by faulty hardware.
In this thesis, we develop techniques that can help ensure the security of hardware platforms. We develop automated techniques for analyzing hardware designs and detecting flaws that may result in security breaches. Our techniques are based on the derivation of abstractions. An abstraction is a simplified model of a system that preserves essential detail but is easier to reason about. By developing techniques to construct abstractions in a principled and automated manner, we show that large hardware designs can be subject to automated security analysis. This can help detect flaws that may result in security vulnerabilities. Our techniques have have helped find security flaws in commercial systems.