Develop System Design Methods

Recent trends in computing have prompted users and organizations to store an increasingly large amount of sensitive data at third party locations in the cloud outside of their direct control. In order to protect this data, it needs to be encrypted. However, traditional encryption systems lack the expressiveness needed for most applications involving big and complex data.

The Complementary Metal Oxide Semiconductor (CMOS) based security primitives typically suffer from area/power overhead, sensitivity to environmental fluctuations and limited randomness and entropy offered by Silicon substrate. Spintronic circuits can complement the existing CMOS based security and trust infrastructures. This project explores ways to uncover the security specific properties of the magnetic nanowire and capture them in detailed circuit model.

Informally speaking, Secure Multi-Party Computation (SMPC) allows two or more parties to jointly compute some function on their private inputs in a distributed fashion (i.e., without the involvement of a trusted third party) such that none of the parties learns anything beyond its dedicated output and what it can deduce from considering both this output and its own private input. Since its inception in 1982 by Yao, SMPC has advanced greatly and over the years a large body of work has been developed.

Deep Learning (DL)-powered personalization holds great promise to fundamentally transform the way people live, work and travel, but poses high risk to people's individual privacy. This project will address the privacy risks arising in DL-powered contextual mobile services by developing solutions that facilitate the use of personal information while maintaining explicit user control over use of the information.

Many challenging real world problems, e.g., voting and blind auction, require computation over sensitive data supplied by multiple mutually-distrustful entities. Elegant cryptographic theories have been developed to solve these problems without relying on a mutually-trusted third party. Practitioners also built prototypes capable of securely computing set intersection, AES encryption, Hamming distance, etc. However, many other applications, such as data mining and running universal machines, are far more complex than what can be supported by the state-of-the-art techniques.