CRII: SaTC: Programming and Compilation Framework for Fast, Portable Computation on General Fully Homomorphic Encryption

Cloud outsourcing has become critical for computing vast amounts of data for emerging applications, such as machine learning, bio-medical analysis, private information retrieval, etc. However, sharing data to the cloud may leak sensitive information, leading to various societal issues, including identity theft, financial loss, reputational damage, and legal consequences. Fully homomorphic encryption (FHE) is a post-quantum cryptography framework that supports computations directly on encrypted data, providing strong protection for sensitive data that remains encrypted during data transmission and cloud services. Despite the benefits, adopting FHE in real-world applications is challenging in (1) transforming application data and operations into encrypted ciphertexts and operations with restricted formats using various complex algorithmic schemes and (2) porting and optimizing on different computing hardware platforms that are critical for ensuring a practical runtime. The project's broader significance and importance are: (1) boosting the development and development of FHE to advance the national welfare by protecting data privacy in broader domains, and (2) promoting the progress of computer science and engineering to improve the availability and performance of privacy-preserving data sharing and analytics.

This project develops a novel programming and compilation framework for exploiting FHE in general privacy-preserving applications on real systems. This project's novelties are (1) a new domain-specific language that improves the programmability of general FHE applications with different algorithmic schemes and (2) an optimized compiler infrastructure to generate high-performance FHE programs on various emerging hardware platforms. This project combines concepts and techniques at different levels in the full system stack of privacy-preserving computing, including privacy-preserving applications, programming languages, and computer systems. The proposed research aims to provide an innovative end-to-end solution that drives fundamental advancements in various aspects of privacy-preserving computing, aiming to proliferate the adoption of practical privacy-preserving computing in broader areas.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Award Number: 2451744
Principal Investigator: Minxuan Zhou
Funds Obligated: $174,999
State: IL