CSR: Small: System Support for Fast, Scalable, and Verifiable Fully Homomorphic Encryption

The University Of Central Florida Board Of Trustees, Orlando FL

Investigators

Abstract

Digital platforms like cloud computing provide significant opportunities for services such as transaction processing and machine learning but also expose users to privacy risks. Fully Homomorphic Encryption (FHE) offers a promising solution by enabling computations directly on the encrypted data without the need to decrypt it, ensuring privacy and compliance with national regulations. However, while FHE guarantees confidentiality, it does not guarantee computation integrity and verifiability, leaving users vulnerable to errors or tampering. The project plans the development of a system for verifiable FHE that ensures both integrity and verifiability with low-performance overhead, enabling safer and broader applications in sensitive domains. The project's broader significance and importance are: (1) advancing the field of verifiable and private data sharing and analysis, as emphasized in the US national strategy to advance privacy-preserving data sharing and analytics; (2) deepening the understanding of interactions between verifiable algorithms, privacy-supporting computing, and trusted hardware design; and (3) enriching computer engineering curricula while encouraging STEM participation. The project develops system-level supports for efficient, verifiable, privacy-preserving data sharing and analysis with algorithm-hardware-system co-optimized solutions. It focuses on three coherent research thrusts: (1) designing near-zero-cost verifiable algorithms for linear operations in FHE-enabled encrypted computations; (2) integrating an integrity-only Trusted Execution Environment (TEE) with the first thrust to strategically support verification for both linear and non-linear FHE operations; and (3) redesigning hardware with a clean-slate TEE architecture to address side-channel vulnerabilities and minimize encryption overhead. These innovations aim to significantly enhance the scalability and trustworthy of privacy-preserving technologies in real-world applications. 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.