ABSTRACT

This project designs and develops Secrecy+, a novel data analytics system that uses secure multi-party computation (MPC) and enables data holders to perform relational analytics on their collective private data with provable and configurable security guarantees. The project's core novelties include a principled optimization framework for outsourced relational MPC and a library of parallel secure operators that scale to much larger datasets than the current state-of-the-art. Secrecy+ targets use cases where multiple data holders are willing to contribute their private data towards a joint analysis (e.g., for profit, social good, or improved services), provided that the data remain siloed from untrusted or unauthorized entities. The research is grounded in two case studies: (i) secure cloud-based analytics on mobile health data, and (ii) secure cross-site analytics on datacenter logs.

The project involves three sets of tasks that span the areas of cryptography, database query evaluation, and distributed systems. First, the investigators define a unified cost model for secure operations across different MPC protocols and physical deployments. Second, they develop a relational MPC query processor that supports parallel query execution and employs optimizations that reduce MPC costs while retaining the full security guarantees of the cryptographic protocols. Third, the investigators design and implement high-level user interfaces and tools for seamless integration with existing cloud infrastructure. Project results have the potential to fundamentally change how private datasets are used by organizations, researchers, and policy makers in accordance with data privacy regulations and can pave the way for new marketplaces in the cloud.

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.

Please report errors in award information by writing to: awardsearch@nsf.gov.