Core Concepts
The authors propose Taypsi, a language that statically enforces privacy policies in MPC applications, eliminating dynamic overhead. This approach improves performance significantly.
Abstract
Taypsi introduces a novel method to enforce privacy policies in MPC applications, enhancing performance and scalability. The language decouples privacy concerns from program logic, offering considerable performance improvements over existing strategies.
The content discusses the challenges faced in implementing secure multiparty computation techniques and introduces Taypsi as a solution. By transforming programs into semantically equivalent versions that statically enforce user-provided privacy policies, Taypsi eliminates the overhead associated with dynamic enforcement.
The paper outlines the contributions of Taypsi, including the introduction of Ψ-types for modular translation of non-secure programs into efficient, secure versions. Experimental evaluations demonstrate exponential performance improvements over previous state-of-the-art solutions like Taype.
Overall, Taypsi offers a promising approach to address the complexities of enforcing privacy policies in MPC applications involving structured data and complex requirements.
Stats
Our experimental evaluation demonstrates considerable performance improvements on a variety of MPC applications.
The resulting system features exponential improvements on complex benchmarks.
Secure version of filter produced by Taype takes more than 5 seconds to run with an oblivious list list≤ with sixteen elements.
The tape semantics strategy used by Taype results in exponential slowdowns for complex applications.