I-Corps: Lightweight Multi-Party Computations with Purpose Control

Purdue University, West Lafayette IN

Investigators

Abstract

The commercial potential of this I-Corps project is to provide a computational platform that facilitates multiple entities to transact where data owned by different stakeholders is used but not disclosed to anyone, including the platform. The resulting transaction would allow entities to make informed decisions about not only choosing from a pool of potential partners but also explore avenues to generate value from their potential partnership(s). This platform eliminates any possibility of losing ownership of data as well as any misuse/abuse of data in transactions. Consequently, it would motivate partnerships among disparate entities, leading to value enhancements and new products. It can be a game changer in allowing a range of businesses to create new products/services, especially when such businesses are complimentary to each other. Moreover, this platform would allow personalized service deliveries to end customers, involving different entities, who can collaborate without exchanging data. This I-Corps project uses a suite of building blocks within its computational platform to perform mutually-agreed computations. Each building block is designed to perform a specific arithmetic or a logical operation using a single external server (SES), wherein the server learns nothing about the inputs. The entities using these building blocks learn only the outputs of the computation but not their collaborators' inputs. Thus, computations executed using these building blocks preserve input confidentiality and data ownership for every involved entity. There are existing technologies that achieve some of the stated functionalities but they either use multiple external servers or computationally intensive cryptographic techniques. These requirements substantially affect the performance overhead and scalability, which make it difficult to deploy them in real-world applications. In contrast, this SES-based building blocks lead to substantial reductions in performance overhead as compared to the existing technologies and are easily scalable.