CAREER: Trading Security for Efficiency in Secure Computation

点击翻译按钮获取中文摘要

Secure computation allows multiple parties, each holding private data, to perform arbitrary computations on the joint data without revealing anything other than the desired output. In the last fifteen years, secure computation has moved from being a purely theoretical area of research, to one that is being actively deployed in a variety of settings. However, technical barriers to adoption still exist: the communication and computational costs are quite high. At the same time, improvements in the last fifteen years have brought us quite near to the limit of what is feasible under standard definitions of security. This project investigates new techniques for improving the efficiency of secure computation beyond the known theoretical limits.This project explores new security relaxations, and new methods of utilizing existing security relaxations, in order to help scale secure computation to larger volumes of data, and a larger number of participating parties. The investigators envision a fully decentralized data plane in which user data is accessible by service providers only at the discretion of the data owner. Computations performed on this data would be subject to appropriate user-specified controls, and the results of the computations would reveal only what is minimally necessary to meet the goals of the data owner. Service providers could still monetize the data, but without ever seeing it, and only with the explicit agreement of the owner. The project has outreach activities that introduce secure computation, privacy and cryptography to various communities in the DC area.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.