Dr. Arkady Yerukhimovich of the computer science department has received a four-year National Science Foundation grant for the project “New Approaches for Large Scale Secure Computation.” This collaborative project with George Mason University (GMU) and The United States Naval Academy (USNA) aims to enable large scale computations while preserving privacy.
According to Dr. Yerukhimovich, in today's connected world, large, distributed computations such as Tor, blockchain, and large sensor networks support thousands, if not millions, of parties, but privacy concerns often impede their collaboration in such computations because the parties involved are not willing to share their inputs.
Dr. Yerukhimovich explains the situation through a common scenario. Data about the world is collected and owned by a large number of different parties. For example, hospitals have records about their patients; banks have information about their customers; Tor relays have information about their users, and so on. In all these scenarios, it is often critical to be able to measure properties of the global state of the world such as the prevalence of a disease, or the stability of financial markets, but there are also strong incentives—both legal and social—for the companies to protect the privacy of the data they collect.
“Enabling such measurements while preserving the privacy of individual inputs will greatly improve our ability to understand the state of the world and the large (medical, financial, technological) systems that are a major part of the modern world,” says Dr. Yerukhimovich.
Specifically, his project with GMU and USNA aims to develop new protocols for secure multi-party computation to enable large numbers of parties to perform joint computations on private data without having to disclose that data to each other.
“In particular, we will develop protocols that scale gracefully to thousands or millions of parties without incurring the expensive per-party costs of prior solutions,” Dr. Yerukhimovich explains. “To achieve this, we propose a new framework for committee-based secure computation that distributes the computation among many small committees to maximize utilization of available resources—for example, bandwidth and computation—while preserving privacy. We hope that this new approach to building secure computation will enable its use in today's large scale distributed computations.”
Dr. Yerukhimovich is excited about the project and the chance to delve deeper into his area of expertise. “Secure multi-party computation [MPC] is a very exciting research area that combines techniques from information theory and cryptography,” he notes. “Designing new, efficient, MPC protocols requires one to build on top of some of the most exciting developments in cryptography over the last thirty years.”