/sites/default/files/styles/banner_image/public/default_images/inside-page-banner_2_1.jpg?itok=Er8q0C-3
Beckman Fellow 2011-12

Manoj Prabhakaran

Computer Science

Prabhakaran imageTheory and Practice of Secure Multi-party Computation

Secure Multi-party Computation (SMC) is a powerful concept in cryptography that allows mutually distrusting parties to collaborate without compromising the privacy of their data. A challenge in bringing SMC to practice is to find the right balance between good security and practicality.

Professor Prabhakaran proposes to develop SMC constructions that are efficient enough to be adapted for practice while also offering theoretically sound security guarantees. Recently, along with his collaborators, he has developed an approach to take efficient SMC protocols that are secure only in semi-trusted settings and use them to build protocols that are fully secure without degrading the efficiency too much. This approach promises greater efficiency and is suited to meeting other practical requirements, such as handling floating point arithmetic and being non-interactive.

A second part of Professor Prabhakaran’s project studies fundamental properties of various SMC tasks. His group will investigate how the cryptographic complexity of a task (a measure of the task’s cryptographic sophistication) is related to the computational complexity of problems that must be exploited for meeting its security requirements, under a new framework dubbed intractability abstractions.

In the current Internet paradigm in which users visit web servers (e.g., browsing, online shopping at trusted websites) or interact with each other for communication (e.g., email, instant messaging), security guarantees are only to protect the communicating parties against external parties. Wide application of SMC could upend that paradigm by allowing users to carry out computational tasks directly with each other, such as conducting online business negotiations involving private data, where the users need to protect themselves against their collaborators.