Executive Summary : | Secure multi-party computation (MPC) protocol is a privacy-preserving technology where a set of parties interact with each other over insecure channels to compute a function on their private data. At the end of the protocol, each party learns the output of the computation, while their inputs remain private. The protocol guarantees the correctness of the result and the privacy of honest parties' inputs. Starting from the seminal work by Yao [Yao82] and subsequently by Goldreich, Micali, and Wigderson [GMW87], there has been significant progress in improving the asymptotic and concrete efficiency of two-party and multi-party computation protocols. However, to utilize the full power of MPC in developing privacy-preserving technologies on a large scale more research and development are required. In this project, our primary goal is to develop efficient techniques for secure arithmetic computation. In MPC literature most of the works consider boolean circuits as their model of computation. However, in many applications, it is natural to express the computation in terms of an arithmetic circuit. One can always transform an arithmetic circuit into an equivalent boolean circuit, but that incurs significant communication overhead. Constructing protocols that securely compute an arithmetic circuit directly might be an efficient alternative in those cases. More specifically on the theoretical side, we would like to study and design more efficient protocols for Oblivious Linear Function Evaluation (OLE), which is considered to be the basic building block for secure arithmetic computation. On the practical side, we aim to use those protocols to design a platform for Secure Collaboration, where multiple organizations can use their confidential data to collaborate and find some necessary information or statistics without revealing any other information about that data. This kind of system can be useful in many health sector applications, where multiple hospitals or healthcare research institutions can collaborate and use their sensitive patient data to find useful statistics without leaking any information about individual patients. |