Secure Multiparty Computation Programming Framework

Secure multi-party computation allows a group of untrusted parties to jointly run a function on their private inputs without revealing any information beyond the result.It's an important research field of modern cryptography.With people's increasing emphasis on privacy,secure multi-party computation has become more and more widely used in academia,industry,and government.Starting from Yao's "Millionaire Problem",protocols for secure multi-party computation have been researched and developed for decades,and many excellent protocols have been proposed.Based on these protocols,many general-purpose multi-party computation programming frameworks that can perform arbitrary Turing complete function have been constructed.We found that these frameworks all have problems such as poor ease of use,difficulty in interacting with existing programming languages,and most of them aren't suitable for big data computation.In order to solve these problems,an ORAM(Oblivious RAM)based hybrid protocol secure multi-party computation framework is designed,and secure multi-party computation compiler construction schemes based on JSR 269 and the LLVM compiler framework are proposed,in which Java and C/C++ programs can be compiled into secure multiparty computation application on the framework proposed in this thesis.Specifically,the main contributions of this thesis are as follows:1.We evaluate the existing commonly used general secure multi-party computation frameworks through multiple evaluation indicators such as ease of use,functionality,interactivity,etc.And the problems of the existing frameworks such as poor usability and their causes are summarized and discussed.2.We design an ORAM-based hybrid protocol secure multi-party computation framework,which has the sub-linear circuit and communication complexity provided by ORAM for big data computation.What's more,it provides more possibilities for the expressiveness and functionality of the front-end language based on the hybrid protocol of Yao's garbled circuits,Boolean sharing circuits and arithmetic sharing circuits.3.We propose and implement secure multi-party computation compiler construction schemes based on JSR 269 and LLVM compiler framework,in which Java and C/C++ programs can be compiled into secure multi-party computation application on the framework proposed in this thesis.These compilers are tested through several typical secure multi-party computation application scenarios.Through experiments,the results show that the framework and compiler construction schemes proposed in this thesis are better than existing frameworks in terms of ease of use,functionality,and interactivity.