Supporting Secure Multi-party Computation for R Framework

Abstract

The trade-off between sharing and privacy is an important issue in data analysis. Secure multi-party computation (MPC) involves all parties jointly performing a computation without revealing their own data. R is a popular domain-specific programming language for data analysis. This paper develops an MPC-enabled R framework that allows data from different data owners to be analyzed and computed without revealing the data themselves. A programming model is proposed to support MPC on an R framework. A third-party library, ABY, is used to provide the ability for MPC. The interface between C/C++ and R is studied in order to integrate the ABY library. In addition, an ABY wrapper is developed to alleviate programmers’ workloads. An experimental application demonstrates that the proposed programming model can correctly complete secure computation for the tested benchmark programs.