Privacy-Preserving Secret Shared Computations Using MapReduce
Shlomi Dolev, Peeyush Gupta, Yin Li, Sharad Mehrotra, Shantanu Sharma
Abstract
Data outsourcing allows data owners to keep their data at \\emph{untrusted}\nclouds that do not ensure the privacy of data and/or computations. One useful\nframework for fault-tolerant data processing in a distributed fashion is\nMapReduce, which was developed for \\emph{trusted} private clouds. This paper\npresents algorithms for data outsourcing based on Shamir's secret-sharing\nscheme and for executing privacy-preserving SQL queries such as count,\nselection including range selection, projection, and join while using MapReduce\nas an underlying programming model. Our proposed algorithms prevent an\nadversary from knowing the database or the query while also preventing\noutput-size and access-pattern attacks. Interestingly, our algorithms do not\ninvolve the database owner, which only creates and distributes secret-shares\nonce, in answering any query, and hence, the database owner also cannot learn\nthe query. Logically and experimentally, we evaluate the efficiency of the\nalgorithms on the following parameters: (\\textit{i}) the number of\ncommunication rounds (between a user and a server), (\\textit{ii}) the total\namount of bit flow (between a user and a server), and (\\textit{iii}) the\ncomputational load at the user and the server.\\B