TY - JOUR TI - Putting secure computation to work DO - https://doi.org/doi:10.7282/T3R213C5 PY - 2015 AB - The demand for solutions that enable secure computation in distributed systems is only increasing. However, the current state of secure computation "in the wild'' is highly unsatisfactory: provably secure solutions receive little attention, while untested security technologies with questionable security claims are being broadly marketed and deployed. The classical approach to secure computation is secure multi-party computation (MPC) protocols, which allow a set of parties to jointly compute any given function while provably preserving the privacy of inputs and correctness of the output. Due to their generality, MPC protocols subsume a wide range of secure computation scenarios. However, MPC's adoption rate in the real world is extremely low. One reason is that MPC protocols have complex and differing security definitions; another is that MPC protocols typically do not map cleanly onto existing network application architectures. Another promising research direction is secure protocols tailored for specific types of computations, such as text search and database access control. Such protocols attempt to strike a balance between efficiency and security, often by allowing a quantified amount of leakage. However, the practical security level of many of these protocols is not well understood. Our research aims to make secure computation more deployable and trustworthy by bridging the gap between theoretical and applied secure computation. This dissertation presents 1) a systematization of MPC protocols that helps clarify their security and efficiency properties, 2) new, efficient protocols for access control in private databases, and 3) new security analyses, including statistical attacks, for currently used searchable encryption technologies. KW - Computer Science KW - Computer security KW - Electronic data processing--Distributed processing LA - eng ER -