Secure function evaluation (SFE) is a cryptographic technique for evaluating functions among multiple parties while preserving the privacy of parties' inputs. Traditional SFE techniques can impose high performance penalties compared to non-privacy preserving protocols, and SFE has largely been a theoretical curiosity without much practical application. In this thesis, we demonstrate techniques for improving the performance of SFE. We present a new protocol for SFE using Ordered Binary Decision Diagrams, and we also show how to design optimized protocols for dynamic programming problems. Our techniques are demonstrated on k-means clustering and the Smith-Waterman gene sequence alignment algorithms. All of our protocols are implemented and evaluated for real-world performance, and are available for download.
展开▼
