Forecasting Run-Times of Secure Two-Party Computation

Abstract

Secure computation (SC) are cryptographic protocols that enable multiple parties to perform a joint computation while retaining the privacy of their inputs. It is current practice to evaluate the performance of SC protocols using complexity approximations of computation and communication. Due to the disparate complexity measures and constants this approach fails at reliably predicting the performance. We contribute a performance model (PM) for forecasting run-times of secure two-party computations. We show the correctness of our PM by an empirical study on the problem of secure division which is relevant for many real world SCs, e.g., k-means clustering or supply chain optimization. We show that our PM can be used to make an optimal selection of an algorithm and cryptographic protocol combination, as well as to determine the implicit security tradeoffs. The predictions of our PM can be used to design or select more efficient or more secure protocols.