Performance of Shuffling - Taking It to the Limits.

In this paper, we propose further performance improvements for Wikstrom’s shuffle proof. Compared to an implementation based on general-purpose exponentiation algorithms, we show that the overall shuffle operation can be accelerated by approximately one order of magnitude. The improvements result partly from applying special-purpose algorithms for fixed-based and product exponentiations in the right way and partly from other optimization techniques. Given that shuffling is often one of the most time-consuming tasks in cryptographic voting protocols, the achieved speed-up is significant for practical implementations of electronic voting systems.