Fast Oblivious AES\\A dedicated application of the MiniMac protocol.

We present actively secure multi-party computation of the Advanced Encryption Standard AES. To the best of our knowledge it is the fastest of its kind to date. We start from an efficient actively secure evaluation of general binary circuits that was implemented by the authors of [DLT14]. They presented an optimized implementation of the so-called MiniMac protocol [DZ13] that runs in the pre-processing model, and applied this to a binary AES circuit. In this paper we describe how to dedicate the pre-processing to the structure of AES, which improves significantly the throughput and latency of previous actively secure implementations. We get a latency of about 6 ms and amortised time about 0.4 ms per AES block, which seems completely adequate for practical applications such as verification of 1-time passwords.