Remote cache-timing attacks against AES

We present a cache-timing attack on the Advanced Encryption Standard (AES) [14] with the potential to be applied remotely and develop an evaluation framework for comparing the relative performance of the attacks under various simulated network conditions. We examine Bernstein's original AES cache-timing attack [3], and its variants [6, 12, 10]. We conduct an analysis of network noise and develop a hypothesis fishing concept in order to reduce the number of samples required to recover a key in our implementation of the attacks of [3]. Our rough estimate for the number of samples required is about 2 × 109 which is comparable to the estimate 4 × 109 of our month-long experiment using Bernstein's technique [3].