Boosting Profiled Cache Timing Attacks With A Priori Analysis

Abstract

The vulnerability of cryptographic devices to side-channel attacks is of interest in the domain of information security. The success of a side-channel attack depends on the crypto-algorithm implementation, the platform being attacked, and the attack strategy. While the former two parameters are generally beyond the adversary's control, the choice of the attack strategy is solely with the adversary. However, there is no unique “best attack strategy.” The attack strategy that works best for one platform may not be the best for another. Further there is no systematic way to choose the best attack strategy from the available pool. In this paper, we analyze a category of side-channel attacks known as profiled cache-timing attacks and develop a methodology by which an adversary capable of limited number of side-channel measurements can choose the best strategy prior to the actual attack. The methodology is tested on several platforms and cipher implementations and shows that the best attacking strategy can be estimated closely, without the requirement of an exhaustive search.