Template attacks on ECC implementations using performance counters in CPU

Abstract

We demonstrate a new set of template attacks on ECC implementations using the performance counters in CPU. Template attacks are powerful mechanisms that can combine statistical intelligence for modelling side channel leakages and can thus compromise complex crypto implementations. Automated attack phases add to the efficiency of this approach. We introduce a new approach of using CPU counter values for constructing templates to carry out attacks on crypto implementations, which also opens up the possibility of many other template attacks that have been demonstrated earlier using power analysis, to be feasible on the modern day architectures. The values obtained from multiple CPU counters are used to generate templates, which are further matched with the traces obtained on the target implementation for guessing the bit values of the secret scalar used during the scalar multiplication function of the ECC implementation. We demonstrate the feasibility of the attack approach on virtualized environments such as the resources of the cloud computing infrastructure. We further model the attack, define the evaluation metrics and demonstrate the effectiveness of the attack approach.