Ridge-based profiled differential power analysis

Abstract

Profiled DPA is an important and powerful type of sidechannel attacks (SCAs). Thanks to its profiling phase that learns the leakage features from a controlled device, profiled DPA outperforms many other types of SCA and are widely used in the security evaluation of cryptographic devices. Typical profiling methods (such as linear regression based ones) suffer from the overfitting issue which is often neglected in previous works, i.e., the model characterizes details that are specific to the dataset used to build it (and not the distribution we want to capture). In this paper, we propose a novel profiling method based on ridge regression and investigate its generalization ability (to mitigate the overfitting issue) theoretically and by experiments. Further, based on cross-validation, we present a parameter optimization method that finds out the most suitable parameter for our ridge-based profiling. Finally, the simulation-based and practical experiments show that ridge-based profiling not only outperforms ‘classical’ and linear regression-based ones (especially for nonlinear leakage functions), but also is a good candidate for the robust profiling.