Title: Encoding Power Traces as Images for Efficient Side-Channel Analysis

Abstract: Side-Channel Attacks (SCAs) are a powerful method to attack implementations of cryptographic algorithms. State-of-the-art techniques such as template attacks and stochastic models usually require a lot of manual preprocessing and feature extraction by the attacker. Deep Learning (DL) methods have been introduced to simplify SCAs and simultaneously lowering the amount of required side-channel traces for a successful attack. However, the general success of DL is largely driven by their capability to classify images, a field in which they easily outperform humans. In this paper, we present a novel technique to interpret 1D traces as 2D images. We show and compare several techniques to transform power traces into images, and apply these on different implementations of the Advanced Encryption Standard (AES). By allowing the neural network to interpret the trace as an image, we are able to significantly reduce the number of required attack traces for a correct key guess.We also demonstrate that the attack efficiency can be improved by using multiple 2D images in the depth channel as an input. Furthermore, by applying image-based data augmentation, we show how the number of profiling traces is reduced by a factor of 50 while simultaneously enhancing the attack performance. This is a crucial improvement, as the amount of traces that can be recorded by an attacker is often very limited in real-life applications.