Encryption Chip Template Attack Based on ResNet Convolutional Neural Network

Abstract

Aiming at the efficiency of traditional template attacks in actual attacks, this paper focuses on the ResN et network model with excellent feature extraction capabilities in the field of image recognition, and proposes a template attack method based on the ResN et network model. In order to adapt the neural network to the characteristics of the side channel data, the network model structure was adjusted appropriately. In the experiment, the traditional multivariate Gaussian distribution and the ResNet-based convolutional neural network model were used to establish templates. The final experimental results show that the success rate of template attacks based on the ResNet model is 94.5%, which is 9.4% higher than the traditional template attacks.