摘要：The majority of recently demonstrated Deep-Learning Side-Channel Attacks (DLSCAs) use neural networks trained on a segment of traces containing operations only related to the target subkey. However, when the number of training traces is restricted such as in the ASCAD database, deep-learning models always suffer from overfitting since the insufficient training data. One data-level solution is called data augmentation, which is to use the additional synthetically modified traces to act as a regularizer to provide a better generalization capacity for deep-learning models. In this paper, we propose a cross-subkey training approach which acts as a trace augmentation. We train deep-learning models not only on a segment of traces containing the SBox operation of the target subkey of AES-128, but also on segments for other 15 subkeys. We show that training a network model by combining different subkeys outperforms a traditional network model trained with a single subkey, and prove the conclusion on two well-known datasets.
关键词：Side-Channel AttackDeep LearningAESCross-Subkey Training
