Profiled Power Analysis Attacks by Efficient Architectural Extension of CNN Implementation

Soroor Ghandali1, Samaneh Ghandali2, Sara Tehranipoor1
1Santa Clara University, 2Google


In a recent line of works, several masking and unmasking AES design have been proposed to secure hardware implementations against power analysis techniques and machine learning-based attacks. Although Deep learning profiling techniques like Convolutional neural networks have been successful in security testing during the last years, evaluation of security still requires a suitable leakage model against profiled side-channel attacks. In this paper, we propose an improved profiling method to exploit the power consumption of complex cryptographic functions based on Deep learning. In order to learn the 256- class Deep neural network of an AES-128, we build successful convolutional neural networks to break its implementation. It has been shown by our experiments that our model achieved a success rate of 99% even with a single trace using Keras library with Tensorflow. For the sake of completeness, we investigate the correct ”key rank” according to the number of traces and as a further performance measure, we use ”recall” metric when attacking the fourth AES SBox. Our model reaches the key rank <10 with the recall metric > 0.99.