Deep Learning Power Analysis for Midori128 with Hiding Countermeasure

Midori128 is a block cipher suitable for loT devices, which enables not only to operate encryption processing with low power consumption but also to be secured computationally. However, the vulnerability against the power analysis attack, which is a malicious attack targeted hardware-implemented cryptography, is pointed out. Therefore, Midori128 oriented countermeasure has been reported. On the other hand, recent power analysis attacks have achieved high accuracy by using deep learning. Furthermore, this analysis has proven to be effective against countermeasure cryptographic circuits. For this background, it is important to evaluate the tamper resistance of Midori128 with countermeasures to deep learning power analysis attacks. Therefore, this study proposes a deep learning power analysis attack for Midori128 with hiding countermeasure. Also, this study evaluates the tamper resistance of Midori128 with countermeasures using an evaluation board. The contribution of this study is to reveal the vulnerability of the countermeasure Midori128.