On the attack of misaligned traces by power analysis methods

Random clock armed cryptosystems inject an intended misalignment in captured power traces while the system is running encryption or decryption. This is a powerful counter-measure against Correlation Power Analysis (CPA) in practice. In order to remove such a countermeasure, a trace preprocessing is necessary before mounting power attacks. In this paper, we consider both the established attacking method CPA and the recently proposed approach Power Amount Analysis (PAA) as well as novel methodologies for removing this countermeasure in the power traces in order to improve the attacks. Therefore, we introduce first a new feature to the PAA attack, which can effectively neutralize the clock frequency effects resulting from a random clock featured cryptosystem running at higher clock frequency. Then we suggest a novel peak position based trace alignment. At the end, a complete framework for preprocessing misaligned power traces is proposed. The application example shows that the PAA attack can effectively neutralize the clock frequency effects and the complete framework is able to handle randomly misaligned power traces very well thus eventually improving the performance of both the CPA and the PAA attacks.