As an unsolved issue for embedded crypto solutions, side channel attacks are challenging the security of the Internet of things. Due to the advancement of chip technology, the nature of side channel leakage becomes hard to characterize with a fixed leakage model. In this work, a new non-linear collision attack is proposed in the pursuit of the side channel distinguishers with minimal assumption of leakage behavior. The attack relies on a weaker assumption than classical DPA: it does not require a specific leakage model. The mechanism of collision generation enables independent recovery of partial keys so that for the first time the collision attack can be fairly compared with other standard side channel distinguishers. The efficiency of this attack has been verified by experiments on an unprotected microcontroller implementation of AES. Its immunity to modeling errors is confirmed through simulation of a broad range of leakage functions.
CITATION STYLE
Ye, X., Chen, C., & Eisenbarth, T. (2014). Non-Linear collision analysis. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 8651, 198–214. https://doi.org/10.1007/978-3-319-13066-8_13
Mendeley helps you to discover research relevant for your work.