Higher-order masking : from block cipher design to secure and efficient implementations
(2018)
Files
journault_very_high_order_masking.pdf
Open Access - Adobe PDF
- 991.6 KB
Details
- Authors
- Supervisors
- Standaert, François-Xavier
- Abstract
- Introduced in the late 90's, side-channel attacks are an important threat for cryptographic implementations. They exploit observations of physical features leaked by a device while it computes cryptographic algorithms. These leakages can provide a source of information about sensitive variables such as keys. A common countermeasure against such attacks is masking. The main idea of masking is to split any sensitive variable manipulated by the leaking device in shares and to perform computations on these shares only. As a consequence an adversary must combine the leakages several shares to recover any sensitive information, which is a significantly harder task. This security guarantee come at the cost of important performance overheads, usually dominated by the cost of the non-linear operations (e.g., multiplications). In this thesis, we explore different solutions to improve the efficiency and security of masked implementations. Firstly, we introduce a new family of block cipher dedicated to masking, which improves both the performances and the security of previously published designs, essentially by reducing the number of multiplications needed to implement the cipher. Secondly, we implement and optimize different state-of-the-art masking schemes in order to compare their respective performances in different implementation contexts. Thirdly, we describe a very high order implementations (with 32 shares) of different block ciphers, which leads to reasonable performances. Finally, we analyze the side-channel security of these very high order implementations by introducing a multi-model approach which allows stating quantitative security bounds.
- Affiliations
Citations
Journault, A. (2018). Higher-order masking : from block cipher design to secure and efficient implementations. https://hdl.handle.net/2078.5/57416