Representation-, Leakage- and Cipher-Dependencies in Algebric Sidde-Channel Attacks

By combining the extraction of side-channel information leakages with classical cryptanalysis techniques, the recently introduced Algebraic Side- Channel Attacks trade a part of the data complexity in standard DPA attacks for more computations. But predicting the success rate of such attacks is made harder because of the numerous parameters that come into play when solving large algebraic systems of equations. In this paper, we study the impact of three of these parameters empirically, along with the metrics needed to quantify them. First, we analyze the efficiency of different representations of the side-channel information as low degree boolean equations. Second, we investigate the impact of different types of information leakages on the attack resolution times. Third, we discuss how these conclusions depend on the target ciphers. From simulated experiments performed in various contexts, we finally provide some more general intuitions for the security of leaking devices.