On Distinct Known Plaintext Attacks
Abstract
Among statistical attacks, we often make a distinction between attacks in the linear context, for which the knowledge of plaintexts and corresponding ciphertexts is enough to perform the attack, and attacks in the differential context, for which the plaintexts are chosen. Such attacks are usually referred as known-or chosen-plaintext attacks. It is commonly believed that attacks in the known-plaintext model are more practical than attacks in the chosen-plaintext model. Nevertheless , it is usual in the literature, to only compare the data, time and memory complexity of these attacks without considering the model. In this paper, we reconsider some known-plaintext attacks, by considering them in the distinct-known-plaintext model. We explain and develop the statistical model for the multiple zero-correlation linear cryptanalysis, multidimensional linear cryptanalysis, as well as for the key-difference-invariant-bias related-key attack introduced at ASIACRYPT 2013. Based on these models validated by experiments, we improve attacks on some ciphers.
Origin : Files produced by the author(s)
Loading...