We use large but sparse correlation and transition-difference-probability submatrices to find the best linear and differential approximations respectively on PRESENT-like ciphers. This outperforms the branch and bound algorithm when the number of low-weight differential and linear characteristics grows exponentially which is the case in PRESENT-like ciphers. We found linear distinguishers on 23 rounds of the SPONGENT permutation. We also found better linear approximations on PRESENT using trails covering at most 4 active Sboxes which give us 24-round statistical saturation distinguishers which could be used to break 26 rounds of PRESENT.
Lecture Notes in Computer Science: 15th International Conference, Seoul, Korea, November 28-30, 2012, Revised Selected Papers, 2012, p. 368-382
Main Research Area:
Lecture Notes in Computer Science
15th Annual International Conference on Information Security and Cryptology (ICISC 2012)Information Security and Cryptology