Abstract: |
The conventional world of cryptology does not consider the possibility of heuristic cryptanalysis being effective against modern standard ciphers like Rijndael, the new Advanced Encryption Standard (AES). However, this paper presents evidence that evolutionary computation should be expected to be able to develop approximations (in an attacker's model) that seriously reduce the effective security of any targeted cipher. This clash of security paradigms is analysed, prospects for the success of EC in cryptanalysis are discussed and solutions to the attendant problems are offered. |