Meaningful number of people believe that in AES is a secret backdoor (e.g. http://forums.hak5.org/index.php?showtopic=13355, http://www.infoworld.com/d/security/aes-proved-vulnerable-microsoft-researchers-170218). It may be supposed that the backdoor can be generated by the KeyExpansion routine and by the process of executing many times cryptographic transformations. As it is known, the element of the AES algorithm, KeyExpansion routine, processes an entered by the user secret key of 4*Nk bytes, Nk equal either 4, 6 or 8 and generates an expanded key w containing 16*(Nr+1) bytes, Nr denoting the number of so-called rounds equal 10 if Nk = 4, equal 12 in case Nk=6 and equal 14 in case Nk is 8. Such intentional manipulation on the secret key can breed the elimination of many of its bytes, and the eliminated bytes of the secret key may be replaced by the values known to initiates. In the presented application this hypothetic thread is entirely eliminated: the KeyExpansion routine is not used, the value of Nr is equal to 1, thus, the expanded key w contains 32 random bytes, dependent on the user password only. This way the presented implementation can be considered as a secure AES with 256 bit key, which is many times more faster than the conventional algorithm.
