Identification of Keys and Cryptographic Algorithms Using Genetic Algorithm and Graph Theory

Abstract

This paper describes genetic algorithms that use the Calisnki-Harabasz index as its evaluation function and graphs techniques that are both used to identify patterns in cryptograms generated by cryptographics algorithms certified by NIST (National Institute Standard Technology), namely AES, RC6, MARS, Twofish and Serpent. Evidence of patterns or "signatures" generated by the algorithms under test were detected, thus corroborating the results of other studies quoted here. The results obtained with these two techniques are compared with results reported in [4] and [5], showing superiority in the accuracy of class generation.