Application of Genetic Algorithms for Cruptanalysis

Author: Zurab Kochladze
Co-authors: L. Beselia, M. Khunjgurua
Keywords: Genetic algorithms, Cruptanalysis
Annotation:

Application of Genetic Algorithms for Cryptanalysis Zurab kochladze zurab.kochladze62@gmail.com Departement of Computer Science, TSU, 3 University, Tbilisi, 1086 Lali Beselia lalibeselia@hotmail.com Faculty of mathematic and computer science SSU, Tbilisi, 1086 Mariam Khunjgurua m.khunjgurua@gmail.com Departement of Computer Science, TSU, 3 University, Tbilisi, 1086 The emergency learning models, that includes also genetic algorithms, execute imitation of evolution of living organisms – the most elegant and powerful form of adaptation. the evolution is a very simple process by its nature. Following environmental impact, gradually, among the most powerful representatives through simple changes of some features and through thin out of a relatively unsuccessful representatives (SPECIMEN) the adaptation ability of population increasing. Similarly, genetic algorithms provide more accurate solution of the task through certain operations on solution candidate populations. Application of these algorithms for cryptographic analyses of modern crythographic algorithms should give good results, due to the fact, that during the analysis of cryptographic algorithms, we are dealing with a variety of possible keys, to find one, the real key. In addition, the process is iterating, the gradual specification of key based on new information. The paper describes the genetic algorithm developed by us, which attacks the famous merkli - helmanos open key system on open text basis. The attack on open text basis means, that both open, as well as encryphion texts are known, and the aim of the attack is to find the encryption key, or to create algorithm, that would give means to restore open text without key. The Algorithm, we created, work on the above described principle. In addition, the fitness function of Algorithm is very simple. There is ongoing comparison of decrypted text with open text and Hemings distance between texts is calculate. It should be noted that Algorithm is promptly recovers open text, if the cryptographic key is small. However, when key length gets closer to the real one and the selected versions are growing, the work of the algorithm becomes less effective. While, when the key length becomes the length of key utilized in real system , the algorithm can not decode text anymore. This indicates that we have developed a fitness function is not an optimal. Work in this direction will be continued in the future