Cryptanalysis of classical ciphers with ant algorithms

Abstract

Cryptanalysis of classical ciphers has been tackled using metaheuristic optimisation techniques, including ant colony optimisation (ACO) and particle swarm optimisation (PSO). Several variants of ACO algorithms have been developed to improve their search process, but their performance on the same problem may vary significantly according to the pheromone update rule they apply. In this paper, we introduce two versions of ACO algorithms for the cryptanalysis problem of classical ciphers, namely virtual ant system (VAS) and virtual MAX-MIN ant system (VMMAS). We investigate their performance on a large set of ciphertexts obtained by six different encryption methods, including Feistel, Vigenere, simple substitution, affine, carre de Polybe, and transposition algorithms, and compare their performance against that of ant colony system (ACS) and elitist ant system (EAS). The experimental results demonstrate the overall superiority of VMMAS over the other ant algorithms. Moreover, comparison of VMMAS against a genetic algorithm (GA) establishes its advantage.