Cryptanalysis of the RC4 Stream Cipher using Evolutionary Computation Methods




Ferriman, Benjamin

Journal Title

Journal ISSN

Volume Title


University of Guelph


In order to protect the actions and information on the Internet, secure protocols are implemented to facilitate secure communication between clients and services. The most common protocol used is the the Secure Socket Layer protocol that wraps around HTTP to obfuscate network data from prying eyes. RC4 is the most widely used encryption algorithm over this protocol and has held strong to many attempts of cryptanalysis over its twenty-six year existence. The following Thesis will outline the RC4 algorithm and present a new approach for cryptanalysis of the cipher by attacking RC4s state register which is required for all the algorithms communication. We intend to improve the exhaustive search process for a permutation using genetic and particle swarm optimization algorithms from the evolutionary algorithm family. It will be shown that thus far, the genetic algorithm has been the most successful of the two methods and on several occasions, has been able to recover 25% and on average, 10% of the keystream with a candidate solution. The attack can theoretically on average recover the state register in 2^121.5 generations.



Cryptanalysis, RC4, Evolutionary Algorithms, Genetic Algorithms, Particle Swarm Optimization