Date on Master's Thesis/Doctoral Dissertation
Computer Engineering and Computer Science
Cryptography; Telecommunication systems--Security measures
In today's world of private information and mass communication, there is an ever increasing need for new methods of maintaining and protecting privacy and integrity of information. This thesis attempts to combine the chaotic world of cellular automata and the paranoid world of cryptography to enhance the S-box of many Substitution Permutation Network (SPN) ciphers, specifically Rijndael/AES. The success of this enhancement is measured in terms of security and performance. The results show that it is possible to use Cellular Automata (CA) to enhance the security of an 8-bit S-box by further randomizing the structure. This secure use of CA to scramble the S-box, removes the "9-term algebraic expression"   that typical Galois generated S-boxes share. This cryptosystem securely uses a Margolis class, partitioned block, uniform gas, cellular automata to create unique S-boxes for each block of data to be processed. The system improves the base Rijndael algorithm in the following ways. First, it utilizes a new S-box for each block of data. This effectively limits the amount of data that can be gathered for statistical analysis to the blocksize being used. Secondly, the S-boxes are not stored in the compiled binary, which protects against an "S-box Blanking"  attack. Thirdly, the algebraic expression hidden within each galois generated S-box is destroyed after one CA generation, which also modifies key expansion results. Finally, the thesis succeeds in combining Cellular Automata and Cryptography securely, though it is not the most efficient solution to dynamic S-boxes.
Luckett, William Matthew, "Cellular automata for dynamic S-boxes in cryptography." (2007). Electronic Theses and Dissertations. Paper 863.