An Evaluation and Enhancement of Seredynski-Bouvry CA-based Encryption Scheme

TL;DR

This paper evaluates a cellular automata-based encryption scheme, identifies its weaknesses in statistical tests, and proposes an improved version that better satisfies cryptographic criteria with fewer iterations.

Contribution

It critically assesses the original Seredynski-Bouvry CA encryption scheme and introduces an enhanced version with improved security and efficiency by redefining neighborhood rules.

Findings

Original scheme fails some NIST tests with small neighborhood radius.

Careful CA-rule selection is crucial for security.

Improved scheme meets strict avalanche and NIST criteria faster.

Abstract

In this paper, we study a block cipher based on cellular automata, proposed by Seredynski and Bouvry in \cite{semabo04} against \emph{plain-text avalanche criteria} and \emph{randomness} tests. Our experiments shows that Seredynski-Bouvry encryption scheme does not pass some NIST statistical tests by neighborhood radius less than three. It also showed that if the CA rule is selected carelessly, it weaken the security of scheme. Therefor, the selection of CA-rule as part of key can not be left to the user. Hence, cryptographic properties such as balancedness and non-linearity should be considered in the selection of CA-rules. This approach is more compliant with Kerckhoffs principle. So security should depend just on security of final data. We also improve Seredynski-Bouvry encryption scheme to satisfy strict avalanche criteria and NIST statistical test suite in about half number of iterations comparing to original scheme. This improvement is achieved by change in the definition of neighborhood.