A Chaos-RC4 Encryption Algorithm based on Memristive Neural Network

Abstract

This paper proposes an algorithm based on memristive neural network to realize Rivest Cipher 4 (RC4) encryption, which effectively releases the correlation between the input key and the output key stream of the RC4 algorithm. At the same time, this design improves both Key Schedule Algorithm (KSA) and Pseudo Random Generation Algorithm (PRGA) of RC4 stream cipher. In the KSA process, a neural network with chaos is constructed by utilizing the non-linearity characteristics of the memristor, and the generated chaotic sequence having good randomness performance, is used as the input key of the RC4 algorithm to realize “one time one secret” dynamic encryption. In the PRGA process, another sequence generated randomly by memristive neural network is used for a dynamic chaotic permutation on the data table of the S-box. Experiment results demonstrate the effectiveness of the algorithm, and show it has larger key space, higher security and stronger robustness than the traditional algorithm.