Memristive Coupled Neural Network Based Audio Signal Encryption

Abstract

This paper proposes a multi-layer audio encryption scheme based on the Memristive Coupled Neural Network (MCNN), S-box, and the Fibonacci Q-matrix. Initially, a pseudo-random key is generated using the MCNN system and XORed with the original audio data. Subsequently, an S-box, created using the OpenSSL Pseudo-random Number Generator (PRNG), is applied to the cipher. Finally, the Fibonacci Q-matrix is used to produce the final encrypted audio. The proposed scheme was evaluated using various metrics, including Peak Signal-to-Noise Ratio (PSNR), Number of Sample Change Rate (NSCR), correlation coefficient, and information entropy. The results demonstrate excellent performance and robust resistance to multiple types of attacks. Additionally, the scheme features a vast key space of 2<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2126</sup>, showcasing significant resistivity to brute-force attacks.