A more random and secure image encryption method: a novel true random number generator based on a W/Ta2O5/Ag memristor

Abstract

True random number generators (TRNGs) are essential for hardware security, gaining critical importance in the era of mobile computing and the Internet of Things. With advantages in integration, power consumption, and resistive switching (RS) speed, memristors show significant potential in TRNG applications. However, research on memristor-based TRNG circuits for image encryption remains limited, which is crucial for enhancing data security. Here, we present a TRNG that utilizes the random state switching of a memristor's resistance as a natural entropy source. Random bits are generated by cycling the SET and RESET voltages of the W/Ta₂O₅/Ag structure memristor, reading various SET voltages under RS. Consequently, a TRNG circuit was constructed, using the binary random numbers generated as keys to encrypt the MNIST handwritten dataset images. This approach effectively scrambled the pixel correlations of the original images, providing a high-security encryption. The correlation coefficients of the encrypted images in all directions were significantly reduced by an average of 86% compared to the original images. This work demonstrates that TRNG circuits based on W/Ta₂O₅/Ag structure memristors hold great potential in image encryption applications.