Chaos-based diffusion for highly autocorrelated data in encryption algorithms

Abstract

In a single substitution box, the same regions (pixels) of an image are encrypted to one unique symbol. To reduce this type of autocorrelation in data, chaos has been extensively applied over the last decade. By using chaotic maps, a single substitution box has been replaced with multiple substitution boxes for the encryption of autocorrelated data. The technique of multiple substitution boxes is becoming very popular among cryptographic algorithm designers to overcome the drawbacks of a single substitution box. In this paper, however, we found that replacing a single substitution box with multiple substitution boxes cannot provide a general solution for highly autocorrelated data. To address this issue, we propose a novel technique by adding chaotic diffusion to the existing substitution process. Extensive security analyses show that the proposed algorithm achieves both higher-level security and fast encryption time when compared with existing algorithms.