Article Preview
Top1. Introduction
With the rapid developments of computer network and multimedia processing, information security has increasingly become an important research field (Qiu & Memmi, 2015; Schmitz et al., 2014). The traditional encryption methods, e.g, Advanced Encryption Standard (AES) and Data Encryption standard (DES), are suitable only for text information, and not effective for encrypting images and video(Zhou & Wang, 2020).
To solve the issues, chaotic systems were introduced to design image encryption algorithms, as its outstanding performances of sensitivity to initial values, ergodicity, and so on. Matthews firstly proposed an encryption method based on the chaotic system(Mao et al., 2004). Many studies (Özkaynak, 2018; Xiao et al., 2009; Zhang, 2020) have shown that the encryption methods based the chaotic systems have high security, low complexity. However, the encryption algorithms still exist some common shortcomings, such as the fragility of resisting chosen- differential attack, short periodic, and long encryption time which is harmful to secure transmission.
Recent years, chaotic systems couple with other algorithms to build hybrid encryption systems. Due to the excellent dynamic performance of the DNA sequence operations, many hybrid encryption methods (Chai, Gan, Yang et al, 2017; Chai et al., 2019; Farah et al., 2020; Torkaman et al., 2012; Wen et al., 2020) based on chaos and DNA are proposed. Chai et al (Chai et al., 2019) proposed a novel encryption scheme, which uses 2D Logistic-adjusted-sine map and DNA sequence operations. Based on the standard encryption algorithm of AES, many hybrid encryption methods (Çavuşoğlu et al., 2018; Grari et al., 2019; Manangi et al., 2010; Musa et al., 2003; Vimalathithan & Valarmathi, 2012) use different chaotic systems to improve dynamic performance. Çavuşoğlu et al(Çavuşoğlu et al., 2018) used Zhongtang chaotic system to improve the dynamic performance of the S-AES.
The above algorithms have excellent security performance, however, the algorithms do not have fast encryption speed. Therefore, a variety of encryption algorithms based on parallel computing are proposed(Luo et al., 2018; Mirzaei et al., 2012). Mirzaei et al(Mirzaei et al., 2012) proposed an parallel encryption algorithm, which uses two chaotic systems to improve security. To accelerate the encryption speed, GPU devices are introduced in encryption algorithms. As GPU has many computing cores, it is more suitable to compute the image process compared to CPU. In addition, many performance analysis results (Luo et al., 2018) show that GPU performance is better for the CPU, in terms of speed-up ratio, complexity, computational efficiency, and redundancy. These algorithms mainly focus on encryption speed, however, the security of the algorithms is not outstanding
According to the above analysis, an excellent image encryption algorithm should have high security to resist attacks and low encryption time. In this study, we proposed a hybrid image encryption algorithm based on a chaotic system and S-AES. The algorithm is implemented on the CPU-GPU platform to accelerate encryption speed. The main contributions of this study can be summarized as:
- •
A dynamic permutation method based on the chaotic system and SHA-256 is proposed;
- •
The proposed encryption method uses the CPU-GPU platform to accelerate the encryption process using the Matlab-CUDA technology;
- •
Results and analysis indicate the proposed algorithm has high security and low encryption time.