Can chaotic cryptographic systems be simple and secure?

In the past decade many chaos‑based cryptosystems have appeared, yet few have reached real‑world use. The main obstacle is that most rely on overly complex designs that hide security and hinder analysis. This paper shows the opposite is possible: chaotic ciphers can be built with well‑known design paradigms without losing speed or strength. First, the authors present a minimal image‑encryption scheme based on a substitution‑permutation network (SPN). Low‑dimensional chaotic maps remove statistical bias, and swapping the map has negligible effect. Next, they introduce a chaotic Feistel block cipher (CFBC) within a generalized Feistel network, allowing standard security assessment such as differential cryptanalysis. To test it, they applied CFBC in CBC mode to image encryption and compared it with existing chaotic image ciphers. Results indicate CFBC achieves statistical performance on par with state‑of‑the‑art image ciphers. In short, the study suggests chaotic cryptosystems need not be tangled constructions; using established principles makes them more analyzable and robust.

The study fits within chaos‑based cryptography, which exploits sensitivity to initial conditions and unpredictability of nonlinear dynamical systems. Low‑dimensional chaotic maps used (e.g., logistic or Henon map) were examined to ensure the absence of statistical correlations that could leak plaintext information. The proposed SPN architecture employs a single S‑box derived from the chaotic transformation, followed by a bit permutation that guarantees full diffusion in few rounds. For the CFBC, the round function combines the chaotic map with XOR and rotation operations, preserving the classic Feistel structure that allows inversion with the same key. Security is assessed via differential analysis: probabilities of input‑output differences (Δ‑in, Δ‑out) were computed for all possible inputs, showing the maximum probability remains close to 2⁻ⁿ, where n is the block size. Additionally, entropy tests, pixel‑pixel correlation, and color value distribution analyses were performed, all exceeding thresholds required for image ciphers. Using CBC mode adds another randomization layer, as each image block is XOR‑ed with the previous ciphertext block, mitigating replay attacks. Finally, comparison with ciphers based on more complex chaotic maps (e.g., Lorenz map) shows that added complexity yields no significant security gains but raises computational cost. This supports the paper's central claim: design simplicity, coupled with established cryptographic principles, is key to making chaotic systems truly analyzable and practical.