Development of a novel chaos-based approach for image encryption in the internet of things

The rapid growth of visual data within the Internet of Things (IoT) ecosystem necessitates the development of encryption solutions that are both secure and lightweight for resource-constrained edge devices. Aiming to establish an optimum balance between high security and low computational cost, this thesis study develops the "Knit Scrambling" (KS) framework, inspired by textile knitting structures, which is proposed by me for the first time in the literature. Designed as a specific implementation of this original framework, the "Triple-Check Pattern" (TCP) provides a deterministic and reversible permutation layer that effectively eliminates spatial redundancy. This newly developed hybrid image encryption architecture integrates the KS-based permutation structure with a chaotic diffusion layer based on a one-dimensional (1D) Logistic map, whose parameters are dynamically derived via SHA-256 using a secret key and a nonce. Furthermore, to maximize the complexity in the diffusion process, a Dual-Stream XOR-Mod-Additive (DXA) mechanism is employed. Comprehensive experimental analyses demonstrate that the algorithm achieves information entropy values quite close to the theoretical maximum (7.9994) and exhibits superior performance in terms of differential security criteria with average NPCR and UACI values of 99.61% and 33.46%, respectively. Exhibiting high robustness against noise and data loss attacks, this method presents a secure, fast, and practical solution for real-time image encryption applications in resource-constrained IoT environments, thanks to its linear time complexity of O(8MN).