@Article{cmc.2025.066331,
AUTHOR = {Gang Liu, Guosheng Xu, Chenyu Wang, Guoai Xu},
TITLE = {Hyper-Chaos and CNN-Based Image Encryption Scheme for Wireless Communication Transmission},
JOURNAL = {Computers, Materials \& Continua},
VOLUME = {84},
YEAR = {2025},
NUMBER = {3},
PAGES = {4851--4868},
URL = {http://www.techscience.com/cmc/v84n3/63188},
ISSN = {1546-2226},
ABSTRACT = {In wireless communication transmission, image encryption plays a key role in protecting data privacy against unauthorized access. However, conventional encryption methods often face challenges in key space security, particularly when relying on chaotic sequences, which may exhibit vulnerabilities to brute-force and predictability-based attacks. To address the limitations, this paper presents a robust and efficient encryption scheme that combines iterative hyper-chaotic systems and Convolutional Neural Networks (CNNs). Firstly, a novel two-dimensional iterative hyper-chaotic system is proposed because of its complex dynamic behavior and expanded parameter space, which can enhance the key space complexity and randomness, ensuring resistance against cryptanalysis. Secondly, an innovative CNN architecture is introduced for generating the key stream for the cryptographic system. CNN architecture exhibits excellent nonlinearity and can further optimize the key generation process. To rigorously evaluate the encryption performance, extensive simulation analyses were conducted, including visualization, statistical histogram, information entropy, correlation, differential attack, and resistance. The method has shown a high NPCR (Number of Pixel Change Rate) of 99.642% and a UACI (Unified Average Changing Intensity) value of 33.465%, exhibiting powerful resistance to differential attacks. A series of comprehensive experimental tests have illustrated that the proposed scheme exhibits superior distribution characteristics, which underscores the robustness and efficacy of the image encryption, and helps for communication security.},
DOI = {10.32604/cmc.2025.066331}
}