Effect of electromagnetic radiation on double-loop neural networks and its application to image encryption

Neurons often exhibit complex chaotic phenomena when they are electrically stimulated, and this property provides an important theoretical basis for the study of neural dynamics. In this paper, a novel double-loop neural network model is proposed to simulate electromagnetic radiation by introducing a simple quadratic function memristor, which acts on different neurons in the neural network, and systematically investigates the differential effects of electromagnetic radiation on the kinetic behaviour of neurons. It is found that the system exhibits rich dynamical phenomena, such as the coexistence of chaotic attractors and amplitude modulation, as the target neurons are changed. When electromagnetic radiation is applied to a specific neuron, the chaotic attractor breaks down with the change of a key parameter. The physical realizability of the theoretical model is verified by a digital circuit platform built with a microcontroller, and the experimental results are demonstrated. In addition, an efficient bit-level image encryption algorithm is designed based on the chaotic properties of this neural network model. The algorithm obfuscates the image pixel dimensions by a parity hopping diffusion operation and combines with chaotic sequences to randomize the arrangement of pixel positions, which significantly improves the security of the encryption scheme. Finally, the encryption performance of the algorithm is verified by various evaluation means.