Memristive multi-wing chaotic Hopfield neural network for LiDAR data security

By applying the synapse-like electrical element, memristor, complex chaotic dynamics can be generated in Hopfield neural networks. However, the multi-wing butterfly chaotic attractor generated by the memristive Hopfield neural network remains undiscovered. In this paper, we introduce a novel chaotic multi-wing butterfly generation method within the Hopfield neural network (HNN). Our proposed approach incorporates a piecewise linear memristor to establish coupling between two neurons in a three-neuronal HNN. This design allows straightforward control over the number of butterfly wings by adjusting the memristor parameters. We conduct a comprehensive numerical analysis of the chaotic butterfly dynamics using phase portraits, Lyapunov exponent spectra, state variable bifurcation diagrams, and bi-parameter dynamical maps. Furthermore, the proposed model is implemented based on the digital circuit FPGA platform and its correctness is verified through experiments. Moreover, we leverage the developed chaotic multi-wing butterfly to construct a secure LiDAR point cloud system. The system employs a chaotic permutation and diffusion algorithm based on the proposed multi-wing butterfly. Security performance and time efficiency are evaluated using multiple numerical methods, and the results demonstrate the effectiveness of the proposed LiDAR data secure system.