Grid Multi-Butterfly Memristive Neural Network With Three Memristive Systems: Modeling, Dynamic Analysis, and Application in Police IoT

Abstract

Nowadays, the Internet of Things (IoT) technology has been widely applied in the police security system. However, with more and more image data that concerns crime scenes being transmitted through the police IoT, there are some new security and privacy issues. Therefore, how to design a safe and efficient secret image sharing solution suitable for police IoT has become a very urgent task. In this work, a grid multibutterfly memristive Hopfield neural network (HNN) with three memristive systems is constructed and its complex dynamics are deeply analyzed. Among them, the first memristive system is modeled by emulating a self-connection synapse, the second memristive system is modeled by coupling two neurons, and the third memristive system is modeled by describing external electromagnetic radiation. Dynamic analyses show that the proposed memristive HNN can not only generate two kinds of 1-directional (1-D) multibutterfly chaotic attractors but also produce complex grid (2-D) multibutterfly chaotic attractors. More importantly, by switching the initial states of the second and third memristive systems, the grid multibutterfly memristive HNN exhibits initial-boosted plane coexisting multibutterfly attractors. Moreover, the number of butterflies contained in a multibutterfly attractor and coexisting attractors can be easily adjusted by changing memristive parameters. Based on these complex dynamics, an image security solution is designed to show the application of the newly constructed grid multibutterfly memristive HNN to police IoT security. Security performances indicate the designed scheme can resist various attacks and has high robustness. Finally, the test results are further demonstrated through Raspberry Pi-based hardware experiments.