Switchable memristor-based Hindmarsh-Rose neuron under electromagnetic radiation

Abstract

Memristors are prevalently used to simulate biological neuronal synapses due to their unique memductance plasticity and memory effects. A new switchable memristor, which can be configured as a nonvolatile discrete memristor, a nonvolatile continuum memristor or a volatile memristor by adjusting its internal parameter, is proposed to mimic the autapse of the Hindmarsh-Rose (HR) neuron. In the meantime, a flux-controlled memristor is introduced to simulate the effect of external electromagnetic radiation on the HR neuron, thus, an improved 4D HR neuron model without equilibrium points is developed in this study. The hidden firing activities related to the strength of autapse and the electromagnetic radiation intensity are revealed through phase diagrams, time series, bifurcation diagrams, Lyapunov exponent spectrums, and two-parameter dynamical maps. More interestingly, it is found that the memory attributes of memristive autapse play an important role in the firing activities of the neuron, which can induce the mutual transition among periodic spiking with different frequencies and chaotic firing. Additionally, the transition between periodic and chaotic firing induced by the initial value of the switchable memristor is also discovered when it is configured as three different types of memristors. Finally, a neuron circuit is designed with the current-mode devices to improve accuracy and reduce power consumption. The Multisim simulation results are provided to validate the correctness of the neuron model and the effectiveness of numerical analysis.