Threshold-Switching Memristor Neurons With Spike-Frequency Adaptation

Le résumé fourni par la source

Threshold-switching memristors (TSMs) exhibit volatile turn-on at a threshold voltage and self-reset when the device voltage drops below the hold voltage, naturally providing the trigger-and-reset functions required for spiking neurons. This letter presents a compact SPICE-compatible phenomenological macro-model that emulates the threshold-switching window defined byVthandVholdand is directly compatible with CMOS co-simulation. Using this model, a minimal TSM-based leaky integrate-and-fire (LIF) neuron is demonstrated, showing monotonic tuning of firing frequency with input voltage and control resistance. A low-overhead adaptive module that converts spike activity into a slowly varying control voltage is further introduced to realize an adaptive LIF (ALIF) neuron with spike-frequency adaptation and recovery. Energy-per-spike scaling with membrane capacitance is investigated forC1ranging from 0.25 to 8 nF underVin=3.25, 3.50, and 3.75 V. Robustness against key parameter variations is further evaluated using Monte Carlo simulations to quantify the efficiency and sensitivity of the proposed circuit.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.