Exploring the Active Realization of Analog Multi-Level Memristors for Neuromorphic Applications

Abstract

This article presents a novel method for identifying analog memristor functions that enable multiple resistance levels with tailored switching parameters. These multi-level memristors hold potential applications in multi-bit memory systems, chaotic oscillators, and multi-level logic design. The proposed approach is based on constructing a Parameter vs. State Map (PSM), which can be designed to contain a specific number of stationary points along its curvature. The regions surrounding these stationary points serve as discrete resistance levels. Using this method, analog multi-level memristors can be realized with predefined switching characteristics, such as resistance margins and switching thresholds. Unlike conventional memristor emulators, which typically rely on monotonic memristance/conductance functions, the proposed approach enables precise control over resistance states. As a result, memristors developed through this method can replace complex,  discrete-valued memristors in various applications. To validate the proposed concept, an OTA (Operational Transconductance Amplifier)-based emulator circuit has been designed and verified through PSPICE simulations, demonstrating the feasibility of the approach.