Memristor Device Modeling & Stochasticity
Variability-aware and probabilistic models of resistive switching devices, including stochastic resonance and Markov jump process frameworks.
Resistive switching (memristive) devices exhibit significant cycle-to-cycle and device-to-device variability rooted in the stochastic nature of conductive filament formation and dissolution. Rather than treating this variability purely as a reliability concern, this research line investigates how stochastic phenomena can be characterized, modeled, and ultimately harnessed for computing purposes.
A central contribution is the development of Markov jump process models for resistive switching dynamics, which capture the probabilistic transitions between high- and low-resistance states in a physically grounded yet computationally tractable framework. These models enable the study of stochastic resonance in memristive circuits — the counterintuitive phenomenon whereby an optimal level of noise enhances weak signal detection — and its potential exploitation in bio-inspired sensing and computing architectures.
Variability-aware compact models developed within this research are directly applicable to circuit simulation of large-scale crossbar arrays, supporting the design of robust memristive systems. This work bridges device physics, stochastic processes, and circuit engineering, providing tools for co-optimization of device characteristics and system-level computing performance at the edge.