Low-power switching of memristors exhibiting fractional-order dynamics

2025-07-24

AI Summary

Fractional-order memristors face challenges in achieving low-power switching due to inherent energy dissipation during state transitions. Method: This work establishes a state-evolution model based on Caputo-type fractional-order differential equations and systematically analyzes the Joule loss mechanism under current-pulse excitation. It identifies that the optimal switching strategy hinges on the matching between the fractional derivative order α and the exponent β in the state equation. Accordingly, an adaptive dual-mode pulse control scheme is proposed: wide, low-amplitude pulses for α/β ≈ 1; narrow, high-amplitude, zero-current pulses when α/β deviates significantly from 1—extended further into a multi-pulse cooperative control framework. Results: Experiments and simulations demonstrate 37–62% energy reduction compared to integer-order baseline strategies. This work provides a novel theoretical paradigm and practical design guidelines for energy-efficient fractional-order memristive devices in neuromorphic computing.

Technology Category

• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

In this conference contribution, we present some initial results on switching memristive devices exhibiting fractional-order behavior using current pulses. In our model, it is assumed that the evolution of a state variable follows a fractional-order differential equation involving a Caputo-type derivative. A study of Joule losses demonstrates that the best switching strategy minimizing these losses depends on the fractional derivative's order and the power exponent in the equation of motion. It is found that when the order of the fractional derivative exceeds half of the power exponent, the best approach is to employ a wide pulse. Conversely, when this condition is not met, Joule losses are minimized by applying a zero current followed by a narrow current pulse of the highest allowable amplitude. These findings are explored further in the context of multi-pulse control. Our research lays the foundation for the advancement of the next generation of energy-efficient neuromorphic computing architectures that more closely mimic their biological counterparts.

Problem

• Switching memristors with fractional-order dynamics efficiently
• Minimizing Joule losses via optimal pulse strategies
• Advancing energy-efficient neuromorphic computing architectures

Innovation

• Switching memristors with current pulses
• Using fractional-order differential equations
• Optimizing Joule losses via pulse strategy