CNN-Based Automated Parameter Extraction Framework for Modeling Memristive Devices

TL;DR

This paper introduces a CNN-based automated method for extracting parameters of RRAM models directly from device I-V data, significantly reducing manual effort and improving adaptability for modeling memristive devices.

Contribution

The work presents a novel CNN-driven framework combined with heuristic optimization for rapid, automated parameter extraction of RRAM models from experimental I-V characteristics.

Findings

Achieves low error in parameter estimation across diverse RRAM devices.

Faster and more reliable than manual tuning methods.

Effective in modeling key NVM metrics such as set/reset voltages and hysteresis.

Abstract

Resistive random access memory (RRAM) is a promising candidate for next-generation nonvolatile memory (NVM) and in-memory computing applications. Compact models are essential for analyzing the circuit and system-level performance of experimental RRAM devices. However, most existing RRAM compact models rely on multiple fitting parameters to reproduce the device I-V characteristics, and in most cases, as the parameters are not directly related to measurable quantities, their extraction requires extensive manual tuning, making the process time-consuming and limiting adaptability across different devices. This work presents an automated framework for extracting the fitting parameters of the widely used Stanford RRAM model directly from the device I-V characteristics. The framework employs a convolutional neural network (CNN) trained on a synthetic dataset to generate initial parameter estimates, which are then refined through three heuristic optimization blocks that minimize errors via adaptive binary search in the parameter space. We evaluated the framework using four key NVM metrics: set voltage, reset voltage, hysteresis loop area, and low resistance state (LRS) slope. Benchmarking against RRAM device characteristics derived from previously reported Stanford model fits, other analytical models, and experimental data shows that the framework achieves low error across diverse device characteristics, offering a fast, reliable, and robust solution for RRAM modeling.