A Realistic Simulation Framework for Analog/Digital Neuromorphic Architectures

2024-09-23

arXiv.org

AI Summary

Developing mixed-signal neuromorphic chips for extreme-edge applications faces a critical challenge: hardware-software behavioral mismatch—arising from device mismatch, analog noise, and insufficient simulation fidelity—leading to low reliability in spiking neural network (SNN) deployment. To address this, we propose ARCANA, a high-fidelity, circuit-level SNN simulation framework tailored for mixed-signal neuromorphic hardware. ARCANA uniquely integrates statistical device mismatch modeling, precise dynamical simulation of spiking neurons and synapses, automatic-differentiation-based parameter optimization, and CUDA-accelerated GPU computation. By tightly coupling algorithmic design with physical hardware constraints, ARCANA bridges the gap between software abstraction and silicon reality: it faithfully reproduces measured chip responses and achieves <3.2% prediction error for software-trained SNN behavior prior to hardware deployment. This significantly enhances both development efficiency and deployment reliability of edge neuromorphic systems.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Hardware-aware ML
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Web data generation and simulation
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

Developing dedicated mixed-signal neuromorphic computing systems optimized for real-time sensory-processing in extreme edge-computing applications requires time-consuming design, fabrication, and deployment of full-custom neuromorphic processors. To ensure that initial prototyping efforts, exploring the properties of different network architectures and parameter settings, lead to realistic results, it is important to use simulation frameworks that match as best as possible the properties of the final hardware. This is particularly challenging for neuromorphic hardware platforms made using mixed-signal analog/digital circuits, due to the variability and noise sensitivity of their components. In this paper, we address this challenge by developing a software spiking neural network simulator explicitly designed to account for the properties of mixed-signal neuromorphic circuits, including device mismatch variability. The simulator, called ARCANA (A Realistic Simulation Framework for Analog/Digital Neuromorphic Architectures), is designed to reproduce the dynamics of mixed-signal synapse and neuron electronic circuits with autogradient differentiation for parameter optimization and GPU acceleration. We demonstrate the effectiveness of this approach by matching software simulation results with measurements made from an existing neuromorphic processor. We show how the results obtained provide a reliable estimate of the behavior of the spiking neural network trained in software, once deployed in hardware. This framework enables the development and innovation of new learning rules and processing architectures in neuromorphic embedded systems.

Problem

• Simulating mixed-signal neuromorphic hardware accurately
• Addressing variability and noise in analog/digital circuits
• Optimizing neuromorphic architectures for edge-computing applications

Innovation

• Realistic simulation for mixed-signal neuromorphic circuits
• Autogradient differentiation for parameter optimization
• GPU acceleration for efficient neural network simulation