Memristor-Based Neural Network Accelerators for Space Applications: Enhancing Performance with Temporal Averaging and SIRENs

2025-09-02

AI Summary

To address severe neural network performance degradation caused by resistive random-access memory (RRAM) non-idealities—namely device variability, conductance drift, and hard failures—in spaceborne AI applications, this work proposes a robust and energy-efficient RRAM-based hardware accelerator. The method innovatively integrates inter-layer temporal averaging, bit-sliced weight encoding, and the SIREN (Sinusoidal Representation Network) activation function: temporal averaging mitigates conductance drift and variability; bit-slicing enhances fault tolerance; and SIREN improves function approximation fidelity under low-precision constraints. Evaluated on onboard navigation control and asteroid geodesy tasks, the accelerator reduces inference errors from 0.07 to 0.01 and from 0.3 to 0.007, respectively—approaching state-of-the-art accuracy. This study constitutes the first systematic validation of RRAM’s feasibility and practicality for radiation-hardened, high-energy-efficiency intelligent computing in space environments.

Technology Category

• Machine Learning: Hardware-aware ML
• Intelligent Robots: Learning & Optimization for ROB
• Computer Vision: Remote Sensing / Geospatial AI

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Memristors are an emerging technology that enables artificial intelligence (AI) accelerators with high energy efficiency and radiation robustness -- properties that are vital for the deployment of AI on-board spacecraft. However, space applications require reliable and precise computations, while memristive devices suffer from non-idealities, such as device variability, conductance drifts, and device faults. Thus, porting neural networks (NNs) to memristive devices often faces the challenge of severe performance degradation. In this work, we show in simulations that memristor-based NNs achieve competitive performance levels on on-board tasks, such as navigation & control and geodesy of asteroids. Through bit-slicing, temporal averaging of NN layers, and periodic activation functions, we improve initial results from around $0.07$ to $0.01$ and $0.3$ to $0.007$ for both tasks using RRAM devices, coming close to state-of-the-art levels ($0.003-0.005$ and $0.003$, respectively). Our results demonstrate the potential of memristors for on-board space applications, and we are convinced that future technology and NN improvements will further close the performance gap to fully unlock the benefits of memristors.

Problem

• Addressing memristor non-idealities in space AI accelerators
• Improving neural network reliability for spacecraft navigation tasks
• Reducing performance degradation in radiation-hardened computing systems

Innovation

• Using bit-slicing to enhance memristor neural network reliability
• Applying temporal averaging to neural network layers
• Employing periodic activation functions like SIRENs