Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design

To meet the high-reliability requirements of real-time on-orbit tasks, this paper proposes a fault-tolerant reinforcement design method for spaceborne intelligent processing algorithms based on convolutional neural networks (CNNs). This method is built on a CNN accelerator using Field-Programmable G...

Full description

Saved in:
Bibliographic Details
Main Authors: Yingzhao Shao, Junyi Wang, Xiaodong Han, Yunsong Li, Yaolin Li, Zhanpeng Tao
Format: Article
Language:English
Published: MDPI AG 2024-12-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/1/69
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850147081538240512
author Yingzhao Shao
Junyi Wang
Xiaodong Han
Yunsong Li
Yaolin Li
Zhanpeng Tao
author_facet Yingzhao Shao
Junyi Wang
Xiaodong Han
Yunsong Li
Yaolin Li
Zhanpeng Tao
author_sort Yingzhao Shao
collection DOAJ
description To meet the high-reliability requirements of real-time on-orbit tasks, this paper proposes a fault-tolerant reinforcement design method for spaceborne intelligent processing algorithms based on convolutional neural networks (CNNs). This method is built on a CNN accelerator using Field-Programmable Gate Array (FPGA) technology, analyzing the impact of Single-Event Upsets (SEUs) on neural network computation. The accelerator design integrates data validation, Triple Modular Redundancy (TMR), and other techniques, optimizing a partial fault-tolerant architecture based on SEU sensitivity. This fault-tolerant architecture analyzes the hardware accelerator, parameter storage, and actual computation, employing data validation to reinforce model parameters and spatial and temporal TMR to reinforce accelerator computations. Using the ResNet18 model, fault tolerance performance tests were conducted by simulating SEUs. Compared to the prototype network, this fault-tolerant design method increases tolerance to SEU error accumulation by five times while increasing resource consumption by less than 15%, making it more suitable for spaceborne on-orbit applications than traditional fault-tolerant design approaches.
format Article
id doaj-art-5c75605a47374ac7aefead3d5d033bca
institution OA Journals
issn 2072-4292
language English
publishDate 2024-12-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-5c75605a47374ac7aefead3d5d033bca2025-08-20T02:27:39ZengMDPI AGRemote Sensing2072-42922024-12-011716910.3390/rs17010069Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance DesignYingzhao Shao0Junyi Wang1Xiaodong Han2Yunsong Li3Yaolin Li4Zhanpeng Tao5State Key Laboratory of Integrated Services Networks, Xidian University, Xi’an 710071, ChinaSchool of Computer Science and Technology, Xidian University, Xi’an 710071, ChinaChina Academy of Space Technology (Xi’an), Xi’an 710100, ChinaState Key Laboratory of Integrated Services Networks, Xidian University, Xi’an 710071, ChinaSchool of Computer Science and Technology, Xidian University, Xi’an 710071, ChinaSchool of Computer Science and Technology, Xidian University, Xi’an 710071, ChinaTo meet the high-reliability requirements of real-time on-orbit tasks, this paper proposes a fault-tolerant reinforcement design method for spaceborne intelligent processing algorithms based on convolutional neural networks (CNNs). This method is built on a CNN accelerator using Field-Programmable Gate Array (FPGA) technology, analyzing the impact of Single-Event Upsets (SEUs) on neural network computation. The accelerator design integrates data validation, Triple Modular Redundancy (TMR), and other techniques, optimizing a partial fault-tolerant architecture based on SEU sensitivity. This fault-tolerant architecture analyzes the hardware accelerator, parameter storage, and actual computation, employing data validation to reinforce model parameters and spatial and temporal TMR to reinforce accelerator computations. Using the ResNet18 model, fault tolerance performance tests were conducted by simulating SEUs. Compared to the prototype network, this fault-tolerant design method increases tolerance to SEU error accumulation by five times while increasing resource consumption by less than 15%, making it more suitable for spaceborne on-orbit applications than traditional fault-tolerant design approaches.https://www.mdpi.com/2072-4292/17/1/69spaceborne computingsingle-event upsetconvolutional neural networkfield-programmable gate arrayfault-tolerant design
spellingShingle Yingzhao Shao
Junyi Wang
Xiaodong Han
Yunsong Li
Yaolin Li
Zhanpeng Tao
Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
Remote Sensing
spaceborne computing
single-event upset
convolutional neural network
field-programmable gate array
fault-tolerant design
title Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
title_full Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
title_fullStr Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
title_full_unstemmed Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
title_short Research on Spaceborne Neural Network Accelerator and Its Fault Tolerance Design
title_sort research on spaceborne neural network accelerator and its fault tolerance design
topic spaceborne computing
single-event upset
convolutional neural network
field-programmable gate array
fault-tolerant design
url https://www.mdpi.com/2072-4292/17/1/69
work_keys_str_mv AT yingzhaoshao researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign
AT junyiwang researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign
AT xiaodonghan researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign
AT yunsongli researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign
AT yaolinli researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign
AT zhanpengtao researchonspaceborneneuralnetworkacceleratoranditsfaulttolerancedesign