A Lightweight Mitigation Technique for Resource- Constrained Devices Executing DNN Inference Models Under Neutron Radiation
- 27 March 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Nuclear Science
- Vol. 70 (8) , 1625-1633
Abstract
Deep neural network (DNN) models are being deployed in safety-critical embedded devices for object identification, recognition, and even trajectory prediction. Optimised versions of such models, in particular the convolutional ones, are becoming increasingly common in resource-constrained edge-computing devices (e.g., sensors, drones), which typically rely on reduced memory footprint, low power budget and low-performance microprocessors. DNN models are prone to radiation-induced soft errors, and tackling their occurrence in resource-constrained devices is a mandatory and substantial challenge. While traditional replication-based soft error mitigation techniques will likely account for a reasonable performance penalty, hardware solutions are even more costly. To undertake this almost contradictory challenge, this work evaluates the efficiency of a lightweight software-based mitigation technique, called Register Allocation Technique (RAT), when applied to a convolutional neural network (CNN) model running on two commercial Arm microprocessors (i.e., Cortex-M4 and M7) under the effects of neutron radiation. Gathered results obtained from two neutron radiation campaigns suggest that RAT can reduce the number of critical faults in the CNN model running on both Arm Cortex-M microprocessors. Results also suggest that the SDC FIT rate of the RAT-hardened CNN model can be reduced in up to 83% with a runtime overhead of 32%.Keywords
Funding Information
- CAPES
- CNPq (317087/2021-5)
- Fundação de Amparo à Pesquisa do Estado do Rio Grande do Sul (22/2551-0000570-5)
- U.K. EPSRC (EP/R513088/1)
- MultiRad through Région Auvergne-Rhône-Alpes under Project PAI
- IRT Nanoelec through French Programme d’Investissements d’Avenir (ANR-10-AIRT-05)
- UGA/LPSC/GENESIS Platform
- Ministerio de Ciencia e Innovación (PID2019-106455GB-C22)
This publication has 34 references indexed in Scilit:
- Selective Hardening for Neural Networks in FPGAsIEEE Transactions on Nuclear Science, 2018
- Analyzing and Increasing the Reliability of Convolutional Neural Networks on GPUsIEEE Transactions on Reliability, 2018
- Ares2018
- MATIC: Learning around errors for efficient low-voltage neural network accelerators2018
- On the Reliability of Linear Regression and Pattern Recognition Feedforward Artificial Neural Networks in FPGAsIEEE Transactions on Nuclear Science, 2017
- Understanding error propagation in deep learning neural network (DNN) accelerators and applications2017
- JEDEC Standards on Measurement and Reporting of Alpha Particle and Terrestrial Cosmic Ray Induced Soft Errors2010
- Neutron-induced soft errors in advanced flash memories2008
- Radiation-induced soft errors in advanced semiconductor technologiesIEEE Transactions on Device and Materials Reliability, 2005
- Basic concepts and taxonomy of dependable and secure computingIEEE Transactions on Dependable and Secure Computing, 2004