A Rapid Evaluation Technology for SEU in Convolutional Neural Network Circuits

Chen, Kai; Chen, Xin; Zhang, Ying; Zhang, Zhiwei

doi:10.1109/iccs52645.2021.9697197

Cited by 2 publications

(3 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to the typical SEE evaluation scheme [19,20], the proposed scheme provides a rapid solution to analyze the vulnerable modules in the logic designs before implementation [21].…”

Section: Seu Evaluation Platformmentioning

confidence: 99%

Implementation of Highly Reliable Convolutional Neural Network with Low Overhead on Field-Programmable Gate Array

Chen,

Xie,

Huo

et al. 2024

Electronics

Self Cite

View full text Add to dashboard Cite

Due to the advantages of parallel architecture and low power consumption, a field-programmable gate array (FPGA) is typically utilized as the hardware for convolutional neural network (CNN) accelerators. However, SRAM-based FPGA devices are extremely susceptible to single-event upsets (SEUs) induced by space radiation. In this paper, a fault tolerance analysis and fault injection experiments are applied to a CNN accelerator, and the overall results show that SEUs occurring in a control unit (CTRL) lead to the highest system error rate, which is over 70%. After that, a hybrid hardening strategy consisting of a finite state machine error-correcting circuit (FSM-ECC) and a triple modular redundancy automatic hardening technique (TMR-AHT) is proposed in this paper to achieve a tradeoff between radiation reliability and design overhead. Moreover, the proposed methodology has very small workload and good migration ability. Finally, by full exploiting the fault tolerance property of CNNs, a highly reliable CNN accelerator with the proposed hybrid hardening strategy is implemented with Xilinx Zynq-7035. When BER is 2 × 10−6, the proposed hybrid hardening strategy reduces the whole system error rate by 78.95% with the overhead of an extra 20.7% of look-up tables (LUTs) and 20.9% of flip-flops (FFs).

show abstract

“…Compared to the typical SEE evaluation scheme [19,20], the proposed scheme provides a rapid solution to analyze the vulnerable modules in the logic designs before implementation [21].…”

Section: Seu Evaluation Platformmentioning

confidence: 99%

Implementation of Highly Reliable Convolutional Neural Network with Low Overhead on Field-Programmable Gate Array

Chen,

Xie,

Huo

et al. 2024

Electronics

Self Cite

View full text Add to dashboard Cite

show abstract

“…proposed a fast single event upset evaluation platform capable of algorithmic level fault injection in convolutional neural networks. [ 8,9 ] There have also been many studies on SEEs based on the Geant4 platform. For example, Truscott et al.…”

Section: Introductionmentioning

confidence: 99%

“…KaiChen et al proposed a fast single event upset evaluation platform capable of algorithmic level fault injection in convolutional neural networks. [8,9] There have also been many studies on SEEs based on the Geant4 platform. For example, Truscott et al examine the performance of the Geant4 radiation transport toolkit for the simulation of energy deposition from proton-and neutron-nuclear interactions in silicon microelectronics.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Linear Energy Transfer Based on Geant4 Simulation and Neural Networks

Zhang

Chen

Liu

et al. 2023

Advcd Theory and Sims

View full text Add to dashboard Cite

Most of the traditional studies based on single event effects (SEEs) favor the analysis of electrical mechanisms of semiconductor devices. Professional simulation software in microelectronics requires researchers to have a solid knowledge of microelectronics theory, and the modeling threshold of the software is relatively high, the simulation speed is slow, and accurate simulation of inter‐particle and particle–material interactions is lacking. SEEs are related to linear energy transfer (LET), in this paper, a method is proposed to obtain LET datas to predict SEEs of particles incident on silicon materials by using the Geant4 Monte Carlo toolkit in combination with a dense convolutional network to accurately and rapidly estimate the energy deposition characteristics of the particles. The proposed network structure has a high prediction accuracy with a mean square error （MSE） of only 1.77 × 10−4. Compared with Geant4, which takes 1 min to compute a set of data, the proposed network structure takes only 0.0817 s. The method explores the feasibility of using Geant4 to model semiconductor devices combined with deep learning algorithms, providing a new research perspective for the prediction of microelectronic devices and making it possible to explore the influence of integrated circuits by SEEs.

show abstract

A Rapid Evaluation Technology for SEU in Convolutional Neural Network Circuits

Cited by 2 publications

References 8 publications

Implementation of Highly Reliable Convolutional Neural Network with Low Overhead on Field-Programmable Gate Array

Implementation of Highly Reliable Convolutional Neural Network with Low Overhead on Field-Programmable Gate Array

Prediction of Linear Energy Transfer Based on Geant4 Simulation and Neural Networks

Contact Info

Product

Resources

About