A Lightweight Hardware-Friendly CNN Accelerator Security Reinforcement Method
Electronics · 19.08.2026 · 15 (16) · 3707
DOI: 10.3390/electronics15163707 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 19.08.2026. Verfasst von Ying Zhang und 2 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Convolutional neural networks (CNNs), as an important branch of artificial intelligence, have a wide range of applications in security-critical scenarios, and their hardware accelerators are also constantly evolving. However, hardware failures can cause parameter bit flipping, and carefully designed bit-flipping attacks (BFAs) can also severely degrade their classification accuracy. This paper proposes a hardware-friendly security-reinforcement framework, which utilizes the representation characteristics of floating-point numbers on memory units to propose modulation and demodulation reinforcement methods for weights. On the computing unit, a data-threshold constraint reinforcement method is proposed, and a bounded ReLU is introduced to mitigate the impact of BFAs by imposing an upper bound on the activation function. Experimental results demonstrate that when the BER reaches 10−2, the accuracies of LeNet-5, LeNet-3D, and ResNet-18 increase from 13.17%, 12.01%, and 10% to 93.21%, 59.12%, and 62.37%, respectively.
Zitieren
APA
Zhang, Y., Yang, H., & Wang, Z. (2026). A Lightweight Hardware-Friendly CNN Accelerator Security Reinforcement Method. Electronics, 15(16), 3707. https://doi.org/10.3390/electronics15163707
Vancouver
Zhang Y, Yang H, Wang Z. A Lightweight Hardware-Friendly CNN Accelerator Security Reinforcement Method. Electronics. 2026;15(16):3707. doi:10.3390/electronics15163707
BibTeX
@article{Zhang2026,
author = {Zhang, Ying and Yang, Hai and Wang, Zixiao},
title = {A Lightweight Hardware-Friendly CNN Accelerator Security Reinforcement Method},
journal = {Electronics},
year = {2026},
volume = {15},
number = {16},
pages = {3707},
doi = {10.3390/electronics15163707}
}
EUVIMED