Trojan-Net Feature Extraction and Its Application to Hardware-Trojan Detection for Gate-Level Netlists Using Random Forest

Kento HASEGAWA; Masao YANAGISAWA; Nozomu TOGAWA

doi:10.1587/transfun.E100.A.2857

Trojan-Net Feature Extraction and Its Application to Hardware-Trojan Detection for Gate-Level Netlists Using Random Forest

Kento HASEGAWA, Masao YANAGISAWA, Nozomu TOGAWA

Full Text Views

0

Share
Cite this

Summary :

It has been reported that malicious third-party IC vendors often insert hardware Trojans into their IC products. How to detect them is a critical concern in IC design process. Machine-learning-based hardware-Trojan detection gives a strong solution to tackle this problem. Hardware-Trojan infected nets (or Trojan nets) in ICs must have particular Trojan-net features, which differ from those of normal nets. In order to classify all the nets in a netlist designed by third-party vendors into Trojan nets and normal ones by machine learning, we have to extract effective Trojan-net features from Trojan nets. In this paper, we first propose 51 Trojan-net features which describe well Trojan nets. After that, we pick up random forest as one of the best candidates for machine learning and optimize it to apply to hardware-Trojan detection. Based on the importance values obtained from the optimized random forest classifier, we extract the best set of 11 Trojan-net features out of the 51 features which can effectively classify the nets into Trojan ones and normal ones, maximizing the F-measures. By using the 11 Trojan-net features extracted, our optimized random forest classifier has achieved at most 100% true positive rate as well as 100% true negative rate in several Trust-HUB benchmarks and obtained the average F-measure of 79.3% and the accuracy of 99.2%, which realize the best values among existing machine-learning-based hardware-Trojan detection methods.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E100-A No.12 pp.2857-2868

Publication Date: 2017/12/01

Publicized

Online ISSN: 1745-1337

DOI: 10.1587/transfun.E100.A.2857

Type of Manuscript: Special Section PAPER (Special Section on VLSI Design and CAD Algorithms)

Category

Authors

Kento HASEGAWA
  Engineering, Waseda University
Masao YANAGISAWA
  Engineering, Waseda University
Nozomu TOGAWA
  Engineering, Waseda University

Keyword

hardware Trojan, gate-level netlist, Trojan-net feature, random forest, machine learning

Cite this

Copy

Kento HASEGAWA, Masao YANAGISAWA, Nozomu TOGAWA, "Trojan-Net Feature Extraction and Its Application to Hardware-Trojan Detection for Gate-Level Netlists Using Random Forest" in IEICE TRANSACTIONS on Fundamentals, vol. E100-A, no. 12, pp. 2857-2868, December 2017, doi: 10.1587/transfun.E100.A.2857.
Abstract: It has been reported that malicious third-party IC vendors often insert hardware Trojans into their IC products. How to detect them is a critical concern in IC design process. Machine-learning-based hardware-Trojan detection gives a strong solution to tackle this problem. Hardware-Trojan infected nets (or Trojan nets) in ICs must have particular Trojan-net features, which differ from those of normal nets. In order to classify all the nets in a netlist designed by third-party vendors into Trojan nets and normal ones by machine learning, we have to extract effective Trojan-net features from Trojan nets. In this paper, we first propose 51 Trojan-net features which describe well Trojan nets. After that, we pick up random forest as one of the best candidates for machine learning and optimize it to apply to hardware-Trojan detection. Based on the importance values obtained from the optimized random forest classifier, we extract the best set of 11 Trojan-net features out of the 51 features which can effectively classify the nets into Trojan ones and normal ones, maximizing the F-measures. By using the 11 Trojan-net features extracted, our optimized random forest classifier has achieved at most 100% true positive rate as well as 100% true negative rate in several Trust-HUB benchmarks and obtained the average F-measure of 79.3% and the accuracy of 99.2%, which realize the best values among existing machine-learning-based hardware-Trojan detection methods.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1587/transfun.E100.A.2857/_p

Copy

@ARTICLE{e100-a_12_2857,
author={Kento HASEGAWA, Masao YANAGISAWA, Nozomu TOGAWA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Trojan-Net Feature Extraction and Its Application to Hardware-Trojan Detection for Gate-Level Netlists Using Random Forest},
year={2017},
volume={E100-A},
number={12},
pages={2857-2868},
abstract={It has been reported that malicious third-party IC vendors often insert hardware Trojans into their IC products. How to detect them is a critical concern in IC design process. Machine-learning-based hardware-Trojan detection gives a strong solution to tackle this problem. Hardware-Trojan infected nets (or Trojan nets) in ICs must have particular Trojan-net features, which differ from those of normal nets. In order to classify all the nets in a netlist designed by third-party vendors into Trojan nets and normal ones by machine learning, we have to extract effective Trojan-net features from Trojan nets. In this paper, we first propose 51 Trojan-net features which describe well Trojan nets. After that, we pick up random forest as one of the best candidates for machine learning and optimize it to apply to hardware-Trojan detection. Based on the importance values obtained from the optimized random forest classifier, we extract the best set of 11 Trojan-net features out of the 51 features which can effectively classify the nets into Trojan ones and normal ones, maximizing the F-measures. By using the 11 Trojan-net features extracted, our optimized random forest classifier has achieved at most 100% true positive rate as well as 100% true negative rate in several Trust-HUB benchmarks and obtained the average F-measure of 79.3% and the accuracy of 99.2%, which realize the best values among existing machine-learning-based hardware-Trojan detection methods.},
keywords={},
doi={10.1587/transfun.E100.A.2857},
ISSN={1745-1337},
month={December},}

Copy

TY - JOUR
TI - Trojan-Net Feature Extraction and Its Application to Hardware-Trojan Detection for Gate-Level Netlists Using Random Forest
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2857
EP - 2868
AU - Kento HASEGAWA
AU - Masao YANAGISAWA
AU - Nozomu TOGAWA
PY - 2017
DO - 10.1587/transfun.E100.A.2857
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E100-A
IS - 12
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - December 2017
AB - It has been reported that malicious third-party IC vendors often insert hardware Trojans into their IC products. How to detect them is a critical concern in IC design process. Machine-learning-based hardware-Trojan detection gives a strong solution to tackle this problem. Hardware-Trojan infected nets (or Trojan nets) in ICs must have particular Trojan-net features, which differ from those of normal nets. In order to classify all the nets in a netlist designed by third-party vendors into Trojan nets and normal ones by machine learning, we have to extract effective Trojan-net features from Trojan nets. In this paper, we first propose 51 Trojan-net features which describe well Trojan nets. After that, we pick up random forest as one of the best candidates for machine learning and optimize it to apply to hardware-Trojan detection. Based on the importance values obtained from the optimized random forest classifier, we extract the best set of 11 Trojan-net features out of the 51 features which can effectively classify the nets into Trojan ones and normal ones, maximizing the F-measures. By using the 11 Trojan-net features extracted, our optimized random forest classifier has achieved at most 100% true positive rate as well as 100% true negative rate in several Trust-HUB benchmarks and obtained the average F-measure of 79.3% and the accuracy of 99.2%, which realize the best values among existing machine-learning-based hardware-Trojan detection methods.
ER -