Trojan-net feature extraction and its application to hardware-Trojan detection for gate-level netlists using random forest

    Research output: Contribution to journalArticle

    2 Citations (Scopus)

    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 hardwareTrojan detection gives a strong solution to tackle this problem. HardwareTrojan 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 Trojannet 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.

    Original languageEnglish
    Pages (from-to)2857-2868
    Number of pages12
    JournalIEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
    VolumeE100A
    Issue number12
    DOIs
    Publication statusPublished - 2017 Dec 1

    Fingerprint

    Random Forest
    Feature Extraction
    Learning systems
    Feature extraction
    Hardware
    Classifiers
    Machine Learning
    Hardware security
    Classify
    Classifier
    Strong Solution
    Design Process
    Optimise

    Keywords

    • Gate-level netlist
    • Hardware Trojan
    • Machine learning
    • Random forest
    • Trojan-net feature

    ASJC Scopus subject areas

    • Signal Processing
    • Computer Graphics and Computer-Aided Design
    • Electrical and Electronic Engineering
    • Applied Mathematics

    Cite this

    @article{e9c07fc47764463ea3d4ed061a5321a6,
    title = "Trojan-net feature extraction and its application to hardware-Trojan detection for gate-level netlists using random forest",
    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 hardwareTrojan detection gives a strong solution to tackle this problem. HardwareTrojan 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 Trojannet 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 = "Gate-level netlist, Hardware Trojan, Machine learning, Random forest, Trojan-net feature",
    author = "Kento Hasegawa and Masao Yanagisawa and Nozomu Togawa",
    year = "2017",
    month = "12",
    day = "1",
    doi = "10.1587/transfun.E100.A.2857",
    language = "English",
    volume = "E100A",
    pages = "2857--2868",
    journal = "IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences",
    issn = "0916-8508",
    publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
    number = "12",

    }

    TY - JOUR

    T1 - Trojan-net feature extraction and its application to hardware-Trojan detection for gate-level netlists using random forest

    AU - Hasegawa, Kento

    AU - Yanagisawa, Masao

    AU - Togawa, Nozomu

    PY - 2017/12/1

    Y1 - 2017/12/1

    N2 - 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 hardwareTrojan detection gives a strong solution to tackle this problem. HardwareTrojan 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 Trojannet 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.

    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 hardwareTrojan detection gives a strong solution to tackle this problem. HardwareTrojan 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 Trojannet 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.

    KW - Gate-level netlist

    KW - Hardware Trojan

    KW - Machine learning

    KW - Random forest

    KW - Trojan-net feature

    UR - http://www.scopus.com/inward/record.url?scp=85038213176&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85038213176&partnerID=8YFLogxK

    U2 - 10.1587/transfun.E100.A.2857

    DO - 10.1587/transfun.E100.A.2857

    M3 - Article

    AN - SCOPUS:85038213176

    VL - E100A

    SP - 2857

    EP - 2868

    JO - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

    JF - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

    SN - 0916-8508

    IS - 12

    ER -