Vulnerability Information Sharing Platform for Securing Hardware
Supply Chains
Kento Hasegawa
1 a
, Katsutoshi Hanahara
2
, Hiroshi Sugisaki
1
, Minoru Kozu
1
,
Kazuhide Fukushima
1 b
, Yosuke Murakami
1
and Shinsaku Kiyomoto
1
1
KDDI Research, Inc., Japan
2
KDDI Foundation, Japan
Keywords:
Supply Chain, Design, Hardware Trojan, Detection, Industry.
Abstract:
The rise of complex global supply chains has increased the risk of malicious actors attempting to insert mali-
cious functions, called hardware Trojans (HTs), into hardware components and devices. Although many HT
detection methods have been proposed over a decade, implementing them in industries may take a long time
due to concerns about these methods. In this paper, we propose a repository system to manage vulnerability
information for securing hardware supply chains and investigate the demand and barriers to introducing hard-
ware Trojan detection schemes in the industry. First, we design a scheme to share the results of HT detection
methods. Second, we design questionnaires to investigate the actual situation of the industry’s awareness of
the threat of HTs and other hardware security issues. We conclude that there is a gap between academics and
the industry, whereas many business operators are concerned about the threat of HTs.
1 INTRODUCTION
The rise of complex global supply chains has in-
creased the risk of malicious actors attempting to
insert malicious functions, called hardware Tro-
jans (HTs), into hardware components and de-
vices (Francq and Frick, 2015; Xiao et al., 2016). HTs
may leak internal confidential information, degrade
performance, or alter the original functionality of
hardware components. Many HT detection methods
have been proposed for over a decade to defend hard-
ware products from the insertion of HTs. However,
these methods were proposed from the viewpoint of
academic research. The feasibility of HTs in the real
world and the industry’s awareness of the threat posed
by HTs have not been sufficiently assessed. Almeida
et al. investigated the feasibility of ransomware at-
tacks as HTs (Almeida et al., 2022). This feasibility
study demonstrated that there are no barriers for an
adversary to devise hardware ransomware in a hard-
ware product. Although this result suggests that HTs
can be easily realized, the actual awareness of busi-
ness operators across various industries has not been
investigated.
a
https://orcid.org/0000-0002-6517-1703
b
https://orcid.org/0000-0003-2571-0116
In this paper, we focus on the actual situation
of the industry as it relates to semiconductor supply
chains. First, we propose a scheme to share the re-
sults of HT detection methods in the hardware de-
sign phase. In terms of software management, vul-
nerability information for software is publicly shared
on several platforms, such as the National Vulnera-
bility Database (NVD) and Common Vulnerabilities
and Exposures (CVEs). Inspired by such platforms,
we design a framework to share vulnerability infor-
mation for hardware. Second, we design question-
naires to investigate the actual situation of the indus-
try’s awareness of the threat of HTs. Through the
questionnaires, we clarify the demand and concerns
with respect to HT detection methods and security as-
sessment schemes in industries.
The contributions of this paper can be summarized
as follows:
We propose a repository operation scheme to
share vulnerability information of hardware prod-
ucts. The scenario using the repository is consid-
ered a use case for HT detection methods in in-
dustries.
We designed questionnaires to confirm the needs,
effectiveness, and concerns regarding the reposi-
tory operation scheme and conducted a survey tar-
Hasegawa, K., Hanahara, K., Sugisaki, H., Kozu, M., Fukushima, K., Murakami, Y. and Kiyomoto, S.
Vulnerability Information Sharing Platform for Securing Hardware Supply Chains.
DOI: 10.5220/0012351800003648
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 10th International Conference on Information Systems Security and Privacy (ICISSP 2024), pages 403-410
ISBN: 978-989-758-683-5; ISSN: 2184-4356
Proceedings Copyright © 2024 by SCITEPRESS Science and Technology Publications, Lda.
403
geting 15 companies.
We found that HT detection methods are attractive
in industries, and the repository scheme will be
enhanced if the scheme covers the whole supply
chain, i.e., the whole lifecycle of semiconductor
products.
2 PRELIMINARIES
2.1 Hardware Trojan (HT)
HTs are malicious modifications inserted into inte-
grated circuits (ICs) during the manufacturing pro-
cess with the intention of causing harmful impacts for
manufacturers or IC users. They can be inserted by a
malicious adversary with access to the design or fab-
rication process or by a rogue employee within hard-
ware supply chains. HTs can perform various mali-
cious functions, such as leaking confidential internal
information, disabling critical functions, or causing
the device to malfunction. A specific trigger, such as
a particular input or a specific time, can activate HTs.
HTs are usually composed of small circuits, mak-
ing them difficult to detect during the design phase.
HTs can also be designed to avoid detection by test-
ing methods and to remain dormant until a specific
trigger activates them. Adversaries can insert HTs in
any phase in hardware supply chains. Particularly, in-
serting HTs in the design phase is relatively easy and
effective because one modification in a design spreads
many products.
2.2 Hardware Trojan Detection
In this paper, we focus on HT detection methods in the
design phase to detect HTs earlier in supply chains. A
hardware design is described in hardware description
languages (HDLs). An adversary may alter the de-
sign described in HDL to insert HTs or may provide
intellectual property that is infected with HTs.
There are several HT detection approaches in the
design phase. A promising method is a structural
feature-based approach, in which HT-specific features
are identified from a tested hardware design (Yang
et al., 2020). In a structural feature-based approach,
the structure of a circuit is extracted based on the
hardware design. According to (Oya et al., 2015),
specific features appear in HT circuits. For example,
a trigger circuit in an HT has many fan-ins to imple-
ment a rarely triggered condition. The method pro-
posed in (Oya et al., 2015) suggests that feature values
that help identify HTs can be extracted from hardware
designs.
Based on the findings, HT detection methods us-
ing machine learning have been proposed (Hasegawa
et al., 2017b; Hasegawa et al., 2017a; Li et al., 2020;
Huang et al., 2020). There are two phases for machine
learning-based HT detection: the feature extraction
phase and the machine learning phase. In the fea-
ture extraction phase, a set of possible features that
can help identify HTs are extracted from hardware
designs (i.e., program codes written in HDL). Then,
a model is trained with the extracted feature values in
the machine learning phase. The trained model is ex-
pected to identify HTs from a given hardware design.
However, there are barriers to implementing HT
detection in industries. A major barrier is the cost of
HT detection. Performing HT detection will consume
some time, thus delaying manufacturing processes.
Many vendors may doubt if the benefit of HT detec-
tion is greater than the cost.
In this paper, we survey the awareness of HT de-
tection in industries from the perspective of these
points. First, to make HT detection beneficial, we
propose a repository operation scheme to share the
vulnerability information between business operators.
Specifically, the list of hardware components and
their HT detection results are stored in the repository
system. Using this repository system and its opera-
tion scheme, we conduct questionnaires to investigate
the feasibility and effectiveness.
3 METHODOLOGY
In this section, we propose a repository system to
share vulnerability information, such as the reports of
HT detection. This repository aims to enhance the se-
curity of hardware supply chains, particularly in the
design phase.
3.1 Overview of the Repository
Operation Scheme
We propose a scheme to share the results of HT detec-
tion as a demonstration. Figure 1 shows an overview
of the repository operation scheme that we designed.
Based on the HT detection methods, we design the
scheme to share the hardware vulnerability informa-
tion between business operators. There are three roles
for the repository users: administrators, manufactur-
ers, and inspection agencies. Administrators manage
the users and contents of the repository. Manufactur-
ers can register their product information in the repos-
itory and view the design information registered by
other manufacturers. Inspection agencies can regis-
ter the inspection results of a product that has been
ICISSP 2024 - 10th International Conference on Information Systems Security and Privacy
404
Repository
Manage the repository
Inspection AgencyManufacturer
Administrator
Register inspection results
Register designs
/ View entries
Users
Permissions are granted based on the user’s role.
Request inspection
Figure 1: Overview of the repository operation scheme.
Repository
User
Management
User
Table
Affiliation, role, etc.
Product
Management
Manager
Product
Table
Inspection
Table
Product information, hash
values for attached files,
etc.
Designer
Inspection
Management
Inspector
Session Management
Add / Edit / Delete
View
Add / Edit / Delete
Add / Edit / Delete
Inspection results of HT
detection, etc.
Relation
* UI: User Interface
UI
UI
UI
Relation
Viewer
+Manage
Only viewing
Only viewing
+Manage
Figure 2: Overview of the repository architecture.
registered in the repository.
3.2 Repository Architecuture
Figure 2 shows an overview of the repository architec-
ture. In the repository, the database includes mainly
three tables: the user, product, and inspection ta-
bles. Since the repository targets the management of
the hardware design phase, the product here indicates
hardware design products.
The user table stores information regarding the
repository users, such as their names, affiliations,
email addresses, and roles. Administrators assign a
role to a user, and users without the administrator role
cannot change their role. The details of the user roles
are explained later.
The product table includes information regarding
hardware design products, such as the designs for pro-
cessors and interfaces. Users assigned to the man-
ufacturer role can register their product information
to the product table, and other users can only see
the product information. Users who are authorized
can register product information, such as the prod-
uct name, product version, and application category
of the product. Documents, including datasheets and
hardware design deliverables, can be attached to the
product information. The product management func-
tion in the repository system provides a user interface
that shows detailed product information and the part
lists used in the product. Since some attached files
cannot be disclosed, the user who registers product
information can determine whether the information is
disclosed.
The inspection table includes information regard-
ing the vulnerabilities found in hardware designs.
Users assigned to the inspection agency role can reg-
ister their inspection reports to the inspection table.
The inspection report includes the results of design
testing, such as HT detection and formal verification,
and is associated with a product record. The inspec-
tion record includes the date of inspection and inspec-
tion result.
3.3 Scenario of Repository Operation
In this repository scheme, the following scenario is
considered.
Step 1. A manufacturer registers their hardware
design products to the repository and requests an
inspection agency to inspect their products.
Step 2. An inspection agency assesses the vulner-
ability of the products using HT detection meth-
ods and reports the evaluation results.
Step 3. The evaluation results are attached to the
product on the repository. These results can be
seen by repository users.
When a manufacturer hopes to use an IC chip,
they can refer to the repository and check the vulner-
ability information. If they find a vulnerability or risk
information, they can stop using the product. Other-
wise, they can feel secure in using the product.
4 EVALUATION
4.1 Implementation Result
We implemented the repository system as a web
application. The tables are stored as a MySQL
database, and the backend of the web application is
implemented using a Python framework. Using this
web application, users, including hardware designers,
manufacturers, and inspection agencies, can refer to
the information stored in the repository and edit in-
formation within their authority.
Vulnerability Information Sharing Platform for Securing Hardware Supply Chains
405
Figure 3: Example of a web form of the repository system.
Figure 4: Illustration of the part composition on the reposi-
tory system.
Figure 3 shows a form of the repository system
to register design information. Since the repository
is implemented as a web application, users can edit
records on a browser.
Figure 4 shows the part composition of a design
registered in the repository system. The repository
system visualizes the relationship of the parts.
4.2 Questionnaire Survey
To investigate the feasibility of the repository oper-
ation scheme, we designed questionnaires and con-
ducted a survey based on the designed questionnaires.
There are five categories in the designed question-
naires: (a) basic information, (b) current situations,
(c) needs, (d) effectiveness and feasibility, and (e) risk
measures in supply chains. The questions are de-
scribed in Tables 1–6 along with the responses cor-
responding to the questions.
The questionnaires were conducted through a con-
sulting firm. For a questionnaire survey, we col-
Table 1: Questions and responses to QA basic informa-
tion.
QA1 Business description.
Electronics or Vehicle
QA2 Position in the supply chain.
Semiconductor manufacturers, fabless ven-
dors, EDA vendors, or vehicle manufacturers
QA3 Revenue.
More than 100 M USD: 12
Less than 100 M USD: 3
QA4 Number of employees.
More than 10000: 13
Less than 10000: 2
QA5 Country of headquarters.
North America: 5, EU: 4, Asia: 6
QA6 Department of the respondent.
Design, Development, Solution or Marketing
QA7 Position of the respondent.
Manager or Section head
lected 15 respondents from around the world, includ-
ing North America, Europe, and Asia. We covered
three business types: integrated manufacturer, hori-
zontally specialized manufacturer, and implementa-
tion vendor. When asking the questions, the purpose
of the survey and the repository scheme proposed in
Section 3 were explained to the respondents. The
requester of the questionnaires was not disclosed to
avoid bias. The survey was conducted from August
2022 to January 2023. In the following section, we
provide the questions and responses. Due to page lim-
itations, the representative responses are shown in the
response tables.
4.2.1 QA: Basic Information
In this part, we asked for basic information about the
respondents’ affiliation and themselves.
Table 1 shows the summary of the responses to the
QA section. The business areas covered semiconduc-
tor manufacturers, fabless vendors, EDA vendors, and
vehicle manufacturers. The headquarters’ countries
cover major regions, including North America, Eu-
rope, and Asia. To collect feedback from production
and development fields, the respondents were mainly
stakeholders from the development and solution de-
partments. Since the questionnaires are related to ac-
tual design or manufacturing processes, the respon-
dent is chosen from a manager or a section head who
understands actual situations.
ICISSP 2024 - 10th International Conference on Information Systems Security and Privacy
406
Table 2: Questions and responses to QB – current situations.
QB1 Does your company investigate the IPs and parts from a security viewpoint before introducing them
to the design?
We do not verify IPs introduced to products. Instead, we verify the final products.
We rigorously inspect products when purchasing. We also certify vendors. However, we use the evalu-
ation metric of quality (i.e., defective rate), not of security.
QB2 Are there any schemes to share alerts when doubt about some IPs or products occurs?
A scheme has been established to respond to an issue and to address the issue with internal and external
parties.
QB3 Are there any schemes to share alerts when doubt about your product occurs?
We have not experienced any problems regarding our products. If this happens, we leverage the scheme
to share information about the issues of products.
4.2.2 QB: Current Situations
In this part, we confirmed the current situation regard-
ing the industry’s awareness of HTs. Specifically, we
aim to clarify that IPs are investigated from a security
viewpoint. In general, hardware design houses aim to
provide hardware designs that have sufficient perfor-
mance and functionality according to specifications.
However, the awareness of security incidents, such as
being infected with HTs, has not been investigated.
We expect the questions in this part to reveal the cur-
rent attitude toward security perspectives.
Table 2 shows the summary of the responses to
the QB section. Few companies investigate IPs when
introducing them. Instead, the final products are suf-
ficiently verified in terms of satisfying the specifica-
tions and quality. However, such verification aims to
confirm the quality of the products, not in terms of
security. Remember that HTs are stealthily inserted
into products. Thus, it is difficult to detect HTs under
current situations.
Most companies have schemes to share issues
with their products. When an issue is found, the in-
formation is shared with the whole company through
the scheme. The scheme can be utilized if a problem
in hardware designs, such as an HT, is found.
In summary, currently, there is a scheme to share
vulnerability information with the whole company.
However, a security-specific investigation has not
been sufficiently conducted.
4.2.3 QC: Needs
In this part, we confirmed the need for HT detection
methods without considering feasibility. Since hard-
ware supply chains are exposed to geopolitical risks,
hardware vendors and manufacturers may pay atten-
tion to such security concerns. Security assessment
processes, such as HT detection, can be a solution to
this concern. In this situation, the cost of introducing
a security assessment process can be a problem. This
section clarifies these questions.
Table 3 shows the summary of the responses to the
QC section. Most companies were interested in secu-
rity assessment tools for hardware designs. In partic-
ular, identifying counterfeit products is desired. To
ensure the neutrality and reliability of the security as-
sessment, many companies expect third-party organi-
zations to perform the security assessment. However,
the cost of the security assessment is a major prob-
lem. Due to the competition over price, design houses
cannot cover the cost. Instead, it would be better for
business operators or industry organizations to cover
the cost.
4.2.4 QD: Effectiveness and Feasibility
In this part, we exhibited several examples of HT
detection methods and asked the respondents for an
evaluation regarding the efficiency and feasibility of
these methods. Although many HT detection meth-
ods have been studied in recent years, the feasibil-
ity of such methods has not been discussed. In this
section, we collected opinions on HT detection from
business operators.
In the QD4 question, we asked what level of de-
sign information your company can provide for the
security assessment process. Although the security
assessment process requires the analysis of hardware
designs, hardware design information is confidential
in most cases. We provided several examples for the
question to clarify the possible resources to be dis-
closed. Table 4 shows the possible resources for the
QD4 question. Since hardware designs written in
HDL (the QD4-a option) are highly confidential in-
formation for most vendors, we provide QD4-b as an-
other option. The QD4-b option shows feature ex-
traction techniques from a hardware design. The ex-
tracted features can be used for HT detection, and
original designs cannot be restored using the extracted
Vulnerability Information Sharing Platform for Securing Hardware Supply Chains
407
Table 3: Questions and responses to QC – needs.
QC1 Are you interested in security assessment tools for hardware designs?
Security assessment tools are interesting.
The market expects to identify counterfeit products everywhere in the supply chain.
QC2 Does your company investigate the delivered products from the security viewpoint?
We do not conduct any security-oriented assessment. Instead, we verify and test our products before
shipment.
Specialized verification processes are conducted for security-related products.
QC3 Who should operate the security assessment process?
Third-party agencies should operate the security assessment process. It is difficult to maintain the
reliability of the assessment in the case of in-house assessment.
QC4 How do you consider the cost of the security assessment process?
Business operators or industry organizations should cover the cost of the security assessment tools.
Table 4: Resources for the QD4 question.
# Item
QD4-a Hardware designs written in HDL.
QD4-b Feature values extracted from hardware
designs. Specifically, there are three
types of feature values: structural fea-
ture values (Oya et al., 2015; Dong et al.,
2020), SCOAP feature values (Salmani,
2017; Tebyanian et al., 2021), and graph
neural network-based embedding (Yu
et al., 2021; Hasegawa et al., 2023).
QD4-c Product information, such as the cate-
gory of IC.
QD4-d Application information, such as for
communication devices or for industrial
devices.
QD4-e Bill of materials or the component list of
the product.
QD4-f Report of inspection.
QD4-g Others (free comment).
features.
Table 5 shows the summary of the responses to
the QD section. Although the repository scheme cov-
ers the design process, most companies expect that
the repository scheme should cover all the processes,
including the manufacturing process. Since skill-
ful knowledge is required for the security assessment
process, many companies hope that an impartial and
professional organization should operate the reposi-
tory scheme. Specifically, government or industry or-
ganizations would be appropriate candidates for the
operator.
Since the hardware design information is confi-
dential, providing designs, bills of materials, and in-
spection reports is extremely difficult. However, in-
terestingly, the features extracted from hardware de-
signs (QD4-b explained in Table 4) can be disclosed
for security assessment if necessary. In this case, the
extraction process does not affect the existing produc-
tion process and is sufficiently supported by an extrac-
tion tool vendor. Although the design process should
be highly confidential, feature extraction is a possi-
ble way to adapt security assessment to semiconduc-
tor supply chains. Some companies commented that
the scope of the disclosure must be restricted when
the information is disclosed.
From the results, it can be seen that the demand
for identifying counterfeit products is increased. To
meet the demand, The repository operation scheme
covering the whole supply chain is expected. In the
repository operation scheme, the feature extraction
process is interesting. Information regarding its cost
and benchmarks is useful for considering the process.
In terms of legal regulations, most companies are
concerned about geopolitical issues. These issues sig-
nificantly affect business processes. Business opera-
tors in the automotive industry comply with interna-
tional standards that are more rigorous compared to
other industries.
4.2.5 QE: Risk Measures in Supply Chains
In this section, we investigated the concerns and risk
measures in semiconductor supply chains in terms of
legal regulations.
Table 6 shows the summary of the responses to
the QE section. For any company, the challenge lies
in balancing compliance with local laws and regula-
tions and maintaining stable production. Many com-
panies consider decentralizing production bases and
multi-sourcing of supply sources to address the risks
in semiconductor supply chains.
In case vulnerability assessment in hardware de-
vices is standardized, there is a concern about the
burden on business operators in terms of the cost and
measures they must take.
ICISSP 2024 - 10th International Conference on Information Systems Security and Privacy
408
Table 5: Questions and responses to QD – effectiveness and feasibility.
QD1 What information is helpful when using the repository scheme?
Tools or schemes to identify counterfeit products are interesting.
It is useful if the repository can cover the manufactured products to identify counterfeit products.
QD2 Who should operate the repository scheme?
Institutions considering introducing a repository operation scheme.
The government or industry organizations should design the system, and an impartial and professional
organization should operate it.
QD3 Is it acceptable if hardware designs are requested for the security assessment process?
Generally, no companies provide circuit design information.
It is unavailable due to license issues with IP vendors.
QD4 What level of design information can you provide for the security assessment process? (Possible
resources are listed in Table 4)
If the repository scheme is operational, we can provide the resources QD4-c and QD4-d.
QD4-b can be introduced, but it is necessary to examine issues such as the introduction burden of the
feature extraction tool regarding QD4-b.
QD4-a and QD4-b are necessary for the repository operation scheme. However, QD4-a is not available.
QD4-b needs to be considered. For others, we think that QD4-c or QD4-d can be provided.
QD5 What kind of business operators would use the repository scheme?
The communication infrastructure market is likely to have a high demand for security.
Table 6: Questions and responses to QE – risk measures in supply chains.
QE1 What are the risks in semiconductor supply chains?
Counterfeit products.
Geopolitical issues.
QE2 What measures do you take to avoid supply chain risks?
Purchasing semiconductor materials from multiple companies and decentralizing production bases.
We are responding to export control measures in each country.
Strengthening public relations, especially by enhancing the relationship with the government.
QE3 What will affect your business if vulnerability assessment in hardware devices becomes mandatory
or is standardized?
The concern is the burden on semiconductor manufacturers, such as the cost and measures taken.
If the vulnerability assessment is standardized, we respond if it is necessary for our business.
The major barriers to implementing a security as-
sessment process in semiconductor supply chains are
commercialization and showing the benefit of security
assurance. Although most companies are concerned
about security problems and interested in security as-
sessment tools, they do not clearly find the benefit of
security assessment. Considering the price competi-
tion of semiconductor products, the cost of security
is unacceptable. It would be difficult to change this
mindset in the semiconductor market. Therefore, if
security assessment is necessary, legal regulation is a
solution to solve the problem.
4.3 Discussion
To secure hardware supply chains, a method using a
blockchain or smart contract platform has been pro-
posed (Chang and Chen, 2020). In supply chain man-
agement, traceability, transparency, and stakeholder
involvement are major concerns. A platform using
blockchain or smart contracts is expected to address
the concerns. The major feature of the blockchain
technology is decentralization. This technology al-
lows peer-to-peer exchange or transactions, such as
digital currency, without trusted authorities. This
mechanism also applies to a supply chain context.
However, implementing a blockchain-based mecha-
nism faces several issues from the perspective of reli-
ability, throughput, and computational cost.
A supply chain management method using a
knowledge graph has been proposed (Zhang et al.,
2019). In this method, a knowledge graph is con-
structed to analyze the risks of a supply chain. Based
on the constructed knowledge graph, the connection
Vulnerability Information Sharing Platform for Securing Hardware Supply Chains
409
between business operators and the bottleneck in the
supply chain can be visualized. However, collecting
standardized and reliable information from multiple
business operators is difficult in the real world.
5 CONCLUSION
In this paper, we focus on the demand and aware-
ness of security assessment for hardware designs in
industries. First, we propose a repository opera-
tion scheme to share security assessment results with
inter-companies. Next, we design questionnaires to
confirm the needs, effectiveness, and concerns based
on the repository operation scheme and summarize
the results of the questionnaires.
From the survey, we found that HT detection
methods are attractive for protecting semiconductor
supply chains. However, the barriers to implementing
HT detection in industries lie in commercialization
and in cultivating awareness of security in the indus-
try. Specifically, identifying counterfeit products and
covering the whole supply chain are expected. For ex-
pectations, future work should cover the whole supply
chain and clarify the benefits of the scheme in indus-
tries.
ACKNOWLEDGMENTS
The questionnaire survey was conducted by Omdia,
a part of Informa Tech as a consulting project for
KDDI Research, Inc. in 2022. The copyright of
the original questionnaire survey belongs to Omdia.
The results reported in this paper were obtained from
“The Contract of Research for Detection Techniques
of Hardware Vulnerabilities” (Ministry of Internal Af-
fairs and Communication, Japan in FY2022).
REFERENCES
Almeida, F., Imran, M., Raik, J., and Pagliarini, S. (2022).
Ransomware attack as hardware trojan: A feasibility
and demonstration study. IEEE Access, 10:44827–
44839.
Chang, S. E. and Chen, Y. (2020). When blockchain meets
supply chain: A systematic literature review on cur-
rent development and potential applications. IEEE Ac-
cess, 8:62478–62494.
Dong, C., Liu, Y., Chen, J., Liu, X., Guo, W., and Chen,
Y. (2020). An unsupervised detection approach for
hardware trojans. IEEE Access, 8:158169–158183.
Francq, J. and Frick, F. (2015). Introduction to hardware
trojan detection methods. In 2015 Design, Automa-
tion & Test in Europe Conference Exhibition (DATE),
pages 770–775.
Hasegawa, K., Yamashita, K., Hidano, S., Fukushima, K.,
Hashimoto, K., and Togawa, N. (2023). Node-wise
hardware trojan detection based on graph learning.
IEEE Transactions on Computers, pages 1–13.
Hasegawa, K., Yanagisawa, M., and Togawa, N. (2017a).
Hardware trojans classification for gate-level netlists
using multi-layer neural networks. In Proc. IEEE In-
ternational Symposium on On-Line Testing and Ro-
bust System Design (IOLTS), pages 227–232.
Hasegawa, K., Yanagisawa, M., and Togawa, N. (2017b).
Trojan-feature extraction at gate-level netlists and its
application to hardware-trojan detection using random
forest classifier. In Proc. IEEE International Sympo-
sium on Circuits and Systems (ISCAS), pages 1–4.
Huang, Z., Wang, Q., Chen, Y., and Jiang, X. (2020). A
survey on machine learning against hardware trojan
attacks: Recent advances and challenges. IEEE Ac-
cess, 8:10796–10826.
Li, S., Zhang, Y., Chen, X., Ge, M., Mao, Z., and Yao, J.
(2020). A xgboost based hybrid detection scheme for
gate-level hardware trojan. In Proc. IEEE Joint Inter-
national Information Technology and Artificial Intel-
ligence Conference (ITAIC), volume 9, pages 41–47.
Oya, M., Shi, Y., Yanagisawa, M., and Togawa, N. (2015).
A score-based classification method for identifying
hardware-trojans at gate-level netlists. In 2015 De-
sign, Automation & Test in Europe Conference & Ex-
hibition (DATE), pages 465–470.
Salmani, H. (2017). Cotd: Reference-free hardware trojan
detection and recovery based on controllability and
observability in gate-level netlist. IEEE Transactions
on Information Forensics and Security, 12:338–350.
Tebyanian, M., Mokhtarpour, A., and Shafieinejad, A.
(2021). Sc-cotd: hardware trojan detection based on
sequential/combinational testability features using en-
semble classifier. J. Electron. Test., 37(4):473–487.
Xiao, K., Forte, D., Jin, Y., Karri, R., Bhunia, S., and Tehra-
nipoor, M. (2016). Hardware trojans: Lessons learned
after one decade of research. ACM Transactions on
Design Automation of Electronic Systems, 22.
Yang, Y., Ye, J., Cao, Y., Zhang, J., Li, X., Li, H., and
Hu, Y. (2020). Survey: Hardware trojan detection
for netlist. In 2020 IEEE 29th Asian Test Symposium
(ATS), pages 1–6.
Yu, S.-Y., Yasaei, R., Zhou, Q., Nguyen, T., and Faruque,
M. A. A. (2021). Hw2vec: a graph learning tool for
automating hardware security. In 2021 IEEE Interna-
tional Symposium on Hardware Oriented Security and
Trust (HOST). IEEE.
Zhang, W., Liu, Y., Jiang, L., Shah, N., Fei, X., and Cai,
H. (2019). The construction of a domain knowledge
graph and its application in supply chain risk analy-
sis. In IEEE International Conference on E-Business
Engineering, pages 464–478.
ICISSP 2024 - 10th International Conference on Information Systems Security and Privacy
410