Visible to the public Property Based Formal Security Verification for Hardware Trojan Detection

TitleProperty Based Formal Security Verification for Hardware Trojan Detection
Publication TypeConference Paper
Year of Publication2018
AuthorsQin, Maoyuan, Hu, Wei, Mu, Dejun, Tai, Yu
Conference Name2018 IEEE 3rd International Verification and Security Workshop (IVSW)
Date PublishedJuly 2018
PublisherIEEE
ISBN Number978-1-5386-6544-2
KeywordsComplexity theory, composability, Coq, fine grained gate level information flow model, formal representations, formal security verification, formal specification, formal verification, formal verification method, golden reference design, Hardware, hardware designs, hardware security, hardware trojan, industrial property, information flow analysis, Integrated circuit modeling, intellectual property security, invasive software, Logic gates, malicious Hardware Trojans, modern computer hardware, policy-based governance, pubcrawl, register transfer level information flow security models, resilience, Resiliency, secret information, security, security behavior, security properties, Semantics, theorem proving, third-party intellectual property cores, Trojan horses
Abstract

The design of modern computer hardware heavily relies on third-party intellectual property (IP) cores, which may contain malicious hardware Trojans that could be exploited by an adversary to leak secret information or take control of the system. Existing hardware Trojan detection methods either require a golden reference design for comparison or extensive functional testing to identify suspicious signals. In this paper, we propose a new formal verification method to verify the security of hardware designs. The proposed solution formalizes fine grained gate level information flow model for proving security properties of hardware designs in the Coq theorem prover environment. Compare with existing register transfer level (RTL) information flow security models, our model only needs to translate a small number of logic primitives to their formal representations without the need of supporting the rich RTL HDL semantics or dealing with complex conditional branch or loop structures. As a result, a gate level information flow model can be created at much lower complexity while achieving significantly higher precision in modeling the security behavior of hardware designs. We use the AES-T1700 benchmark from Trust-HUB to demonstrate the effectiveness of our solution. Experimental results show that our method can detect and pinpoint the Trojan.

URLhttps://ieeexplore.ieee.org/document/8494858/
DOI10.1109/IVSW.2018.8494858
Citation Keyqin_property_2018