Hardware Trojan Detection and Localization through Digital Twin of the Integrated Circuits
Abstract
Due to the globalization of the Integrated Circuit (IC) manufacturing supply chain, untrusted stages introducing00 03/17/17hardwar""e Trojan has become a threat to the security (confidentiality, integrity, and availability) of mission criticalsystems (such as Nav""y ships, airplanes, etc.). Current Trojan detection and localization methodologies rely ontechniques that do not incorporate the fu""ll spectrum of data available throughout the life-cycle of the IC, and hence areincapable of providing an effective solution. There""fore, in this proposal, we are proposing a novel methodology ofcreating and maintaining a virtual representation of the IC which ca"n be used to both detect and localize the hardwareTrojan without having to either tamper the IC or stop the real time processing of" the IC when further testing andsimulations are required.This virtual representation, also known as the Digital Twin, takes advant""age of the large amount of data present duringthe manufacturing process, on-chip and off-chip sensors, as well as historical data f""rom previous version of the chips,or chips from different batches, to create a virtual model. This model is then constantly updated"" to detect presence ofknown Trojans, predict unknown Trojans, and localize the area of the Trojan for further diagnosis of the infe"cted IC tostrengthen the future IC. The major techniques in creating the Digital Twin lies in processing the large data (Big Data)"to acquire meaningful features (explore new ones as well), fuse multiple features, linking multiple models (functional,behavioral m""odels with the analog emissions such as power and thermal emissions) to create a model capable ofreasoning, that can analyze relati"on between data from various stages of the IC to detect anomalous behavior andlocation of the such behavior in the IC. For proof of" concept, we will demonstrate the applicability of our method inFPGA, and demonstrate how Digital Twins can aid in better detection"" of Hardware Trojans.If successful, the proposed methodology will be able to aid in effective Trojan detection and localization wit""hout theneed for intrusive or destructive testing procedure. Furthermore, a single Digital Twin may be able to provide data toothe""r Digital Twins to better localize the affected regions of the IC. Furthermore, since these Digital Twins evolvethroughout the life"#NAME?
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 05, 2017
- Source ID
- N000141712499
Entities
People
- Mohammad Al Faruque
Organizations
- Naval Information Warfare Center Pacific
- Office of Naval Research
- United States Navy