Event
CALCE Webinar - A PHM Solution for Zero-Trust Hardware
Tuesday, July 14, 2020
11:00 a.m.-12:00 p.m.
Abstract
Software attacks are becoming more frequent across a wide range of industries. However, not all attacks on a system are caused by or related to software; the physical components of a system can also be compromised and in some cases may create anomalies that could appear as a cyber-attack. This is especially a concern because, for most systems produced today, components, manufacturing, and assembly are often outsourced to offshore facilities. The result is that anomalous behavior can be due to defective components, counterfeit components, or Trojan horses, where there is a malicious insertion of a Trojan that can be triggered. Process reliability Trojans are a new class of Trojans that alter neither the circuitry nor functionality of an Integrated Circuit (IC), but they negatively impact IC reliability through acceleration of wearout mechanisms for transistors. Process reliability Trojans are introduced through malicious modifications of IC fabrication process steps and can result in ICs with a much shorter lifetime. These modifications are extremely difficult to detect, as they are made at the wafer level and only manifest as shortened lifetimes of infected ICs. Manufacturing at trusted foundries is a safe, although expensive option and hence, there is growing interest to move away from trusted foundries towards operation and development in zero-trust environments. Zero-Trust Architecture is a concept that helps prevent successful data breaches by eliminating the concept of trust from an organization’s network architecture and is designed to protect digital environments. To extend the concept of Zero-Trust Architecture from the network to the hardware domain, this paper lays out a deep-learning based approach to detect anomalous behavior in integrated circuits, that further distinguishes anomalous behavior into component faults or process reliability Trojans. This approach ensures that infected hardware is detected during manufacturing testing or during field usage, thus negating the necessity of relying on trusted foundries.
About the Presenter: Varun Khemani received his B.E. degree in Instrumentation Engineering from the University of Mumbai, and his M.S. in Industrial Engineering from North Carolina State University. He’s currently pursuing his Ph.D. in Reliability Engineering at the University of Maryland. He has work experience with Stanley, Black and Decker, Towson, MD and Aker, Mumbai, India. His research interests include reliability and cybersecurity of electronic circuits and EDA Modeling and optimization of high dimensional circuits. He’s a student member of IEEE, ASME, IEEE EDS, PHM Society, and ACES. He was the recipient of the ASME Petroleum Division Scholarship in 2016 and finalist for the Qualcomm Innovation Fellowship in 2019.
