Detecting Hardware Trojans Using Structural Features for Hardware Security and Reliability
Mostafa M. Helmy, Mohamed A. Y. Abdalla, Ahmed K. F. Khattab · 2023
Hardware security is a very important aspect in systems design and manufacturing. Globalization allows third parties to contribute to the design of some ICs in a large system-on-chip (SoC). This gives untrusted third parties the opportunity to implant malicious hardware trojan (HT) in their design. This paper proposes a complete framework to detect HTs at the gate-level netlist using machine learning techniques. The paper presents efficient techniques to address the complexity of extracting the structural features of the gate-level netlist, which are used as the dataset for the machine learning model. We exploit eXtreme Gradient Boosting (XGBoost) machine learning as a classifier for detecting HTs. We apply the XGBoost classifier to the gate-level benchmarks in Trust-HUB which result in a high true positive rate (TPR) of 99.3%.