Using Path Features for Hardware Trojan Detection Based on Machine Learning Techniques

Chia-Heng Yen1, Jung-Che Tsai1, Kai-Chiang Wu2
1National Yang Ming Chiao Tung University, 2Department of Computer Science, National Chiao Tung University


As outsourcing processes of design and fabrication to third parties becomes common in the IC industry, the consciousness of hardware security has been rising these years. In this paper, we propose a method of hardware Trojan detection using specific path features at gate level. Path classifiers are trained with SVM and RF algorithms using the path features extracted from the circuit designs. An average of 0.96 on the F1-score in path classification results demonstrates that paths can be easily classified into Trojan paths and Trojan-free paths with the trained path classifiers. Moreover, the intersections between paths are favorable for locating Trojan gates precisely. We obtain high TPRs for locating the Trojan gates with the proposed scoring mechanism while keeping the FPRs low to prevent normal gates from misclassifying into Trojan gates.