Graph Centrality Algorithms for Hardware Trojan Detection at Gate-Level Netlists

The rapid growth in the supply chain of electronic devices has led companies to purchase Intellectual Property or Integrated Circuits from unreliable sources. This dispersion design fabrication stages IP/IC new attacks called hardware Trojans. Hardware Trojans can bargain information, reduce performance, cause failure. Various methods have been introduced detect prevent Machine learning are one these. Selecting type and number input variables algorithm an important role performance model. Some previous Trojan detection studies used structural gate-level features create data sets for machine models. In this paper, a method based on directed graphs extracting is proposed. proposed use Graph Centrality Algorithm features. To examine importance impact extracted with method, three types created as model made XGBoost. trained models these show that graph-based improved F1-score by 10% ROC 22%. combination 17.5% 38.5%.