Single-triggered hardware Trojan identification based on gate-level circuit structural characteristics

Abstract

Feature-based hardware Trojan (HT) detection method has gained its popularity recently because it does not need any test patterns or golden models. However, HT feature database is to be expanded due to emergence of new HTs. We present in this paper a method based on the structural characteristics of single-triggered HTs. Various structure templates are extracted from gate-level netlists, and a scoring approach and an outlier determination algorithm are proposed. Gates with a high (outlier) score, i.e. highly matched to the templates, are considered as Trojan candidates. Experimental results show that the proposed method can detect 39 out of 40 single-triggered HTs from TrustHub, with low false positive rate (average 0.04%) and a time overhead of 0.02s ~ 2000s. The proposed method makes a good supplement to existing Trojan feature database.