Genetic Algorithm for hardware Trojan detection with ring oscillator network (RON)

Securing integrated circuits against malicious modifications (i.e., hardware Trojans) is of utmost importance, as hardware Trojans may leak information and reduce reliability of electronic systems in critical applications. In this paper, we use ring oscillators (ROs) to gather measurements of ICs that may contain hardware Trojans. To distinguish between Trojan-inserted ICs and Trojan-free ICs, we investigate several classification approaches. Furthermore, we propose a novel feature selection approach based on the Genetic Algorithm (GA) and evaluate its performance compared to several popular alternatives. The proposed method is an improvement over principal component analysis (PCA) in terms of accuracy and equal error rate by 30% and 97% respectively.