Machine Learning–Based Hardware Trojans Detection in Integrated Circuits: A Systematic Review

A purposefully inserted additional circuit known as the Hardware Trojan (HT) is implanted inside original integrated circuits during the designing or manufacturing stages. It has the potential to manipulate circuit performance or acquire underlying information. Due to machine learning’s (ML) exceptional results across a range of learning domains, the academic and business community are now looking at how Hardware Trojan (HT) attacks can be strengthened by employing conventional methods. Only a few survey studies have thoroughly evaluated the achievements and covered the unresolved issues in this subject. The literature for methods of defining HT concerns centered on machine learning is being reviewed in this research. Specifically, we first classify all known HT attacks and later analyze the evolution of the latest machine learning models in five separate areas of HT detection: reverse engineering, side-channel analysis, and golden model-free analysis, circuit feature analysis and classification approaches. Based on the review, we analyze the lessons learned and obstacles that have emerged from prior investigations. HT Defense Studies discusses the pros and cons of Supervised and unsupervised ML. Finally, a comparison of machine learning-based and nonmachine learning–based HT detection approaches is shown and current challenges with future work are also suggested.