A Review: Machine Learning Based Hardware Trojan Detection

Abstract

The Hardware Trojan (HT) is a deliberately inserted extra circuit, which is placed inside the integrated circuits (ICs) during the design or production phase. It has the ability to extract hidden information or alter circuit activity. Different methods have been designed to deter and identify these malicious circuits so far, based on traditional or machine learning approaches, in order to limit the dangers connected with them. Machine learning's (ML) outstanding performance in a number of learning disciplines has prompted academic and commercial communities to investigate its capability to enhance hardware Trojan (HT) attacks. Although several publications have been released over the last ten years, a couple of survey articles have thoroughly assessed the accomplishments and discussed the remaining concerns in this field. This paper reviewed the literature for approaches defining HT threats based on machine learning. In specifically, at first different machine learning models evolved so far are discussed and later these models are analyzed for HT detection mainly in three different domains a) reverse engineering b) circuit feature analysis and c) side channel analysis.