XG Boost Algorithm based Hardware Trojan Detection in Hardware Circuits

Abstract

Hardware security is a protection that comes in the form of a physical device. A hardware security module provides security for sensitive data. Hardware trojan is an intentionally altered circuit of an Integrated Circuits chip. It may be during the chip formation or fabrication that is the chip is widened without the probable knowledge of the person who designed it. The Trojan leaks the information and malfunctions during crucial operations to getting around security precautions. They can also capture keystrokes or record the passwords entered by the users frequently. The trojans can be detected effectively by developing the machine learning models. Machine learning algorithms can achieve much higher accuracy while making predictions or classifying labelled data. XGBoost in machine learning algorithm is incredibly quick as it uses tree-based models and strives for best class accuracy. The feature set is trained using the XGBoost algorithm, which classifies the circuit into suspected Trojan circuits and regular circuits. The accuracy score achieved while making predictions using the Gradient Boosting algorithm is 99.43%.