Identification and Localization of Quantum Electromagnetic Fields of Hardware Trojan Attacks Using QDM-Based Unsupervised Deep Learning

Ensuring the reliability and trustworthiness of elec-tronic systems heavily relies on maintaining the integrity and se-curity of the semiconductor integrated circuit (IC) supply chain. Traditional hardware trojan detection methods often depend on side-channel analysis, which presents a drawback due to the limited availability of golden references. Although golden-free approaches can detect trojans, they lack automatic localization capabilities, necessitating complex reverse engineering of the entire IC. This paper presents a potential approach to efficiently identify points of interest (POIs) for trojan detection in ICs using Quantum Diamond Microscope (QDM) image analysis and unsupervised deep learning, without relying on golden references. The proposed method minimizes the need for extensive reverse engineering and offers a promising direction for future research in hardware trojan detection. By leveraging QDM magnetic field images, this approach has the potential to enhance the trustworthiness of the semiconductor IC supply chain through targeted trojan identification.