A Novel Zero-Trust–enabled Threat Intelligence Framework for Secure Cyber–physical Systems

Abstract Cyber–Physical Systems (CPS) have emerged as critical infrastructure enabling smart cities, intelligent transportation, industrial automation, and connected healthcare. However, the integration of heterogeneous devices, legacy components, and high-speed communication surfaces significantly increases vulnerability to advanced cyber threats. Traditional perimeter-based security architectures are insufficient to handle multi-vector attacks, supply-chain compromises, and insider threats. This study proposes a novel Zero-Trust–Enabled Threat Intelligence Framework (ZT-TIF) designed to continuously validate access requests, enforce micro-segmentation, and integrate real-time adversarial behavior analytics. A hybrid machine learning model combining Bi-LSTM and Random Forest (RF) is employed to detect anomalies and predict attack patterns without relying on static signatures. The framework is evaluated using the ToN-IoT and UNSW-NB15 datasets, demonstrating improvements in detection accuracy, false-positive reduction, and scalable policy enforcement. Additionally, comparative analysis (Tables 1–3) shows ZT-TIF outperforming existing Zero Trust and behavioral detection systems.