AI-Driven Anomaly Detection and Zero Trust Frameworks

The Industrial Internet of Things (IIoT) 's convergence with hybrid cloud infrastructures transforms industrial operations through scalable, real-time data processing and automation. However, this evolution introduces complex cybersecurity risks, including broadened attack surfaces and dynamic trust boundaries. This chapter proposes a cybersecurity framework that integrates AI-driven anomaly detection with Zero Trust Architecture (ZTA) to address these challenges. The framework spans edge, fog, and cloud layers and leverages models like Autoencoders, LSTMs, and Isolation Forests for real-time, contextual threat detection. ZTA principles enforce microsegmentation, identity validation, and adaptive access control. Empirical use case based on the SWaT and WADI datasets validates the framework's efficacy in detecting cyber-physical anomalies. Additionally, the solution aligns with standards such as NIST 800-207, IEC 62443, and ISO/IEC 27001, offering practical guidance for secure, adaptive IIoT deployments.