Date of Publication : September 2026

Abstract: The rapid growth of the Industrial Internet of Things (IIoT) systems has changed the landscape of automation and smart manufacturing and has presented intricate security issues. Traditional intrusion detection systems are unable to accommodate the dynamism, heterogeneity, trust sensitivity of IIoT systems. This review discusses the development of trust-sensitive artificial intelligence (AI) models that increase the detection accuracy using behavioral trust modeling, relationship analysis, and federated learning. The main goal is to analyse the potential to enhance AI-based detection models using trust metrics which are based on the reputation of the device, social relations, and interaction patterns to prevent malicious actions and data poisoning. A variety of benchmark datasets is tested to model a wide range of IIoT security scenarios: the SIoT Network is a social IoT interaction and topology-based trust construct; the WSN Secure and Energy-Efficient one focuses on the energy optimization and data integrity balance; Edge-IIoTset focuses on centralized and federated building designs; ToN-IoT provides a combination of protocol-rich traffic and multiple attack vectors; and BoT-IoT provides large-scale botnets as a way to benchmark. It was found in the analysis that the combination of trust evaluation mechanisms and AI-based intrusion detection does contribute to accuracy, energy efficiency, and interpretability significantly. These methods can be shown to be highly promising in edge deployments and federated deployments where scalability and privacy are of key concern. The authors further find that the dynamism, explainability, and trustfulness of AI security frameworks are vital to achieve robust security protection against the Industry 4.0 and next-generation IIoT ecosystems.

Keyword : Trust-Aware AI, Industrial IoT Security, Intrusion Detection Systems, trust metrics, Federated Learning, Edge Computing, Social IoT Networks.

Reference :