Integrated blockchain-federated learning framework for IOT security

Abstract

The rapid expansion of the Internet of Things (IoT) introduces complex, multi-layered security challenges that centralized security models and conventional intrusion detection systems struggle to address. This paper presents a multi-layer security framework that integrates Blockchain for decentralized trust management, Federated Learning (FL) for privacy-preserving intelligence, and Generative Adversarial Networks (GANs) for class-imbalance mitigation, with each technique mapped to a specific IoT architectural layer. The framework is evaluated on the NSL-KDD intrusion detection dataset, using GAN-based augmentation to improve detection of rare attack classes, and is further validated through case studies in healthcare and smart-agriculture deployments. The proposed GAN-FL-Blockchain model achieves 95% detection accuracy, 94% precision, 96% recall, and a 94.5% F1-score, improving detection rate by 13% and reducing latency by nearly 50% (150 ms vs. 300 ms) relative to a SMOTE+SVM baseline. These results demonstrate a scalable, regulation-compliant approach to layered IoT security. Breach identification, ML framework development, and integration of individual modules in a seamless and synchronized manner. Tailored countermeasures, ML integration, experimental validation with the auto encoder-based anomaly detection, integration of Graph Neural Network with the Generative Adversarial Network and Federated Learning procedure. Enhanced breach detection, practical implementation guidance with the validation accuracy value of 98%. The integrated model will be targeted to have overall latency minimization in the future.