Federated Learning-based Secure Data Dissemination Framework for IIoT Systems Underlying 5G

This paper introduces a novel framework for enhancing the security and privacy of Industrial Internet of Things (IIoT) systems using federated learning. Addressing critical security concerns inherent in IIoT environments, the framework employs a decentralized approach to machine learning model training across multiple programmable logic controllers. By decentralizing the training process, sensitive data is preserved, ensuring privacy while mitigating risks associated with cyber threats such as Denial of Service and data breaches. Extensive empirical validation demonstrates significant improvements in model accuracy and robustness against various attacks, with training accuracies ranging from 88.14% to 89.51% and validation accuracies ranging from 88.20% to 89.48%. These results underscore the practical applicability and effectiveness of the proposed approach in enhancing IIoT system security.