Balancing data privacy and sharing in IIoT: Introducing the GFL-LFF aggregation algorithm

Abstract

As the Industrial Internet of Things (IIoT) continues to evolve, the need for effective data aggregation schemes that balance the imperatives of data sharing and individual data privacy becomes paramount. Traditional data aggregation methods often entail risks of sensitive information exposure during data transmission and storage. This paper proposes a novel Graphical Federated Learning-based Local Fennec Fox (GFL-LFF) algorithm to ensure secure aggregation by minimizing the revelation of individual data. In this study, a Graph Neural Network (GNN) with federated learning is employed to enhance privacy protection for IIoT's data aggregation scheme. A Local search-based fennec fox optimization is utilized for tuning the parameters of GNN. The GFL and LFF approaches are integrated to enhance the security of the data that can be shared through the IIoT system. The MNIST dataset, CIFAR-10 dataset, and LFW dataset were taken to conduct the experimentation results, and the evaluation measures, such as throughput, end-to-end delay, network lifetime, latency, and energy consumption, are utilized to evaluate the proposed GFL-LFF method, and these results are compared with other approaches. The GFL-LFF method achieved a throughput of 0.98 Mbps, an end-to-end delay of 1.2 s, a network lifetime of 5610 rounds, a latency of 1.6 s, and an energy consumption of 0.2 mJ. The experimental results illustrate the effectiveness of the proposed GFL-LFF method for IIoT's data aggregation scheme in privacy protection.