A Distributed Framework for the Ocean IoT Network

The rising adoption of IoT devices in marine environments has led to an increased interest in edge computing research. Distributed Learning (DL) has emerged as a key solution for addressing privacy concerns in distributed edge computing. However, existing DL frameworks may not fully account for resource limitations and unique maritime IoT network topologies. In this paper, a novel Federated Learning (FL) framework for marine wireless communications was proposed, focusing on modeling the communication environment, designing a practical wireless Ocean Wireless Federation (OWF) framework, and optimizing resource allocation. We develop a comprehensive wireless signal propagation model for marine environments and design an iterative FL algorithm that addresses marine communication challenges. Additionally, we propose a resource scheduling and allocation scheme for efficient bandwidth, energy, and computation utilization. Extensive experiments validate the properties of our OWF algorithm and demonstrate superior performance in accuracy, resource utilization, and convergence speed for practical network parameters.