Federal Hierarchical Mutual Assistance Based on Maritime Intelligent Decision Recommendation

With the rapid development of artificial intelligence, edge computing and distributed collaborative processing, more effective solutions have been provided to solve the problems in the field of maritime communications. The surge in maritime activities and the growing number of vessels have precipitated an escalating demand for the real-time processing of oceanic data. However, the vast ocean environment and limited communication resources make the management and intelligent decision-making of heterogeneous data complex. In handling the collaborative processing of large-scale maritime fleet tasks, such as emergency responses to marine pollution, the network nodes of individual vessels are dispersed, and data timeliness is poor, which significantly impacts decision-making for ship operators. Additionally, the information exchanged between cooperating vessels must maintain privacy, making it difficult to develop personalized intelligent communication strategies. To address these potential issues, we propose a Federated Hierarchical Mutual Assistance (FHMA) algorithm for maritime recommendation systems. By integrating federated learning with layered privacy protection techniques, this approach ensures data privacy while also enabling model personalization. Additionally, it introduces compute power sharing and reward distribution mechanisms, incentivizing client participants to actively contribute their computational resources, thereby improving network efficiency. This algorithm not only efficiently handles large-scale, dynamically changing spatiotemporal data but also maintains high-quality recommendation results in parallel and distributed data processing environments, providing new directions for intelligent decision-making in the maritime communication field.