Data Space Privacy Model with Federated Learning Technique for Securing IoT Communications in Autonomous Marine Vehicles
Taher M. Ghazal, Mohammad Kamrul Hasan, Kola Narasimha Raju, Muhammad Attique Khan, Ahmad Sami Al‐Shamayleh, Mohammed Wasim Bhatt, Munir Ahmad
Abstract
The deployment of the Internet of Things (IoT) raises significant concerns regarding the security and integrity of sensitive data communicated across various heterogeneous devices, resources, applications, and users in autonomous marine vehicles. The existing heterogeneous nature often struggles with the diversity of different device types' communication features, and the dynamic nature of IoT environments also compromises privacy due to unauthorized users' marine vehicle devices. A novel data space privacy model, supported by connected federated learning, is utilized for secure authentication in the communication of IoT marine vehicles, thereby enhancing security and privacy. The proposed model ensures user data privacy by enabling localized model training using synchronous connectivity for resource verification and data space privacy and reducing the risk of data exposure. The learning is periodically updated with the synchronous fitness value to retain similar authentication throughout the sharing intervals. By incorporating real-time resource verification, optimizing resource allocation, and mitigating communication drops. The results demonstrate that this proposed model reduces communication drop by 11.21% across varying users, 12.06% regarding sharing space value and time, 11.9% of transaction values, and 12.17% across multiple devices, with an improved authentication success rate and data integrity compared to existing models. The research implications offer valuable insights for developing more secure IoT communication protocols and policy frameworks that enhance data privacy and authentication in marine IoT networks.