Advanced Federated Learning for Integrated Terrestrial and Non-Terrestrial IoT Networks With Security by Design

The evolution of technologies and standardization in next-generation systems is essential for building a ubiquitous communication infrastructure, enabling the integration of air-space-ground-maritime (ASGM)-based architecture that unify non-terrestrial networks (NTNs) and terrestrial networks (TNs) across diverse layers. However, this integration introduces heterogeneity due to varying network frameworks, the increasing number of internet-of-things entities, and the expected surge in data volumes, which pose challenges in ensuring network security and maintaining resilience in hybrid dynamic network environments. To address these challenges, this paper first provides a comprehensive review and analysis of the heterogeneities and obstacles within the integrated ASGM system, and then presents a robust wireless federated learning framework designed to enable privacy-preserving service and secure aggregation to achieve resilient networks. This framework incorporates security-enhanced algorithmic strategies while considering spatial and temporal variations with device selection mechanisms to meet diverse network requirements, resulting in reliable collaborative learning across multiple network components for further improvement in privacy, security, and resilience in seamless NTN-TN integration.