Bridging Terrestrial and Non-Terrestrial Networks: A Novel Architecture for Space-Air-Ground-Sea Integration System
Xin Wang, Bo Yi, Saru Kumari, Chien‐Ming Chen, Shalli Rani, Keqin Li, Jianhui Lv
Abstract
With the rapid evolution of wireless technology and the expansion of human activities, the integration of terrestrial and non-terrestrial networks, encompassing 5G/6G and the Internet of Things (IoT), is becoming essential for future networking paradigms. These integrated networks must support extensive spatial and content coverage, serving diverse environments from urban landscapes to remote regions such as mountains, deserts, oceans, underground areas, and airspace. While 5G/6G technologies offer significant improvements, their widespread deployment faces challenges, including high infrastructure costs and difficulties in covering extremely remote or inaccessible areas. To address these challenges, we design and propose a novel space-air-ground-sea (SAGS) integration architecture that builds upon terrestrial networks and supplements them with non-terrestrial networks, aiming to provide ubiquitous, intelligent, collaborative, and efficient information support across vast spatial domains. Our approach focuses on three critical aspects: global situation awareness, leveraging reinforcement learning, graph convolutional networks, and multimodal data fusion to enhance situational awareness and decision-making; reliable transmission, ensuring robust data transmission by mitigating environmental conflicts and optimizing communication pathways across space, air, ground, and sea; and dynamic time-varying scheduling, formulating a multi-objective scheduling optimization model to minimize uncovered areas, energy consumption, and operational spans, adapting to the time-varying nature of services in the SAGS environment. Key contributions of this work include a comprehensive SAGS architecture that integrates advanced AI techniques to optimize network performance and experimental validation, demonstrating that our proposed SAGS outperforms state-of-the-art methods by specific percentages in terms of convergence efficiency, latency, and throughput, which highlights the system's feasibility and effectiveness.