AI at the core of 6G : Integrating multi-orbit LEO satellites and terrestrial networks

초록

The integration of multi-orbit Low Earth Orbit (LEO) satellites with 6G Terrestrial Network (TN) represents a transformative milestone in achieving ubiquitous, reliable, and scalable global connectivity. Leveraging the extensive coverage and low-latency capabilities of LEO satellites, this study examines critical challenges in Non-Terrestrial Networks (NTNs), including frequent handovers, spectrum coexistence, interference, and efficient resource allocation. Artificial Intelligence (AI) emerges as a key enabler, employing advanced techniques such as predictive modeling with Large Language Models (LLMs), Service Continuity Graphs (SCG), Reward-Based Reinforcement Learning (RBRL), and Prioritized Experience Replay (PER). This paper provides a comprehensive review of current advancements, identifies unresolved challenges, and proposes research directions for scalable and robust AI-driven NTN frameworks. By emphasizing collaborative innovations and software-defined implementations, the study outlines a roadmap for enabling an efficient 6G NTN infrastructure, ensuring global connectivity and enabling next generation applications such as Internet of Things (IoT), Augmented Reality (AR)/Virtual Reality (VR), and autonomous systems.