A Survey on Reinforcement Learning Methods for Seamless Connectivity in 6G Non-Terrestrial Networks: Algorithms and Key Findings

Abstract

The vision of sixth-generation (6G) communications aims to achieve global hyperconnectivity by integrating terrestrial, aerial, and maritime domains. Non-terrestrial networks (NTNs) are key enablers of this vision but face inherent challenges such as high mobility, long propagation delays, and limited onboard computation. To address these challenges, reinforcement learning (RL) has emerged as a promising solution. This paper presents a comprehensive tutorial on RL-based methods for seamless 6G NTN connectivity. It adopts a dual perspective: (i) how RL addresses core NTN problems such as handover, beamforming, and resource allocation and (ii) how NTN constraints inspire the development of advanced RL techniques, including quantum RL, graph-based RL, and federated RL. Based on Third Generation Partnership Project (3GPP) Technical Report (TR) 38.811, the paper identifies 14 optimization challenges across four categories of NTN characteristics—platform dynamics, altitude-dependent propagation, system design constraints, and network architecture—surveys state-of-the-art RL approaches, and analyzes their suitability under real-world NTN conditions. It further discusses performance evaluation results and future research directions, identifying open problems and emerging paradigms that will shape the next phase of intelligent NTN systems. By bridging the gap between theory and practical deployment, this work positions RL as a foundational technology for intelligent, adaptive, and resilient 6G NTN systems.