Intelligent Spectrum Allocation for TN-NTN Integration: A Q-Learning Approach

Sixth-generation (6G) wireless networks integrate terrestrial networks (TN) with non-terrestrial networks (NTN), including satellites and aerial platforms, to enable global connectivity. However, dynamic topology variations, interference, and heterogeneous quality-of-service requirements significantly complicate efficient spectrum allocation. This paper proposes a self-directed channel assignment framework based on Q-learning, where agents learn optimal policies by observing network load, interference conditions, and temporal traffic dynamics within a Markov decision process. A multi-objective reward function is designed to jointly optimize system throughput, user fairness, and interference mitigation, while an ϵ-greedy strategy facilitates effective exploration. Simulation results over 5,000 training episodes with 10 channels, 15 TN, and 8 NTN users demonstrate stable convergence, achieving an average reward of 37.5 and a throughput of 28.5 Mbps. Moreover, the proposed approach achieves Jain's fairness index of 0.75 and reduces interference by 26.3% compared to random allocation through adaptive response to dynamic traffic patterns.