Abstract : In this paper, we present a novel approach to enhance the throughput of 6G Non-Terrestrial Networks (NTN) by incorporating deep learning-based channel estimation, Doppler pre-compensation, and compensation techniques. We propose a new framework for accurate and efficient channel estimation in 6G-NTN systems, leveraging neural networks to improve channel estimation performance, leading to enhanced throughput and link performance. Furthermore, we introduce Doppler pre-compensation and compensation techniques to address the challenges posed by high mobility scenarios in 6G-NTN. Extensive simulations demonstrate the effectiveness of our approach, showing significant improvements in mean squared error (MSE), throughput, and robustness to Doppler effects under high mobility scenario in NTN systems. The training data for the convolutional neural network (CNN) model, developed specifically for DM-RS channel estimation, demonstrates a Mean Squared Error (MSE) of 1.4175 at a transonic speed of 1,000 km/h and an altitude of 10 km in the NTN environment. The implementation of both Doppler pre-compensation and compensation techniques effectively neutralizes the Doppler shift. This results in a comparable bit error rate (BER) performance, achieving link reliability with a spectral efficiency of 3.325 bps/Hz at an NTN mobility of 1,000 km/h and an altitude of 10 km. The proposed framework has the potential to significantly impact the performance of 6G-NTN systems, paving the way for reliable and efficient wireless communication in challenging environments.
Index terms : 6G, Non-terrestrial networks (NTN), Deep Learning, Channel Estimation, Doppler-pre Compensation, Unmanned Aerial Vehicle