2DLAM: Joint Delay-Doppler Estimation in UAV mmWave System via Large AI Model
Keywords: Channel estimation, Vision Transformer, UAV mmWave systems
Abstract: Title: 2DLAM: Joint Delay-Doppler Estimation in UAV mmWave System via Large AI Model
Abstract: Unmanned aerial vehicle (UAV) has been recognized as a promising platform for fulfilling high-rate data communication in the sixth-generation (6G) wireless networks, due to the benefits of strong line-of-sight (LoS) link probability, controlled mobility, and on-demand deployment. In UAV-enabled communication systems, channel estimation plays a crucial role and has been facing increasing challenges, especially in high frequency millimeter-wave (mmWave) band. This is because the channel coherence time in UAV mmWave systems will be significantly shortened by the severe delay and Doppler effect arised from high-speed movement of the UAV, making the traditional channel estimation approaches designed for low-speed scenarios less applicable. To address this issue, in this paper, we propose a novel channel estimation algorithm suitable for UAV-enabled communication systems, in which a terrestrial base station equipped with a large-scale array communicates with a UAV target of high mobility. In particular, the proposed algorithm leverages the large artificial intelligence model (LAM) to jointly estimate the time-varying parameters in the delay-Doppler domain within the shortened coherence time, thus improving the accuracy of channel estimation in high-speed scenarios. More specifically, we characterize the time-varying channel matrices as two-dimensional (2D) images, thus allowing us for incorporating the pre-trained imageGPT (iGPT) model to handle the joint delay and Doppler parameters estimation. Through comparisons with various benchmarks, we demonstrate how our proposed algorithm can accurately estimate the delay and Doppler parameters in the considered UAV-enabled communication systems with a relatively small training cost.
Submission Number: 8
Loading