Answers are grounded in this paper and cite their sources.
Accurate channel estimation (CE) is critical for ensuring information transmission reliability and flight safety in unmanned aerial vehicle (UAV) systems. Nevertheless, the high mobility of UAVs results in the rapid variations in wireless communication scenarios, which significantly degrades the CE accuracy in UAV-assisted wireless communication systems. Although deep learning (DL)-based CE methods improve estimation accuracy to some extent, they still encounter severe challenges in highly dynamic UAV communication scenarios, as new environments require modifications to neural network (NN) architectures and retraining of network parameters. To tackle these challenges, a perception-assisted CE enhancement method is proposed in this paper. Based on the perception of the number of echo paths, this method predicts the processing capability of the NN used in DL-based CE methods, thereby forming a branch processing framework. By using the residual-enhanced CE network (ReCENet) as an example, this branch framework enhances the NN’s processing adaptability in highly dynamic UAV communication scenarios by employing the trained ReCENet directly or applying it iteratively. Consequently, without the need for architectural modifications, network retraining, or hyperparameter tuning, the proposed method adaptively aligns the processing capability of the trained ReCENet so as to improve CE accuracy. The experimental results demonstrate that, compared with baseline methods, the proposed method significantly enhances the CE accuracy in UAV-assisted wireless communication systems. Under parameter variations, the proposed method demonstrates robustness.
No takes yet. Share an insight, caveat, or question.
Qing et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: