A UAV Air-to-Ground Channel Estimation Algorithm Based on Deep Learning

Unmanned Aerial Vehicles (UAVs) with mobility and flexibility enhance wireless transmission performance in various mobile communication scenarios by acting as a mobile base station or relay. However, the high-speed movement of UAV results in the difficulties of channel estimation because of the fast time-varying channel. In this paper, we propose a novel channel estimation algorithm based on Long Short-Term Memory (LSTM) for UAV air-to-ground transmission to obtain Channel State Information (CSI). To estimate the current slot CSI, we construct the input, forget, and output gates to learn the time correlation of UAV channel. We also define a memory function to formulate the useful information retained by the forget and the input gates, in which the forget gate discards the previous slot CSI and the input gate updates received signal of the current slot. The current slot CSI is estimated through the memory function and output gate. Compared with Least Square (LS) and Minimum Mean Square Error (MMSE) algorithm, the simulation results show that the proposed algorithm obtains more accurate CSI and higher robustness in different UAV mobile scenarios.