Neural Network-based Vegetation Attenuation Measurement and Modeling for Air-to-Ground Channels

Jiajia Zhou, Ruonan Zhang, Xiaoqian Yi, Marie Siew, Bin Li, Daosen Zhai

Keywords: Vegetation Attenuation, Channel Measurement, Channel Modeling, Multilayer Perceptron, UAV

TL;DR: This paper conducts multi-scenario channel measurements and proposes a vegetation attenuation prediction model based on a MLP.

Abstract: Vegetation, such as roadside trees and landscape forests, can obstruct the line-of-sight path between unmanned aerial vehicles (UAVs) and ground mobile terminals, causing fading in air-to-ground (A2G) communication links and significantly impacting the data rate and reliability of UAV communication and radio control systems. Therefore, accurate prediction of vegetation-induced fading is essential for UAV deployment and applications. This paper develops an A2G large-scale channel fading measurement system and conducts measurements and modeling across various regions and types of vegetation. According to the measurement results, we propose a vegetation attenuation prediction model based on a multilayer perceptron (MLP) for estimating attenuation in different scenarios. Experimental results demonstrate that, compared with the ITU-R model, the proposed model exhibits high prediction accuracy, offering an effective tool for optimizing UAV communication in vegetated environments.

Submission Number: 27