Channel Measurements and Characterizations for Low-Altitude Communications via an AI-Empowered Multi-Node Sounding System
Kai Mao, Hangang Li, Qiuming Zhu, Hanwen Xu, Zhangfeng Ma, Boyu Hua, Petros S. Bithas, Qihui Wu
Abstract
Low-altitude (LA) intelligent network is a key component of integrated space-air-ground networks, where understanding radio propagation channels is crucial for the design and optimization of reliable communication links. In this paper, we present a high efficient and low-interference LA channel measurement scheme for multiple nodes. To avoid self-interferences and cancel cross-interference, a series-parallel switching method is developed, where sounding signals are specially designed via a genetic algorithm (GA). It can effectively reduce the multi-node interference and improve the efficiency. In order to reduce the data size of multi-link channels and release the data transmission and postprocessing burden, a real-time data reduction method is proposed, which can extract the valid multi-path components (MPCs) and small-scale fading (SSF) characteristics. On this basis, a prototype of four-node LA channel measurement system is implemented. The prototype is applied to perform LA channel measurements in a campus scenario, and measured channel characteristics for air-to-air (A2A), air-to-ground (A2G), and ground-to-ground (G2G) links are presented. Then, the deep neural network (DNN) and the variational autoencoder (VAE) are utilized to characterize the LA channels, which can effectively predict and generate new channel characteristics for real applications.