An Accurate Beam-Tracking Algorithm with Adaptive Beam Reconstruction via UAV-BSs for Mobile Users

TL;DR

This work addresses real-time, high-precision beam tracking for mobile users in UAV-based mmWave networks by combining a cooperative GDCSA-based localization of U-UAVs, a GPR-driven MU-angle predictor, and a TIAM-enabled adaptive beam reconstruction. The proposed BAB-AR algorithm enables bi-directional angle-aware beam tracking with adaptive beam reconstruction, achieving angle prediction errors within $0.2$ for UAV-BS localization and MU-angle errors $\leq 0.025$, while delivering higher SNR and data rates than baseline methods. The key contributions are the GDCSA localization for GPS-free U-UAV positioning, the GPR-based MU angle prediction, and the TIAM for dynamic beam updates, all integrating into a two-stage beamforming architecture (2D UAV-to-UAV and 3D UAV-to-MU). The results show substantial gains in beam gain, energy efficiency, and resilience to mobility, highlighting practical impact for reliable, high-capacity UAV-enabled wireless networks in dynamic environments.

Abstract

Unmanned aerial vehicles (UAVs) with flexible deployment contribute to enlarging the distance of information transmission to mobile users (MUs) in constrained environment. However, due to the high mobility of both UAVs and MUs, it is challenging to establish an accurate beam towards the target MU with high beam gain in real-time. In this study, UAV base stations (UAV-BSs) consisting of position-known assisted UAVs (A-UAVs) and position-unknown assisted UAVs (U-UAVs) are employed to transmit data to MUs. Specifically, a bi-directional angle-aware beam tracking with adaptive beam reconstruction (BAB-AR) algorithm is proposed to construct an optimal beam that can quickly adapt the movement of the target MU. First, the angle-aware beam tracking is realized within the UAVBSs using a proposed global dynamic crow search algorithm without historical trajectory. Furthermore, the Gaussian process regression model is trained by A-UAVs to predict the azimuth and elevation angles of MUs. Meanwhile, we focus on the beam width and design a time interval adjustment mechanism for adaptive beam reconstruction to track high-speed MUs. Finally, the performance of the BAB-AR algorithm is compared with that of benchmark algorithms, and simulate results verifies that the BAB-AR algorithm can construct an accurate beam capable of covering high-speed MUs with the half power beam width in a timely manner.

Figures (20)

• Figure 1: Communication model between UAVs and MUs.
• Figure 2: Time frame of the BAB-AR algorithm.
• Figure 3: Flow chart of BAB-AR algorithm.
• Figure 4: $k$ A-UAVs jointly locating U-UAVs.
• Figure 5: Beam reconstruction with fixed time interval.