Robust beam control for terahertz drone networks

Official URL: https://dx.doi.org/10.1109/TAES.2025.3581533

Ensuring reliable, high-capacity links in Terahertz (THz) drone networks is challenging due to the highly directional nature of THz signal transmission, which utilizes narrow pencil-like beams to mitigate high path loss. This paper investigates the challenge of beam control for drone-to-drone links in the 0.5-1 THz frequency range, where movement and environmental factors can hinder reliable beam alignment between the communicating drones. We consider planar antenna arrays employed on the drones to achieve narrow, asymmetric beams in 3D and introduce a sectored antenna gain model to control beamwidth by dynamically adjusting the number of active antenna elements. For addressing drone mobility uncertainties, we first propose a beamwidth optimization method using an Upper Confidence Bound (UCB) approach that maximizes the average gain. We, then, propose a Contextual Multi-Armed Bandit (CMAB) framework for beam control that combines beam steering with beamwidth optimization to enhance link performance. Our simulations considering real drone mobility scenarios show that UCB achieves over 10 times higher capacity, in both obtained maximum and average ergodic capacity results, as compared to the fixed beamwidth technique. Moreover, CMAB provides tens of Gbps ergodic capacity, outperforming beam steering via tracking approach. All presented performance results reveal the effectiveness and significance of the proposed robust beam control for next-generation drone networks.