An optimal transmit power allocation scheme using UAV position estimation in MmWave NTNs

Non-terrestrial networks (NTNs) typically consist of UAV swarms equipped with multiple sensors that generate massive data, requiring real-time communication to a gateway for further processing. Hence, millimeter-wave (mmWave) communication technologies operating above 24 GHz emerge as a suitable solution for enabling high-speed intra-UAV swarm communication. However, mmWave communication technologies use multiple antenna-based directional beamforming for improved coverage, which leads to higher power consumption and frequent beam training overhead, affecting swarm endurance. To address this, we propose a novel optimal transmit power allocation scheme that enhances swarm endurance and improves throughput by reducing beam training overhead. Firstly, the proposed scheme uses the Kalman filter (in this paper, but not limited to) at the transmitting UAV to estimate the real-time position of the receiving UAV. The estimated position is utilized to calculate path loss and select the optimal transmit power level needed to meet the required received signal power threshold at the receiving UAV. To reduce outages from errors in UAV position estimation, the proposed scheme also adjusts the transmit power level by incorporating an additional buffer distance around the estimated position, thereby enhancing reliability with a minimal increase in transmit power. The performance analysis shows that the proposed scheme achieves an average reliability of more than 99% and power savings of up to 49.4% while increasing the throughput under saturated traffic conditions, thus establishing its effectiveness in mobile UAV swarms. Also, the proposed scheme is compared with three popular mechanisms existing in the literature: 1) baseline approach where constant transmit power is utilized, 2) deep learning (long short-term memory, LSTM) based transmit power allocation, and 3) power allocation using $\alpha -\beta -\gamma$ filter for receiving UAV position estimation. The performance comparison shows that the proposed scheme offers superior performance in terms of power savings, reliability, throughput, and computational complexity.