Gradient-based optimization of terrestrial cellular networks for aAerial highways

Unmanned aerial vehicles (UAVs), owing to their ability to navigate 3D space and encounter line of sight (LoS) conditions, often experience high channel gains across multiple cells. This phenomena results in comparable signal power from various neighboring cells, causing significant interference and reduced signal-to-interference-plus-noise ratio (SINR), which complicates their integration into terrestrial networks. In this paper, we investigate how prior knowledge of planned aerial highways (AHs) can be leveraged by network operators to efficiently optimize their infrastructures for seamless UAV connectivity. In particular, we propose a gradient-based optimization algorithm to adjust the vertical tilts of 4G long term evolution (LTE) cells, ensuring optimal power distribution between ground and sky, while meeting minimum connectivity requirements along the AHs. Moving towards 5G new radio (NR), we also investigate how multi-antenna systems can be leveraged, together with the information of the planned AHs, to provide further enhanced coverage and performance. In this line, we propose an optimization framework that utilizes gradient computations to design optimal coverage and data beams from NR network cells. Simulation results show significant enhancement in coverage, SINR, and data rates for UAVs. Specifically, considering LTE, our approach yields up to 7 dB in SINR along with a fivefold increase in rates, allowing to satisfy requirements and eliminate coverage gaps. While within NR, our solution achieves up to 13 dB SINR gains and a sixfold improvement in UAV data rate, thereby overcoming previous-generation limitations and further improving aerial connectivity. These findings provide valuable recommendations for network operators aiming to support UAVs services with existing terrestrial infrastructures.