Cellular Network Design for UAV Corridors via Data-driven High-dimensional Bayesian Optimization

Mohamed Benzaghta, Giovanni Geraci, David López-Pérez, and Alvaro Valcarce · 2025

Read Cellular Network Design for UAV Corridors via Data-driven High-dimensional Bayesian Optimization by Mohamed Benzaghta, Giovanni Geraci, David López-Pérez, and Alvaro Valcarce (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-05176.txt.

Introduction

Mohamed Benzaghta and three colleagues design mobile networks that serve drones flying along set air corridors, in a 2025 paper. Ground networks aim their signals down, so drones above often get weak and patchy coverage.

The team used data-driven search methods that tune antenna tilt and beam width together. Some methods raised the median signal quality along drone routes by over twenty decibels, with little harm to ground users. Knowledge from one scene sometimes carried to a new scene, and sometimes failed. A test on a real network more than doubled data rates along the drone routes. The search also mapped trade-offs between drone coverage and ground speed.

Catalog record

CREATORMohamed Benzaghta, Giovanni Geraci, David López-Pérez, and Alvaro Valcarce
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE76 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-05176

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Cellular Network Design for UAV Corridors via Data-driven High-dimensional Bayesian Optimization Thanks: M. Benzaghta is with Universitat Pompeu Fabra (UPF), Spain. G. Geraci is with Telefónica Research and UPF, Spain. D. López-Pérez is with Universitat Politècnica de València, Spain. A. Valcarce is with Nokia Bell Labs, France. Thanks: This work was supported by a) HORIZON-SESAR-2023-DES-ER-02 project ANTENNAE (101167288), b) the Spanish Ministry of Economic Affairs and Digital Transformation and the European Union NextGenerationEU through actions CNS2023-145384, CNS2023-144333, and the UNICO 5G I+D SORUS project, c) the Spanish State Research Agency through grants PID2021-123999OB-I00 and CEX2021-001195-M, d) the Generalitat Valenciana, Spain, through the CIDEGENT PlaGenT, Grant CIDEXG/2022/17, Project iTENTE, and e) the UPF-Fractus Chair on Tech Transfer and 6G. Thanks: Some of the results in this paper were presented at IEEE Globecom’23 [ 1 ] . Mohamed Benzaghta, Giovanni Geraci, David López-Pérez, and Alvaro Valcarce Abstract We address the challenge of designing cellular networks for uncrewed aerial vehicles (UAVs) corridors through a novel data-driven approach. We assess multiple state-of-the-art high-dimensional Bayesian optimization (HD-BO) techniques to jointly optimize the cell antenna tilts and half-power beamwidth (HPBW). We find that some of these approaches achieve over 20 dB gains in median SINR along UAV corridors, with negligible degradation to ground user performance. Furthermore, we explore the HD-BO’s capabilities in terms of model generalization via transfer learning, where data from a previously observed scenario source is leveraged to predict the optimal solution for a new scenario target. We provide examples of scenarios where such transfer learning is successful and others where it fails. Moreover, we demonstrate that HD-BO enables multi-objective optimization, identifying optimal design trade-offs between data rates on the ground versus UAV coverage reliability. We observe that aiming to provide UAV coverage across the entire sky can lower the rates for ground users compared to setups specifically optimized for UAV corridors. Finally, we validate our approach through a case study in a real-world cellular network, where HD-BO identifies optimal and non-obvious antenna configurations that result in more than double the rates along 3D UAV corridors with negligible ground performance loss. I Introduction I-A Background and Motivation Robust and reliable connectivity will be essential for the development of the uncrewed aerial vehicles (UAVs) ecosystem, especially in high-impact applications including delivery services and advanced urban air mobility. Private companies and government bodies are increasingly turning to mobile networks to support UAV command and control links and data payload transmission [ 2 , 3 , 4 , 5 ] . Traditional terrestrial cellular base stations (BSs) are optimized for 2D ground-level connectivity. Consequently, UAVs are often limited to receiving signals through the weaker upper antenna sidelobes, resulting in significant signal instability during flight. Additionally, when UAVs fly above buildings, they frequently face interference from line-of-sight (LoS) signals from nearby BSs [ 6 ] , which degrades their signal-to-interference-plus-noise ratio (SINR) [ 7 , 8 ] . The coverage and capacity of cellular networks are greatly affected by the configuration of BS antennas, where adjustments in parameters like the down-tilt angle and half-power beamwidth (HPBW) play a crucial role in optimizing signal strength and minimizing interference. Such an optimization, also known as cell shaping, is complex due to the inter-dependencies of settings across cells. Furthermore, optimizing for UAVs flying at high altitudes, requires directing some of the radiated energy upwards towards the sky, conflicting with the needs of legecy ground users (GUEs), which benefit from down-tilted cells pointing towards the ground. To achieve 3D connectivity for end-devices allowed to fly at different heights, it may be necessary to re-engineer the cellular network originally designed for GUEs. Traditionally, cell shaping is carried out through trial and error, using radio frequency planning tools. This approach lacks scalability, calling for automated optimization techniques that leverage recent advances in data-driven models. In this paper, we propose using high-dimensional Bayesian optimization (HD-BO) to design and optimize a cellular network for 3D connectivity. I-B Related Work Our research community has been dedicated to enable UAV cellular connectivity while maintaining optimal performance for ground networks. Short-term solutions, such as time and frequency separation, offer only limited scalability for managing cellular-connected UAVs effectively. While these methods can temporarily improve connectivity by allocating specific time slots or frequency bands to UAVs, they lack the capacity to handle a large and growing number of UAVs in the sky [ 9 , 10 ] . Additional strategies for achieving ubiquitous aerial connectivity involve dedicating specific infrastructure for aerial services, increasing network density, and incorporating non-terrestrial networks such as LEO satellites [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 ] . These approaches aim to enhance coverage, capacity, and reliability for UAVs, supporting seamless connectivity even in areas with limited ground-based infrastructure. Although these techniques are promising, they require new deployments or signal processing enhancements and still face challenges in achieving ubiquitous connectivity for a large number of UAVs. Another line of research focus on optimizing decisions and actions on the UAV side, such as trajectory planning, aiming at maximizing UAV coverage, and/or rate, while simultaneously minimizing the impact on GUE performance [ 20 , 21 , 22 ] . Also, there exists a line of research utilizing stochastic geometry for UAV deployment optimization, which emphasizes user-centric positioning of UAVs relative to network demand [ 23 , 24 ] . The aforementioned studies assume UAV operations without spatial restrictions, requiring networks to ensure connectivity across the entire 3D space. However, akin to terrestrial vehicles and aircraft, UAVs are expected to operate within designated aerial paths or UAV corridors , as established by authorities [ 25 , 26 ] . With most UAVs traveling along corridors, network operators might focus on providing reliable connectivity within these specific areas rather than across the entire sky. This vision spurred research into optimizing UAV trajectories to better align with optimal network coverage [ 21 , 27 , 28 ] . Nonetheless, the establishment of UAV corridors is expected to be driven by safety and logistics considerations, not communication needs, offering limited flexibility for altering UAV paths and instead underscoring the need for specialized 3D cellular service. Recent studies have attempted to adjust cellular network configurations to better serve UAV corridors, mainly through ad-hoc system-level optimizations of simplified setups [ 29 , 30 , 31 , 32 ] or through fundamental theoretical analysis [ 33 , 34 ] . Despite these promising advancements, the community still faces the challenge of establishing a scalable optimization framework that effectively harnesses available data to enhance practical key performance indicators (KPIs), which are often mathematically complex and difficult to optimize. Our goal is to bridge this gap through a large-scale, data-driven approach designed to tackle such shortcomings. I-C Approach and Contribution We propose a methodology based on Bayesian optimization (BO) for designing cellular BS antenna settings and provide reliable service to both GUEs and UAVs in designated aerial corridors. Although BO [ 35 ] has previously proven useful in addressing coverage/capacity tradeoffs, optimal radio resource allocation, and mobility management [ 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 ] , it faces limitations due to the number of decision variables it

Read the complete text