Performance Evaluation of UAV-enabled Cellular Networks with Battery-limited Drones

Yujie Qin, Mustafa A. Kishk, Mohamed-Slim Alouini

Abstract

Unmanned aerial vehicles (UAVs) can be used as flying base stations (BSs) to offload Macro-BSs in hotspots. However, due to the limited battery on-board, UAVs can typically stay in operation for less than 1.5 hours. Afterward, the UAV has to fly back to a dedicated charging station that recharges/replaces the UAV's battery. In this paper, we study the performance of a UAV-enabled cellular network while capturing the influence of the spatial distribution of the charging stations. In particular, we use tools from stochastic geometry to derive the coverage probability of a UAV-enabled cellular network as a function of the battery size, the density of the charging stations, and the time required for recharging/replacing the battery.

Introduction

Unmanned Aerial Vehicles (UAVs) are increasingly being considered as flying base stations to provide cellular coverage in high-demand hotspots. Because they fly at high altitudes, they easily establish Line-of-Sight (LoS) links with ground users. However, there is a massive logistical bottleneck: battery life. Most commercial UAVs can only stay airborne for less than 1.5 hours before needing a recharge. When a drone leaves its designated hotspot to recharge, the users below lose their dedicated coverage and must fall back to terrestrial base stations (TBSs). This paper addresses how the spatial density of charging stations, the size of the drone's battery, and the time it takes to recharge/swap a battery impact the overall cellular coverage of the network.

Methodology

The authors use stochastic geometry to model the spatial distribution of the network components and derive exact mathematical expressions for network performance. User hotspots are modeled using a Matern cluster process (MCP), with a UAV hovering at the center of each hotspot. Terrestrial base stations (TBSs) and UAV charging stations are modeled as independent Poisson Point Processes (PPPs).

The model incorporates a detailed aerodynamic power consumption model to determine the probability that a UAV is "available" (actively serving the hotspot) versus "unavailable" (traveling to/from a charging station or currently charging). The overall network coverage probability is then calculated by conditioning on this availability.

Coverage probability distributions showing the impact of increasing charging time (a) and decreasing battery capacity (b).

The Infrastructure Trade-off: Density vs. Speed

A critical finding is that propulsion energy dominates communication energy. Because the physical energy required to keep a drone airborne and moving vastly outweighs the energy used to transmit RF signals, optimizing wireless communication efficiency won't meaningfully extend flight times. Logistical solutions, such as battery swapping, are mandatory.

There is a direct, mathematical relationship between the quality of charging stations (how fast they charge) and the quantity (how densely they must be deployed). Reducing the turnaround time at the charging station is the most powerful lever for improving network coverage. A system that physically swaps batteries (taking ~5 minutes) is vastly superior to one that recharges them in place (taking ~40 minutes), drastically reducing the physical footprint of charging infrastructure needed.

Results and Quantitative Impact

The analytical models, validated by Monte Carlo simulations, reveal a stark infrastructure trade-off. To achieve a specific target coverage probability, a network using slow charging stations (40 minutes) requires roughly 100 times higher spatial density of charging stations compared to a network using fast battery-swapping stations (5 minutes).

Average coverage probability vs. charging station density for different charging times (a) and battery capacities (b).

Furthermore, while increasing the UAV's battery capacity ($B_{max}$) improves the maximum achievable coverage probability, these gains are severely bottlenecked if the charging time remains high. As shown in the figures, the coverage probability saturates as the density of charging stations increases, with the saturation level determined by the battery capacity and charging efficiency.

Conclusion

This work provides a robust spatial framework for evaluating UAV-enabled networks. While the model currently assumes immediate availability of charging slots and simple fallback to TBSs, it highlights the critical role of charging infrastructure. Future work could extend this framework to include multi-UAV coordination, queuing theory at charging pads, or hybrid networks integrating tethered drones.

Summary updated

Abstract

Unmanned aerial vehicles (UAVs) can be used as flying base stations (BSs) to offload Macro-BSs in hotspots. However, due to the limited battery on-board, UAVs can typically stay in operation for less than 1.5 hours. Afterward, the UAV has to fly back to a dedicated charging station that recharges/replaces the UAV's battery. In this paper, we study the performance of a UAV-enabled cellular network while capturing the influence of the spatial distribution of the charging stations. In particular, we use tools from stochastic geometry to derive the coverage probability of a UAV-enabled cellular network as a function of the battery size, the density of the charging stations, and the time required for recharging/replacing the battery.
BibTeX
@article{qin2020performance,
  author = {Yujie Qin and Mustafa A. Kishk and Mohamed-Slim Alouini},
  title = {{Performance Evaluation of UAV-enabled Cellular Networks with Battery-limited Drones}},
  journal = {IEEE Communications Letters},
  year = {2020},
  url = {https://openreview.net/forum?id=vbDyvRCQBue},
}
Figure