PaperlayerView exact revision on arXiv

Source-linked AI summary

Mobile-Network Connected Drones: Field Trials, Simulations, and Design Insights

Xingqin Lin, Richard Wiren, Sebastian Euler, Arvi Sadam, Helka-Liina Maattanen, Siva D. Muruganathan, Shiwei Gao, Y. -P. Eric Wang, Juhani Kauppi, Zhenhua Zou, Vijaya Yajnanarayana

arXiv:1801.10508v2cs.NI

TL;DR

Mobile networks may provide connectivity for low-altitude drones, though interference and mobility pose challenges. The article combines commercial-LTE field measurements with simulations to examine that potential. It concludes that existing terrestrial LTE networks can support initial low-altitude drone deployment, with challenges remaining.

  • Problem

    Practical evidence on mobile-network support for drones is limited, particularly for mobility, latency, and cell association, despite the need for connectivity for drone operation and management.

    Source: INTRODUCTION · p. 1

  • Method

    The article analyzes field measurements from drone flights in a commercial LTE network and simulations based on current 3GPP channel, traffic, antenna, and network models.

    Source: MEASUREMENT SETUP · p. 2, CONCLUSIONS AND FUTURE WORK · p. 9

  • Results

    Existing mobile LTE networks targeting terrestrial usage can support the initial deployment of low-altitude drones, though interference and mobility challenges remain.

    Source: CONCLUSIONS AND FUTURE WORK · p. 9

  • Takeaways & Limitations

    The measurements and simulations provide design insights that can guide future work on mobile-network connectivity for drones.

    Source: INTRODUCTION · p. 2

    The study focuses on drones up to 300 m; higher-altitude connectivity may require vertical beamforming or up-tilted base-station antennas and further analysis, simulations, and field measurements.

    Source: CONCLUSIONS AND FUTURE WORK · p. 9

Abstract

from arXiv · show

Drones are becoming increasingly used in a wide variety of industries and services and are delivering profound socioeconomic benefits. Technology needs to be in place to ensure safe operation and management of the growing fleet of drones. Mobile networks have connected tens of billions of devices on the ground in the past decades and are now ready to connect the drones flying in the sky. In this article, we share some of our findings in cellular connectivity for low altitude drones. We first present and analyze field measurement data collected during drone flights in a commercial Long-Term Evolution (LTE) network. We then present simulation results to shed light on the performance of a network when it is serving many drones simultaneously over a wide area. The results, analysis, and design insights presented in this article help enhance the understanding of the applicability and performance of providing mobile connectivity to low altitude drones.

INTRODUCTION

Mobile networks offer a potential way to connect and manage low-altitude drones, but mobility and interference remain challenges. This article combines field measurements and simulations to provide practical findings on cellular connectivity for drones.

INTRODUCTION

  • Mobile networks could support drone command, control, and payload communications, while reusing terrestrial networks may avoid new technologies and large investments.The introduction also identifies interference and mobility as challenges.Source: INTRODUCTION · p. 1
  • Figure 1 maps cell sites and sector orientations, the drone route, and a car route used as a ground reference.The flight test took place in Masala, Finland.Source: Figure 1 · p. 2
  • The article presents field measurements and simulations to provide design insights into mobile-network connectivity for low-altitude drones.It also reports cell-association patterns in the sky and explains increased radio link failures.Source: INTRODUCTION · pp. 1–2

MEASUREMENT SETUP

Field measurements compared cellular data from a drone at two heights with a car route on the ground in a suburban commercial LTE network in Masala, Finland.

MEASUREMENT SETUP

  • Measurements used a consumer-grade DJI Phantom 4 Pro carrying an LTE phone with TEMS Pocket 16.3; the drone's maximum flight time was about 30 minutes.The phone was mounted on the drone.Source: MEASUREMENT SETUP · p. 2
  • The comparison used ground, 50 m, and 150 m routes; 150 m is Finland's maximum permissible drone height, while 50 m approximates rural base-station height.The 50 m scenario also reflects the emerging use case of drone-based cell-site inspection.Source: MEASUREMENT SETUP · p. 2
  • The drone flew at about 18 km/h along routes closely following the ground car route, whose speed varied from 20 to 40 km/h.Although other bands were measured, only 800 MHz data are presented for brevity.Source: MEASUREMENT SETUP · p. 2

SERVING CELL RESULTS AND ANALYSIS

Field measurements show that drone height affects serving-cell power, signal quality, and uplink throughput. Sky measurements had stronger serving-cell RSRP and higher uplink throughput than ground measurements, but lower RSRQ and RS-SINR.

SERVING CELL RESULTS AND ANALYSIS

  • 18.9 dB and 15.6 dB higher 50th-percentile RSRP at 50 m and 150 m, respectively, than at ground level.The article attributes the stronger sky RSRP to near-free-space propagation despite reduced antenna gains from low-gain main-lobe parts or sidelobes.Source: SERVING CELL RESULTS AND ANALYSIS · p. 2, Figure 2 · p. 3
  • 2.8 dB and 4.1 dB lower 50th-percentile RSRQ at 50 m and 150 m, respectively, and 7.7 dB and 10.2 dB lower RS-SINR, respectively, than at ground level.The article links the lower signal quality to stronger downlink interference from nonserving cells; signal quality decreases as height increases over the studied range.Source: SERVING CELL RESULTS AND ANALYSIS · p. 3, Figure 2 · p. 3
  • ~18.5 Mbps and ~18.8 Mbps 50th-percentile uplink throughput at 50 m and 150 m, respectively, versus ~12.5 Mbps at ground level.Uplink-throughput variances were also much smaller at both heights; performance depends on factors including network load and scheduling.Source: SERVING CELL RESULTS AND ANALYSIS · p. 3, Figure 2 · p. 3

NEIGHBOR CELL RESULTS AND ANALYSIS

The field measurements show that neighbor-cell signals become stronger and more similar in strength with altitude, while the serving cell is more often weaker than the strongest neighbor. The article also describes enhanced measurement reporting and notes the limits of field trials as motivation for simulations.

NEIGHBOR CELL RESULTS AND ANALYSIS

  • The median RSRP difference between the strongest and fourth-strongest neighbors narrows from 11.4 dB on the ground to 7.4 dB at 50 m and 6.5 dB at 150 m.At 50 m and 150 m, even neighbors other than the strongest have median RSRPs above -86 dBm; on the ground, they are below -105 dBm.Source: NEIGHBOR CELL RESULTS AND ANALYSIS · p. 4, Figure 3 · p. 4
  • At 150 m, the median serving-cell-to-strongest-neighbor RSRP gap is 1.6 dB, versus 2.8 dB at 50 m and 6.5 dB on the ground.The gap is negative in about 11% of ground measurements, 21% at 50 m, and 33% at 150 m; the increase is likely related to antenna-pattern rolloff.Source: NEIGHBOR CELL RESULTS AND ANALYSIS · p. 4
  • Release-15 LTE allows measurement reports to be triggered when multiple cells’ RSRPs, RSRQs, or RS-SINRs exceed configured thresholds.This is intended to give the network more control over measurement reports from drone UEs.Source: NEIGHBOR CELL RESULTS AND ANALYSIS · p. 5
  • Field trials are limited in the number of drones tested simultaneously and in the features and configurations available in a commercial network, so simulations complement them for wide-area, many-drone performance.Source: NEIGHBOR CELL RESULTS AND ANALYSIS · p. 5

LATENCY SIMULATION RESULTS AND ANALYSIS

Reliable, bounded-latency command-and-control communication is important for remote drone operation; this section presents LTE radio-interface latency simulations for that traffic.

LATENCY SIMULATION RESULTS AND ANALYSIS

SIMULATION SETUP

The simulations follow the 3GPP aerial-vehicle study assumptions and evaluate drone command-and-control traffic across heights and radio-resource allocations.

SIMULATION SETUP

  • The evaluation uses periodic 1250-byte packets every 100 ms, a 50 ms latency bound, and a 19-site urban-macro deployment with three cells per site.It assumes 10 MHz bandwidth at 2 GHz and orthogonal frequency resources for drone and terrestrial traffic; drones can still experience interference from neighboring cells.Source: SIMULATION SETUP · pp. 5–6
  • Figure 5 maps maximum-received-power-based cell associations at ground level and at 50, 100, and 300 m, with colors indicating the associated site.Source: Figure 5 · p. 6
  • The simulation includes five drone UEs per cell, with heights of 1.5, 30, 50, 100, and 300 m and allocations of 6 PRBs or 15 PRBs.These allocations correspond to 1.08 MHz and 2.7 MHz, respectively.Source: SIMULATION SETUP · p. 6

CONNECTED DRONES ALLOCATED WITH 6 PRBS

With 6 PRBs, rising altitude coincides with lower PDSCH SINR and high resource utilization, and even ground-level latency does not meet the 50 ms bound with high confidence.

CONNECTED DRONES ALLOCATED WITH 6 PRBS

  • Even at 1.5 m, the 50 ms latency bound cannot be met with high confidence, such as 90%.PDSCH SINR drops significantly with height, especially at 50 m or above.Source: CONNECTED DRONES ALLOCATED WITH 6 PRBS · p. 6
  • Resource utilization is about 41%, 57%, 90%, 95%, and 96% at 1.5, 30, 50, 100, and 300 m, respectively.The authors relate the high utilization aloft to poor geometry, lower spectral efficiency, and stronger inter-cell interference.Source: CONNECTED DRONES ALLOCATED WITH 6 PRBS · p. 6

CONNECTED DRONES ALLOCATED WITH 15 PRBS

With 15 PRBs, latency simulations show that the 50 ms bound is met with high confidence at most tested heights, while higher altitude increases resource use and interference. The results reflect a high command-and-control traffic demand and a simulated network configuration.

CONNECTED DRONES ALLOCATED WITH 15 PRBS

  • ~99% confidence meets the 50 ms latency bound at 1.5 m, 30 m, 50 m, and 100 m; at 300 m, confidence is ~92%.With 15 PRBs serving drone traffic, SINR still drops with height, though it is much higher than when only 6 PRBs are used.Source: CONNECTED DRONES ALLOCATED WITH 15 PRBS · p. 6
  • ~11%, ~11%, ~23%, ~30%, and ~47% resource utilization at 1.5 m, 30 m, 50 m, 100 m, and 300 m, respectively.The paper links utilization not exceeding ~30% at the lower heights to moderate downlink interference, versus stronger interference at 300 m.Source: CONNECTED DRONES ALLOCATED WITH 15 PRBS · p. 7
  • The latency results show a tradeoff between latency performance and PRBs used for drone command-and-control traffic, under high traffic demand.The paper notes initial deployment may have much lower command-and-control demand.Source: CONNECTED DRONES ALLOCATED WITH 15 PRBS · p. 7
  • The simulations model 57 cells with five drone UEs per cell; the authors could not validate these results with a field campaign of that scale.They report that results in 3GPP Tdoc R1-1720571 match their simulations and help validate their accuracy.Source: CONNECTED DRONES ALLOCATED WITH 15 PRBS · p. 7

MOBILITY SIMULATION RESULTS AND ANALYSIS

At higher altitudes, sidelobe coverage and increased line-of-sight interference complicate drone mobility: cells may be fragmented and serving links can degrade before handover triggers. Simulations and field measurements show more radio link failures aloft, while flight-route information is identified as a possible aid to mobility management.

MOBILITY SIMULATION RESULTS AND ANALYSIS

  • Higher-altitude cell coverage becomes fragmented according to base-station antenna lobes, and field measurements above 300 m show that a UE may connect to faraway rather than closest cells.Increased line-of-sight to base stations also raises uplink and downlink interference, which can degrade SINR and contribute to RLFs or handover failures.Source: MOBILITY SIMULATION RESULTS AND ANALYSIS · pp. 7–8
  • In a simulated 300 m flight at 30 km/h, the serving-cell RSRP falls 7 dB within 4 s, but no neighbor reaches the 3 dB handover-report trigger before RLF at 7 s.The trace describes movement away from a base-station antenna sidelobe; poor serving-cell SINR leads to the RLF.Source: MOBILITY SIMULATION RESULTS AND ANALYSIS · p. 8
  • At 150 m, the field trial measured more than 50% more radio link failures than on the ground, consistent with the mobility simulations.Figure 8 presents the ground and 150 m field-trial RLF counts.Source: MOBILITY SIMULATION RESULTS AND ANALYSIS · p. 8, Figure 8 · p. 8
  • The simulation illustrates sudden signal drops between sidelobe coverage areas, small serving-to-neighbor RSRP gaps, and persistently low serving-cell SINR from neighbor interference.The observations are consistent with field RSRP and SINR measurements; the default handover procedure may be too slow to execute successfully.Source: MOBILITY SIMULATION RESULTS AND ANALYSIS · pp. 8–9
  • When flight routes are available, the network can use them to facilitate more robust drone mobility management; Release-15 LTE supports requesting flight-path information from the UE.The paper presents this as a potential use of supplemental route data.Source: MOBILITY SIMULATION RESULTS AND ANALYSIS · p. 9

CONCLUSIONS AND FUTURE WORK

Commercial LTE field measurements and simulations support initial deployment of existing mobile networks for low-altitude drones, while identifying interference and mobility challenges. The conclusions outline research needs in higher-altitude coverage, air-ground channel characterization, and integrating drone-based base stations.

Conclusions

  • Existing terrestrial LTE networks can support initial low-altitude drone deployment, but interference and mobility remain challenges.The conclusion draws on commercial-network measurements and simulations; it calls for further validation and work to address the challenges.Source: CONCLUSIONS AND FUTURE WORK · p. 9

Future work

Higher-altitude drones

  • Drones above the studied 300 m scope may need vertical beamforming or up-tilted base-station antennas, with further analysis, simulations, and field measurements.The article focused on low-altitude drones up to 300 m.Source: CONCLUSIONS AND FUTURE WORK · p. 9

Air-ground channel characterization

  • Air-ground channels differ from terrestrial channels, and the article identifies this difference as a root cause of interference and mobility challenges.More empirical measurements could support more accurate statistical air-ground channel models; Doppler effects were implicitly captured through speed-dependent results.Source: CONCLUSIONS AND FUTURE WORK · p. 9

Drones as BSs

  • The article studies drones as user equipment, while drone-mounted base stations for event coverage or disaster response remain a separate research direction.Integrating supplemental drone-based networks with existing mobile networks needs further analysis, simulations, and field measurements.Source: CONCLUSIONS AND FUTURE WORK · p. 9

BIOGRAPHIES

The biographies describe a team whose experience spans mobile-network research and standardization, radio engineering, and drone operations. Several authors focus specifically on aerial connectivity, air-to-ground communications, or related LTE and 5G work.

Author backgrounds

  • Xingqin Lin leads 4G/5G research and standardization on drones and satellites and contributed to the 3GPP Release-15 study of enhanced LTE support for aerial vehicles.Source: BIOGRAPHIES · p. 10
  • Richard Wiren combines mobile-network deployment and optimization experience with registered RPAS pilot credentials in Finland and Sweden; Arvi Sadam works on 3D radio design for UAV missions and holds a CPL(H) license.Source: BIOGRAPHIES · p. 10
  • Sebastian Euler works on extending LTE and 5G NR standards for aerial vehicles; Helka-Liina Määttänen and Siva D. Muruganathan have aerial-network standardization and research roles.Määttänen was Rapporteur of the LTE Rel-15 WI on Aerials; Muruganathan was Rapporteur for the 3GPP Release-15 study on LTE Aerials.Source: BIOGRAPHIES · p. 10
  • Shiwei Gao researches multi-antenna technologies and their standardization, while Y.-P. Eric Wang has contributed to more than 150 U.S. patents and 50 IEEE articles.Source: BIOGRAPHIES · p. 10
  • Juhani Kauppi brings more than 30 years in telecom, while Zhenhua Zou researches air-to-ground communications and wireless control, optimization, and learning.Source: BIOGRAPHIES · p. 10
  • Vijaya Yajnanarayana is an Ericsson Radio Research Group researcher in Stockholm.Source: BIOGRAPHIES · p. 10

Related papers

Loading 1801.10508v2…