Source-linked AI summary
Massive MIMO for Drone Communications: Case Studies and Future Directions
Prabhu Chandhar, Erik G. Larsson
TL;DR
Drone applications require long-range, high-throughput, low-latency links for fast-moving drones and whole swarms, needs current Wi-Fi connectivity does not meet. The paper makes the case for massive MIMO, with a ground station antenna array serving single-antenna drones, as an approach for these networks.
Problem
Drone applications need long-range, high-throughput, low-latency links and simultaneous connectivity for swarms, while existing wireless technologies are limited to short ranges and low mobility.
Source: §I · p. 1Method
The paper makes the case for massive MIMO, with the ground station estimating channels from drone pilots and applying signal processing while drones use single antennas.
Source: §I · p. 1, §IV · p. 5Results
Massive MIMO is identified as an enabler for drone-swarm connectivity, with high throughput, spatial multiplexing, coverage extension, and high-mobility support.
Source: §VII · p. 14Takeaways & Limitations
A ground station array can serve lightweight, single-antenna drones simultaneously in applications requiring long-range, high-throughput links.
Source: §I · p. 1, §VII · p. 14
Abstract
from arXiv · showhide
Unmanned aerial vehicles (UAVs), also known as drones, are proliferating. Applications, such as surveillance, disaster management, and drone racing, place high requirements on the communication with the drones in terms of throughput, reliability, and latency. The existing wireless technologies, notably Wi-Fi, that are currently used for drone connectivity are limited to short ranges and low-mobility situations. New, scalable technology is needed to meet future demands on long connectivity ranges, support for fast-moving drones, and the possibility to simultaneously communicate with entire swarms of drones. Massive multiple-input and multiple-output (MIMO), the main technology component of emerging 5G standards, has the potential to meet these requirements.
I. DRONE TECHNOLOGY IS PROLIFERATING
Drone applications are expanding, including missions that stream high-quality imagery, while current and emerging use cases demand substantial throughput, reliability, and low latency. Massive MIMO is presented as a way to connect many drones simultaneously, but drone communications also face mobility, line-of-sight polarization, and image-data-rate challenges.
I. DRONE TECHNOLOGY IS PROLIFERATING
- Drone use is expanding from military operations to civilian missions such as monitoring, search and rescue, inspection, and goods transport.The paper also notes routine use of lightweight quadcopters for aerial photography and drone racing.Source: §I · p. 1
- Drones can serve as cellular user equipment, streaming imagery to ground stations, rather than acting as flying base stations.This article focuses on drones acting as user equipment.Source: §I · p. 1
- Massive MIMO can provide reliable, high-rate links to drone swarms using an array at the ground station and a single antenna on each drone.Many small drones can cover a large area quickly, while single-antenna drones remain lightweight.Source: §I · p. 1
II. DRONE COMMUNICATION REQUIREMENTS AND CHALLENGES
- Image and video data rates depend on mission requirements and onboard processing; a 4K, 60-FPS video stream requires 64 Mbps per drone, or 1.28 Gbps for 20 drones.Raw imagery may be needed when image processing is performed in the cloud.Source: §II · p. 3
- Higher spatial resolution requires a higher-throughput link, and image-transmission rate depends on target GSD and image overlap.Figure 3 plots required rate against target GSD for two image-overlap values.Source: §II · p. 3
- Drone links must support simultaneous connectivity, high throughput, low latency, and mobility, while 3D movement can cause frequent polarization mismatch in line-of-sight channels.The passage contrasts this with cellular channels, where multipath makes complete mismatch less likely.Source: §II · pp. 1, 3
A. Inadequacy of existing technologies
Existing Wi-Fi, XBee-PRO, and cellular networks have limitations for drone links, including restricted range or altitude coverage, mobility-related connectivity problems, and shared-resource throughput loss. The paper argues for a dedicated technology designed around aerial communication requirements.
A. Inadequacy of existing technologies
- Under line-of-sight conditions, Wi-Fi reaches 300–500 m at 10 Mbps for one drone, while XBee-PRO reaches 1 km at 250 kbps.When multiple drones share resources, per-drone throughput decreases proportionally because of MAC-protocol limitations.Source: A. Inadequacy of existing technologies · p. 3
- Cellular coverage is limited at low altitudes and elevation angles, while neighboring-cell interference, infrastructure availability, and drone polarization mismatch constrain aerial links.The cited measurements report limited coverage below 100 m; cellular base stations may also be unavailable in emergencies or remote environments.Source: A. Inadequacy of existing technologies · p. 4
- A dedicated drone technology could design PHY- and MAC-layer protocols from a clean slate for aerial communication requirements.The paper identifies high mobility and large drone counts as settings where Wi-Fi and XBee-PRO perform poorly.Source: A. Inadequacy of existing technologies · pp. 3–5
III. MASSIVE MIMO IS A SUITABLE TECHNOLOGY FOR DRONE COMMUNICATIONS
Massive MIMO is presented as suitable for drone communications because array gain, spatial multiplexing, and channel reciprocity support coverage, many simultaneous links, and mobility. The paper also notes conditions relevant to pilot use and implementation maturity.
III. MASSIVE MIMO IS A SUITABLE TECHNOLOGY FOR DRONE COMMUNICATIONS
- Massive MIMO combines array gain for coverage extension, spatial multiplexing for serving many terminals on the same time-frequency resource, and reciprocity-based handling of high mobility.The signal-processing complexity resides substantially at the base station, keeping terminals low-complexity; the technology works in both rich-scattering and line-of-sight environments.Source: §III · p. 5
- With high coherence bandwidth and low drone density, mutually orthogonal pilots can be afforded, making pilot contamination not a significant issue in the described scenarios.This assumes the antenna array is directed upwards into the sky.Source: §III · p. 5
- High-mobility field trials and efficient hardware implementations have been demonstrated, supporting the feasibility of massive-MIMO ground stations built from maturing technology.A cited implementation performs zero-forcing for 8 terminals with 128 base-station antennas over 20 MHz in real time at about 50 milliwatts.Source: §III · p. 5
- A master aircraft carrying an antenna array can serve as a control station for single-antenna drones in hard-to-reach areas or during disasters.For urban high-altitude coverage, the array may be placed on a high-rise building or tilted skyward on a cellular tower.Source: §III · p. 5
IV. WHAT AND HOW MUCH MASSIVE MIMO CAN PROVIDE FOR DRONE COMMUNICATIONS?
This section outlines Massive MIMO training and TDD operation for drone links, then examines capacity under LoS and rich-scattering channel models. It also considers coherence time, angular interference, and frame-resource allocation.
IV. WHAT AND HOW MUCH MASSIVE MIMO CAN PROVIDE FOR DRONE COMMUNICATIONS?
- The GS estimates drone channels from simultaneous orthogonal uplink pilots, then uses TDD channel reciprocity for downlink operation.The section states reciprocity applies as long as the uplink…Source: §IV · p. 5
- At 2.4 GHz, a 3 MHz coherence bandwidth and 30 m/s drone speed yield τ = 6250 samples per coherence interval.The coherence interval depends on coherence time and bandwidth.Source: §IV · p. 5
- TDD frame symbols are divided among control, uplink and downlink pilots, and data; one nondirectional downlink pilot suffices.Control carries synchronization, scheduling, and trajectory information; uplink pilots number at least K, and uplink data uses channel inversion power control.Source: §IV · p. 5
- The examples use an M-antenna GS array, single-antenna drones, and maximum-ratio combining, with LoS channels incorporating path loss, polarization, distance, spacing, and wavelength.The channel model includes free-space path loss and polarization mismatch loss and antenna gain.Source: §IV · pp. 5–6
- Pure-LoS capacity is significantly lower than i.i.d. Rayleigh capacity because drone channel vectors are highly correlated; practical air-to-ground channels are close to LoS.The authors characterize the LoS assumption as a worst-case design for Massive MIMO drone communications.Source: §IV · p. 6
- Interference varies with drones’ angular separation; at 60 GHz, the mainlobe is 25 times narrower than at 2.4 GHz for the stated equal-aperture comparison.The aperture is 6.25 m at both carriers; element spacing differs between the two frequencies.Source: Figure 8 · p. 7, §IV · p. 7
Simultaneous high-throughput transmission from a large number of drones
Massive MIMO sum throughput first rises and then falls as more drones share the link, while pilot overhead reduces data resources and interference limits gains from higher SNR. The section also gives a lower-bound rate expression for estimating practical deployment throughput.
Simultaneous high-throughput transmission from a large number of drones
- A lower bound on per-drone uplink ergodic rate is given for identically oriented GS elements and a specified drone distribution, with antenna and polarization effects represented by χwc and κ.The rate expression also depends on bandwidth, downlink symbols, and the drone-motion coherence interval.Source: Simultaneous high-throughput… · p. 7
- The rate expression remains applicable to other drone-position distributions by changing Ω, and Ω is zero when element spacing is an integer multiple of half a wavelength.The authors state that the expression can estimate throughput in practical deployment scenarios.Source: Simultaneous high-throughput… · p. 7
- Sum throughput rises with drone count up to a point, then falls as pilot overhead reduces uplink data samples; throughput is nearly unchanged between 0 dB and 10 dB data SNR as interference limits gains.The example uses 20 MHz bandwidth and 100 GS antennas.Source: Simultaneous high-throughput… · p. 7
Pseudo-randomly oriented circularly polarized antenna elements enable stable link conditions
Pseudo-randomly oriented antenna elements reduce polarization-mismatch losses and improve coverage probability across drone locations and orientations. Array geometry and element spacing also affect throughput performance.
Pseudo-randomly oriented circularly polarized antenna elements enable stable link conditions
- With a 100 mW power budget, coverage probability is 78% for identically oriented, linearly polarized antennas and 99.99% for circularly polarized, pseudo-randomly oriented antennas.The comparison assumes a target SNR of 0 dB.Source: Pseudo-randomly oriented circularly… · pp. 7–8
- Pseudo-random orientation reduces polarization-mismatch losses; identically oriented elements can produce effective gain below −30 dB, while pseudo-randomly oriented elements keep it within 10–20 dB.The passage reports gain varying from −30 dB to 20 dB for identically oriented elements.Source: Pseudo-randomly oriented circularly… · p. 8
- For uniformly distributed drones inside a spherical shell with MRC, throughput is maximized when element spacing is an integer multiple of one-half wavelength; other distributions generally have different optima.A rectangular array is preferred for its 3D resolution and reduced space occupancy.Source: Pseudo-randomly oriented circularly… · p. 8
High mobility support
In the stated over-water and mountainous settings, increasing drone speed from 0 to 30 m/s has little effect on uplink throughput. Predictable LoS phase shifts may also reduce channel-estimation overhead for drones following known trajectories.
High mobility support
- For 100 drones at 2.4 GHz with 3 MHz coherence bandwidth and 90% of samples used for uplink, throughput reduction is less than 2% as speed rises from 0 to 30 m/s in over-water and mountainous settings.The passage notes that speed-related loss is more significant in other environments and at larger carrier frequencies.Source: High mobility support · p. 8
- The throughput effect of speed is conditional: the less-than-2% reduction applies to the specified over-water and mountainous scenario, while other environments and larger carrier frequencies can incur greater loss.Source: High mobility support · p. 8
- For a drone moving along a known trajectory in LoS, predictable phase shifts may allow channel prediction over many wavelengths of motion without additional pilots.The passage says this could increase the samples available for data transmission.Source: High mobility support · p. 9
Sub-6 GHz versus mmWave frequencies
The paper contrasts sub-6 GHz coverage with mmWave bandwidth: lower carrier frequencies support longer ranges, while mmWave can provide high sum throughput but may have limited coverage under drone power constraints.
Sub-6 GHz versus mmWave frequencies
- At 60 GHz rather than 2.4 GHz, required transmit power is 625 times higher for the stated target-SNR comparison, assuming unit antenna gains and perfect polarization match.The passage attributes lower mmWave energy efficiency to decreased effective antenna areas.Source: Sub-6 GHz versus mmWave frequencies · p. 9
- The coverage range is inversely proportional to carrier frequency under the stated Friis-based assumptions of unity antenna gains.The figure’s range comparison concerns control-channel coverage distance.Source: Sub-6 GHz versus mmWave frequencies · p. 9, Figure 12 · p. 9
- For 40 Mbps per drone with 20 MHz bandwidth and 20 dBm drone transmit power, control-channel range is 14 km at 2.4 GHz and 370 m at 60 GHz.This control-channel range is the single-antenna-system coverage distance and remains constant with the number of GS antennas; data-channel range depends on antenna count through coherent beamforming gain.Source: Sub-6 GHz versus mmWave frequencies · pp. 9–10
- Abundant mmWave bandwidth can support sum throughput on the order of 10 Gbps, but drone transmit-power limits may constrain coverage range.Source: Sub-6 GHz versus mmWave frequencies · p. 10
Coverage extension removes the need for multi-hop solutions
Massive MIMO’s coherent beamforming can extend data-channel coverage, while channel and interference hardening reduce SINR variation. The emergency-response case study illustrates the scale of a drone swarm and its uplink data requirements.
Coverage extension removes the need for multi-hop solutions
- With M GS antennas and accurate channel estimates, coherent beamforming extends data-channel coverage distance by a factor of M, making Massive MIMO a promising alternative to multi-hop solutions.The passage contrasts multi-hop solutions’ reliability, latency, and coordination issues.Source: Coverage extension removes the need… · p. 10
Channel and interference hardening make low latency communication possible
- Channel hardening makes effective channel gain deterministic in LoS and fading, while interference hardening occurs with many antennas even in LoS; reduced SINR variation helps meet low-latency requirements.Channel variations in wireless links lead to frequent retransmissions and increased latency.Source: Channel and interference hardening… · p. 10
V. CASE STUDIES
- The case-studies section presents three different studies.Source: §V · p. 10
A. Case study I: Emergency response and disaster management
- For a 4 km × 4 km area at 20 cm GSD, about 23 drones working in parallel can complete the emergency-response survey in 20 minutes.A single drone moving at 20 m/s would take more than seven hours, with small-drone flight times often limited to 10–30 minutes.Source: A. Case study I: Emergency response… · p. 10
- A 4K compressed video at 2 cm GSD requires 1.46 Gbps sum throughput, and more than 200 GS antennas are required depending on the environment.The stated compression ratio is 200:1.Source: A. Case study I: Emergency response… · p. 10
B. Case study II: Real-time video streaming at sports events
The sports-event case study considers 20 drones streaming 4K 360-degree video to a stadium access point, requiring 18.4 Gbps in total. Under the stated 60 GHz design conditions, 227 antennas are required to meet that sum-throughput target.
B. Case study II: Real-time video streaming at sports events
- The case study places 20 or more quadcopters with 4K 360-degree cameras at a large sports event, sending footage to a massive-MIMO access point for real-time virtual reality.The stadium-rooftop array-to-drone distance is estimated at 50–100 m.Source: B. Case study II: Real-time video… · pp. 10, 12
- Figures 14 and 15 plot required antenna count against target throughput per drone at 2.4 and 60 GHz under different stated speed, coherence-bandwidth, and bandwidth conditions.Figure 14 specifies 30 m/s, 300 KHz coherence bandwidth, 20 MHz bandwidth, and 0 dB SNR; Figure 15 specifies 20 m/s, 3 MHz coherence bandwidth, and 0 dB SNR.Source: Figure 14 · p. 11, Figure 15 · p. 11
- A 4K 360-degree stream at 60 frames per second requires 920 Mbps per camera after 200:1 compression, or 18.4 Gbps for 20 drones.The per-camera estimate uses 16384 × 8192 pixels and 24 bits per pixel for a 90-degree horizontal angle of view.Source: B. Case study II: Real-time video… · p. 12
- 227 antennas are required for 20 drones to achieve 18.4 Gbps sum throughput at 60 GHz, with 0 dB SNR possible at 160 m using 1 W transmit power.The 60 GHz estimate uses over 300 MHz bandwidth; the stated 100 mW range for 0 dB SNR is 50 m.Source: B. Case study II: Real-time video… · p. 12, Table II · p. 12
C. Case study III: Drone Racing
For drone racing, the paper discusses low-latency control and the throughput needed to transmit raw video, then outlines scheduling and MAC-layer research directions. It also identifies channel conditions and compatibility constraints that shape those directions.
C. Case study III: Drone Racing
- Racing drones can exceed 100 km/h, while control needs latency of tens of milliseconds; current analog transmission limits races to eight drones, whereas massive-array service could reach hundreds on one time-frequency resource.Standard compression introduces about 50 ms latency, and a quadcopter moving at 40 m/s travels 2 meters in that interval.Source: C. Case study III: Drone Racing · p. 12
- Because compressed frames depend on earlier frames, frame errors can prompt retransmissions and increase latency; raw video avoids compression and decompression, and a high-throughput massive-MIMO link can make end-to-end latency negligible.The stated raw-video requirement for 25 drones is 1.84 Gbps with latency below 70 ms.Source: C. Case study III: Drone Racing · pp. 12–13
- 415 antennas are estimated to support 1.84 Gbps of raw-video sum throughput for 25 racing drones, with latency below 70 ms.The paper links raw transmission to avoiding video compression and decompression modules.Source: C. Case study III: Drone Racing · p. 13
Scheduling, resource optimization and 3D trajectory optimization
- When drone channels change slowly or positions are close, the ground-station array may not resolve them spatially, making spatial multiplexing impossible and requiring separate time-frequency scheduling.Joint trajectory planning and scheduling is proposed as an alternative when bandwidth is limited and throughput requirements are high.Source: Scheduling, resource optimization and… · p. 13
New MAC layer design
- Known mobility patterns can reduce the need for frequent CSI pilots, and flexible TDD framing could account for array geometry, speed, altitude, range, and uplink-heavy traffic.The passage states that effective-interference variation depends on array geometry, flying speed and altitude, and range.Source: New MAC layer design · p. 13
- Drone networks can use flexible MAC designs without backward-compatibility constraints, and video or image traffic may benefit from cross-layer design alongside massive MIMO.The paper presents these as possible design approaches for improving network performance.Source: New MAC layer design · p. 13
Air-to-ground channel models
The section describes aerial-channel measurements and their implications for coherence bandwidth, spectrum use, and future network design. It notes measurement gaps and frames massive MIMO as an enabler for high-throughput links to drone swarms.
Air-to-ground channel models
- Aerial channels are generally line-of-sight with Rician K-factors typically above 25 dB; reported C-band RMS delay spreads correspond to coherence bandwidths of 3–20 MHz.Reported average RMS delay spread is about 10 ns over water and in hilly or mountainous terrain, and 10–60 ns in suburban and near-urban settings.Source: Air-to-ground channel models · p. 13, Table III · p. 14
- Measurements are limited to linearly polarized antennas without changing drone-antenna orientation, motivating campaigns across polarizations, frequencies, and environments.The paper notes that mmWave measurements are unavailable and expects higher coherence bandwidths there because of fewer multipath components.Source: Air-to-ground channel models · p. 13
Spectrum management
- Spectrum needs vary with environment, range, drone density, frequency, duplex mode, power, and throughput; high-density applications may need dedicated spectrum, while massive MIMO's spatial multiplexing can reduce spectrum requirements.Low-density applications may opportunistically share spectrum with cellular and satellite systems; unlicensed spectrum is another possibility.Source: Spectrum management · p. 13
- TDD requires at least one uplink pilot symbol per drone per coherence interval, while FDD requires at least the number of ground-station antennas plus drones; beam tracking may reduce CSI acquisition needs.The paper says detailed comparison of duplex modes across environments and applications is still required.Source: Spectrum management · p. 13
Networking challenges
- Future aerial networks may mix high-altitude, medium-range, short-range, and mini-UAVs, with antenna arrays on drones or distributed across a large region in some applications.At mmWave frequencies, drone-mounted arrays are identified as one possible way to achieve additional capacity gains.Source: Networking challenges · p. 13
VII. SUMMARY
- Massive MIMO is presented as an enabler for long-range, high-throughput swarm connectivity, using a large ground-station array while each drone can have a single antenna.The paper identifies high throughput, spatial multiplexing, coverage extension, and high-mobility support as relevant features.Source: §VII · p. 14