Drone Site Inspection
Overview
Drones have already changed how sites are surveyed, but the link between aircraft and controller has not kept up. Most inspection flights still use a proprietary point-to-point radio between aircraft and controller, which ties the flight to line of sight from one operator position, keeps the video on the pilot's screen rather than in front of the people who need to see it, and offers no path to the aircraft from anywhere else.
Private 5G replaces that link with the site's own network. Video and telemetry arrive at the Onyx Edge and are distributed to whoever needs them (site office, remote engineer, client) in real time. Flights can be conducted from a fixed ground control station rather than from wherever the radio reaches. And because the network belongs to the site, coverage is designed for the flight envelope instead of being whatever the nearest public cell happens to provide at altitude.
The applications follow directly: structural and facade inspection, stockpile and earthworks volumetrics, progress capture from angles no fixed camera can hold, thermal survey of roofs and electrical infrastructure, and inspection of assets that would otherwise need scaffolding, a rope team, or a shutdown.
Aviation regulation governs this workflow, and it governs it first. Rules for command-and-control links, beyond-visual-line-of-sight operation, flights over people, and the use of cellular networks for C2 differ by jurisdiction and change frequently. Nothing on this page authorises an operation. Confirm the applicable national aviation authority's requirements, and hold the necessary approvals, before planning a flight around any network capability described here.
How It Works
Dual-bearer architecture
What each component does
Inspection aircraft
Visual, thermal, or LiDAR payload, with a 5G module and antenna sited clear of the airframe
Onyx RU
The cell serving the flight envelope. Its antennas are downtilted for ground users, so the aircraft is served by sidelobes
FHM / shared cell
The aircraft crosses the site without handover, so neither video nor C2 suffers a handover gap
Video bearer · 5QI 71
GBR, 150 ms budget, 15–40 Mbps uplink. Bursty and large
C2 bearer · 5QI 84
Delay-critical GBR, 30 ms budget, 50–200 kbps symmetric. Small and time-critical
Post-flight bearer · 5QI 8/9
Non-GBR. 5–50 GB per flight, scheduled and out of the way of live traffic
Onyx Edge
Local breakout to the ground control station and to viewers; photogrammetry and volumetrics on site; imagery stays under site control
Ground control station
On the site network rather than tied to wherever a proprietary radio reaches
Live viewers
Site office, client, and the remote engineer: the capability that most changes how inspection works
Onyx Portal
Link performance logged per flight, so a degraded inspection is explained rather than guessed at
The flight path, hop by hop
1
Payload → encode
H.265
Encoded to the worst point in the envelope, not to the camera's maximum
2
Aircraft → Onyx RU
5G NR Uu, two bearers
Video on 5QI 71; C2 and telemetry on 5QI 84. Separate, always
3
RU → FHM → Onyx Edge
eCPRI, aggregated fronthaul
No handover as the aircraft crosses the site
4
UPF → Ground control station
N6 local breakout
Pilot's C2 path, monitored independently of video
5
UPF → Live viewers
Local breakout + distribution
Remote engineer directs the pilot to positions of interest in real time
6
Post-flight
Non-GBR bulk upload
Full-resolution imagery transfers on a low-priority bearer
7
→ Photogrammetry
On the edge
Point clouds, orthomosaics, stockpile volumes; findings tagged to asset and raised into the inspection or CMMS workflow
Running independently: the aircraft's own link-loss failsafe: return to home, hover, or land. That, not the network, is the safety mechanism.
Why air coverage is a separate design problem
Ground coverage does not imply air coverage, and this is the most consequential technical point on the page.
RU antennas are downtilted to serve users on the ground. An aircraft at 60 m is therefore served by antenna sidelobes, at lower gain, from further away, and from more cells simultaneously than a ground device would see. It both suffers more interference and causes more of it.
The practical consequence: survey the envelope by flying it. Fly a test profile with a logging device and record RSRP, SINR, and achievable uplink throughput through the volume. Identify altitudes and positions where coverage is marginal and treat them as no-stream zones or exclude them from the plan. A ground survey tells you nothing useful here.
Why C2 and video never share a bearer
Video is bursty and large; C2 is small and time-critical. On a shared bearer, a 40 Mbps video burst delays the control message, and it does so at exactly the moment the aircraft is doing something demanding, because that is when the video bitrate peaks. Separate 5QIs, separate bearers, verified under full video load rather than in isolation.
Network Requirements
Uplink per aircraft, 1080p H.265
15–25 Mbps
Uplink per aircraft, 4K H.265
25–40 Mbps
Command-and-control telemetry
50–200 kbps, symmetric
Aggregate, 4 concurrent aircraft
60–160 Mbps uplink
Post-flight bulk upload
5–50 GB per flight
One-way latency, video
< 50 ms
One-way latency, C2
< 20 ms
Jitter
< 15 ms
Packet error rate
10-5
C2 link-loss tolerance
Aircraft-defined; typically 1–3 s before failsafe
Availability within flight envelope
99.99%
Video QoS
5QI 71: GBR, live uplink streaming, 150 ms PDB, 10-6 PER
C2 QoS
5QI 84: delay-critical GBR, 30 ms PDB, 10-5 PER
Post-flight upload QoS
Non-GBR 5QI 8 / 9
Altitude band to validate
Ground level to the ceiling of the intended envelope
TDD pattern
Uplink-weighted, DSUUU
Bands
n77, n78, n48 (CBRS), n40
Deploying It
Establish what is permitted first
Confirm the authorizations required for the intended operation, including any specific approval to use a cellular network for command and control. Complete the operational risk assessment the jurisdiction requires. Confirm insurance and pilot competency for the specific operation.
Pre-flight each time
Confirm coverage at the planned positions on the day: site structures change, and a facade in coverage last month may be shadowed by a new frame. Confirm weather, airspace, and site conditions against the agreed stop criteria. Brief ground crew. Verify the video path end to end to every intended viewer before the aircraft leaves the ground.
Devices & Ecosystem
5G-equipped inspection aircraft: either a factory 5G option or an integrated module, with the antenna sited clear of the airframe.
Payloads: high-resolution visual for facade and defect inspection; thermal for roofs, electrical assets, and heat loss; LiDAR where the deliverable is a point cloud rather than imagery.
Ground control station: on the site network, reaching the aircraft through local breakout.
Remote viewing: the site office, the client, and the remote engineer, all receiving the same live feed.
Photogrammetry and volumetrics: hosted on the edge, producing orthomosaics, models, and stockpile volumes.
Inspection and CMMS integration: so a finding becomes a work item rather than a note in a flight report.
Ground survey reference: control points for volumetric accuracy where the output is used commercially.
Design Considerations
Regulation is the first design input, not a compliance step at the end. What is permitted, particularly for beyond-visual-line-of-sight operation and for cellular command and control, determines the operation. Establish it before designing anything else.
Air coverage is a separate design problem from ground coverage. Downtilted antennas serve altitude through sidelobes. An aerial device sees more cells, at lower gain, from further away, and both suffers and causes more interference than a ground device. Survey the envelope by flying it.
Never share a bearer between C2 and video. This is the single most important network decision here. Video is bursty and large; C2 is small and time-critical. On a shared bearer, the video burst delays the control message. Separate 5QIs, separate bearers, verified under full video load.
The aircraft's failsafe is the safety mechanism, not the network. Design the network to keep the link up, and design the operation to be safe when it does not. Verify link-loss behavior by inducing loss, on the ground, before the first flight, and again after any configuration change.
Encode to the worst point in the envelope. A stream configured for the camera's maximum will fail where coverage is weakest, which, on a facade or structural inspection, is precisely where the aircraft needs to be.
Make the flight profile repeatable. Most of the value in inspection is comparative: this month's facade against last month's, this week's stockpile against last week's. A repeatable profile makes comparison automatic; an improvised flight makes every inspection a one-off.
Give post-flight upload its own low-priority path. Fifty gigabytes moving off the aircraft after landing should not contend with a camera contribution feed or an AGV control loop. Non-GBR, scheduled, and out of the way.
Validation Checklist
Next steps
Size it. Take these requirements through Plan Your Network: capacity, coverage, architecture, and a preliminary bill of materials.
Compare it. Network Requirements at a Glance puts every workflow side by side, for sites carrying more than one.
Talk to us. For a validated design and a quote, contact GXC.
Related reading
Last updated
Was this helpful?