The day your operations leave visual range, the datalink stops being a component and becomes infrastructure — with availability statistics, degradation policies and a starring role in your BVLOS approval. Here's how teams actually build C2 architectures that regulators sign and networks don't betray.
What changes at BVLOS
Within sight, a lost link is an inconvenience backed by a pilot's eyeballs. Beyond it, the link is the operation: command, telemetry, video for the remote crew, and conformance data all ride on it. Three design consequences:
- The C2 function must be quantified — availability, latency, coverage along the route — not vibes. Your SORA will ask.
- Lost-link behaviour is a designed, tested aircraft function (continue, hold, return, land — per mission phase; see failsafes).
- Dissimilar redundancy beats more of the same — two links that fail for different reasons (cellular + sub-GHz private, cellular + satcom) multiply availability instead of sharing an outage.
Cellular: the default, with asterisks
LTE/5G gives you national coverage, encryption, and modems that cost less than a prop set. The asterisks, learned in the field:
- Altitude coverage ≠ ground coverage. Antenna patterns point down at ground users. At 100–120 m AGL an aircraft often sees many cells at similar strength → interference and handover churn. Symptoms: brief stalls at cell boundaries. Mitigations: modems allowing operator/band steering, route-level coverage surveys (fly it with a logging modem before promising it), aeronautical connectivity offerings from carriers where available.
- Latency: 30–80 ms typical LTE round trip, spiking during handovers — fine for supervisory autonomy, marginal for manual flight, which is one more reason BVLOS aircraft should fly themselves (see onboard autonomy).
- SIM strategy: multi-IMSI or dual-SIM across carriers is cheap dissimilarity; roaming SIMs can silently route traffic through another country (latency + data-protection surprises).
- Contract reality: consumer SIMs can be throttled or deprioritised; for commercial operations use M2M/IoT plans with defined terms.
Satcom: when the map runs out
LEO broadband (Starlink-class flat panels, and emerging small-terminal services) changed the trade: single-digit-kilogram terminals with usable bandwidth now fit larger UAVs. Legacy L-band (Iridium/Inmarsat) remains the low-bandwidth, high-reliability C2 fallback with tiny antennas. Current working split:
| Option | Terminal mass/power | Bandwidth | Role |
|---|---|---|---|
| Iridium (L-band) | ~50–300 g, 1–5 W | kbps-class | Last-resort C2 + position, global incl. poles |
| LEO broadband panel | 1.5–5 kg, 40–100+ W | 10s–100s Mbps | Primary link offshore / remote, larger airframes |
| GEO L/Ku legacy aero | Heavy, costly | Varies | Legacy large-UAV territory |
The mass and power lines explain why satcom starts making sense on the same aircraft classes where alternative power does: size solves both.
Private networks and mesh
Fixed corridors (ports, mines, rail, energy sites) reward owned infrastructure: private LTE/5G on local spectrum, or sub-GHz/microwave mesh along the route. You trade capex and spectrum paperwork for a network whose availability statistics you control — often the cleanest evidence story for a repetitive-route BVLOS approval, and immune to a carrier's Tuesday maintenance window.
Architecture pattern that keeps passing reviews
- Primary: cellular (multi-carrier) carrying C2 + video + data, VPN/TLS end-to-end.
- Secondary: dissimilar — private sub-GHz along the route, or Iridium-class C2 — carrying command and position only.
- Tertiary: the aircraft itself: autonomy that completes or safely aborts the mission with no link at all, because the only link with 100% availability is not needing one.
- Evidence: log link metrics on every flight (RSRP/SINR, latency, dropouts, failovers); the histogram from 200 route flights is your availability claim.
Remote operations centres fail too: the GCS's internet, the VPN concentrator, the operator's building power. High-availability BVLOS treats the ground segment with the same redundancy discipline as the airborne one — dual ISPs, UPS, and a rehearsed "ops centre offline" procedure in the operations manual.
Frequently asked questions
Can drones use 4G/5G for BVLOS control?
Yes — cellular is the workhorse of small-UAS BVLOS today. Caveats: networks are optimised for ground users, so coverage at 100 m altitude differs from ground maps (often more towers visible, more interference and handover churn); latency is fine for supervisory control but not manual piloting at range; SIM/roaming and network-slicing arrangements matter for reliability.
Do I need satellite communications for BVLOS?
Only when the mission leaves cellular coverage — offshore, remote areas, or as a dissimilar backup for high-assurance operations. Modern LEO services make aircraft-sized terminals workable, but weight, power, cost and per-MB pricing still favour cellular wherever it exists.
What C2 link reliability do regulators expect for BVLOS?
EASA's SORA doesn't set a single number; it requires you to define C2 performance, degradation behaviour and lost-link procedures, with rigour scaling with risk (SAIL). Demonstrated link statistics from a flight-test campaign are typically part of the evidence.