Non-Terrestrial Networks for Automotive and IoT

David Turner
Jul 24
6 min read

Updated: Aug 10

Non-terrestrial networks for automotive and IoT

NTNs in Automotive and IoT: Trade-Offs and Deployment Decisions

Non-terrestrial networks are a recurring subject I've been asked about over the last year or two. Interest amongst Automotive and IoT teams has been piqued by Starlink, and high-return space stocks like RKLB and ASTS have pulled NTNs into the public consciousness. So to answer these questions I'll skip the full telco perspective here and focus on practical deployment.

What is a non-terrestrial network?

An NTN is reasonably straightforward. It's a mobile network using satellites orbiting the earth instead of ground-based cell towers and base stations.

The advantage of an NTN is coverage where terrestrial cellular does not exist. Cell towers are expensive, and if there's no commercial or regulatory case for deployment, mobile network operators won't build them. This covers mountain ranges, deserts, oceans, and remote terrain. A secondary benefit matters more than it appears - areas affected by environmental disaster or conflict often lose towers and cables first. Satellites often operate regardless.

Three orbital types exist. LEO operates at 400–500 km altitude with low latency and high throughput, completing an orbit every 90 minutes. MEO sits at 20,000 km and dominates navigation systems (GPS, Galileo, Glonass) and dedicated enterprise broadband like SES OB3, which offers up to 10 GB/s at 150 ms round-trip time. GEO is synchronised to a single ground position at 36,000 km altitude, covering large areas with a 24-hour orbit but higher latency.

Most commercial connectivity uses LEO (Starlink included). At LEO altitudes, satellites travel at 7.6 km/s (27,000 km/h). A Starlink dish uses phased-array antennas to direct radio beams at target satellites, each passing overhead in roughly two minutes before handover to the next in the constellation to arrive.

What type of architecture do NTNs use?

The network architecture differs in one obvious way between terrestrial and non-terrestrial networks: satellites replace cell towers. In the first-generation NTN standard - Transparent Payload - base stations, core network, and operations systems remain on the ground. The satellite acts as an orbiting repeater. The next generation, Regenerative Payload architecture, places a base station onboard, allowing satellites to route data between themselves via laser inter-satellite links before sending it back to earth. Operating and provisioning systems such as OSS/BSS and UDM/HSS remain on ground.

Do NTNs require dedicated SIMs and devices?

For mobile access, you don't need to buy a dedicated satellite SIM. Instead, you roam onto a constellation operator like Starlink via a carrier holding an agreement with them. Broadband is different - proprietary hardware (a dish) handles network registration directly. Handsets work with direct-to-cell networks like Starlink or AST SpaceMobile through standard LTE-capable phones; these networks appear to devices as ordinary 4G. For official 5G satellite bands (S-band n256, L-band n253/n255), your device needs 3GPP Release 17/18 compliance.

Non-terrestrial networks for Automotive and IoT

IoT is more constrained. Ruggedised chipsets from vendors like Qualcomm or MediaTek are required, and 3GPP standards matter more, especially for automotive. Minimum: Release 17. Preferred for automotive: Release 18, which adds improved handover between terrestrial and satellite networks and network-based location verification for emergency routing. Release 19 goes further, adding GNSS-independent positioning, store-and-forward telematics for areas with no coverage, and support for regenerative payloads.

Do NTNs support Emergency calling?

Emergency calls (112 eCall) can route over NTN. The mechanism depends on your country's NG112 rollout status. Where NG112 exists, eCall routes as a true emergency call over the NTN to the PSAP. Where direct-to-cell is used, the call routes via terrestrial network.

The practical problem with LEO eCall is border routing. A satellite at that altitude covers multiple countries at once. An accident in Sweden cannot route to a Danish PSAP. The satellite must verify device location and route to the correct emergency service in the correct country. For those of us who have lived experience of circuit-switched eCall systems, it’s a familiar problem.

As radio signals don’t respect political borders any more than a satellite does, calls were often made from the scene of an accident in one country, with the Cell ID of a tower a couple of hundred metres across the border leading to a call placed with another country's PSAP. The PSAP operator then has to forward the call back over to the neighbouring country’s emergency services to respond. This is why Release 18 is essential for automotive - it supports network-verified location.

The NTN operator is responsible for securing interconnects and ground gateways. The OEM and lead-carrier meanwhile must configure the SIM profile to support emergency roaming before any subscription exists - an emergency IMS registration must be supported under 112.

With the advent of SGP32 and SGP42 eSIM standards, provisioning the correct profile at point of production has become simpler than ever. Only a few years back I was leading programs where architecting global eUICC solutions for automotive OEMs meant dealing with bootstraps and in-market profile swaps upon delivery. The introduction of IPA (IoT Profile Assistant) and IFPP (In-factory profile provisioning) has since reduced complexity and enhanced robustness of those processes.

Technical Constraints

Several technical constraints sit beneath every NTN decision:

Power budget. Phased-array antennas draw several watts continuously in automotive installations, compared to under 1 W for conventional cellular modems. For battery-powered IoT - agricultural sensors, asset trackers - this is a decisive constraint. Release 18's reduced-capacity mode (RedCap) helps, but overhead remains significant compared to terrestrial LTE.

Antenna size. Those arrays are physically large. A Starlink terminal is roughly pizza-box sized. Automotive installations hide the antenna in roof panels, but the mechanical envelope is fixed. Compact IoT sensors face hard constraints - a small node cannot mount hardware needed for reliable LEO acquisition and tracking.

Signal propagation. LEO satellites typically use higher frequencies (Ku/Ka, S/L-band) than terrestrial networks. These frequencies suffer rain attenuation and atmospheric absorption, tightening link budget in adverse weather. Urban canyon effect is real - tall buildings block sky view, and maintaining lock on a fast-moving satellite becomes unreliable. GEO offers better weather resilience and sky visibility from a stationary position, but trades latency and complexity.

In summary, these constraints mean NTN works best in open terrain at elevation, where power is available and sky view is guaranteed. Dense urban, indoor, and power-constrained environments are less forgiving.

Permanent roaming and Data sovereignty.

You can’t ‘permanently roam’ using an NTN (it’s viewed as an extension of your home network PLMN). If you cross into a jurisdiction with data sovereignty laws though, your IoT traffic must still route to a local core. If the current provisioned profile doesn’t allow that, then you must swap to one that does. Ensuring you have the correct profiles specified for this scenario and the technical and commercial ability to swap as required is a must.

Should you use NTN for automotive or IoT?

Whether you should use non-terrestrial networks for Automotive and IoT depends on whether your use-case has a real coverage gap or a perceived one. Where coverage is sparse - mountains, oceans, deserts - NTN solves a concrete problem. For asset tracking, agriculture, industrial monitoring, and emergency services in remote regions, it's worth the engineering effort and cost trade-offs. Chinese OEMs like Geely and Zeekr moved first via proprietary LEO constellations (Geespace).

European manufacturers favour 3GPP-compliant approaches through the 5GAA, partly to sidestep regulatory friction around GDPR and NG112. Tesla will likely choose a proprietary Starlink integration according to filed patents, and transition away from their current 4G/5G connectivity.

From experience it’s hard to see an outright winner amongst these approaches in current form. It’s certainly possible to imagine a shared European LEO constellation emerging, although how fast and who funds it are open questions. Similar debates around v2x rumbled on for years without a clear ‘winner’, so I wouldn’t be overly surprised to see different solutions exist in parallel here either.

Where coverage already exists via terrestrial networks, NTN is currently a redundancy feature, not a fundamental capability. Whether that justifies the cost and complexity is a business decision, not a technical one.

Standards will mature, coverage will improve, and costs will compress in accordance with all the usual laws of scale and economics. What remains uncertain is the timeline this happens across, and whether proprietary or standardised approaches dominate in the long-term.

David Turner is the founder of Kói, an independent technology consultancy advising investors, founders, and boards on technology assessment for automotive, telecoms, and IoT infrastructure.

You can reach him at: enquiries@dkoi.design