A fleet platform assumes a connection; the vehicle keeps losing one. Tunnels, remote highways, borders, underground car parks — the engineering problem is not the dashboard, it is keeping data flowing and correct across coverage the vehicle will inevitably lose, and behaving sensibly while it is gone.
| Bearer | Strength | Where it fits |
|---|---|---|
| 4G / 5G | Throughput, low latency, video | Live telematics and camera streams |
| LTE-M | Mobility and power balanced | Moving assets, firmware updates |
| NB-IoT | Deep coverage, lowest power | Slow or fixed sensors on the vehicle |
| NTN satellite | Works with no cellular at all | Remote, maritime and dead-zone fallback |
| Wi-Fi at the depot | Bulk offload when parked | Dumping logs and video cheaply |
| Store-and-forward | No live link needed | Correctness across every gap above |
A fleet management platform is a database with a map on top, and it assumes the data arrives. The hard engineering sits one layer down, on the device in the vehicle, where the connection is intermittent by nature. When the link drops, the device cannot simply stop: it has to keep timestamping events against a trusted clock, keep estimating position by dead-reckoning through the GNSS gap, and buffer everything so that when coverage returns the platform receives a correct, ordered history rather than a hole.
That turns connectivity into a tiering decision rather than a single choice of radio. Live safety events — a harsh brake, a panic button, a crash pulse — have to go now, over whatever bearer is up, even a satellite link that costs per byte. Routine telemetry can wait, batch and compress, and bulk data like video can wait for Wi-Fi at the depot. A fleet that streams everything live over cellular pays for bandwidth it did not need and still loses data in the dead zones; the design is about matching each class of data to the cheapest bearer that meets its deadline.
Two constraints then shape the rest. Vehicles cross regions, so the device needs a roaming strategy — an eUICC or multi-network SIM — that keeps it legal and connected across borders without a truck roll to swap hardware. And a vehicle spends much of its life with the ignition off, so the device has to persist on its own power budget, staying reachable enough to report a theft or a tow without flattening the battery. None of this is visible on the dashboard, which is exactly why it is where the work is.
Because coverage, cost and power do not allow it. 5G is excellent where it exists, but a fleet drives through tunnels, remote routes and borders where it does not, and streaming everything live is expensive and still loses data in the gaps. The right design tiers bearers — live cellular for urgent events, low-power cellular for telemetry, satellite for dead zones, Wi-Fi for bulk — rather than assuming one radio covers every case.
The device treats the link as unreliable by design. It timestamps every event against a trusted clock, estimates position by dead-reckoning while GNSS is unavailable, and buffers everything locally. When the connection returns it forwards an ordered, gap-free history, so the platform sees what actually happened rather than a hole followed by a jump.
With a roaming strategy built into the device — typically an eUICC or multi-network SIM that can attach to a compliant local network as the vehicle moves between regions, keeping it both legal and connected. The alternative, swapping SIMs or hardware per region, does not scale to a moving fleet.
Written for engineers. Share it with one.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.