In November 2025, TRAI drove a test rig 297.8 km down the Raipur–Jagdalpur highway with SIMs from all four national operators. On Airtel, 20.86% of the samples it logged were poor signal. On Jio, 24.58%.
The same campaign's city drive through Durg and Bhilai, that same month, found poor signal in no more than 3.18% of samples on any network. TRAI published both figures side by side.
Same operators, same month, same test equipment. The only thing that changed was the road. That is almost always why a fleet dashcam feed goes black: fleet dashcam connectivity fails on the route, not on the camera — and the route is the one part of the system nobody in the fleet office chose.
The camera is still recording
Start with the thing that matters most, because it changes how worried you should be. When the live tile goes black, the recorder in the vehicle almost always keeps writing video to its own card or drive. What died is the live stream — the small, separate feed sent over the cellular link so someone can watch in real time.
The two are different streams by design. A current two-channel fleet dashcam documents recording at 6 to 25 Mbps per channel but streams live view at roughly 600 kbps, a small fraction of the recording rate. The arithmetic of what fits down a moving link is a post of its own; the point here is simpler. Losing the live feed and losing the footage are two different events, and most dead zones cause only the first.
So the real question is not "is the camera broken?" It is "why does a link that works everywhere I test it keep failing where the trucks actually drive?"
Coverage maps are drawn for people standing outside
Most coverage maps and coverage statistics answer the question: could a phone, held outdoors at this spot, get a usable signal? A dashcam modem is not outdoors. It sits inside a steel cab, frequently behind the dashboard.
The UK regulator is one of the few that publishes the difference, and the gap is not small. Ofcom's Connected Nations 2025 report puts 4G coverage on UK B roads at 95–97% per operator outside a vehicle — and 81–83% inside one. On motorways and A roads it is 98–99% outside, 90–93% inside.
The reason is in Ofcom's methodology: a road counts as covered in-car only where the signal is 10 dB stronger than the outdoor threshold, −95 dBm against −105 dBm. Ten decibels is a tenfold difference in received power, and it is roughly what the vehicle body is assumed to cost. Two caveats belong with those numbers: they are modelled from operators' own predictions rather than measured, and they describe the UK, one of the better-covered road networks anywhere.
The truck is part of its own dead zone. A stretch that looks green on a coverage map can be marginal inside the cab — and marginal is exactly the condition under which a live stream gives up first.
Which raises the obvious question: what do real routes look like when someone actually drives them?
What a route actually looks like
TRAI's independent drive tests are the closest thing to an answer, because the regulator drives the routes itself with handsets from every operator and publishes per-route results. Four tests show the range.
Route (TRAI test) | What was measured | Result by operator |
|---|---|---|
Durg–Bhilai city drive, Nov 2025 | Samples with poor signal | Airtel 0.87% · Vi 0.87% · BSNL 2.36% · Jio 3.18% |
Raipur–Jagdalpur highway, 297.8 km, Nov 2025 | Samples with poor signal | Vi 7.16% · Airtel 20.86% · Jio 24.58% · BSNL 26.29% |
Jodhpur–Ahmedabad rail route, 456.9 km, May 2026 | Samples with poor signal | Jio 2.6% · Airtel 7.0% · Vi 19.9% · BSNL 31.2% |
Munabao–Tanot highway (NH-70), 311.4 km, Jun 2026 | Samples with NO coverage | Jio 6.0% · Airtel 40.5% · Vi 81.8% · BSNL 83.6% |
Percentages are computed from TRAI's published sample counts. The NH-70 route runs along the Rajasthan border desert and is an extreme case — but it is a national highway that trucks use, and it is where the difference between operators stops being academic.
Two things stand out. First, highways are an order of magnitude worse than cities on the same networks, which is the whole of the symptom. Second, no operator wins everywhere. Vi had the cleanest signal on the Raipur–Jagdalpur highway and among the weakest coverage on NH-70. Jio was the weakest of the private networks on one highway and by far the strongest on the other.
For a fleet, that makes the SIM a route decision, not a procurement line item. The operator that works at head office tells you nothing about the operator that works on the corridor your trucks run every night.
Why the feed dies instead of degrading
A reasonable objection: a weak signal should mean a blurrier picture, not a black one. Sometimes it does. On a moving vehicle it usually does not, for three reasons.
Upload is the scarce direction. Live video from a vehicle flows up, and cellular networks are built for traffic flowing down. One of the most thorough recent driving studies — a 5,700 km coast-to-coast campaign across three US operators — found uplink throughput an order of magnitude lower than downlink, and noted that operators often keep uplink traffic on slower LTE and low-band layers even when faster 5G is available. Its 2024 repeat saw median uplink rise between 2.25× and 5.8× depending on operator — while the bottom quarter of samples on one operator's LTE layer still sat at or below about 1 Mbps.
The link keeps changing underneath the stream. The same study counted a median of one to three handovers per mile while driving. Each handover is brief, but a live stream is a continuous flow: every interruption drains the player's buffer, and a real-time stream keeps that buffer deliberately small so the picture stays live.
Recovery is not instant when the signal returns. Real-time video adapts its bitrate down as the link weakens — bandwidth estimation is why a call downgrades before it drops. But once the link actually drops, the session has to find a path again, and on mobile networks that path very often has to go through a relay, for the reasons carrier NAT creates. The decoder then waits for a fresh keyframe before it can draw anything. A two-second outage can easily look like a ten-second black tile.
Add these together and the pattern fleet managers describe — fine in town, black on the highway, slow to come back — is exactly what the physics predicts. None of it is a camera fault.
What a fleet manager can actually change
You cannot engineer the network. You can stop being surprised by it. Four things are in the fleet's control, none of which require touching the architecture between vehicle and command centre.
- Know your routes, not your city. Check TRAI's drive-test results for the corridors you actually run. If your recorders or telematics units log signal strength against GPS, you already have your own drive test, with far more miles in it than TRAI's.
- Choose the SIM per route. The table above is the argument. Where recorders support dual SIMs, pair operators whose coverage fails in different places, not the two cheapest.
- Get the antenna out of the cab. If the vehicle body costs roughly 10 dB, a roof-mounted external antenna is the one hardware change that attacks the problem directly rather than working around it.
- Set the expectation honestly. On a moving vehicle, live view is best-effort. The recording is the record. Anyone promising uninterrupted live video on rural highways is describing a network that does not exist yet.
That last point moves the question from "how do we keep the feed alive?" to "what happens to the footage the network drops?" — which has a proper engineering answer.
The Bottom Line
A black tile on a fleet dashboard is almost always a coverage event, not a hardware fault. Regulator data shows highways losing signal an order of magnitude more often than cities on the same networks, coverage figures are measured outside the vehicle rather than inside it, and no single operator is best on every route.
Treat live view as best-effort, choose SIMs by corridor, get the antenna outside the cab, and design for the gap rather than against it.
What's Next
The gap itself is the next problem: store-and-forward for fleet video covers how recorders buffer through a dead zone, what still gets lost, and how the missing minutes are reconciled afterwards. And if the question is whether you needed the live feed at all, when a fleet actually needs a live feed works through it use case by use case.
Frequently Asked Questions
Why does my fleet dashcam go offline on highways?
Because highway coverage is far patchier than city coverage, and the camera's modem sits inside the vehicle, where signal is weaker still. In TRAI's November 2025 drive test, 21–25% of samples on Airtel and Jio were poor signal on the Raipur–Jagdalpur highway, against under 3.2% on any network in the same month's city drive. The camera is usually still recording locally; only the live stream has dropped.
Is footage lost when a fleet dashcam loses connection?
Usually not. Most fleet dashcams and MDVRs record continuously to a local card or drive, and the live stream is a separate, much smaller feed. A dead zone interrupts watching, not recording. What can be lost depends on how the recorder buffers and uploads afterwards, which is the store-and-forward question.
Which mobile network is best for fleet dashcam connectivity in India?
It depends on the route, and the regulator's own data shows why. On one 2025 highway test Vi had the cleanest signal; on a 2026 border highway Vi had no coverage for 82% of samples while Jio was missing for 6%. Check TRAI's drive-test results for your corridors, or your own recorders' signal logs, rather than choosing on head-office coverage.
Does 5G fix fleet dashcam connectivity?
It raises upload speed where 5G exists, but it does not fill coverage gaps on rural highways, which is where feeds die. Driving studies also show operators often keep upload traffic on LTE and low-band layers. Treat 5G as a faster link in covered areas, not as a cure for dead zones.
How do I improve fleet dashcam connectivity?
Choose SIMs by route using drive-test data, pair operators whose coverage fails in different places, and use a roof-mounted external antenna so the cab is not attenuating the signal. On the platform side, Samvyo is one option: based on SFU architecture, it keeps the media path, TURN relays and recording on infrastructure you choose, so relays can sit close to the regions your fleet actually runs in.
Can live fleet video work in a dead zone?
No platform can stream through an absent network. What the platform can control is what happens around the gap: how quickly a feed recovers, where the relays sit, and whether the footage recorded during the outage reaches the command centre afterwards. Platforms that keep recording on infrastructure you control, Samvyo among them, let you decide where that post-outage footage lands and how long it is kept.
