Technology Trends

Both technologies find things inside buildings. They measure different physical quantities to do it, and that is where the accuracy gap begins. Ultra-Wideband (UWB) times how long a pulse takes to travel. Bluetooth Low Energy (BLE) historically inferred distance from how loud a signal arrived, and now also measures the angle it arrived from and the phase it arrived with.
The published numbers are closer than most buying guides admit, and the thing that separates them is not the radio. It is the wall.
Which is more accurate, UWB or BLE?
UWB is more accurate in hard conditions, and roughly equal in easy ones. With clear line of sight, UWB two-way ranging has been measured at 8.2 to 10.6 cm RMSE, and calibrated Bluetooth Channel Sounding at 6 to 10 cm mean error. Once line of sight breaks, UWB holds up better.
| Property | Ultra-Wideband | Bluetooth Low Energy |
|---|---|---|
| Primary measurement | Time of flight of a wideband pulse | Signal strength, angle, or phase |
| Line-of-sight error, published tests | 8.2 to 10.6 cm RMSE | 6 to 36 cm, method dependent |
| Through a wall, uncorrected | Up to 85.3 cm RMSE | Degrades, peaks in metres |
| Radio in a visitor’s own phone | Some models | Effectively all models |
| Legal transmit ceiling | -41.3 dBm per MHz | 100 mW EIRP |
Read that table as two different questions. “Which is more precise under ideal conditions” is now close to a tie. “Which one still works when a forklift, a fire door and forty people are between the tag and the anchor” is not close at all.
Why does UWB measure distance better?
Because it measures time, not strength, and it does so across a very wide slice of spectrum. A transmitter only counts as UWB if it occupies at least 500 MHz of bandwidth, or a fractional bandwidth of 0.20 or more, under the definitions in 47 CFR 15.503 (current edition).
That bandwidth buys time resolution. A wide pulse is short in time, so the direct path and the reflection that bounced off a steel shelf arrive as two separable events rather than one blurred sum. Signal strength cannot do this. A strength reading has no way to tell a tag three metres away from a tag one metre away behind a body.
The ranging itself is standardised. IEEE 802.15.4z-2020, published 25 August 2020, amended the low-rate wireless networking standard with enhanced UWB physical layers, extra coding options and MAC support for time-of-flight ranging procedures, with the stated aim of raising the integrity and accuracy of a range measurement.
Integrity matters as much as accuracy. If a measurement can be spoofed, it cannot be used for a safety interlock or an access decision, however many centimetres it claims.
How accurate is BLE now that it has direction finding and Channel Sounding?
Better than its reputation, and the reputation is ten years out of date. Bluetooth 5.1 added Angle of Arrival and Angle of Departure, and Bluetooth Core Specification 6.0 added Channel Sounding, which combines phase-based ranging with round-trip time instead of guessing from signal strength.
The measured results are worth reading closely.
| Method | Reported error | Condition |
|---|---|---|
| BLE 5.1 Angle of Arrival | 0.28 degrees, 22 cm | Outdoor, line of sight |
| Bluetooth Channel Sounding | 6 to 10 cm mean | Lab, no channel collisions |
| Bluetooth Channel Sounding | Up to 331 cm peak | 24 colliding 2.4 GHz channels |
| UWB two-way ranging | 8.2 to 10.6 cm RMSE | Three line-of-sight anchors |
| UWB two-way ranging | 85.3 cm RMSE | Two anchors behind walls |
The Angle of Arrival figures come from an experimental study of commercial Bluetooth 5.1 hardware, arXiv:2103.04121 (preprint, 6 March 2021), which reported an average angular error of 0.28 degrees and 22 cm average positioning error outdoors, and angular error about 73% worse indoors. An earlier empirical study of the same mechanism, arXiv:1909.08063 (ACM WiNTECH, 2019), found sub-metre accuracy reachable with two antennas per receiver while centimetre accuracy stayed out of reach, and demonstrated that a malicious device could tamper with packet structure to alter the measured angle.
Channel Sounding is the bigger change. A 2026 preprint on connectionless Channel Sounding, arXiv:2605.17094 (16 May 2026), describes it as phase-based ranging plus round-trip time across 72 RF channels at 1 MHz spacing, and reports mean absolute errors of 6 cm and 10 cm for collision-free device pairs, with 90th-percentile errors of 16 and 20 cm.
Then the same authors stressed it. With 24 of the channels deliberately occupied by overlapping traffic, mean error rose modestly, to 13 and 14 cm. Peak error reached 309 and 331 cm.
That is the trade-off in one line. BLE now reaches centimetre-class accuracy in a quiet 2.4 GHz band, and the 2.4 GHz band in a real warehouse is not quiet.
What does a wall do to each one?
It costs you most of your accuracy, whichever technology you picked. A controlled study of a DecaWave DW1000 UWB system in Sensors 22(21), article 8247, published 27 October 2022, measured error area by area as line of sight was removed.
With three line-of-sight anchors, RMSE was 10.6 cm in one run and 8.2 cm in another. With one anchor behind a wall, raw error rose to 25.2 and 36.4 cm. In the area with only two line-of-sight anchors it reached 85.3 cm, and in the area with none it sat near 40 cm because the geometry happened to be kinder.
Software recovered much of it. Adding a wall-delay model and weighted least squares brought those same areas back to 11.1 to 26.5 cm.
So the honest ranking is this: line of sight first, anchor geometry second, radio technology third. A UWB grid with anchors mounted where the cable tray was convenient will lose to a thoughtfully placed BLE grid, and no datasheet will warn you.
Why does UWB need more anchors than BLE?
Because regulators cap UWB power far below BLE power, and range follows power. UWB survives alongside licensed services by staying under the noise floor, which is a deliberate design decision in the rules rather than a limitation of the hardware.
| Rule | Ceiling | Where it is written |
|---|---|---|
| UWB, United States | -41.3 dBm per MHz, average EIRP | 47 CFR Part 15 Subpart F |
| 2.4 GHz BLE, Europe | 100 mW EIRP, 10 dBm per MHz | ETSI EN 300 328 V2.2.2 |
| UWB in Europe | 30 MHz to 10.6 GHz, location tracking part | ETSI EN 302 065-2 V2.1.1 |
The United States limit for the 3100 to 10600 MHz band is listed as -41.3 dBm in the emission table of 47 CFR Part 15 Subpart F (2024 edition). Spread across a 500 MHz channel that totals about -14.3 dBm of radiated power. Europe’s 2.4 GHz band permits 100 mW, which is 20 dBm, under ETSI EN 300 328 V2.2.2 (July 2019). The gap is roughly 34 dB.
European UWB location tracking has its own harmonised standard, ETSI EN 302 065-2 V2.1.1 (November 2016), which covers location tracking devices specifically under the Radio Equipment Directive.
Budget for the consequence, not the radio price. A denser anchor grid means more mounting points, more power drops or more battery-powered anchors, more cable, and more commissioning time per square metre.
What do power and battery life cost you?
Update rate decides battery life, far more than the choice of radio does. The Channel Sounding preprint modelled a CR2032 cell at 240 mAh doing four measurements per cycle, and the spread between a fast and a slow update interval is two orders of magnitude of calendar time.
| Update interval | Modelled CR2032 life | Basis |
|---|---|---|
| 1 second | About 4 months | Four ranges per cycle, 0 dBm assumed |
| 30 seconds | About 5 years | Same model, fewer cycles |
Those figures are modelled, not measured on a shipped tag, and the authors state the 0 dBm number is an assumed halving of their measured +8 dBm case (arXiv:2605.17094, 16 May 2026). Treat them as the shape of the curve rather than a warranty.
The same curve applies to UWB, with a steeper slope, because a two-way ranging exchange costs several packets per fix rather than one advertisement. Ask any vendor for battery life at your update rate, in your anchor density. A battery figure without an update interval attached is not a number.
When is UWB the right choice?
Choose UWB when a wrong answer of half a metre costs you something. That is the test, and it is narrower than it sounds. Pallet-level inventory in racking, tool tracking on an assembly line, robot and person separation on a shop floor, and anything where a ranging measurement feeds a safety or access decision all pass it.
Secure ranging is the underrated half. Because 802.15.4z extends the MAC for time-of-flight procedures and tightens measurement integrity, UWB is the one of the two that can defend a distance claim, which is why it ended up in car keys.
Also choose UWB when you control the tag. If you issue the badge, the asset tag and the wristband, you decide what radio is in it.
When is BLE the right choice?
Choose BLE when the thing being located is a phone you do not own, or when there are thousands of tags and a tight budget. Visitor wayfinding is the clearest case. BLE is in essentially every phone in the building, and UWB is in a minority of models.
Scale is the second case. Tag cost, anchor count and installation labour all favour BLE, and the accuracy you lose is often accuracy you had no use for. Finding a wheelchair to the correct corridor is a solved problem at two metres.
The UWB ecosystem also remains split across implementations. A 2022 survey of UWB standards and organisations, arXiv:2202.02190, treats cross-vendor interoperability as an open question rather than a settled one.
Should you run UWB and BLE together?
Usually yes, and the split follows the money. BLE carries discovery, identity and the low-rate heartbeat from every tag on site. UWB is switched on for the zones and the assets where centimetres change a decision.
This is not a compromise. It is a cost model. A hybrid design lets you pay for an anchor grid only over the area that needs one, and it keeps visitor-facing features on the radio that visitors already carry.
The hard part is the handover logic: deciding when a tag earns a UWB fix, and reconciling two position estimates with different error characteristics without letting the less accurate one drag the better one around.
How do you decide for your own building?
Start from the decision the position feeds, not from the accuracy number. Write down the worst wrong answer you can tolerate, how often you need a fix, who owns the device being tracked, and how much of the floor has line of sight. The technology falls out of those four answers.
| If you must | Choose | Because |
|---|---|---|
| Locate a pallet to 15 cm | UWB | Time of flight separates direct path from reflections |
| Guide a visitor’s own phone | BLE | The radio is already in their pocket |
| Track thousands of tags cheaply | BLE | Tag cost and battery life dominate |
| Defend a distance for access or safety | UWB | 802.15.4z targets ranging integrity |
| Cover a large floor on a fixed budget | BLE | Higher legal power, fewer anchors |
Then survey before you buy. The DW1000 study above moved from 10 cm to 85 cm inside one building, and nothing about the hardware changed between those two measurements.
If you are weighing UWB against BLE for a specific site, the useful next step is a conversation about that site: its floor plan, its wall construction, what you need to locate, and how often you need to know. Mapsted works on indoor positioning and wayfinding deployments and can walk through the trade-offs for your building. Book a scoping call.
Frequently asked questions
Is UWB always more accurate than BLE?
No. With clear line of sight the published figures overlap: UWB two-way ranging at 8.2 to 10.6 cm RMSE, calibrated Bluetooth Channel Sounding at 6 to 10 cm mean error. The gap opens when walls, metal racking and crowds remove line of sight, where UWB’s wide bandwidth lets the receiver separate the direct path from its reflections.
Does Bluetooth Channel Sounding make UWB unnecessary?
Not yet. Channel Sounding measures phase across 72 one-megahertz channels, so it needs those channels reasonably clear. In a test with 24 channels deliberately collided, mean error stayed near 13 cm but peak error reached 331 cm (arXiv:2605.17094, May 2026). A busy 2.4 GHz environment looks more like the stressed test than the clean one.
Why does a UWB deployment need more anchors?
Transmit power, not accuracy. UWB is held to -41.3 dBm per MHz of average EIRP in the 3100 to 10600 MHz band under 47 CFR Part 15 Subpart F (2024 edition), while 2.4 GHz equipment in Europe may radiate 100 mW. Less power means shorter usable range, so the anchor grid is denser for the same floor area.
Can both technologies run on the same site?
Yes, and it is the common design. BLE handles discovery, identity and the heartbeat from large tag populations, while UWB covers the zones and assets that need centimetre certainty. The real design question is which assets justify a UWB anchor grid, and on most sites the honest answer is a minority of them.