Bluetooth Channel Sounding explained.
In our view, Channel Sounding is the most significant change to Bluetooth positioning in years, and also the most over-claimed. This is what it measures, what independent testing actually records, and the constraint that decides whether it belongs in your building.
What it is
Channel Sounding arrived in Bluetooth Core Specification 6.0; the Bluetooth SIG announced it on 3 September 2024. It lets two Bluetooth devices measure the distance between them directly, rather than inferring closeness from signal strength.
One point to settle before any vendor conversation: Channel Sounding is an optional feature of the specification. A device advertised as Bluetooth 6.0 does not necessarily support it, and it generally requires Channel Sounding-capable silicon; check support with the chip vendor rather than assuming a firmware update will add it. Both ends of the measurement need capable radios.
How it measures distance
Two methods run together, and they do different jobs.
Phase-Based Ranging (PBR) is the precision measurement. The two devices exchange unmodulated tones across up to 72 physical channels, and distance is derived from how phase shifts with frequency. The specification deliberately does not mandate the estimation algorithm, which is why results differ measurably between silicon vendors using the same standard.
Round-Trip Timing (RTT) is coarser but carries the security. Cryptographically randomised packets are timed out and back, which bounds the distance in a way an attacker cannot shorten by amplifying a signal. It also cross-checks the phase measurement for consistency.
What accuracy to actually expect
This is where published figures diverge most, so it is worth setting them side by side.
| Source | Figure | Nature of the claim |
|---|---|---|
| Bluetooth SIG | “Centimeter-level accuracy”; “typically ±50 cm up to 100 meters” | Standards body positioning |
| Ghent University / University of Antwerp (IEEE Access, 2025), Nordic nRF54L15, single antenna | Outdoors, more than 90% of 800+ measurements within 2 m; indoors to 20 m, the best algorithm kept errors below 2.5 m in 87% of cases | Peer-reviewed measurement study |
| Graz University of Technology (EWSN 2026), Nordic and Silicon Labs kits | Relative orientation shifted the calibration offset by up to 1.38 m; single-antenna outlier rates of about 10–18%; dual antennas cut outliers to about 1–4% and improved RMSE by 47.5% | Peer-reviewed measurement study |
The honest summary: sub-metre ranging is realistic at short range, in line of sight, with good antenna design, and dual-antenna hardware makes a large difference. The Ghent/Antwerp authors conclude that single-antenna setups currently behave more like zone-based systems than precise indoor localisation. Centimetre-level results in a working building are not yet an established independent result, and device orientation alone can move results by more than a metre on the same hardware.
If a vendor quotes 10 cm, the right question is not whether they are lying. It is which of those rows they are quoting.
It returns a distance, not a position
Channel Sounding produces a one-dimensional range between two devices. That is not a location.
To convert ranges into a position on a floor plan you need at least three anchors with known coordinates and a solver on top, and none of that is in the specification. Any proposal that presents Channel Sounding as a complete replacement for an RTLS platform is describing an architecture that still has to be built.
The constraint nobody quotes
This is the number that decides most business cases, and it rarely appears in vendor material.
Channel Sounding runs over a Bluetooth LE connection between each measuring pair, one to one. Every extra tag an anchor ranges with shares the same radio time. At the time of writing (September 2026), Silicon Labs’ software, for example, supports up to four reflector connections per initiator, and its guidance is to lengthen the procedure interval in proportion to the number of connections, which divides the update rate between them. The Ghent/Antwerp study also reports that transferring the raw measurement data over Bluetooth can take up to 500 ms.
Compare that with ultra-wideband TDoA, where anchors passively receive short transmissions from many tags, and the economics invert. The per-unit price of an anchor is not the number that matters; the anchor count is. For dense asset tracking, the connection model pushes that count up. Anyone presenting Channel Sounding as the low-cost route to dense tracking has not modelled the anchor count.
Where it is genuinely strong: security
Proximity built on signal strength is trivially defeated. An attacker amplifies the signal and the receiver believes a device is closer than it is, which is how relay attacks against keyless entry work.
Channel Sounding was designed against that. Round-trip timing bounds distance in a way amplification cannot shorten, tone and packet sequences are randomised, and the two measurement methods are cross-checked for inconsistency. For access control, digital keys and any case where proximity authorises something, this is a real improvement rather than a marketing one.
The fair caveat is that it is designed to be resistant, not immune, and implementations differ between silicon vendors.
Where it fits, and where it does not
Good fit: one-to-one proximity where the distance authorises something — keyless entry, device unlock, tool-to-operator presence, find-my. Room-level occupancy. Retrofitting onto an existing Bluetooth estate where the alternative is signal-strength guesswork.
Poor fit: dense asset tracking, high update rates, anything needing centimetre accuracy, safety cases such as forklift and pedestrian separation, and any deployment where the anchor count drives the budget.
There is also a timing question. Channel Sounding's advantage is meant to be device ubiquity, but that depends on handsets shipping it with access for third-party applications. Android 16 added a Ranging API that includes Channel Sounding, but support still depends on each handset’s hardware. A business case that rests on phone support should be tested on the devices your people actually carry.
What we would advise
Do not choose between Channel Sounding and ultra-wideband on accuracy alone. Work out what the location data has to authorise or trigger, how many tagged things there are, and how often you need a fresh position. Those three answers usually settle it before any radio comparison begins.
If a vendor proposes Channel Sounding for dense tracking, ask for the anchor count, the tags-per-anchor figure and the resulting update rate in writing. If a vendor dismisses it entirely, ask how they handle relay attacks on proximity.
We hold no vendor relationships and take no referral commission, so our answer is whichever of these costs you less to run. The full comparison against UWB is here, and the wider picture is in our technology comparison.
Sources
Standards and vendor facts on this page come from the public sources below, checked on 25 September 2026. Vendor statements are reported as the vendor’s own claims.
- Bluetooth SIG — Bluetooth Channel Sounding press release (3 Sep 2024)
- Bluetooth SIG — Channel Sounding: how it works (PBR, RTT; “typically ±50 cm up to 100 m”)
- Wieme et al., “Performance Evaluation of Bluetooth Channel Sounding on Commercial Hardware”, IEEE Access, Oct 2025 (Ghent University / University of Antwerp, imec)
- Dzafic et al., “Orientation Matters: An Empirical Study on Ranging Performance of Bluetooth 6.0 Channel Sounding”, EWSN 2026 (Graz University of Technology)
- Silicon Labs — Channel Sounding developer guide: capabilities and multi-connection scheduling
- Android Developers — Range between devices (Android 16 Ranging module: UWB, Bluetooth Channel Sounding, Wi-Fi RTT)
Last updated: 25 September 2026