UWB vs BLE Channel Sounding After Bluetooth 6.0 Silicon Shipped

What shipped with Bluetooth 6.0, and what did not?

The Bluetooth SIG adopted Core 6.0 in September 2024, and Channel Sounding was the headline feature: two-way ranging that combines phase-based measurement with round-trip timing across up to 72 channels in the 2.4 GHz band. Through 2025 the silicon arrived. Nordic's nRF54L15 ships with Channel Sounding support, dev kits are in distribution, and module makers are sampling tags.

Here is what did not ship: an upgrade path for your installed base. Channel Sounding requires a radio that can step channels and measure phase to tight tolerances. A BLE 5.x gateway bolted to your ceiling cannot gain that through firmware, and neither can the beacons already in your racking. Every anchor that ranges against a tag must be new 6.0-capable hardware.

That flips the most common argument for BLE positioning. The pitch was always to reuse the Bluetooth infrastructure you already have. For Channel Sounding, that infrastructure does not exist in any warehouse yet. You are buying and cabling new anchors either way, which puts UWB back on equal footing for the infrastructure line of the budget. Your installed beacons will never do channel sounding.

How accurate is BLE channel sounding between steel racking?

Indoor ranging accuracy is set less by the protocol than by bandwidth, because bandwidth determines how finely a receiver separates the direct path from reflections. The rule of thumb is the speed of light divided by bandwidth. A UWB channel under IEEE 802.15.4z is 500 MHz wide at 6.5 or 8 GHz, so it resolves reflected paths about 0.6 m longer than the direct one. Channel Sounding sweeps roughly 80 MHz of the 2.4 GHz band, so any reflection within about 3.7 m of extra path length blends into the direct signal. A warehouse aisle lined with steel uprights and shrink-wrapped pallets produces exactly those reflections.

The consequence shows up in reported numbers:

• Channel Sounding, line of sight: vendor demos in the 10 to 50 cm range.
• Channel Sounding, between racking: plan for 0.5 to 1 m typical error, with occasional multi-meter outliers when the direct path is blocked.
• UWB, same environment: 10 to 30 cm holds, because the roughly 2 ns pulse lets the receiver pick the first-arriving path even without line of sight.

If your use case tolerates meter-class error, Channel Sounding clears the bar. Bandwidth is physics, and 80 MHz is narrow.

Where channel sounding wins: the tag you already ship

Tag economics are where Bluetooth 6.0 lands its punch. A Channel Sounding tag is one SoC, one 2.4 GHz antenna, and a coin cell, and that SoC is the same part already handling the tag's communication, firmware updates, and sensors. Ranging rides along at near-zero incremental BOM.

A UWB tag carries that same BLE SoC for provisioning and control, plus a UWB transceiver such as Qorvo's DW3110 or NXP's Trimension SR040, a second antenna tuned for 6.5 or 8 GHz, and its matching front end. At volume that adds roughly 3 to 5 dollars per tag. Tag 50,000 pallets and totes and the delta is 150,000 to 250,000 dollars of BOM before you count the UWB radio's higher transmit energy per fix.

So ask what question each asset class actually needs answered. Which aisle is this pallet in, and did this tote leave the staging zone, are meter-class questions, and most warehouse assets only ever get asked meter-class questions. Meter-class questions deserve meter-class radios.

Where UWB keeps winning: vehicles, people, and sub-30 cm

Some warehouse problems do not tolerate a meter of error. Forklift anti-collision needs to know whether a pedestrian tag is 1.5 m or 3 m from the vehicle, and it needs that answer many times per second. Bay-level putaway verification has to distinguish adjacent pallet positions about 1.4 m apart on standard racking. AMR docking and pick-to-position confirmation sit in the same tier.

UWB holds these jobs for two reasons:

• Accuracy under multipath. The 500 MHz channel keeps error at 10 to 30 cm even with bodies and racking in the way.
• Update rate. A double-sided two-way ranging exchange under 802.15.4z completes in a few milliseconds, so anchors can service fast-moving tags at 10 Hz or better. A Channel Sounding procedure steps across dozens of channels per measurement, which costs more airtime per fix and crowds the same 2.4 GHz band your WiFi and existing BLE traffic already occupy.

Nothing in Bluetooth 6.0 changed either constraint. Fast metal needs fast, wideband ranging.

So which radio wins after the silicon shipped?

Neither, outright. The honest answer after Bluetooth 6.0 hardware shipped is that the decision moved from which radio to which radio for which asset class. A split that holds up in practice:

• Channel Sounding: pallets, totes, returnable containers, and tools that only need aisle or zone resolution. High tag counts, lowest BOM, one radio per tag.
• UWB: forklifts, AMRs, pedestrian safety tags, and any workflow that verifies position at bay level or tighter. Lower tag counts, higher per-fix precision.

Since neither path reuses installed gateways, plan the anchor layer once. An anchor that carries both a 6.0-capable BLE radio and a UWB transceiver is a straightforward design, and a shared anchor grid feeding one location engine costs far less than two sequential rollouts. The expensive mistake is betting the whole site on one radio in 2026, then re-cabling in 2028 when the other tier of use cases arrives. Segment by accuracy tier, not by radio loyalty.

Why license the positioning layer instead of rebuilding it?

Whichever radio you choose, the radio is the easy part. The hard part is turning noisy ranges into positions a WMS can act on: rejecting multipath outliers, fusing RF ranges with inertial and vision data, and keeping an estimate stable while a tag passes behind a forklift. Teams routinely spend a year or more getting from a ranging demo to positions that survive a live warehouse.

Position Imaging licenses that layer. The portfolio spans hundreds of granted patents in RF ranging, real-time positioning, computer vision, and machine learning, including US 11,774,249, US 12,079,006, US 12,066,561, and US 12,000,947, and it is cited by Apple, Bosch, and other major firms. Licensing buys two things a from-scratch build cannot: estimation methods that work on either radio, so your Channel Sounding and UWB tiers share one engine, and freedom to operate in a field where the core ranging patents predate Bluetooth 6.0 by years. That turns the radio decision above into a configuration choice instead of a second engineering program. License the math, ship the product.