Bluetooth 6.0 Channel Sounding Tag Cost: When to Switch from UWB

What accuracy does Bluetooth 6.0 channel sounding deliver in a real warehouse?

Channel sounding entered the Bluetooth Core 6.0 specification in September 2024. It combines phase-based ranging with round-trip time measurements, hopping across up to 72 channels in the 2.4 GHz band for roughly 80 MHz of aggregate bandwidth. On early silicon such as Nordic's nRF54 series, vendor demonstrations show 30 to 50 cm ranging error with clear line of sight.

A warehouse is not a demo floor. Steel racking, wire-guided forklifts, and stretch-wrapped pallets of canned goods create reflections and blocked paths on most reads. In that environment, plan on:

• 1 to 3 m typical error in racked aisles
• Worse than 3 m in deep NLOS, such as a tag buried mid-pallet in a full bay
• Sub-meter only in open staging and dock areas

That is still a large upgrade over RSSI beaconing, which struggles to do better than 5 to 10 m zones. It reliably answers "which aisle, which staging lane, which dock door." Plan for meters between racks, not centimeters.

Why does UWB still own everything below one meter?

The gap is bandwidth, and bandwidth sets time resolution. An IEEE 802.15.4z UWB channel is 499.2 MHz wide, which lets a receiver separate signal paths arriving about 60 cm apart. The direct path stands out from the rack reflection, so two-way ranging holds 10 to 30 cm even in cluttered aisles.

Channel sounding works with about 80 MHz of aggregated narrowband hops. That resolves paths roughly 3.7 m apart. Phase-based processing interpolates below that, and it works well when the direct path is strong. When a steel upright blocks the direct path and a reflection dominates, the estimate walks, and no firmware update changes the spectrum available in 2.4 GHz.

So the workloads that made warehouses buy UWB in the first place stay UWB workloads: forklift and AMR positioning at 10 to 30 cm, bin-level putaway verification, and pedestrian proximity warning around powered trucks. Bandwidth is the moat, and it is physics.

What does each radio cost per tag at warehouse scale?

Tag economics is where channel sounding wins, because it rides Bluetooth volume measured in billions of units per year.

• Channel sounding tag: built on a standard BLE SoC like the nRF54L15, single radio, coin cell or small pouch cell. Expect roughly $5 to $10 per tag at 10,000-unit volumes, with 3 to 5 years of life at slow update rates.
• UWB tag: dual radio in practice, since most designs pair UWB with BLE for wake-up and configuration, plus a larger battery to feed wider-bandwidth transmissions. Expect roughly $15 to $30 per tag at the same volumes.

Infrastructure narrows the gap less than vendors imply. Both radios need anchor grids on a 15 to 30 m pitch in racked space, and UWB TDoA adds tight time synchronization between anchors. But tags dominate the math: at 20,000 tracked pallets and totes, a $10 to $20 per-tag delta is $200,000 to $400,000 before installation. At fleet scale, the tag line is the budget.

When should you switch from UWB to channel sounding?

Treat it as a per-asset-class decision, not a site-wide one. Switch an asset class to channel sounding when all three hold:

• The location questions are zone level: which aisle, which staging lane, which trailer yard row. 1 to 3 m answers them.
• The class is large, roughly 5,000 tags or more, so per-tag cost and battery servicing dominate operating cost.
• Reads are mostly line of sight or near it: totes on conveyors, pallets in staging, returnable containers at docks.

Stay on UWB when the workload is vehicle guidance, bin-level accuracy, or anything safety-adjacent where a 3 m error is an incident report. And time the move honestly: channel sounding silicon has shipped since late 2024, but multi-vendor interoperability and locate-engine maturity vary, so run a two-radio pilot in your worst aisle before committing a fleet. Your densest racking decides, not the datasheet.

Can you run both radios without doubling the infrastructure?

Yes, and most warehouses above 200,000 square feet should. The pattern that works: UWB on the few hundred assets that move fast and matter most, forklifts, AMRs, high-value tooling, and channel sounding on the tens of thousands of pallets, totes, and returnable containers where $8 tags make universal tagging affordable for the first time.

Hardware supports this. Anchor vendors increasingly ship combo units carrying both a UWB radio and a CS-capable BLE radio on one mount, one PoE drop, one commissioning pass.

The hard part is software. Two ranging modalities with different error profiles must land in one coordinate frame, with fusion logic that weights a 20 cm UWB fix differently from a 2 m channel sounding fix, and hands assets off cleanly when a UWB-tracked forklift picks up a CS-tracked pallet. That fusion layer is where in-house teams burn quarters. Budget for the fusion layer, not the anchors.

How do you switch radios without rebuilding the stack or the patent exposure?

Here is what the radio debate hides: the techniques that make either system work, two-way ranging protocols, phase-based distance estimation, multipath mitigation, fusing RF ranging with inertial or vision data, are patented independently of whether the waveform is 499.2 MHz UWB or 80 MHz of Bluetooth hops. Swapping silicon does not reset your freedom-to-operate analysis. It reopens it.

That argues for licensing the ranging layer rather than rebuilding it per radio. Position Imaging licenses hundreds of granted US patents in real-time positioning and radio-frequency ranging, a portfolio cited by Apple, Bosch, and other major firms, including recent grants such as US 11,774,249 and US 12,079,006. A license gives you proven positioning methods that apply across radios plus a cleared path on the IP, so a UWB deployment today can add channel sounding tags next year without a second clearance cycle. Teams building on licensed IP ship in months instead of re-deriving multipath handling from scratch. License the ranging layer, then swap radios freely.