BLE Channel Sounding Plus UWB: Zone Warehouse Tracking by Accuracy

Your warehouse has three accuracy zones, not one

Walk a 20,000 square meter DC and the accuracy requirement changes by aisle. In reserve racking the question is which bay, and a standard bay runs about 2.7 m wide for three pallet positions, so 1 to 3 m accuracy answers it. In staging and putwall areas, pallets sit shoulder to shoulder near their 1.0 by 1.2 m footprint, so telling neighbors apart takes sub-meter. At dock doors on 3.5 m centers, at AMR handoff points, and inside automated cells, the system has to confirm one position against the next with 10 to 30 cm and sub-second latency, because a wrong-door load costs a truck roll.

Here is the part single-radio specs miss: most tagged inventory is pallets sleeping in racking, the loosest zone. Speccing UWB everywhere prices every pallet for the dock door it may never see. Speccing Channel Sounding everywhere fails the dock and the robot outright. Map the zones before you pick a radio.

What each radio delivers between steel racks

Bluetooth 6.0 Channel Sounding hops across roughly 72 channels covering about 80 MHz of the 2.4 GHz band. That bandwidth resolves time of flight to about 12.5 ns, meaning reflections that travel within roughly 3.7 m of the direct path smear into it. On an open bench, phase-based ranging demos from Nordic (nRF54L15) and NXP (KW47) land in the 0.3 to 0.5 m range. Between steel uprights, wire decking, and shrink-wrapped metal, plan on 1 to 3 m.

802.15.4z UWB transmits on a 499.2 MHz channel, time resolution near 2 ns, so it separates any reflection that travels more than about 60 cm farther than the first path. Qorvo DW3000-class and NXP Trimension radios hold 10 to 30 cm in the same aisles and keep a usable first path through a loaded pallet. Bandwidth is physics, not a firmware setting.

What the hardware bill looks like per 10,000 tags

Rough volume numbers, but they frame the decision.

• Channel Sounding tags ride the 2.4 GHz SoC the tag already carries for telemetry, so finished tags land around $5 to $15.
• UWB tags add a second radio, front end, and antenna, roughly $5 to $10 more BOM, with finished tags commonly $20 to $50.
• UWB anchors need time sync for TDoA and typically sit on 15 to 30 m spacing; BLE locators are cheaper per unit and reuse 2.4 GHz site knowledge.

Work the example: 10,000 pallet tags, and suppose zoning puts 8,500 of them in racking where Channel Sounding suffices. At a $7 per-tag delta that is about $60,000 saved on tags alone. Spend a fraction of it on a denser UWB anchor grid over the 10 to 15 percent of floor that handles docks, staging, and automation. Pay for centimeters only where centimeters pay.

Update rate and tag count decide as much as accuracy

Channel Sounding is a connected procedure. Tag and locator pair up, step across the channel map, and exchange phase measurements, so each fix costs a connection and tens of milliseconds of radio time per locator. A fleet of thousands of tags therefore refreshes on a cycle of minutes, which is fine for a pallet that moves twice a day.

UWB TDoA inverts the model. The tag transmits a sub-millisecond blink, anchors timestamp it on arrival, and no dialog happens. That supports thousands of tags at 0.1 to 1 Hz in one zone, and a forklift or AMR tag at 10 Hz or more with sub-100 ms latency, fast enough to gate a dock-door decision in real time. If any asset in your spec needs a position every 100 ms, the Channel Sounding airtime math does not close no matter how accurate a single fix is. Capacity, not accuracy, kills most hybrid designs.

How to run two radios as one system

A zoned site fails when it becomes two RTLS products with two dashboards. The working pattern:

• One coordinate frame. Survey anchors and locators into the same site map once, so a fix is a fix regardless of radio.
• One estimator. Feed BLE and UWB measurements into the same filter so a pallet's track does not jump when it rolls from racking into staging.
• Dual-radio tags only where they earn it. Forklifts, AMRs, and high-value pallets that cross into precision zones carry both radios; the 85 percent that never leave racking stay BLE-only.
• Per-fix accuracy on the API. Tag every position with its estimated error so downstream logic can, for example, auto-confirm a putaway only when the fix is under 0.5 m.

Zone handoff then becomes a software event, not an operational seam. One engine, two radios, zero seams.

Where licensed IP shortens this build

The radios are the easy part; silicon vendors sell both. Hybrid projects stall in the layer above: ranging across dissimilar RF sources, filtering them into one continuous track, and handing assets between zones without losing identity. That layer is also where granted patents concentrate, which is a freedom-to-operate question before it is an engineering one.

Position Imaging licenses hundreds of granted patents in RF ranging, real-time tracking, and multi-signal positioning, including US 11,774,249 and US 12,079,006, from a portfolio cited in filings by Apple, Bosch, and other major firms. Teams license the proven tracking layer, build their zone logic on top, and reach pilot in months instead of spending a year reinventing fusion math that is already claimed. License the ranging layer and ship sooner.