UWB vs BLE Channel Sounding: Warehouse Infrastructure Tradeoffs

What did Bluetooth 6.0 actually change for warehouse tracking?

The Bluetooth SIG published Bluetooth 6.0 in September 2024 with Channel Sounding, a native two-way ranging protocol that combines phase-based ranging with round-trip timing across up to 72 channels in the 2.4 GHz band. Before that, BLE positioning meant RSSI trilateration with 3 to 5 meters of error, or Bluetooth 5.1 direction finding that required antenna arrays and careful per-site calibration. Channel sounding is different in kind: it measures distance, not signal strength, and it runs on commodity silicon. Nordic's nRF54 series already ships with channel sounding support and more vendors follow through 2026. Vendor demos show 30 to 50 cm in clean line of sight. A warehouse is not clean line of sight. It is steel racking, metal pallets, forklift traffic, and tags buried mid-stack, which is exactly where the comparison with UWB gets interesting. Cheap ranging changes the bakeoff, not the physics.

Accuracy between the racks: 500 MHz beats 80 MHz

Ranging accuracy tracks bandwidth. Time resolution is roughly the speed of light divided by signal bandwidth, so a UWB channel at 499.2 MHz under 802.15.4z can separate signal paths about 60 cm apart. That is why UWB holds 10 to 30 cm in production warehouses even near metal: the direct path and the rack reflection arrive as distinct pulses. Channel sounding hops across roughly 80 MHz of the 2.4 GHz band, so reflections arriving within about 3.7 meters of the direct path smear into one measurement. Phase-based ranging sharpens results in benign conditions, but steel racking generates exactly the short-delay reflections it handles worst. Plan on:

• UWB: 10 to 30 cm, holding up in non-line-of-sight between racks
• Channel sounding: 30 to 50 cm in open line of sight, 1 to 2 meters in dense racking, worse with buried tags

For bay-level inventory, 1 meter is fine. For slot-level putaway or vehicle guidance, it is not. Bandwidth is the budget nobody can negotiate.

Tag cost per asset: where BLE channel sounding wins

Channel sounding runs on the same SoC that already handles BLE connectivity, so the ranging capability is firmware, not added bill of materials. Finished BLE tags land at $3 to $7 in volume, and channel sounding tags price the same way. A UWB tag needs a dedicated radio, typically Qorvo or NXP Trimension silicon, plus a BLE radio for discovery and management, and finished tags run $15 to $40 depending on housing and battery. Across 20,000 tracked pallets that is roughly $60,000 to $140,000 for BLE against $300,000 to $800,000 for UWB, before a single anchor goes up. Battery behavior differs too. A UWB TDoA tag transmits a sub-millisecond blink and sleeps, which supports multi-year life at low blink rates. Channel sounding is a connected two-way exchange, so each fix costs more energy: at one fix per minute both last years, at one fix per second the BLE tag drains first. Tag price is where BLE earns its seat.

Infrastructure is the hidden line item: anchors, sync, concurrency

Tag price is the visible number. Installed system cost is mostly overhead. A UWB TDoA deployment needs anchors every 15 to 30 meters with overlapping coverage, synchronized to nanosecond precision over wired clock distribution or a wireless sync mesh, usually with PoE runs to every unit. Anchors cost $200 to $500 each installed, so a 100,000 square foot facility plans for 25 to 50 of them plus cabling labor, often the largest line on the quote. Channel sounding locators are cheaper at $50 to $150, and because the protocol is two-way ranging they need no tight time sync at all. The catch is concurrency. Channel sounding is connection-oriented: a locator ranges one tag at a time per session, so serving thousands of tags means scheduling, and update rates fall as the tag count grows. A UWB TDoA tag blinks once and every anchor in earshot timestamps it, which is why one UWB cell scales to thousands of tags at 1 Hz. Count the anchors, the cable pulls, and the fixes per second before you count tags. Your real invoice hangs from the ceiling.

Which radio for which warehouse workload?

Split the decision by asset class, not by facility.

• Pallets and returnable totes needing bay-level location: channel sounding. At tens of thousands of tags the $10 to $30 per-tag spread dominates everything else, and 1 meter answers the bay question.
• Forklifts, AMRs, and high-value staging needing 30 cm and fast updates: UWB. A site runs hundreds of vehicles, not tens of thousands, so tag cost barely registers against collision avoidance and slot-level accuracy.
• Hybrid is the default ending. UWB tags already carry BLE for management, so one network layer serves both populations, with UWB anchors over vehicle corridors and channel sounding locators over bulk storage.

The one-radio-per-building framing rarely survives contact with a real operation. The radios answer different questions at different prices, and a mixed fleet of assets deserves a mixed bill of materials. Pick per asset class, not per datasheet.

Should you build the ranging stack or license it?

Either radio still leaves you building the hard layer: multipath rejection, sync solvers, tag scheduling, fusion with inertial or vision data, and the tracking engine that turns raw ranges into positions a WMS can act on. Teams routinely spend 18 to 24 months there, and they build inside a dense patent field. Position Imaging licenses a portfolio of hundreds of granted patents across RF ranging, real-time positioning, and tracking, including US 11,774,249, US 12,079,006, US 12,066,561, and US 12,000,947, IP that firms including Apple and Bosch cite in their own filings. Licensing the positioning layer means your engineers spend their cycles on the warehouse product, the workflows, and the integrations, and you reach the floor in months with freedom to operate instead of re-deriving ranging math that is already claimed. Ship the product, not the patent search.