UWB vs Bluetooth 6.0 for Pallet Tracking: Accuracy and Tag Cost

What does Bluetooth 6.0 channel sounding actually change?

Bluetooth 6.0, adopted by the Bluetooth SIG in September 2024, adds Channel Sounding: a native ranging protocol that combines phase-based ranging with a round-trip time measurement. The radio steps across roughly 80 MHz of the 2.4 GHz band in 1 MHz increments, measures phase at each step, and computes distance from the phase slope. The round-trip time exchange cross-checks that result and defeats the relay tricks that made RSSI ranging untrustworthy.

On shipping silicon such as Nordic's nRF54L15 class parts, vendor line-of-sight tests land around 0.5 m of distance error. Compare that to the 3 to 5 m zone guessing BLE 5 RSSI gave you. One caveat matters for fleets: channel sounding is a two-way procedure between exactly two radios, and a full channel-map exchange takes tens of milliseconds of airtime. That caps how many tags a single anchor can range per second, which matters once a staging area holds 200 pallets.

Still, the shift is real. Ranging is now native to commodity silicon.

Is 1 meter accurate enough to track a pallet?

Do the geometry before the radio selection. A GMA pallet is 48 by 40 inches, about 1.2 by 1.0 m. A standard selective rack bay runs 2.4 to 2.7 m wide and holds two or three pallets. Staging lanes and dock aprons are wider still. So pallet questions come in two sizes:

• Which lane, which bay, which dock door. This needs 1 to 1.5 m accuracy. Channel sounding with reasonable sightlines clears it.
• Which position inside a bay, which pallet is on which forklift. This needs 30 to 60 cm. That is UWB territory, where 802.15.4z two-way ranging delivers 10 to 30 cm.

Most pallet workflows live in the first bucket. Receiving verification, putaway confirmation, staging audits, and dock-door checks all resolve at lane granularity. The second bucket appears when you bill customers by slot or automate picks against a specific position.

Pay for the question you actually need answered. Pallet questions are usually lane questions.

Why multipath, not the spec sheet, sets warehouse accuracy

Spec-sheet numbers assume open air. A warehouse real time location system operates in the opposite: steel racking, metal pallet loads, forklift masts, all reflecting 2.4 GHz energy.

The limit is bandwidth. A receiver can separate two propagation paths only when they differ in length by roughly the speed of light divided by twice the signal bandwidth. At UWB's 499.2 MHz channel width, that resolution is about 0.3 m, so the direct path stands apart from the rack reflection and the receiver can pick the first arrival. At channel sounding's 80 MHz span, resolution is about 1.9 m, so any reflection within two meters of the direct path blends into the phase measurement and pulls the answer.

In practice that means channel sounding holds near 0.5 m in open floor areas, stretches to 1 to 3 m deep in racking, and occasionally reports a reflection as truth. UWB degrades more gracefully, typically staying under 0.5 m even with a pallet load blocking line of sight.

Bandwidth sets the floor, firmware cannot raise it.

What does each radio cost per pallet tag at fleet scale?

A channel sounding tag is one die. An nRF54 class SoC in volume runs around $2 and handles ranging, telemetry, and firmware updates. A built tag with enclosure, coin cell, and antenna lands near $4 to $6 at 10,000 units. A UWB tag needs a UWB transceiver (Qorvo DW3000 class or NXP Trimension) plus a BLE radio for commissioning and wakeup, so silicon alone runs $4 to $6 and built tags commonly land at $12 to $20.

Across a 10,000-pallet distribution center, that is roughly $50,000 in channel sounding tags against $150,000 or more in UWB tags, before anchors. Anchors claw some back: UWB TDoA anchors cover 20 to 30 m spacing passively, while channel sounding spends anchor airtime per tag, so large fleets need denser anchors or slower updates.

Power rarely decides it. Pallets move a few times a day, so an event-driven tag ranging a handful of times per hour runs years on a CR2477 with either radio.

At pallet scale, tag cost is system cost.

The 2026 architecture: tier the radios, license the positioning

The deployments that pencil out in 2026 do not pick one radio. They put $5 channel sounding tags on every pallet for lane-level and bay-level truth, and reserve UWB for the 50 forklifts and the high-value loads where 10 to 30 cm pays for itself. A forklift carrying UWB becomes a moving precision anchor: when it lifts a pallet, short-range channel sounding identifies which tag is aboard, and the forklift's UWB position becomes the pallet's position during transit. You get slot-level accuracy at the moments that matter without paying UWB prices across the whole fleet.

The hard part is not the radios, it is the fusion, filtering, and tracking logic that turns noisy ranges from two different physical layers into one trustworthy pallet history. That math is also where the granted patents sit. Position Imaging licenses hundreds of granted patents in RF ranging, real-time positioning, and tracking, IP cited by Apple and Bosch, so a team can ship a tiered pallet RTLS in months with freedom to operate instead of rediscovering the prior art after launch.

Build the warehouse product, license the positioning.