UWB vs BLE Channel Sounding for Pallet Tracking: The Per-Pallet Math

How accurate does pallet tracking actually need to be?

Start from the rack, not the datasheet. A standard GMA pallet is 48 by 40 inches, and a selective rack bay on 96 inch beams holds two of them with slot centers roughly 1.3 m apart. To say which slot a pallet sits in, you need lateral accuracy better than about 50 cm, plus the rack level, which means height. UWB at 10 to 30 cm clears that bar with margin. Bluetooth 6.0 Channel Sounding, which vendors quote near 0.5 m in clear line of sight and 1 to 3 m in cluttered conditions, identifies the bay or the aisle, not the slot.

But plenty of pallet questions are coarser than a slot:

• Staging lanes run 3 to 4 m wide. Lane-level is enough.
• Dock doors sit on roughly 12 foot centers. Door-level is enough.
• Floor-stacked bulk zones only need zone entry and exit.

If your operation is mostly lanes, doors, and zones, Channel Sounding accuracy is sufficient and the cheaper tag wins. If put-away verification to the slot is the point of the project, it is not. Match the radio to the slot width.

Why metal racking punishes 2.4 GHz phase ranging

Channel Sounding ranges by measuring phase across up to 72 hops in the 2.4 GHz band, synthesizing about 80 MHz of bandwidth. Bandwidth sets multipath resolution: 80 MHz cannot separate two signal paths whose lengths differ by less than about 3.7 m. In a warehouse that matters, because a reflection off a steel upright two meters longer than the direct path blends into the direct path and drags the range estimate off by meters. The estimate also shifts as the environment shifts: a rack face full of shrink-wrapped product reflects differently than the same rack empty, so yesterday's calibration drifts.

UWB under 802.15.4z takes the opposite approach. A 499.2 MHz channel resolves paths about 60 cm apart, and the receiver timestamps the leading edge of a roughly 2 ns pulse. The first-arriving path survives even when a reflected path carries more energy, which is exactly the condition inside racking.

The accuracy gap between the two radios is small in an open lab and large between loaded pallet racks. Benchmark in your building, in peak season, not in a demo hall. Steel does not care about the spec sheet.

The per-pallet math: tags outnumber anchors 200 to 1

Pallet tracking inverts the usual RTLS budget. A 5,000 square meter site can hold on the order of 8,000 pallet positions and needs a few dozen anchors, call it 40 to 60 once you cover every aisle. That is a tag-to-anchor ratio near 200 to 1, so the tag line item swamps everything else.

Run the numbers:

• Tag BOM. A Channel Sounding tag needs one SoC, something in the class of Nordic's nRF54L15. A UWB tag needs a BLE SoC plus a UWB radio in the class of Qorvo's DW3000 family, adding several dollars at volume before antenna and battery.
• Shipping tag prices. Commercial BLE tags commonly land in the $5 to $10 range. UWB tags commonly run $15 to $25. On 8,000 pallets, a $10 delta is $80,000.
• Anchor delta. Even a $300 per-anchor premium across 50 anchors is $15,000, a rounding error next to the tags.

And pallets leave. They ship out under freight, get swapped at the dock, and take their tags with them, so tag spend recurs while anchor spend does not. Tag price is the whole budget.

The hybrid that fixes the math: cheap pallets, precise forklifts

You do not have to pick one radio for the whole building. Pallets spend most of their dwell time stationary, and every move happens on a forklift or pallet jack. That asymmetry is the design.

Put a low-cost BLE tag on every pallet and a UWB tag on every piece of handling equipment. At put-away, the forklift's 10 to 30 cm UWB position, plus fork height from a mast sensor, stamps the exact slot. The pallet inherits that location and holds it, verified by its BLE tag at zone level, until the next lift event updates it. A site with 8,000 pallets and 25 forklifts buys 25 tags at UWB prices instead of 8,000.

The hybrid also fixes the power budget. The pallet tag can sleep and report on motion or on a slow heartbeat, stretching a coin cell across multiple years, while the forklift tag draws from vehicle power and ranges continuously. The hard part moves into software: associating the right pallet with the right lift event, fusing ranging with inertial and height data, and keeping 8,000 location records honest. Track the mover precisely, the load cheaply.

Should you build the ranging stack or license it?

Whichever radio you choose, the differentiating work is the same layer: multipath rejection inside steel racking, fusion of RF ranging with motion and height data, lift-event association, and tag fleet management at thousands of units. Teams routinely spend 12 to 24 engineering months rebuilding that layer, and then discover the field is dense with granted patents on exactly these methods.

That is the case for licensing. Position Imaging licenses hundreds of granted patents in RF ranging, real-time positioning, and computer vision, a portfolio cited by Apple, Bosch, and other major filers, including US 11,774,249 and US 12,066,561 in the ranging and tracking space. A license buys two things at once: proven methods your team does not have to reinvent, and freedom to operate so a competitor's portfolio does not surface mid-launch.

The radio decision stays yours. BLE Channel Sounding for zone-level pallet visibility, UWB where slot-level accuracy pays, or the hybrid that puts each radio where its economics work. The positioning layer underneath is already invented and already granted. License the ranging layer, ship the product.