Cost Per Tracked Pallet: BLE Channel Sounding vs UWB in Warehouses

What accuracy do the two radios actually deliver on a warehouse floor?

Ranging accuracy follows channel bandwidth. UWB under IEEE 802.15.4z uses 499.2 MHz channels near 6.5 and 8 GHz, which lets a receiver resolve signal paths about 30 cm apart. Bluetooth 6.0 channel sounding hops across up to 72 narrow channels in the 2.4 GHz band, roughly 80 MHz of usable spectrum, so paths closer than about 2 m blur together. In a building full of steel racking, that difference decides whether the receiver can separate the direct path from a reflection.

The field numbers follow the physics:

• UWB two-way ranging: 10 to 30 cm typical on a live floor, under 10 cm with tuned anchor geometry.
• BLE channel sounding: vendors demo 30 to 50 cm in open line of sight; expect 1 to 3 m between loaded racks, with multi-meter outliers when the direct path is blocked.

For pallet-level questions like which aisle, 1 to 3 m answers the question. For which rack slot, or whether a forklift is 40 cm from a person, it does not. Bandwidth is destiny here.

Why does cost per tag favor BLE, and by how much?

Tags are where the gap compounds. A channel sounding tag is a standard BLE part; Nordic and NXP shipped channel sounding silicon by 2025, with other vendors following, so a finished tag lands around $5 to $10 and runs 2 to 5 years on a CR2032 coin cell at a few updates per minute. A UWB tag carries a second radio (Qorvo DW3000 class or NXP Trimension) alongside BLE for discovery and control, so finished tags run $15 to $30 and draw more energy per ranging exchange.

Infrastructure widens the gap. BLE locators cost $30 to $100 and often already exist as beacon gateways. UWB anchors run $150 to $300, want four or more per coverage cell, and TDoA modes need tight time sync, wired or wireless, which adds installation labor.

Across 10,000 tracked pallets, tags alone total roughly $50,000 to $100,000 for BLE versus $150,000 to $300,000 for UWB, before a single anchor goes up. At pallet counts, BLE wins the spreadsheet.

Where does channel sounding struggle at warehouse scale?

Channel sounding is connection-oriented. Each measurement is a pairwise exchange that steps through dozens of channels, so one ranging event takes tens of milliseconds of radio time between one tag and one locator. Multiply that by thousands of tags, several locators each for trilateration, and an update every few seconds, and airtime becomes the budget you manage, not accuracy. The 2.4 GHz band is also shared with Wi-Fi and every other BLE device in the building, and warehouse access points are not quiet neighbors.

Multipath is the second wall. Phase-based ranging assumes the strongest arriving path is the direct one. Behind a loaded steel rack it often is not, and the error shows up as a confident, wrong distance rather than an obvious dropout. Filtering helps, but filtering out 2 m of bias takes motion models and time.

Plan for channel sounding to place pallets in an aisle or zone, refreshed every few seconds, not to stream positions. Scale is the ceiling, not the chip.

When is UWB worth three times the tag price?

Pay for UWB when the position feeds a decision that a person or machine acts on in real time:

• Forklift and pedestrian safety. Proximity warning needs 30 cm class accuracy at sub-100 ms latency. A 2 m error is the difference between a slowdown zone and a missed stop.
• Robot and AMR navigation aid. Fusing UWB fixes with wheel odometry holds drift-free position in dark aisles and repetitive racking, exactly where visual localization breaks.
• Rack-position inventory. Knowing a pallet sits in slot B-14-3, not just aisle B, removes the walk-and-scan step from every pick and putaway.
• High-value assets at low counts. For 200 tools or trailers instead of 10,000 pallets, the tag premium is noise against the search time saved.

UWB TDoA also scales differently than two-way ranging: the tag transmits a blink lasting under 1 ms and goes back to sleep, so a battery tag can blink every second for years while the anchors do the math. Where position drives action, buy the bandwidth.

How do you decide, and what does the split deployment look like?

The decision rule for 2026 is to match the radio to the question, not to the hype cycle:

• Aisle or zone answers, thousands of tags, multi-year batteries: BLE 6.0 channel sounding.
• Sub-30 cm answers, safety or navigation, latency under 100 ms: UWB on 802.15.4z.
• Most warehouses past about 100,000 square feet end up running both, BLE on pallets and totes, UWB on 50 forklifts and the AMR fleet, fused into one position feed.

That fusion layer, the algorithms that turn raw RF ranges plus inertial and vision data into one reliable track per asset, is where teams burn 12 to 24 months, and it is also where granted patents concentrate. Position Imaging licenses a portfolio built across radio-frequency ranging and multi-sensor tracking, IP cited by Apple and Bosch in their own filings, so a team can ship on either radio, or both, without rebuilding the tracking stack or guessing at freedom to operate. License the tracking layer once.