Channel Sounding vs UWB: Speccing Asset Tracking Hardware in 2026

What accuracy does each radio actually ship in 2026?

UWB radios built to 802.15.4z range over 500 MHz channels at 6.5 or 8 GHz and measure time of flight directly. In practice that produces 10 to 30 cm accuracy, and it degrades gracefully when the direct path weakens, because 500 MHz of bandwidth can separate signal paths that arrive about 30 cm apart. Bluetooth 6.0 channel sounding works differently. It hops across up to 72 narrow channels inside roughly 80 MHz at 2.4 GHz and reconstructs distance from phase measurements across those hops. On a bench with clear line of sight, that gets you 30 to 50 cm. In a warehouse aisle, where steel racking bounces 2.4 GHz energy everywhere, reflected paths blur into the direct one and reported error grows to 1 to 2 m. Neither result is a vendor tuning problem. Ranging resolution scales with bandwidth, so an 80 MHz system physically cannot pull apart reflections the way a 500 MHz system can. Bandwidth decides who survives multipath.

Which asset tracking chipsets are real, not roadmap?

On the Bluetooth side, Nordic Semiconductor's nRF54L series supports channel sounding in a single low-power SoC, which means a channel sounding tag is close to an existing BLE beacon with a new radio and new firmware. Tag makers already shipping BLE hardware can move without a ground-up redesign. On the UWB side the silicon is mature: Qorvo's DW3000 family and NXP's Trimension line have shipped in volume for years, and Apple has put UWB in every flagship iPhone since the U1 in 2019, with its second-generation UWB chip arriving in iPhone 15. Phone support is the asymmetry to watch. UWB talks to iPhones today through Nearby Interaction. Channel sounding needs Bluetooth 6.0 phone silicon plus OS ranging APIs, and Android's Ranging API arrived before Apple said anything about supporting it. If your product involves a consumer's phone finding a tag, that gap is a launch-date question, not a footnote. Check your phone dependency first.

Ranging security is where the two standards really diverge

Distance measurement can be attacked. A relay or phase manipulation attack convinces a system that a tag is closer than it is, which matters when ranging opens a door, releases inventory, or confirms chain of custody. UWB under 802.15.4z answers with the scrambled timestamp sequence, STS: ranging pulses are derived from an AES-128 key, so an attacker cannot predict or replay them to fake a shorter distance. The FiRa Consortium certifies this behavior, and the Car Connectivity Consortium chose UWB for Digital Key 3.0 largely for this reason. Bluetooth 6.0 channel sounding took a different route. It pairs phase-based ranging with round-trip time checks, hops channels in randomized order, and reports a Normalized Attack Detector Metric, NADM, that flags suspected manipulation. That is credible engineering, but phase-based measurement remains the softer of the two targets. If your tracker only produces dashboards, this barely matters. If it gates access or custody, it decides the radio. Secure ranging is a feature, not a checkbox.

What does the BOM look like per tag and per anchor?

A channel sounding tag is one 2.4 GHz SoC, a coin cell, and an antenna, the same bill of materials as BLE beacons that sell for $5 to $10 at volume, with multi-year battery life at modest ranging duty cycles. A UWB tag needs the UWB radio plus a BLE companion for discovery and wake-up, a larger battery because UWB ranging bursts draw more current, and more careful antenna work at 6.5 to 8 GHz. Volume UWB tags land closer to $15 to $30. Infrastructure follows the same pattern:

• Channel sounding anchors can reuse BLE gateway hardware, mounting points, and site surveys you may already have.
• UWB anchors typically cost over $100 each and want denser placement plus PoE or wired backhaul for clean installs.

For a site tracking 10,000 pallets, the tag delta alone is six figures. For a small site with 40 anchors, infrastructure dominates instead. Run both totals against your real asset count. Count anchors before you count tags.

How do you spec hardware without betting the product on one radio?

The decision rules are short. Pick UWB when the product fails above half a meter of error, when non-line-of-sight is the normal case rather than the exception, or when ranging acts as a security control. Pick channel sounding when 1 m zone accuracy answers the customer's actual question, when tags must run 2 to 5 years on a coin cell, or when the tag count is large enough that a $10 to $20 per-unit delta breaks the business case. Then hedge in hardware. Lay out the tag with a modular radio section or a dual footprint, so a board respin swaps radios without re-certifying the whole product. Keep the location engine radio-agnostic, so range measurements from either PHY feed the same filters and the same anchor geometry logic. Teams that bury radio assumptions in application code pay for the migration twice, once in firmware and once in the backend. Abstract the radio, not the requirement.

The radio choice does not solve the positioning problem

Chipsets give you range measurements. Turning those into positions a customer trusts requires multipath rejection, anchor calibration, motion models, and fusion with inertial or vision data, and that algorithm layer is where most tracking startups burn their first year. It is also where patent exposure concentrates, because RF ranging and RTLS system design were patented heavily long before Bluetooth 6.0 or 802.15.4z existed, and shipping hardware makes you visible. Position Imaging licenses granted patents in RF ranging, real-time positioning, and computer vision, including US 11,774,249, US 12,079,006, US 12,066,561, and US 12,000,947, part of a portfolio cited by Apple and Bosch in their own filings. Licensing means the hard-won algorithm work and freedom to operate arrive together, so your team spends its runway on the product instead of re-deriving filter tuning and answering demand letters. Whichever radio wins your bakeoff, the physics above it is already proven. Ship the tracker, license the physics.