Private 5G vs UWB for Factory Tracking: Release 18 Sidelink Reality

Can private 5G positioning replace UWB on the factory floor in 2026?

The short answer: no, not where accuracy decides the workflow. 3GPP has shipped network-based positioning since Release 16, using downlink and uplink time difference of arrival (DL-TDOA, UL-TDOA), multi-cell round trip time, and angle methods. Release 16 targeted 3 m horizontal accuracy indoors for 80 percent of devices, and Release 17 wrote 0.2 m targets for industrial IoT use cases into the requirements. Field results on private networks tell a different story. Over a single 100 MHz carrier around 3.5 GHz (CBRS in the US, local n78 licenses in Germany), integrators typically report meter-level accuracy on working factory floors, because steel racking, gantry cranes, and moving forklifts smear the arrival time of every reference signal. Meanwhile 802.15.4z UWB systems deliver 10 to 30 cm in the same buildings and have done so in production for years. If your use case is zone-level visibility across a campus, NR positioning already works. If it is bin-level picking or forklift guidance, it does not. Paper targets are not floor measurements.

What does Release 18 sidelink positioning actually add?

Release 18, frozen by 3GPP in mid-2024, added four things that matter for factories:

• Sidelink positioning (SL-PRS). Devices range each other directly over the PC5 interface, with no base station in the loop. Ranging between two tags, or a tag and a vehicle, becomes a native NR feature.
• Carrier phase positioning. Borrowed from the RTK GNSS playbook, it can in principle reach centimeter level, but it needs continuous phase lock and handles integer ambiguities poorly when forklifts keep breaking line of sight.
• LPHAP, low power high accuracy positioning. Signaling changes that let a battery tag sleep between fixes instead of holding a full connection.
• RedCap positioning. Cheaper, narrower-bandwidth modems get access to positioning procedures, which is what an asset tag bill of materials actually needs.

The catch is timing. Rel-18 features reach commercial modules a year or two after freeze, so expect usable SL-PRS silicon and infrastructure support to arrive through 2026 and 2027, not in your Q1 pilot. The spec is frozen; the silicon is not.

Why bandwidth still decides accuracy in metal-heavy plants

Time-of-flight accuracy is set by how finely you can separate the direct path from reflections, and that resolution scales with bandwidth: roughly the speed of light divided by twice the bandwidth. Run the numbers:

• A 100 MHz NR carrier resolves paths about 1.5 m apart.
• An 802.15.4z UWB channel at 499.2 MHz resolves paths about 30 cm apart.

A factory is the worst case for this math. Steel racking, welded mezzanines, and moving vehicles produce reflections that arrive a few nanoseconds behind the direct path, and 1 ns of timing error equals 30 cm of range error. With 100 MHz you cannot see those reflections as separate arrivals, so they bias the estimate instead. Carrier phase methods can beat the bandwidth limit, and Release 18 opens that door, but they degrade exactly when a plant is busiest, because phase tracking resets every time a trailer or a crane blocks the path. More bandwidth, finer resolution. Physics does not negotiate.

Where sidelink genuinely helps a factory rollout

Treat sidelink as a coverage and cost tool, not an accuracy upgrade.

• Anchor economics. TDOA compares arrival times across base stations, so the gNBs need nanosecond-grade sync, and every nanosecond of sync error adds 30 cm of position error. Sidelink ranging between two devices removes network sync from the error budget for relative measurements.
• Coverage gaps. Yards, docks, and the space between buildings rarely justify another gNB. Two Rel-18 devices can still range each other there.
• Relative positioning. Forklift-to-pallet approach, AGV-to-AGV separation, and tool-to-workpiece association are relative problems. Sidelink answers "how far and closing" without ever computing an absolute coordinate.
• Tag-assisted infill. A tag that ranges against a nearby vehicle holding a good fix inherits usable position in areas the network covers poorly.

None of this changes what a lone tag ranging against the network achieves on a 100 MHz carrier. Sidelink fixes coverage math, not accuracy math.

How should you split a 2026 rollout between 5G and UWB?

Decision rules that hold up in 2026:

• Need under 30 cm for autonomous forklift guidance, bin-level picking, or tool control: UWB, full stop. FiRa-certified 802.15.4z hardware and the omlox interoperability standard give you multiple vendors today.
• Need zone or room level across a large campus, and you already run private 5G for connectivity: use NR positioning and skip a second radio network for the low-precision tier.
• Need both, which describes most plants: go hybrid. Put UWB cells over the 10 to 20 percent of floor area where precision earns money, and let the 5G network carry everything else. Tag vendors already combine a UWB radio and a cellular radio in one housing.

Whatever you pick, run the pilot on your own floor. Walk tags along real routes at real speed, past racking and under cranes, and score the 95th percentile error, because the workflow fails at P95, not at the mean. Measure P95 along real routes, then decide.

The ranging and tracking layer is the same build either way

Both radios sit on the same underlying discipline: extracting range from RF signals in heavy multipath, fusing measurements across time, and keeping tracks stable while assets move and occlude each other. That layer is where engineering teams burn 12 to 24 months before the first workflow ships, and it is also where the patent exposure sits, because this field carries two decades of granted claims. Position Imaging licenses hundreds of granted US patents in radio-frequency ranging and real-time tracking, including US 11,774,249, US 12,079,006, US 12,066,561, and US 12,000,947, a portfolio cited by Apple, Bosch, and other major filers in their own applications. Licensing that layer means your team spends 2026 on the product instead of re-deriving solved ranging problems, and your freedom-to-operate review starts from a documented license rather than a blank page. License the ranging IP; spend your year shipping.