UWB anchor placement — the independent design guide.

How vendors over-specify anchor count — and how to spot it.

Vendor reference designs default to 1 anchor per 80–120 m² for production-grade UWB. We have audited Tier-1 automotive sites where 40 anchors hit the same KPI as the vendor 80-anchor design — 50% saving on hardware AND install (€33,000+ per site).

Three structural pressures push density up: SLA buffer, unfalsifiability lock-in, and hardware revenue. All three work against the buyer.

The fix: ask the vendor to model 70% of their proposed anchor count against your KPI. If they refuse, that is the entire answer.

The five anchor-placement rules that actually matter.

1. Ceiling-height verified, not nominal. An anchor designed for 4.2 m mounted at 3.8 m changes coverage cone by 15–25%. Every mount point gets measured.

2. Geometric offset for multipath, not ceiling grid. Metal racking, glass and concrete-aggregate floors create multipath. Placement compensates via geometric offset.

3. Multi-time-of-day survey passes. A 09:00 survey does not see the 14:00 shift cycle. We run minimum three passes.

4. TDoA synchronisation tolerance budgeted into network design. A mis-cabled PoE switch quietly destroys positioning without logging an error.

5. Acceptance test under production load. Vendor commissioning at 06:00 in an empty bay is a different RF environment from 14:00 on a busy shift.

What the audit deliverable looks like.

For a 5,000 m² site, an independent audit produces: predictive RF model against the as-built environment,

device count justification with KPI sensitivity at 70/100/130% of vendor density, mount-point register, multi-time-of-day data, network design with TDoA timing tolerance, acceptance test plan.

Typical fee: €8,000–€18,000 per site. Typical install-bill saving: €20,000–€50,000 single site; €100,000–€400,000 on 5–15 site rollouts. See our infrastructure and install service.

UWB anchor geometry: first path, sync and redundancy over raw count

UWB locating quality is dominated by geometry and first-path visibility, not by buying the maximum anchor count in the quote. In open areas, practical spacing often lands around 8–15 m between anchors at 2.5–3.5 m mounting height; metal-heavy or partitioned spaces need tighter grids. Rules of thumb of “one anchor per X m²” are starting points for a survey — not a substitute for one.

Placement rules that matter. Keep antennas away from large metal backplanes; prefer ceiling or high-wall mounts that look down corridors of motion; ensure each critical zone has four or more anchors in view for redundancy and NLOS resilience; avoid collinear layouts that create poor dilution of precision; and plan power/PoE with OT network segmentation from day one.

TWR vs TDoA implications. Two-way ranging is simpler to deploy but airtime scales poorly with huge tag populations. TDoA scales better for many tags at higher combined rates but needs tight time synchronisation — sync drift silently degrades accuracy. Choose the mode for your tag count and update-rate budget, then design anchors accordingly.

How vendors over-specify. Extra anchors are an easy SLA buffer and an easy margin line. Independent design challenges often remove 20–40% of proposed anchors while holding the production KPI by fixing height, orientation and zone priorities. Demand a density rationale tied to use-case zones, not a uniform carpet.

Commissioning under load. Validate with forklifts, carriers and people moving — static empty-bay walks hide the NLOS events that dominate industrial error. Re-check after racking or cell changes. Document which zones are continuous centimetre-class vs zone-class so the business does not assume uniform accuracy everywhere.

Maintenance and lifecycle for UWB anchor estates

Plan firmware updates, certificate rotation, spare anchors and periodic geometry checks. Cranes, new mezzanines and racking moves invalidate yesterday’s survey. Monitor sync health on TDoA deployments — silent drift is a locating outage that looks like “random inaccuracy.” Document which cells are safety-critical and require faster MTTR than general asset-find zones.

When challenging a vendor BOM, ask which anchors are load-bearing for the KPI versus SLA padding. Remove padding first; never remove redundancy from safety geofences without a documented risk acceptance. Re-run a subset of walk/drive tests after any significant density change before declaring savings banked.

Safety-zone vs asset-find placement differ on purpose

Safety geofences prioritise low false negatives and stable first-path geometry along pedestrian–vehicle interaction paths. Asset-find zones prioritise coverage efficiency and search-time KPIs. Do not use one density rule for both. Mark safety cells on drawings, require four-plus anchors in view where feasible, and test with real vehicle motion. Asset-find cells can often share a leaner grid once safety cells are protected.

Document expected error bands on the drawing

Every UWB layout drawing should state expected accuracy bands by zone under production load — not a single site-wide centimetre claim. Safety cells, asset-find halls and compromised brownfield corners need different numbers. Publishing those bands prevents operators and boards from assuming uniform performance the physics will not deliver.