RF site survey — the independent pilot-preparation checklist.

Why most site surveys fail on day 60, not day 1.

The default vendor survey runs at 09:00 in an empty bay against an architectural drawing. It produces a design that works at commissioning and starts failing six weeks later when the bay fills with racking, guarding, mezzanines and tooling.

Five failure modes account for most accuracy regression: anchor height drift, RF blockers introduced post-survey, multipath from metal and glass, line-of-sight gaps over the production day, and synchronisation timing. Every one is detectable in a properly-scoped survey.

The 12-point pre-survey checklist.

1. Site walkthrough at three different times of day — 06:00, 12:00, 16:00 minimum — capturing different shift states, door cycles and traffic patterns.

2. Floor-plan reconciliation against as-built. Every floor plan we have seen is wrong by 5%+.

3. Ceiling-height map at minimum every 10 m. Variations of 200 mm matter for UWB.

4. Floor surface material map — concrete with rebar, polished concrete, epoxy and raised floor affect RF reflection differently.

5. Metal-density and glass-partition map. Dominant multipath generators.

6. HVAC and ceiling utilities. Ducting, trays and sprinkler runs affect line-of-sight.

7. Mobile equipment cycle log — forklifts, AGVs, doors, presses.

8. Existing wireless interference scan — Wi-Fi, BLE, RFID, DECT, Zigbee, proprietary 433/868/915 MHz already on site.

9. Network topology map — PoE backbone, switch hop count, VLANs, time source.

10. Operations stakeholder interviews — what does the floor team need to see, where do they lose visibility today.

11. KPI baseline measurement — current WIP, search time, dock dwell, OEE.

12. Pre-agreed acceptance criteria. Pass/fail thresholds signed before scoping the pilot.

Per-technology specifics.

UWB: ceiling-height map, metal-density assessment, TDoA timing budget, multipath simulation against the as-built environment.

RAIN RFID: antenna geometry against dock-door geometry, dwell-time analysis, near-field shielding, conveyor integration timing.

BLE-AoA: Locator ceiling grid alignment, multi-antenna orientation, multipath-aware geometry, density vs ceiling height.

LoRaWAN: gateway placement against line-of-sight, rooftop or tower mount safety, lightning protection, GNSS-disciplined clock.

Wi-Fi RTLS: AP density (typically insufficient for RTLS), 802.11mc FTM support, network-refresh integration.

Survey for locating under live load — not for pretty heatmaps at dawn

An RF site survey for locating is not the same as a Wi-Fi coverage survey. You are validating whether the chosen radio can deliver the locating KPI — accuracy band, update rate, read rate, latency — in the geometry and traffic of a real shift. Surveys that only walk an empty floor at commissioning time routinely pass, then fail by day 60 when racking, vehicles and people create multipath and non-line-of-sight.

Before anyone climbs a ladder. Confirm use cases and the smallest distance that changes a decision; map metal, liquids, mezzanines and moving equipment; capture peak-hour schedules; list candidate mount points with power/network constraints; agree the acceptance metrics in writing; and freeze a floor plan version so design and survey refer to the same layout.

During the survey. Test candidate locator/anchor/portal positions with temporary mounts; walk and drive real tagged assets, not foam boards; repeat with doors open/closed and with adjacent equipment present; log first-path vs reflected-path behaviour for UWB; check AoA bearing stability for BLE; measure Passive RFID portal cross-reads between adjacent docks; and record interference sources (other readers, Wi-Fi, industrial noise).

Per-technology notes. UWB: geometry and time sync matter as much as density — prefer height and clear view corridors over packing anchors behind steel. BLE AoA: antenna orientation and ceiling clutter dominate. Passive RFID: door geometry, tag orientation on real SKUs, and shielding between portals. Wi-Fi RTT: AP density and 802.11mc support rarely match RTLS needs without a refresh plan.

Deliverables you should insist on. Annotated drawings with proposed mounts; risk register for multipath/NLOS zones; density recommendation tied to KPIs (not vendor default); bill of materials with spares; commissioning plan under live load; and a clear statement of what the survey did not prove. Re-survey after major layout changes — locating designs are not set-and-forget.

Handover pack: from survey drawings to installable locating design

A survey that ends as a slide deck of heatmaps is incomplete. The handover pack should include mount schedules with heights and orientations, cable and PoE schedules, risk zones with expected accuracy bands, a tagged-asset test matrix, and a commissioning script timed for peak operations. Installers and PMs should be able to execute without re-interpreting RF intent.

Include a change-control note: which layout changes trigger a re-survey, who owns the decision, and how temporary works (construction hoarding, seasonal racking) are handled. Locating programmes decay when facilities moves are invisible to the RF owner.

Finally, store raw survey logs (where tooling allows) alongside the signed acceptance metrics. When day-60 accuracy drifts, you need evidence of what was true at design time — not a reconstructed memory of a green demo.