Real time location services are only as good as the foundations under them. A glossy dashboard can tell you a tag is at Bed 12 or on Pallet Row C, but if the radio layer was guessed instead of measured, the dots will drift, alerts will misfire, and confidence will erode. The site survey is where an RTLS program earns its credibility. When done well, it blends radio engineering, fieldcraft, and operational insight to produce a design that meets accuracy targets without wrecking budgets or workflows.
I have walked hundreds of thousands of square feet with a tripod and a spectrum analyzer, from neonatal wards to frozen warehouses. The patterns repeat, but the details matter. This guide lays out the methods and tools I actually use, why I use them, and the trade-offs to expect when you are building or auditing an RTLS network.
What success looks like in measurable terms
Before arguing over tools, define what “good” means in your context. A surgical instrument tracking project in a hospital and a forklift visibility initiative in a 1.2 million square foot distribution center do not share the same tolerance for error, latency, or cost per square foot.
In practice, I anchor goals to a few measurable targets. Accuracy can be specified as 95th percentile error under certain conditions. For example, “sub 3 meters in patient areas, sub 1 meter in operating rooms” or “sub 50 centimeters on a 10 by 10 meter calibration grid.” Latency is the time from movement to visible location change, often 1 to 3 seconds for asset tracking, faster for safety use cases. Update rate quantifies how often position is refreshed, which affects battery life and interference risk. Location confidence is the system’s own estimate of reliability, which is valuable in exception logic. And coverage is not just radio presence but usable geometry for multilateration or angle of arrival.
These targets drive the survey scope. You cannot validate 50 centimeter accuracy with a one hour hallway walk. You need repeatable measurements, documented geometry, and a plan for corners, alcoves, and the kind of clutter you cannot draw on a floor plan.
Discovery before design
The most successful projects start with discovery hours, not survey hours. Meet the people who live in the space. They will tell you what the floor plans missed. A sterile processing manager will point out stainless walls that bounce 2.4 GHz like a mirror. A warehouse lead will warn you that racking reconfigures every quarter and that a zone near inbound doors stays packed with water pallets.
Bring printed floor plans and a pen. Confirm dimensions with a laser measure in a couple of representative corridors and rooms. Ceiling height is a primary design parameter, not a detail. I have seen facilities with 4 meter ceilings in offices and 12 meter bays in production on the same building grid. That difference changes the number and placement of anchors or beacons and whether you can use existing mounting points.
Power and backhaul matter as much as RF. If you plan PoE anchors for UWB TDoA, you need switch ports in the right closets and a time sync strategy. If your RTLS network rides on enterprise Wi‑Fi, engage the WLAN team about channel plans, roaming, and multicast. Even for dedicated RTLS infrastructure, coordinate on structured cabling routes, spare conduit, and infection control rules for drilling or ceiling access.
Finally, verify local regulations. Maximum transmit power for 2.4 and 5 GHz varies across regions. Ultra‑wideband is not universally permitted or may have band restrictions. The rtls provider should document compliance with FCC, ETSI, ANATEL, or your national authority, but it is your survey that will uncover edge cases like a surgical navigation device that occupies a quiet part of the spectrum you intended to use.
Choosing the right technology for the job
RTLS is a family of techniques, each with different physics and costs. You do not choose technology in a vacuum. You select it to match the environment, accuracy requirement, battery life, and integration needs.
Bluetooth Low Energy is the workhorse for many asset tracking projects. BLE tags are cheap and sip power. With dense beaconing or AoA arrays, you can reach 1 to 3 meter accuracy in open office or patient care settings. BLE has rich ecosystem support and can ride on existing Wi‑Fi APs that support BLE radios, though you lose control of antenna geometry if you rely entirely on AP adjacency.
Wi‑Fi based RTLS uses RSSI or fine timing measurements. It is attractive when you already have pervasive WLAN and want coarse location, for example room level presence. The trade‑off is power and update rate. Wi‑Fi tags drain faster and struggle in high multipath spaces. For fine accuracy, you need more APs and careful channel planning, which can collide with client connectivity requirements.
UWB excels when you need sub‑meter performance in metal‑rich or dynamic environments. Time of flight and TDoA ignore RSSI fluctuations that plague RSSI‑only systems. I routinely see 30 to 50 centimeter 95th percentile error in open bays with 5 to 7 anchors in view, tighter in controlled rooms. The costs are higher anchors, tight time sync, and careful siting. UWB TDoA wants line‑of‑sight and clock discipline on the anchors, often via PTP and GPS disciplined timing at the core.
Passive and active RFID cover choke points well. If you just need to know when an item passes a doorway or dock, do not overbuild a positioning grid. Combine portals with zone beacons and you can mix cheap event detection with area awareness.
Cellular and private 5G lend themselves to outdoor yards, inter‑building coverage, or high mobility assets. Accuracy is typically coarse unless you augment with BLE or UWB, but the backhaul can simplify campus‑wide coverage.
Most real world deployments are hybrids. Do not hesitate to blend BLE for broad coverage, UWB in surgical suites or test bays, and RFID at choke points. The rtls management platform should abstract these inputs so operations see locations and events, not radio types.
Survey methods that actually answer design questions
A good survey is not a single pass. It is a sequence that moves from hypothesis to measurement to validation. You combine predictive modeling with field checks, then iterate.
Here are the common methods and where they fit:
- Predictive survey in software: You import CAD or scaled floor plans, set material types, ceiling heights, and radio parameters, then simulate coverage and dilution of precision. It is fast and cheap. It gets you a first cut on anchor counts, likely problem rooms, and cable routes. It is wrong in the details but right in the trends, which is exactly what you need before drilling holes. Passive RF survey: You listen to the existing spectrum. With a spectrum analyzer and Wi‑Fi/BLE scanner, you map noise floors, channel occupancy, and intermittent interferers. This catches microwave ovens, legacy analog gear, or a baby monitor that saturates 2.4 GHz in a corner room. If you are piggybacking on existing Wi‑Fi, you also confirm AP placements and channel plans. Active survey with “device on a stick”: You place a temporary anchor, AP, or AoA array on a tripod and walk with calibrated tags to measure real RSSI, AoA, or TDoA geometry. This answers whether the predictive placements are practical and whether multipath breaks your model. In high bays, you can simulate final mounting heights with telescoping masts. Validation survey with near‑final placements: After install, you run a grid validation with known points and moving paths. You collect thousands of samples and compute error CDFs, latency distributions, and dropouts. This is where acceptance criteria live or die, and where you tune filtering and beacon rates. Environmental stress checks: You intentionally vary the environment to catch edge cases. Move metal racks, fill a room with IV poles, bring a pallet of bottled water into a test zone, or close elevator doors and see whether your algorithm “teleports” tags. These tests save embarrassment later.
Tools that pay for themselves
Field work depends on reliable tools. A short list covers most needs. For predictive work, Ekahau and iBwave can model Wi‑Fi and BLE with reasonable accuracy. For UWB, vendor tools often provide link budget calculators and PDOP visualization tied to their anchor geometry. None are truth. They are scenario builders.
In the field, a handheld spectrum analyzer like the Tektronix RSA or a Wi‑Spy DS spectrum dongle with software such as Chanalyzer is useful to spot interference peaks. A packet sniffer that can capture BLE advertisements and Wi‑Fi frames helps find channel plans and beacon densities. A calibrated tag kit is essential. Use tags with recent battery replacements, known TX power settings, and firmware matching your production devices. If you test with a lab tag at 0 dBm then deploy production tags at negative 8 dBm, your validation will lie.
Tripods with weighted bases, a telescoping mast, and a stable survey cart prevent hours of frustration. Laser rangefinders, a small label printer for marking test points, and painter’s tape to establish grids all speed validation. In healthcare, bring disposable sleeves and wipes, and respect infection control procedures. In food and pharma, hair nets and cleanroom gear can be required even for a hallway survey.
Do not forget time sync gear if you test UWB TDoA. A small PTP grandmaster or a GPS disciplined time source in the IDF turns a field test from hand‑waving into data.
Running an active survey without disrupting operations
The survey itself is equal parts choreography and diplomacy. Schedule tests during lower traffic periods, not just to reduce collisions but to avoid measuring people as moving absorbers. That said, if your production environment is always bustling, include busy periods in a separate round to see how body loss or vehicle motion changes performance.
For “anchor on a stick” testing, I stage temporary mounts approximately where predictive modeling suggested, then find at least three alternate locations per anchor in case cable routing or safety rules block the first pick. In open warehouses, I target anchors at 8 to 10 meters high to reduce occlusion by racks and moving stock. In hospitals, 2.7 to 3 meters is typical, constrained by ceiling grids and infection control. Angle of arrival arrays need clear space around the antenna panels, so avoid deep soffits or metal housings.
During walks, capture both stationary points and smooth motion. Static points confirm raw position scatter. Motion reveals filter lag and snap‑to‑grid artifacts. For example, a Kalman filter tuned for smoothness can overshoot when a nurse changes direction in a corridor. You want to see that behavior before go‑live.
Record your ground truth. A simple method is to lay a 5 by 5 meter tape grid in a test area and mark labeled points every meter. For larger spaces, use a survey wheel and reference known landmarks on the floor plan. If you can afford it, a laser total station or indoor positioning reference rig accelerates this, but I have produced perfectly acceptable validations with tape, patience, and clear documentation.
Modeling, then adjusting, then modeling again
Back at a desk, feed field results into your model. If BLE RSSI is 6 to 8 dB lower than predicted in a stainless heavy utility corridor, adjust material losses and see the impact on required beacon density. If a UWB anchor placement causes poor GDOP in a bay corner, nudge the anchor by 2 meters and watch the PDOP heatmap improve. This loop is where experience compounds. After a few projects, you start to see that moving an anchor one ceiling tile can cut worst case error by a meter without adding hardware.
Do not get lost in pretty heatmaps. Always tie changes to your acceptance targets and budget constraints. A 10 percent anchor count increase can be worth it to jump from 2.5 meters to 1.5 meters in a revenue critical area, while a general 50 percent increase to chase theoretical performance everywhere is a common way to overspend.
Interference and the uncooperative world
Interference is usually not the cartoonish constant noise people imagine. It is intermittent, local, and tied to human behavior. In a hospital, 2.4 GHz may spike at 11 am near staff lounges where microwaves run. In a factory, an arc welder lights up the spectrum during a maintenance window. In a grocery DC, 915 MHz RFID portals hammer the near field when trailers are unloading. Your passive survey should include time series captures in suspect areas, not just walking sweeps.
Multipath behaves differently by band. BLE in 2.4 GHz is more tolerant of mild reflections, but metal corridors create standing waves that produce pockets of unusually high or low RSSI. UWB tolerates multipath better due to the narrow correlation peaks, but at short anchor spacing and low ceilings you can still confuse a first path detector. Angle of arrival arrays hate reflective ceilings and close metal. Learn to spot architectural enemies: bulkheads with embedded steel, elevator banks, refrigeration rooms with foil insulation.
People are RF absorbers at 2.4 GHz. A crowded waiting area can cut effective range by 30 to 50 percent. Plan for it, or you will watch accuracy degrade during visiting hours. Water pallets are worse, especially stacked high. In warehouses that live on bottled beverages, set anchor heights higher than your tallest static stacks.
Acceptance criteria that operations can live with
You need acceptance criteria that are tough enough to guarantee utility and practical enough to meet. They should be quantitative, location specific, and time bounded. Below is a checklist I often use to frame sign‑off.
- Accuracy: 95 percent of samples within stated error thresholds by area type, measured with at least 1,000 samples per zone. Latency and update rate: Position updates within the target latency for moving tags, sustained over 15 minute motion periods without missed updates beyond a defined gap threshold. Coverage and geometry: At least the minimum number of anchors or beacons in view per zone for the algorithm in use, verified via field logs. Resilience tests: Defined edge cases tested, such as elevator rides, racks moved into line of sight, or doors closed, with documented behavior within acceptable bounds. Power and backhaul: PoE budget and switch port allocations validated, PTP time sync performance within vendor specs where required.
Tie these to the go‑live plan. For example, room level presence may be acceptable in staff support areas, while ICU beds demand sub 2 meters. Map success to risk and revenue, not a single global number.
Documentation is part of the deliverable
A survey that lives only in someone’s head will hurt you during maintenance or expansion. Capture anchor coordinates, heights, azimuths for AoA, and cable runs. Note wall materials that differ from as‑built drawings. Photograph each mounting location and include a labeled panorama for context. Keep calibration files from validation passes and the firmware versions used. All of this should live in your rtls management repository, not on someone’s laptop.
Make it easy for others to repeat your validation. I include a one page quick start with the grid layout, tag IDs used, TX power settings, and how to pull logs. Six months from now, someone will want to check a new wing or verify performance after a remodel. They should not have to reinvent your method.
Common pitfalls I still see
Relying solely on predictive modeling is the top mistake. Simulations do not know about that foil backed insulation behind a drywall segment. Another frequent miss is underestimating time sync for TDoA. Network PTP across a poorly performing switch fabric will create drift that ruins UWB location while seeming fine in general IT monitoring. If you see location “breathing” on stationary tags every few minutes, suspect time.
On BLE, AoA arrays installed at aesthetic heights often look great to architects and terrible to physics. AoA wants line of sight and separation. Tucking arrays into soffits or behind glass with metal frames is a shortcut to poor elevation angles and broken multipath assumptions.
Using different tag firmware or TX power during survey than in production is another classic. Document and lock tag configurations before field work. And do not ignore battery life. Cranking up BLE advertisements to 5 Hz may satisfy accuracy in validation but will crush batteries and create a maintenance nightmare.
Lastly, watch cable routes and IDF capacity. I have seen a perfect anchor layout die because the only path to run new Cat6 crossed a restricted OR corridor no one had permission to touch.
Working with your rtls provider rather than at cross‑purposes
A strong rtls provider will bring playbooks for surveys and validation, calibration tags, and software to visualize performance. Use them. Push for clarity on how their algorithms respond to edge cases you know exist in your facility. If they claim 30 centimeter accuracy, ask under what geometry and at what height, then test that configuration.
Agree on acceptance data formats. Many providers can export per‑sample error distributions if you feed them ground truth. Standardize filenames, time bases, and coordinate systems to avoid arguments later. And insist on training for your team on the rtls management console so validation and adjustments do not rely solely on vendor staff.
Special environments and how I handle them
Elevators: Metals and moving Faraday cages. Do not try to track inside. Treat elevator portals as events and snap position to floors based on last known location and door transitions. Validate by riding with a tag and confirming behavior.
Cold rooms and freezers: Cold kills batteries and changes RF properties. Use tags rated for the temperature range and test at operating temperatures, not at room temp. UWB performs well, but mounting and cabling must handle condensation and frost.
High‑bay warehouses: Anchor height and geometry dominate. Place anchors high and on rack end‑caps or building columns, not mid‑aisle. Expect occlusion when aisles fill. Use cross‑aisle anchors to maintain geometry when longitudinal views are blocked.
Hospitals: Infection control rules change how you work. Plan ceiling access with facilities, use clean covers for equipment, and sanitize gear. Stainless in utility areas is an accuracy killer. In patient rooms, privacy curtains are RF absorbers. Validate with curtains open and closed.
Hazardous locations: If you are near classified areas, use certified enclosures or keep active gear out of the zone. Passive RFID portals often make more sense than powered anchors inside hazardous bays. Work with EHS early.
Budget and trade‑offs you can explain to leadership
Every meter of accuracy costs something: hardware, cabling, or staff hours. I often frame decisions in clear, incremental steps. For instance, moving from 5 meter room level to 2 or 3 meters across a floor might require beacon density to double and tag rates to rise, with a 30 to 40 percent hit to battery life. Moving from 2 meters to sub‑meter in a focused area may require UWB anchors and time sync, with higher install costs per room but minimal impact to the rest of the building.
Show cost per location quality zone instead of a single blended number. Leaders can then decide that the NICU justifies premium infrastructure while admin offices run BLE room presence. This avoids all‑or‑nothing arguments and aligns spend with value.
After go‑live, keep the survey alive
Spaces change. Racks move, walls go up, and new devices enter the spectrum. Build a light maintenance plan. Re‑run a small validation suite quarterly in representative areas. Monitor error metrics in the rtls management platform, not just uptime. An unexplained drift or a rise in low confidence events should trigger a mini‑survey.
Tag batteries will die. Use your survey data to tune advertisement rates to the minimum that preserves accuracy for each use case. Do not leave everything at “high” forever. The cheapest RTLS program I ever ran saved more money in battery labor than it cost in advanced anchors.
A short case story to bring it together
A regional hospital wanted bed and pump tracking with 2 meter accuracy in patient areas and sub‑meter precision in two hybrid ORs. Predictive modeling suggested BLE AoA across the floors and UWB in the ORs. The passive survey found intermittent 2.4 GHz spikes near the dietary service route. We placed AoA arrays away from those corridors and raised a few arrays by 30 centimeters to clear curtain rails. In the ORs, we set UWB anchors at 2.8 meters, synced via PTP with a grandmaster in the same IDF. Validation grids used 1 meter spacing on general floors and 50 centimeter spacing in the ORs, with 2,000 samples per room type.
Results were consistent with targets: 1.7 meter 95th percentile in patient rooms, 45 centimeter in ORs. Latency averaged 1.2 seconds with outliers near elevators, which we excluded by design. Battery models predicted 14 to 18 months for BLE tags at 2 Hz in rooms and 9 to 12 months for high priority equipment with 4 Hz, which the clinical engineering team accepted. Documentation captured array photos, angles, and a set of test scripts so facilities could validate after a planned remodel. Two years later, when the hospital added a pediatric wing, that original survey pack saved weeks.
The throughline
A real time location system is not a black box. It is a radio network tuned to your walls, your ceilings, your https://beaunjob922.huicopper.com/benchmarking-rtls-accuracy-across-environments people, and your operations. A site survey is how you make it yours. Pick the technology that matches the job. Measure, then model, then measure again. Use tools that surface truth rather than hide it behind pretty graphs. Write acceptance criteria you would bet your reputation on. And keep the survey alive as your space evolves.
If you bring that mindset, the rtls network will stop being a science experiment and become part of how your organization runs the day, with fewer lost hours and events that fire when they should. That is the quiet success you want from an RTLS program, where dots on a map stand for trust rather than wishful thinking.
TrueSpot
5601 Executive Dr suite 280, Irving, TX 75038
(866) 756-6656