Indoor Positioning Without GPS: The Real Accuracy Numbers

Indoor positioning systems are the set of methods — radio ranging, Bluetooth signal strength, Wi-Fi fingerprinting, magnetic-field matching, and camera-based matching — used to estimate where a device or person is inside a building, in place of GPS. None of them use satellites at all. GPS itself simply doesn't work reliably once you're under a roof.

Consider a visitor walking from a mall's parking structure into a three-level atrium in Rotterdam. Their phone's blue dot drifts for a few seconds, jumps a floor, then goes quiet. That gap — the moment satellite navigation gives up — is exactly what every technology in this article exists to close. The question isn't whether indoor positioning works. It's which version of "works" you're buying, and how far that number drifts once it leaves the lab.

This piece walks through the main technologies, their published accuracy figures, the field work required to make any of them function at all, and where the rules differ depending on which European market you're deploying in.

Why GPS Checks Out at the Door

GPS works by timing radio signals from satellites roughly 20,000 km overhead. Concrete, steel beams, and even heavy glass attenuate or scatter those signals badly enough that a phone's GPS chip either loses lock or reports a position off by dozens of metres. Indoor Positioning Systems exist precisely because satellite and other outdoor positioning technologies lack precision or fail once you're inside a building. That's not a software bug. It's physics, and it's the same in every country.

Every technology below trades accuracy against installation cost and maintenance burden differently. The gap between a vendor's lab-reported number and what you'll actually see on a crowded mall floor is, honestly, most of this article.

The Core Technologies, and What They Deliver in Practice

Ask five vendors how accurate their system is and you'll get five different numbers, usually measured in a quiet lab with a handful of anchors and zero foot traffic. Real buildings have shelving, escalators, elevator shafts, and crowds. Here's what published research and vendor documentation actually show once those variables show up.

Ultra-wideband (UWB) refers to a radio technique that times very short pulses travelling between a tag and several fixed anchors, converting that time-of-flight into distance. It's the most accurate RF option on the market today. UWB can achieve 10-cm accuracy in line-of-sight conditions, and keeps performing reasonably well even in multipath environments like hallways with reflective surfaces. Under tightly controlled test conditions with multiple anchors, average ranging accuracy reached 4.35 cm, with a standard deviation of 2.33 cm. That's a genuinely impressive number. It also comes from a 4×4 metre test area with no crowd walking through it, which matters once you try to scale the same setup across a 60,000 m² retail floor.

BLE beacons are cheaper and far more common in malls and airports. Client-based Bluetooth positioning, where the phone itself calculates its position from nearby beacons, typically lands in the 1–5 metre range under real conditions, with UWB sitting at roughly 10–50 cm and Wi-Fi around 10 metres on the same comparison scale. Lab tests under ideal, flat, low-noise conditions have hit 10 cm with BLE — but accuracy dropped to meter-scale the moment the setup moved into a real, three-dimensional, noisy environment. Coverage costs more than people expect, too: deployment guidance calls for roughly one to five beacons per 150 square metres of floor, and every one of them eventually needs a battery swap.

Wi-Fi RSSI fingerprinting reuses access points you've probably already installed. It compares a phone's measured signal strengths against a pre-walked fingerprint map of the building, typically landing below 10 metres of accuracy — coarser than BLE, but it adds zero new hardware if your Wi-Fi density is already decent. The catch: every time IT swaps an access point, the fingerprint map goes stale, which is the same freshness problem that haunts most location datasets over time (we've covered why address and POI data decays the same way elsewhere on this site).

Geomagnetic and sensor-fusion positioning skips beacons entirely. It matches a phone's magnetometer and inertial sensors against a magnetic signature map built by walking the building once. Vendors report 1–5 metre accuracy using more than 50 fused data points, with no beacons, no added Wi-Fi dependency, and no on-site hardware required at all. It's the cheapest option to scale across a big footprint, and the most fragile when the building physically changes — a new steel delivery dock or a renovated wing can quietly shift the magnetic fingerprint underneath an otherwise-working system.

Visual positioning (matching a phone camera's view against a 3D reference model) and pedestrian dead reckoning (chaining inertial-sensor steps between known waypoints) both show up as supplementary layers, usually stacked on top of one of the methods above rather than standing alone. Neither has a single standardised accuracy figure worth quoting on its own, because both inherit whatever error already exists in the reference map or starting waypoint they depend on.

Technology	Typical real-world accuracy	Infrastructure needed	Best for / Where it stops
UWB (ultra-wideband)	10–50 cm line-of-sight; worse behind thick obstructions	Fixed anchors every ~10–20 m, powered and calibrated	Best for asset and forklift tracking. Stops at: anchor density makes whole-building retail rollout costly.
BLE beacons	1–5 m typical	~1–5 beacons per 150 m², battery maintenance	Best for consumer wayfinding apps. Stops at: meter-level precision only, degrades in dense 3D layouts.
Wi-Fi RSSI fingerprinting	Below 10 m	Existing APs plus a fingerprint survey	Best for buildings with dense Wi-Fi already. Stops at: map goes stale whenever APs move.
Geomagnetic / sensor-fusion	1–5 m	None — a single mapping walkthrough	Best for large footprints avoiding beacon cost. Stops at: drifts after renovations or big metal fixtures move.

Which one wins? Depends entirely on what you're trying to do with the position once you have it — route a shopper, or track a forklift.

Before You Can Position Anything, You Need an Accurate Map

Here's the part vendors leave out of the brochure: none of the technologies above work without an accurate base map of the building first. A BLE beacon can tell a phone it's three metres from Anchor 14. It can't tell the phone that Anchor 14 sits next to a pillar in the middle of the food court, unless somebody already measured and modelled that pillar.

Simultaneous Localization and Mapping (SLAM) refers to the process a mobile scanner uses to build a 3D map of its surroundings while tracking its own position within that map in real time, without GPS. It's the method that made mobile indoor mapping practical at all. Mobile LiDAR scanning was limited to outdoor work until GPS could be used for precise positioning outside — SLAM is what removed that outdoor-only constraint for interiors.

Three capture methods dominate commercial indoor mapping today, and they trade speed against precision in predictable ways.

Capture method	Typical accuracy	Speed	Best for / Where it stops
Terrestrial tripod LiDAR	Sub-centimetre, survey grade	Slowest — static setup, scan by scan	Best for legal floor-area measurement and as-built archives. Stops at: too slow to capture a whole mall overnight.
SLAM handheld / backpack mobile mapping	A few centimetres — enough for facility digital twins	Roughly 10x faster than terrestrial scanning	Best for mall corridors and retail floors at volume. Stops at: slightly less precise than tripod scanning; motion blur in crowds.
Drone-assisted / image-based (360° camera + LiDAR)	Varies with flight height and camera resolution	Fast coverage of large open volumes	Best for atriums, exteriors, repetitive infrastructure inventories. Stops at: tight corridors still favour handheld SLAM.

In one documented comparison, a SLAM-based mobile scanner captured a building in roughly a tenth of the time a static terrestrial scanner required for the equivalent job. For a large retail floor, that's the difference between finishing a scan overnight and closing a wing for a week.

The broader capture market is leaning the same direction. Mobile mapping vendors increasingly pair LiDAR with high-resolution 360° imagery rather than laser alone, with imagery and photogrammetry workflows increasingly treated as the practical path forward for many mapping-heavy industries. Drones and backpack rigs cover atriums and parking structures fast; handheld SLAM units still win between shelving and in tight service corridors. The LiDAR-in-mapping market itself reflects that momentum, valued at roughly USD 5.3 billion in 2025 and projected to grow at a 25.6% CAGR through 2035, a pace driven partly by exactly this kind of indoor and facility-scale demand.

We've gone deeper on the full capture-to-model pipeline in our piece on what a shopping mall digital twin actually contains and how it gets built, if you want the next layer down from positioning into the full dataset.

A Worked Example: Wayfinding Across a Three-Level Mall

Consider a 60,000 m² shopping centre with 180 stores across three levels, somewhere in the Benelux. Management wants an app that routes a visitor from the parking entrance to a specific shoe shop — accurate enough to say "second corridor on your left," not just "you're somewhere on level two."

That rules out Wi-Fi fingerprinting alone; a 10-metre error radius can put a visitor in the wrong wing entirely. BLE beacons at 1–5 metres get a visitor to roughly the right storefront, which is plenty for consumer wayfinding, and the hardware cost — beacons every 150 m² or so — is manageable at mall scale. UWB would be overkill and far too expensive to blanket three retail levels in anchors. It earns its cost tracking forklifts in a warehouse, not routing pedestrians past a food court.

None of that beacon math works, though, if the underlying floor plan is wrong by even a metre. That's why the field work — the SLAM scan, the floor-by-floor walkthrough, the beacon placement survey — has to happen before a single line of navigation code gets written. Teampl has run exactly this kind of programme: a mall and commercial-space mapping effort across Germany, France, Benelux, and Switzerland, collecting hundreds of scans a day for months, because that's genuinely what it takes to keep a floor plan current across four markets at once.

Does Geography Change the Technology? Mostly No. The Rules Around It? Yes.

The physics of time-of-flight ranging is identical in Lyon, Zurich, or Rotterdam. A UWB pulse doesn't care which country it's in. What differs by market is how you're allowed to capture and retain the signals your positioning system depends on.

Wi-Fi and BLE-based positioning both rely on scanning nearby devices, and in the EU that scanning sits squarely inside GDPR and ePrivacy territory. For a navigation app a visitor has actively opted into, this is mostly a non-issue — the phone is choosing to report its own position. For passive footfall analytics layered on the same beacon infrastructure, it's a genuinely different compliance conversation, one we've written about in more depth when it comes to what retail analytics can still measure from Wi-Fi probe requests given MAC randomisation.

Measurement standards differ less by country than you'd expect. BOMA (Building Owners and Managers Association) floor-area standards, originally American, show up routinely in European facility-management specs because landlords and tenants both want a consistent definition of rentable square metres. The underlying capture technology — SLAM, terrestrial LiDAR, or photogrammetry — doesn't change by jurisdiction. The reporting format it feeds into does, and that's usually a lease-and-compliance question, not a technical one.

Keeping the Position Right After Launch Day

A digital twin captured once and never revisited starts drifting from reality the day a tenant changes its storefront. The same is true for a beacon map, a Wi-Fi fingerprint, or a magnetic signature — all four degrade on different clocks, but all four degrade.

• BLE beacons fail individually — batteries die on a schedule you can at least plan around.
• Wi-Fi fingerprints go stale whenever IT swaps an access point, usually without telling facilities.
• Magnetic maps shift after any large metal fixture moves — a new escalator, a delivery dock, a seasonal pop-up kiosk built from steel framing.
• The base floor plan itself goes out of date the moment a store signs a new lease and knocks down a wall.

Treat the capture as a maintenance programme, not a one-off project — the same discipline we'd apply to any field dataset that decays on its own timeline. Re-scan key zones on a cadence tied to your lease-renewal and renovation calendar, not to a fixed date on the wall calendar.

So Which Technology Should You Actually Pick?

Short version: match the technology to the decision the position needs to support, not to the most impressive accuracy number on a spec sheet.

Routing shoppers to a storefront? BLE or geomagnetic sensor-fusion gets you there at a fraction of UWB's cost. Tracking a forklift around a warehouse full of racking? UWB's centimetre-to-decimetre accuracy earns its anchor density. Just want a floor-level "you are here" and already have decent Wi-Fi coverage? Fingerprinting is close to free.

And whichever you pick, the field operation underneath it — the scanning, the beacon placement, the repeat visits that keep the map honest — matters more than the technology label on the box. That's the part the brochure skips, and the part that actually determines whether your wayfinding app still works on day 400, not just on launch day.

Start a Project

Teampl runs field data collection programs — egocentric video, in-store audits, mystery visits, GIS surveys — designed around a specific research question, not a generic panel. If this topic touches a program you're planning, tell us what you're trying to learn.

Frequently Asked Questions