Does GPS work indoors? Why it fails and what works instead

Does GPS work indoors? Mostly, no. A phone next to a window or under a light roof may get a rough fix, but inside a concrete, steel or multi-storey building the satellite signals are too weak and too scrambled to give a usable position, and GPS has no idea which floor you are on. Indoor positioning therefore uses other signals — Bluetooth, Wi-Fi, ultra-wideband, the Earth’s magnetic field or the camera — and every one of them needs an indoor map to make its coordinates mean something.

Spatial Map is a browser-based indoor mapping platform: it turns floor plans into multi-floor 3D maps and shows the positions your positioning system reports on them. It does not compute positions itself, which is why this guide treats positioning and mapping as two separate jobs.

GPS satellites orbit at roughly 20,200 km. By the time their signal reaches the ground it is far weaker than the background radio noise; receivers only recover it because they know the code pattern to look for. Outdoors with open sky, the US government’s GPS.gov site puts typical smartphone accuracy at about 5 metres. Indoors, four things go wrong at once:

• Attenuation. Concrete slabs, metal roofs, foil-backed insulation and coated glass absorb most of what is left of the signal. A metal-clad warehouse can behave almost like a closed box.
• Multipath. The signal that does get in bounces off walls, racks and floors. The receiver measures a longer path than the true one, so the position drifts or jumps.
• Too few satellites. A fix needs at least four. Next to a window you may see two or three, all from the same side of the sky, which makes the geometry poor even when a fix appears.
• No floor. GPS height is the least accurate part of a fix even outdoors. Indoors it cannot tell the ground floor from the second, which is the question people actually ask.

Indoor positioning systems replace the satellites with something closer. They split into two families: infrastructure-based systems, where you install beacons, anchors or readers, and device-based systems, where a phone works out its position from signals or sensors it already has. The precision figures below are rough orders of magnitude from published vendor and research material; the building, the installation density and the calibration decide what you actually get.

Most real systems combine two or more: BLE or Wi-Fi for a rough position, inertial sensors to smooth it, and a barometer or a map rule to settle the floor. For tagged assets rather than people, infrastructure-based options win, because the tag does not need a screen, an app or a user who remembers to open it. The BLE asset tracking page explains gateways and tags in more depth.

The cheapest system is the one that answers your question and no more. Four questions narrow it down quickly:

1. 1
   What are you locating — people with phones, or things with tags?

   Visitors navigating a venue need device-based positioning in an app. Pallets, trolleys, tools and equipment need tags and infrastructure.

2. 2
   Which answer do you need — building, room, aisle or shelf?

   “Which room is the infusion pump in?” is a BLE question. “Which bay is the forklift in?” may need UWB. “Did this carton leave through dock 4?” is an RFID question.

3. 3
   How often must it update?

   Continuous tracking costs batteries and infrastructure. Many operations only need to know where something was last seen, or when it crossed a door.

4. 4
   Who maintains it?

   Fingerprint surveys go stale when a building changes; anchors and gateways need power and mounting. Budget for the upkeep, not just the install.

If the job is showing many kinds of positions together — BLE tags, UWB vehicles and RFID reads on one floor plan — you need an RTLS software layer that takes positions from each system and plots them in one place, rather than a separate screen per vendor.

A position is a pair of numbers, a zone ID or a reader name. On its own it says nothing a person can act on. It becomes useful when it lands on a map that knows the rooms, doors, racks and floors around it — “Bay 12, mezzanine, next to the fire door” rather than “x 34.2, y 18.9”. That map has to be built from the building’s own floor plans, because no satellite or street-view car has ever been inside.

• Floors and names. Positioning systems report a floor number or level ID; the map turns it into “Level 2, Pharmacy store” and keeps the levels stacked in the right order.
• Real-world coordinates. When the indoor map is placed at true latitude and longitude, outdoor GPS can hand over at the entrance without the dot jumping across the car park.
• Walls and doors. Routes and position smoothing both need to know what is walkable. A dot that crosses a solid wall is a mapping problem as much as a positioning one.
• Context. Live sensors, stock and alerts on the same map are what turn “where is it?” into “what should I do about it?”.

An airport terminal mapped in 3D, ready for positions and live data

In Spatial Map, positions arrive from your RTLS, gateway or middleware through a REST source or a table with a write API, and appear as live markers or heatmaps on the right floor. Directions work without any positioning at all: indoor wayfinding routes between two points you pick, across floors through stairs, lifts and escalators, with Fastest, Step-free and Other route options. There is no blue dot of the visitor’s own phone; that part belongs to the positioning system.

If you are planning indoor positioning, build the map first. It is useful on day one for planning, directions and reports, and it gives the positioning vendor a georeferenced base to calibrate against. The indoor mapping guide explains how indoor maps are made from PDF and DWG floor plans; Spatial Map’s 7-day free trial lets you try it on up to 5 facilities without a card.