Choosing a Positioning Technology: UWB, BLE, RFID or Wi-Fi

Indoor positioning proposals tend to open with an accuracy figure, which is the least useful place to start. All four common technologies will locate an asset indoors. They differ in what they physically measure, and that difference decides both the accuracy you get and the price you pay for it.

RFID does not really measure position at all. A passive tag has no power and no clock; it responds when a reader energises it, so what you learn is that a tag was within range of a specific reader at a specific moment. Placed at a doorway that becomes a boundary crossing, which for a great many inventory problems is the entire requirement. It is also by far the cheapest per item, because the tag is a printed antenna and a chip with no battery.

BLE and Wi-Fi mostly infer distance from received signal strength. Signal strength falls off with distance, so with enough receivers you can trilaterate a position. The difficulty is that signal strength also falls off with a person standing in the way, a metal shelf, a tag's orientation, and reflections arriving out of phase. Accuracy in an empty room is genuinely a few metres; accuracy in a loaded warehouse is worse and, importantly, biased rather than noisy — the error does not average out over time.

UWB measures time of flight with very short pulses, which is a fundamentally different quantity. Distance derived from timing is far less sensitive to obstruction and reflection, which is why 10 to 30 centimetre accuracy survives conditions that degrade signal-strength methods badly. It costs more per tag, needs powered anchors installed with known positions, and requires the site survey that installation implies.

The way to choose is to write down the smallest distinction the application must make, then buy the cheapest technology that makes it reliably. If the question is which room, BLE is enough. If it is whether an item left the building, RFID at the doors answers it for a fraction of the money. If it is which of two adjacent pallets, or whether a key is inside or outside a vehicle, only time-of-flight will do — and in that last case the requirement is not accuracy but security, because a system that can be fooled about distance can be fooled about permission.

Most real sites end up with more than one, and that is the correct outcome rather than a failure to decide. The cost that gets underestimated is not the hardware but reconciling several systems into one view, which is where the schedule usually goes.