Bluetooth in crowded rooms
BLE demos beautifully on a desk with two devices. A venue with thousands of phones is a different radio environment, and the difference shows up late.
Bluetooth Low Energy is the right tool for a lot of proximity problems, and it is unusually good at looking finished before it is. The bench does not reproduce the environment it has to work in.
The spectrum is busy
BLE advertising shares 2.4GHz with every WiFi network, microwave and wireless mic in the building. In a packed venue the noise floor rises, phones are already saturated with radio work, and effective range collapses well below the number in the datasheet.
Plan for far shorter range than the bench suggests, and test in the actual space before committing to a layout.
Phones do not scan the way you expect
Scan behaviour differs between platforms, between OS versions, and between foreground and background. Background scanning is aggressively throttled, and duty cycles vary in ways you do not control.
Anything time-sensitive has to tolerate a device noticing late - or not at all, for that cycle. Design the experience so a missed detection degrades rather than breaks.
Density changes the maths
More transmitters is not linearly better. Past a point, advertising packets collide and detection rates fall.
- Tune advertising intervals rather than maximising them
- Stagger transmitters instead of synchronising them exactly
- Measure detection rate at density, not at a desk
- Plan placement around bodies, which absorb 2.4GHz effectively
Test with real crowds
A room with twenty phones behaves nothing like a room with two thousand. If the deployment is a venue, the only meaningful test is in that venue with people in it - which means budgeting for a site visit before launch, not after the first complaint.