[pizza]

Time-of-flight is like knowing (dt_[1, 1] = 0,) dt_[1, 2], dt_[1, 3,], .. dt_[2, 3], .. dt_[n, n] for all inter-beacon times, and I'm assuming bluetooth signal travels through air in essentially constant time (doing any muxing of bluetooth "channels", if such a thing need exist, to prevent "overlap" of bandwidth?).

So do you have some kind of convex hull program that finds a shape of (dx = v * dt), where any predicted dx_[1,3] >> dx_[2,3] + dx_[1,2] implies some line-of-sight obstruction between beacon 1 and beacon 3? Especially when combined with dB of signal strength versus mere "dx" quantity alone?

I guess I'm wondering what the typical resolution is of a floor plan, how many beacons are necessary, and what kind of algorithm can crunch all those numbers into a neat path-planning-type solution.

edit: I'm quite the fool this morning, having not read the post before commenting! Nicely done. Looks like a sweet product. Hope you're able to grow the company!

[jimiasty]

The time-of-flight is actually possible because of UWB radio. It's an additional chip to Bluetooth that can estimate distance between nodes with inch precision. Then using Bluetooth mesh these data are passed to other beacons and to phone where automapping is performed. Once we know location of nodes standard indoor positioning with Bluetooth beacons is performed.

Bluetooth range is aprox 100m and UWB range 70m, so that's the maximum distance between nodes. The more nodes you have the more accurate the shape is.

For a 1000 sqf office you would neeed probably 12-20 beacons. For a retail store more than a 100.