How does LoRa based geolocation work? How does it measure distance?

Question

The speaker in the video Richard Lansdowne - LoRa Geolocation mentions straight-off that he won't be quoting an accuracy for geolocation using LoRa, and it's of course not the ideal signal you would use.

A comment below the video says:

Excellent Richard, you nailed it! You've succinctly described the IoT asset tracking problem and what we need to do going forward to make this real.

So far I can't figure out, at least from the video, how this might actually be done using standard LoRa and perhaps LoRa-WAN. How would LoRa protocol estimate distance in order to triangulate, assuming that's how this works? What would be an approximate best-case accuracy based on hardware limitations? (of course your milage may vary)

Answer 1

LoRaWAN localization uses TDOA: Time Difference Of Arrival.

Gateways with the precise timing extension are all synchronized via GPS. When a device sends a frame, the very accurate time at which the frame was emitted is unknown, but by recording the exact time each of several gateways receive that frame (compared to GPS time), and knowing the position of gateways, you can calculate the position of the device.

On the gateway side, that requires hardware capabilities beyond what a SX130x offers, a clear view of the sky, some calibration and a clever solver somewhere is a server room.

On the device side, that requires nothing.