In an attempt to try and fly a drone indoors, we worked (and succeeded) on building an indoor positioning system. We used 2 sets of TREK1000 demo kits from DECAWAVE to calculate ranges from each of the anchors to our tag on the robot. We then used an iterative least squared estimation (think linear algebra and GPS) to solve for the postion which then was broadcasted via XBEE.
The eventual goal is to use migrate to the DWM1000 (much cheaper) and using kalman filters and LIDARs/ Barometers to augment the indoor postioning system
I am not sure which approach might be quicker, 4 points giving squares or 5 points giving linear. I'll have to check how you work with just 2 fixed reference points, but I read somewhere that even one is enough. It is just the number of sample points that finally gives the location.
You might find some useful additions to your approach from work being done for a 3D printer: see the log https://hackaday.io/project/13420/log/56335-calibration-with-multilateration-using-w1209, and the associated comments.
I am not sure which approach might be quicker, 4 points giving squares or 5 points giving linear. I'll have to check how you work with just 2 fixed reference points, but I read somewhere that even one is enough. It is just the number of sample points that finally gives the location.