08-30 Custom PLM coefficient results

I set up the custom PLM coefficients for each server and tested the results. Below is a visualisation of the difference in path loss properties based on servers. The calibration walk was done along the same path, so the only factors that influenced the results were the different servers.

{ width=80% }

Problem and solution

Testing the setup with the new changes yielded some promising results. Something that became quite apparent was that the server 38:ae was being picked up from way across the room and used in trilateration, even when there were clients that were much closer. The curve is a clear outlier as shown above in blue, so low (unreliable) measurements were perceived to be relatively close due to the weird coefficients. I am not sure why the server 38:ae got such a weird curve, but after manually modifying the y-offset the system behaved as expected. The approximations seemed reasonable within a sensible margin of error.

Next steps

Next I would like to try out the client splitter that I built last week. Afterwards I should start thinking about the GUI.