Thanks! The lossy compression of the image will reduce the information available in the observation so having the raw data would be a great help. We could download the audio and analyze it but this would save us a step and might also be of use to others.
We’re planning to scan the data for an elongated point spread function. This is a very unique signature (ie. 12.5 khz by x milliseconds every y seconds) that stands out above the background noise. It would also make it easier to see an extremely weak signal detection that is not obvious visually in the waterfall plots. Theoretically it could even be used to automatically vet an observation if you know the expected transmitter bandwidth, packet duration, and tx interval. The dB volume within that psf ellipse compared to the average background noise gives a robust measure of snr for each reception.
We’re still working on a method to demod the proprietary encoding of the XDL Micro transmitter. For now we are just looking at detection versus missed packets. Then we’ll estimate signal strength as it changes during a pass for each station.