Target position bobble

I’ve always noticed on Basestation that sometimes the target on landing approach will go backwards a bit and then continue forward. It was always kind of a bug for me.

I’ve implemented the dump1090 CPR algorithm in my Java tracker now and I noticed the bobble today. Aha! It’s obviously switching back and forth from the global position (odd/even position reports) and the relative position (either an odd or even, and my local reference, or a reference to a previous global).

I need to add a display to show which of the three positions it is using. Might be an interesting statistic. 1) global, 2) relative to last global, and 3) relative to station reference.

It’s not the switch from airborne to surface encoding that’s doing it?

It might be. I’m going to instrument it and find out. Surface does sound more suspect though. Seems like a switch between two references.

I’d be interested in what it is. I’d have thought that if it was the reference position / relative decoding then it would be a much larger effect because the reference position is really just used to disambiguate which 360NM cell (for airborne) the position is in. Global CPR is basically doing the same thing, the combination of the two messages tells you which cell you’re in and then you use the more recent message to find the position within the cell.

It might be a change in precision? The surface encoding has better resolution.

Well I looked at the code, and I only use surface if I detect the aircraft signal it is on the ground. This happens usually on approach (maybe other times but I don’t notice it much maybe). Otherwise the code uses the global relative.

It’s not a huge bobble. Maybe a couple of pixels at most.