You are right, your fork (readsb) does bring noticeably more positions.
However most RPi users are accustomed to Piaware SD Card image and feel ease and confidence in using it “as is”, and that is one of the main reason readsb is “not so popular”. Same is the case of Flightradar24 users who seem to be hooked to their Pi24 image.
So the slice_phase function changes are 99% just trial and error.
I had a hunch about neighboring symbols being assumed zero not being a good assumption and using an assumption of half level might lead to better results.
But as i started tweaking … any theory just went out of the window and i just followed an iterative process.
I’ve set a threshold of 75 for the preamble check for your PR that should keep the CPU the same as before or even a bit lower.
Adjust as you see fit from your own tests.
Only compile tested for the moment you’ll want to test it anyhow i assume.
I’ll have to check why the dump1090-fa mag length logic is different i just used the previous used calculation
uint32_t mlen = mag->validLength - mag->overlap;
In readsb it’s just mag->length but … as i said haven’t checked yet on why the difference.
Let me know how the changes work for you.
My main feeder is out of the top 3,000 now with just over a third of the number of aircraft that the top place has. Out of the top 1,400 feeders I think there’s only two outside the USA.
In my area there do seem to be increasing flights in general. But what really takes the temperature here are the air ambulance and life flight aircraft. We have a fair number of rural hospitals that end up having to get people transported to Albuquerque if they take turns for the worst. It’s sobering to watch them and realize the inbound flights generally have a life in the balance.
Not a lot of change in this region in recent months. Unrelated to COVID trends, we have fewer sunny days to bring out those peak bursts of GA traffic. Commercial aircraft traffic is flat, as seen in the low points on the chart. The one-off drops are holidays.