thanks to prog and all other postings. going from e 7.0 to e7.1(mini) is a surprising increase. (RPI 3B+ and M=12…still can’t do M=20 but working on it). just downloaded wiedehopf performance graphs
1 Like
Going from -e 8.4 to 7.4, reduces single message tracks significantly.
(Still experimenting with the best graph format, averaging is not visible here, only noticeable for 48h/24h/6h/1h graphs)
Please check v1.59 with a new (repurposed) -w <whitelist_threshold> switch for configuring the white list threshold. The default value is 1 valid CRC message for whitelisting the ICAO, but you can set anything between 1 and 10. Note that this is tightly linked to the timeout setting.
This comment by @obj started the exploration:
2 Likes
How does this interact with the -x switch? Previously that resulted in an increase in bad decodes - this will obviously help prevent that, but should there be any advantage of using -x in some or all circumstances?
-x and -w are not mutually exclusive. -x allows more frames to contribute to the whitelisting than just CRC valid ones. The effects can be accumulated.
OK. Something to experiment with I guess. It might make sense to use -x with a higher -w setting, as it would appear that the filtering is happening at a later stage with the -w switch.
Yes, exactly. But -w also has some feedback effect for the next decodes.
I had to wait for the lunch break to SSH into my box, you guys have all the fun…
I have added -x and -w2, I’ll see what happens.
My instalation is a Pi4 runing an Airspy-Mini at 20Mhz with -e 7.2 and -w 2
Like you can see on the picture, the tracks with single message dropped, but the tracks with more than one message dropped too !
I supose that is not good ?
I have try with -w 3 but it’s the same result.
Doesn’t look like the green area changed.
Display is changed so the graphs are stacked now.
Get the new version it has the green on the bottom, easier to compare.
You can also edit /etc/default/graphs1090 and at the end set a fixed upper limit for that graph.
# set custom y-axis upper limits for the individual graphs
# for automatic upper limits leave them blank
ul_message_rate=
ul_aircraft=
ul_tracks=
ul_range=
ul_maxima=
ul_rate_per_aircraft=
ul_adsb_cpu=
ul_range_uat=
ul_tracks=2000 should work fine for you with the new graphs.
If you want to redraw all graphs:
sudo systemctl restart graphs1090
2 Likes
I’ll post my six-hour graph showing pre -w usage along with -w 3, -w 2 and -w 1 in a little bit. In the meantime, it’s clear to me that -w 1 lets a lot of bogus stuff through, which means I’ll likely switch to -w 2 long term.
There is no way I’m copying three of anything on the ground right now, especially with non-american ICAOs.
Well -w 1 is the what the previous version without -w did.
“More or less” 
2 Likes
Changed to -w 2, then enabled -x with -w 2. Looks like a higher setting is required with -x. I’m now trying with -w 3 and -x.
The number of aircraft shown without a position in dump1090 stats is definitely affected by this:
I wouldn’t use -x at all, even with -w …
I will likely turn it off again, but I wanted to see the effect of the different filtering. Having -x switched on makes the decoder more permissive with the trade off of more bad decodes - I wanted to see whether using -w is sufficient to counteract that bad data.
After a brief period, setting -w 3 with -x has about half the single tracks as with -w 2. I’m going to increase to -w 5 now and see how it changes.
Ok today i updated and added -w2 as a start busy results at mo no real change over all except for the OBVIOUS 17:00
@wiedehopf: I was just going by what Prog said here:
Ok, I’m going to switch back to ‘-w 2’.
Here is a screenshot of (a) pre -w, (b) -w 3, (c) -w 2 and (d) -w 1
Mike
The assumption still holds.