Without getting into the weeds, the main thing to avoid is reports from the same skimmer location using the same RBN callsign but with two totally different Rx antennas covering the same frequencies, so spots only a few minutes apart, or within the same minute, from the same skimmer, with the same frequency, show two completely different signal reports.
Each individual SNR report by each Skimmer is very accurately calculated by the SkimSrv software. Though they may vary a lot over time, the MEDIAN of all reports is usually very stable. But calculating a median report when each duplicate report skews too much because of totally different antennas leads to meaningless statistics since it's like reports from two completely different skimmer locations even though they are at the same location.
What you want to avoid it seeing reports like this from the Spots Analysis Tool caused by duplicate posts (graphic may be deleted by this reflector, but see the archives for the attachment.
In the context of multiple antennas & skimmers, I'm curious
about this data analysis use with skimmers.
Maybe it's been discussed before, but the RBN reports of the same
station from the same skimmer vary quite a lot over time in my
observation. I've noticed this in many contests where I may have
two screens of the RBN website feed up - one looking at my signal
and another looking at another guy who I'm interested in comparing
too, to see how we look at a single given skimmer site. In this
case, these reports can vary by 10+ db very easily, especially on
the lower bands.
Why I'm thinking about that flows from your comment on the data
analysis and it reminds me that I have seen some guys talking on
the reflectors about using RBN reports for fine tuning antennas
where they are considering changes that are in the few-dB range
and using RBN as the measurement method. I don't understand how
that can be done with precision given the variability the
atmosphere provides in cases excluding ground wave prop. It seems
like the noise in the data would swamp the relative compare value
for this objective.
Skimmer reports SNR so both the noise floor skimmer sees as well
as the signal(s) of interest needs to be stable over time so that
Skimmer is able to provide a repeatable report (at least within a
short measurement window). Otherwise it seems like a guy would
need to have two very different antennas (for example one skimmer
tied to his TX vertical and another one tied to a low-noise
directional RX) so that the SNR difference would be significant
enough to overcome the prop.
Maybe there is a statistical method that addresses this? Or maybe
I'm missing a key point here that is otherwise obvious?
If you have two skimmer instances monitoring the same
frequencies, they should never post under the same Skimmer
callsign, otherwise the reports will be quite useless for
data analysis, because the signal reports will be bouncing all
over the place and you can't tell which report to use.
Eliminating dupes alone doesn't solve this problem, because
the same station could be reporting with two different signal
strengths every few minutes, depending on which skimmer heard
it first.
If the two SDRs or two Skimmers are monitoring the same
band segments, use 3V8SS-1 for one and 3V8SS-2 for the other.
You need to run two independent copies of Aggregator to make
this work right, since the posted callsign is determined by
the first SkimSrv.ini, not by Aggregator.
For more details explaining how to do this, please review
this post by Aggregator author W3OA:
I use two receivers with separate receiving antennas. Both are
skimming
to Aggregator. How can I ensure Aggregator doesn't send the
same spot
twice if it happens that they come from both receivers?