For my LF opds-32 detector, I currently have 49 entries in the
searchlist: https://dl.dropboxusercontent.com/u/26404526/callsloc.txt
I am regularly inspecting the output for likely false positives, which
are then highlighted by a question mark at the end of the line. Over
several months, their rate of occurrence has been on the order of 2 per
week. Most are associated either with one of the many spurious
carriers, or Opera transmitters which had not yet been included in the
list at the time.
Best 73,
Markus (DF6NM)
-----Ursprüngliche Mitteilung-----
Von: jrusgrove <[email protected]>
An: rsgb_lf_group <[email protected]>
Verschickt: Mi, 4 Feb 2015 3:38 pm
Betreff: Re: LF: OPdynamic false T/A spot
Mike
We agree that OPdynamic spots with ?? should not be uploaded ... but
that isn't the case at the
moment. The R7NT OPdynamic spot of WD2XNS (10 hours after the
transmitter was turned off) shows up
right along with all of the genuine spots - it's indistinguishable.
This is a problem and seriously
undermines the credibility of OPERA.
As for callsigns only being held for a short time in the deep search
list ... I going to have to
disagree. While it may put a bandaid on the serious OPdynamic problem
with false spots it makes it
far too easy on the correlation detector. It's possible that only one
callsign might be on the list!
I'd much rather have the DF6NM system where all possible stations
(about 30 stations last time I
checked) are present all the time. The game sure seems a lot less
'rigged' when the correlation
detector has to analyze 30 callsigns rather than just one or maybe two.
When you think about it, the DF6NM system must be vastly superior to
OPdynamic ... considering the
minute number of false spots with 30 callsigns to pick from. OPdynamic
produces far more false spots
and it only has a few callsigns to choose from at any time. Reducing
OPdynamics choices even further
(by shortening the 'holding time') would reduce false spots ... but
that just points up how inferior
OPdynamic really is. Not a good solution in my opinion.
Jay W1VD WD2XNS WE2XGR/2
----- Original Message -----
From: "Mike Dennison" <[email protected]>
To: <[email protected]>
Sent: Wednesday, February 04, 2015 5:50 AM
Subject: Re: LF: OPdynamic false T/A spot
Jay,
I am inclined to agree with you, but almost all of the Deep Search
spots that have no question marks are genuine. If at all possible,
spots =with= the question marks should not be uploaded to PSK
Reporter as most are false.
Also, I suggested early on that callsigns are on the database for too
long, and should only be a couple of hours after the last known
'real' Op decode.
These steps would of course slightly reduce the number of genuine
weak signals decoded/reported but the result would be far more
accurate and still show a useful increase in sensitivity over Op.
Mike, G3XDV
==========
|
