Graham
I won't be quite a 'charatable' as Mike ...
There were at least two false detections on my transmitted signal (WD2XNS) last night that I saw ...
one into G and one into SV. About the only way to tell they were false detections was the time lag
between actual and the 'Deep Search' output - it wasn't the expected 4 minutes. Had there been no
'real' receptions during that time period one might have actually believed what was reported ... as
you did and reported on the Yahoo reflector.
On receive from here last night from there were false detections of VO1NA.
I saw one detection of a G station in JA make the list! This was clearly a false detection. No doubt
there were plenty of other false detections but after this couple hour 'outing' with 1.5.5 I gave up
on it. It's clearly not ready for prime time and should be pulled from circulation and return to the
drawing board!
Perhaps the author, instead of taking the time to write a 'propaganda' .pdf page about DF6NM's OPDS
and include it with the download, should spend more time actually making his software work as well
as Markus's. In almost a year of using OPDS I have identified only one false detection. This is
severly at odds with Jose's test of OPDS. Since Jose seems to now like the idea of Deep Search,
maybe he should 'borrow' more ideas from Markus ... like the correlation % and dB 'certainty'
indicator, an accurate time indicator and a highly accurate frequency readout. These indicators,
especially when correlated with other OPDS users grabbers, makes the very rare false detection with
OPDS obvious.
At this point Markus's OPDS runs circles around Jose's Deep Search ... regardless of what his
'propaganda' page says.
Jay W1VD WD2XNS WE2XGR/2
----- Original Message -----
From: "Graham" <[email protected]>
To: <[email protected]>
Sent: Thursday, January 08, 2015 9:03 AM
Subject: Re: LF: New version OPERA >> ''Opera Dynamic'' <<
Hi Mike,
It is actually set to just over 5 dB gain over the standard Opera system , the opds s/n
reports can be upto 4 db's lower for the same signal , but interims of ultimate
sensitivity , the 11 dB advantage was not demonstrated in testing ,
Its interesting to note , that the search detection could only improve on the Opera data
decoder by 5 or 6 dB for the same level of reliability , which as a data mode , points to
it being quite efficient
Your right , you have now 2 systems in one package , watching last night , I could see some
miss-decodes from both , but the dynamic list remained valid , which is one of the design
features and will minimise its content on a 8 day rolling schedule
Compared to the initial roll out , when the news groups lit up with 'too much cpu'
and we hate OOK mfsk is better etc , I think the launch went quite well , its really
to the user as to what use is made of the available systems , Jose maintains this is a
hobby ..
Early days yet ,
73-Graham,
G0NBD
--------------------------------------------------
From: "Mike Dennison" <[email protected]>
Sent: Thursday, January 08, 2015 10:54 AM
To: <[email protected]>; <[email protected]>
Subject: Re: LF: New version OPERA >> ''Opera Dynamic'' <<
Having taken a long look at Opera 1.5.5, it seems to have good and
band points, but is useful if used with great caution.
On the plus side, it appears to have some 3dB more sensitivity than
previous versions of Opera. This is not better than DF6NM's OPDS
program, but that is more complex to set up.
On the minus side is the number of believable false decodes which I
think are being underestimated by the program's advocates. For
instance, on day one one third of the times that my transmissions
were reported I was off the air. On day two I transmitted once but
was reported on two occasions. However, all of these false decodes
were one-offs, in other words only one station reported me at that
time, whereas several reported the genuine transmissions.
So, v1.5.5 looks very useful but any decodes that are reported by
just one station has a high likelihood (but not certainty) of being
false. For this reason, I would not advise using this version without
having a web connection, or at least cross-checking later against the
web archive.
Mike, G3XDV
==========
Graham, thanks for the useful info. I have sent you a direct email
with some constructive suggestions.
One snag with false decodes of this type is that you don't know they
are false unless the sender denies he was sending (or propagation is
impossible) at that time, so there will be statistically far fewer
reports of false decodes than there are in reality. I transmitted four
times in the last 24 hours and there were two additional decodes
falsely carryng my callsign in the same period. That's two out of a
total of six, ie 33%.
regards
Mike
=====
> Some points Mike
>
> 200%F is a bodge to get the text beacon via the server , not
> showing in my latest spot listing ?
>
> JA8SCD de SWL-JA8SCD/1 (): Op32 0 mi -41 dB in Tokyo with 0w +
> 16:29 136 JA8SCD de JA8SCD/1 Op32 Deep Search 0 mi -41 dB in
> Tokyo with 0w +
>
> 4 Min processing delay showing in these detection's
>
> 16:15 136 LA3EQ de G4WGT Op32 Deep Search 468 mi -43 dB in
> Chorley, IO83qo, with 0.2w + Marconi-T 16:11 136 LA3EQ de G8HUH
> Op32 605 mi -34 dB F:15% in Somerset with 0.2w + Marconi-T
>
> 17:23 136 LA3EQ de GW0EZY Op32 Deep Search 541 mi -42 dB in
> Welshpool IO82ho with 0.2w + Marconi-T 17:23 136 LA3EQ de 2E0ILY
> Op32 Deep Search 505 mi -42 dB in Shropshire IO82qv with 0.2w +
> Marconi-T 17:19 136 LA3EQ de G8HUH Op32 605 mi -33 dB F:1% in
> Somerset with 0.2w + Marconi-T
>
>
> >> Nothing can be 100% , Dynamic detections , will show 1
> >> minuet after conventional OP8 and 4 minuets after a
> >> conventional OP32 decode , this is due to data processing
> >> time and is a fixed off set , Time can be used as a
> >> reality
> >> check
>
>
> There are now two entirely different systems packaged in the
> one application , the original Opera data mode , which is
> capable of presenting the fade pattern and the s/n , the
> other
> relies on call matching and presents s/n only , hence the
> 'odd' 200%F , the dynamic search has been trailed on normal
> 'ham' kit and provides an extension to the existing system .
>
> Before any variations are attempted , off air stats are
> needed , there was close to 100 instances of V155 in use
> since its launch , with only 1 error that would be 1% .
> Occurrences will
> be statistically related to user numbers
>
> Where it is unfortunate 'real' calls are produced , the
> time delay presents a simple , at a glance spot-check , the
> integrity of the original system has not been compromised, at the
> moment the list has a 8 day life , this could be reduced , but
> with the 7 day operating cycle , 8 provides overlap
>
> 73-G,
>
>
>
> --------------------------------------------------
> From: "Mike Dennison" <[email protected]>
> Sent: Wednesday, January 07, 2015 3:37 PM
> To: <[email protected]>
> Subject: Re: LF: New version OPERA >> ''Opera Dynamic'' <<
>
> > Sorry, Graham, but you are wrong. Last night's results did not
> > "give a good working compromise between real and false
> > detections." There were several very plausible but false decodes.
> > These appeared much more frequently than they do on OPDS which in
> > any case gives enough additional information on signal quality to
> > be able to spot most false decodes. It's a nice idea, but the
> > real/false threshold is set much too far towards 'real', making
> > the resultant information next to useless.
> >
> > And what is the "200%" fade figure all about? I am 1000% sure I
> > don't know.
> >
> > Mike, G3XDV
> > ==========
> >
> >> All,
> >>
> >> Other than the odd false spot , that looked to be quite a
> >> good launch for the new addition to Opera
> >>
> >> Nothing can be 100% , Dynamic detections , will show 1
> >> minuet after conventional OP8 and 4 minuets after a
> >> conventional OP32 decode , this is due to data processing
> >> time and is a fixed off set , Time can be used as a
> >> reality
> >> check .
> >>
> >> Dynamic compares Call Signs with activity , to provide the
> >> lowest possible detection levels , to do this , Opera
> >> changes mode to give a second chance detection , based on
> >> s/n , The detection threshold has been set to give a good
> >> working compromise between real and false detections.
> >>
> >> Noting , each individual transmission is processed
> >> independently , Conventional and Dynamic processes are
> >> running in parallel
> >>
> >> 73-Graham
> >> G0NBD
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> From: Graham
> >> Sent: Tuesday, January 06, 2015 8:38 PM
> >> To: [email protected]
> >> Subject: LF: New version OPERA >> ''Opera Dynamic'' <<
> >>
> >>
> >> New version of opera up-loaded
> >>
> >> Opera Dynamic: Introduces the first, adaptive multi mode
> >> Beacon decoding/detection system , seamlessly taking over from
> >> the Opera data decoder and dynamically configuring the low
> >> level detector.
> >>
> >>
> >>
> >> Opera Dynamic is configured for MF OP8 and LF OP32 ,
> >> providing
> >> a dynamically configured low level detection mode ,
> >> complimenting the Opera Data mode ,
> >>
> >>
> >>
> >> Opera Dynamic forms part of the windows distributed
> >> package ,
> >>
> >>
> >>
> >> 73-Graham
> >>
> >> G0NBD
> >>
> >>
> >>
> >>
> >>
> >> UPLOAD
> >>
> >>
> >>
> >> http://rosmodem.wordpress.com/
> >>
> >>
> >>
> >> SUPPORT
> >>
> >>
> >>
> >> https://groups.yahoo.com/neo/groups/O_P_E_R_A_/info
> >>
> >>
> >>
> >> https://www.facebook.com/groups/operadatamode/
> >>
> >>
> >>
> >> -----
> >> No virus found in this message.
> >> Checked by AVG - www.avg.com
> >> Version: 2015.0.5577 / Virus Database: 4257/8885 - Release Date:
> >> 01/07/15
> >>
> >
> >
> >
>
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2015.0.5577 / Virus Database: 4257/8885 - Release Date:
> 01/07/15
>
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2015.0.5577 / Virus Database: 4257/8885 - Release Date:
01/07/15
|
