"Martin 53N 1W" <ml_news@ddnospamddml1dd.co.uk.dd> wrote in message news:dyi%c.42$b%6.38@newsfe1-gui.ntli.net...

Rob Dekker wrote:

"Martin 53N 1W" <ml_news@ddnospamddml1dd.co.uk.dd> wrote in message

<snipsnap>

You are assuming that there must be something out there and that it must
have been observed.

Yes, I assume there are signals out there. If they have been observed already,
they the will have been discarded by current analysis (scoring) methods.

We have 4 million SETI users. I think it is time for another
screensaver..

And the beauty of Science and now with the support of Boinc, that can
more easily be done.

Cool !

State your theory and prove it.

All right...
I still dont know the real problems with signal analysis, but I assume it's RFI.

First of all, I think we should forget about finding beacon signals (narrowband, deliberately
aimed signals with a message for us) . 1240 Mhz is way to low a frequency for interstellar beacons
(too expensive to operate for ET).

I believe that it is much more likely that ET signals out there detectable with Arecibo - seti@home,
will be CW radar signals (semi-long (10-100sec), narrowband radar pulses), and these
will be the first ones we will pick up.
They will NOT repeat often. They will sound very much like RFI, so the sifting
process is the challenging part.

The sifting is currently done by the seti@home team, not by the seti@home users.
And the sifting is done assuming that we are looking for a beacon signal, not a radar
signal. To find the radar signals from ET in the RFI noise, we need a massive undertaking
in sifting. New algorithms, new strategies. Perfect for distributed computing.

Here are some suggestions :

(1) To rule out RFI you need to recognize it : To automate recognition, you need to
have a database of know RFI signals or some pattern analysis.
To obtain a list of RFI patterns, analyze ALL signals (from the past 5 years)
and set up a signature which includes bandwidth, duration, strength etc, but NOT
direction (in the sky). Sort these signals by re-occurrence, so you obtain a list of repeating
signals which is almost certainly RFI. A screensaver program can do this (correlation work).

(2) Set up a 'top-1000' list of promising signals, make them down-loadable (with their
same-sky point counterparts). Us SETI enthausists to sift through them and correlate
manually. Keep the list at 1000 at all times. Let the manual sifters rate the signals, and
drop signals with low score. Add high-score signals which come fresh out of the
single-unit seti@home analysis, after scoring them through the RFI filters (existing filters
and (1)).

(3) Repetition analysis :
Current screensaver basically returns score of a signal in a frequency/time slot.
Since Arecibo now moved over every point in the sky twice (or three times),
every work unit has couple of counter work units observed at a different time,
from the same point in space. More or less the same frequency.
The correlation analysis done for the 200 interesting signals was +-75Hz or so.
With a large computer farm, a much more thorough correlation can be done
on all same-sky-different-time work units.

Need I continue ?

Rob

Meanwhile, there is still the danger of an overenthusiastic ignorant
media shouting wolf a few too many times and then discrediting all the
seti projects.

As long as we keep the top-1000 list open and steady at 1000, there will
not be any media attention.. Just more seti enthausiasts sifting through prospect signals.

It's when we say that one particular signal is 'promising' is when media are interested.

Regards,
Martin