"P. Backus" <pbackus@seti.org> wrote in message

news:4e30f668.0312102022.3848d69e@posting.google.com...

"Alfred A. Aburto Jr." <aaburto@pacbell.net> wrote in message

news:<8RIBb.69329$7r.38040@newssvr25.news.prodigy.com>...

"David Woolley" <david@djwhome.demon.co.uk> wrote in message

 news:T1070828218@djwhome.demon.co.uk...

In article <0Mtzb.32209$_g4.16482@newssvr29.news.prodigy.com>,
Alfred A. Aburto Jr. <aburto@sbcglobal.net> wrote:

Another point to be raised is that the SNR need not be so "high"
as shown in the FAQ. In fact, an SNR of 3 might be ok in many
cases. I note that Kraus in the Radio Astronomy reference, used

3:1 will produce totally unacceptable false positive rates.

I'm not sure it is "unacceptable" because I have seen FFT time histories
of the Project Phoenix's detection of the Pioneer Probe signal and it
was clearly detectable at low SNR ... there are also examples of the
the detection of low SNR (averaged) signals where during one scan
you could not distinguish the signal (target) but for multiple scans it
stands out very clearly ...

I don't have the logs from our old observing system here, but I think
Pioneer 10 usually registered as SNR ~ 6 to 10 in a single sample (.7
Hz x 1.4 sec) and when integrated for a few minutes was a strong
detection.

I looked for the example I was thinking of, but I did not find it. The
signal was fading in and out alot. You'd see a strong signal on one
scan then it would be gone for a few other scans then back again. But
when you looked at a number of sequential scans the track of the
signal over time and frequency was clearly seen. It just seemed that
there was a very clear detection with low overall SNR.

The signal was very much like that in Figure 11-7 of the Project
Cyclops book (page 130), except there were fewer bright points along
the signal track (pulsar signal in this case).

I think that the SNR threshold is something an individual user/system
should adjust for their own purposes.

The SNR and ultimately the PFA should be set by the overall goal of
the project.  In the case of Phoenix, where we do immediate follow-up
of candidate signals, we need to balance the use of the telescope for
follow-ups vs. covering new stars and frequencies. Our threshold for
CW signals is 8.5 times the mean noise.  For SERENDIP/SETI@Home, you
need a threshold that will generate some number of candidates that can
be reasonably checked in a day or two of dedicated observing about
once a year.  You can't afford to generate lots of false candidate
signals that will use up the limited dedicated telescope time needed
for confirmation observations.

In the sci.astro.seti FAQ, I think originally I went with Kraus's example
(SNR = 3) because for a small radio telescope one needed as much
sensitivity as possible and I sacrificed the PFA ... I'll need to correct
this
situation in the FAQ but yet not hinder flexibility ... (and provide more
references) ...

Thank you for responding!
Al