In article <5uhsd.38700$6q2.26179@newssvr14.news.prodigy.com>,
Rob Dekker <rob@verific.com> wrote:
[

In article <264c4ea4.0412011231.a0b529b@posting.google.com>,
someone with an inappropriate user name wrote:

]

A simple question, if there was an intelligent signal coming from a
distant planet how on earth could we get it out from all the noise?

By using a large antenna, to collect lots of signal; by using low noise 
receivers, to get rid of avoidable noise (still the main source of noise);
by using the smallest bandwidth that will propagate reliably, reduce the
noise bandwidth; and by averaging over an extended observation period,
to reduce the uncertainty in the noise level.  If you look at the FAQ, 
you should discover that this results in adequate sensitivity to detect
signals that we are easily capable of generating, at interstellar
distances (note that the FAQ doesn't give good examples of large antenna,
long averaging time measurements, so doesn't give the ultimate sensitivity).

That's the job of RFI filtering. It is noisy out there, so there are some

RFI filtering is a factor, but the first consideration is natural noise.

Beamforming, by phase-synchronizing multiple receivers will greatly increase
resolution of a receiver, but not its sensitivity. Since SETI searches in

It doesn't improve sensitivity for a given total aperture, but it does
reduce civil engineering costs and is becoming possible because of falling
electronics costs, hence the Allen and SKA systems.

I believe that an 8db SNR ratio is required for seti@home. Much lower and
there will be way too many false-negatives.

The SETI@Home spike detection threshold is 13.4dB (22 linear).
This is relative to the mean noise, including quantisation and other
receiver noise averaged over the whole work unit bandwidth (~10kHz) but
measured in the detector noise bandwidth (variable but down to ~0.07Hz).
This corresponds to -37.8dB in the work unit bandwidth (approximately AM
broadcast bandwidth) and -61.8dB in the full receiver bandwidth (treating
quantisation noise as though it were receiver noise).  I don't know the
exact figures for the other detection modes (these effectively involve
averaging, but in a non-trivial way; the pulse detection may achieve
negative dB SNR's in the detection bandwidth).

Detection SNR values are determined by the allowable false positive rate
on a single observation.  Observations with real time followup can
use lower SNRs, because they can extend the observation time immediately
to get better effective SNRs by averaging.  False positive rates are 
inverse exponentially related to the detection threshold and no threshold
will bring them to zero (all real life measurements are subject to
a possiblity of error).

(SNR's are really (signal+noise to noise ratios), but I've computed
some figures as though they were really signal to noise).

seti@home still detects the Pioneer 10 signal.... It works allright !

S@H does not cover any frequency at which we are allowed to intentionally
transmit, so cannot detect any man made space probe.  Deep space probes
at half a light day have been used as test signals for project Phoenix.
In any case, as the hardware is essentially the same as the deep space
communications network's, and the signal processing is well understood,
success can be predicted.