"Patrick" <[email protected]> wrote in message
news:[email protected] m...
> > The threshold value is normally determined by your noise PDF and by
> > selecting some acceptable "Probability of False Alarm". Most of the
> > analytical theory here is based around Gaussian noise models.
> >
> > Cheers,
> > David
>
> Hello David,
>
> How can I compute the PDF of the noise ?
>
> I've tried this to integrate the output of the matched filter and
> compute the maximal value
>
> Then how to compute the PDF value of the noise ?

Look at the output of the filter with noise and no signal?
Set the threshold so that the false alarm rate is acceptable?
(Because even after you have the PDF, you have to select a point on the PDF
that's acceptable for your application).
To what extent does the noise level vary > suggesting a variable threshold?

Fred