Hi, Claudio.
On Fri, 11 Oct 2002, Claudio Girardi wrote:
I have just finished a new release of glfer, a Linux application combining a
spectrogram viewer and a QRSS/DFCW keyer.
As You are working in computer signal processing it seems result of some
my reseach (not published) should be interesting for You.
Main idea is as folow. Conventional algorithms of signal processing (say
FFT) are optimal if noise is Gauss distributed. But is spheric noise
Gauss distributed realy? My investigation show that this is wrong.
Really spheric noise (QRN) is distributed not by Gauss function
exp(-x*x). Much better aproximation is Lorentz function 1/( 1 + x*x).
But conventional signal processing is not optimal in such a noise!
As it was shown by me optimal signal processing in Lorenz noise require
to solve very complex nonlinear integral equation. But if one
solve the equation by iteration procedure he get as folows. First
iteration yeld conventional linear processing. But next iteration yelds
that optimal is not linear filtration of signal but nonlinear instant
conversion by a function x/(1 + x*x) and THAN conventional linear
processing. Incoming signal level should be controled to give mean
noise level equal to 1.
So it is very simple to realize! The only addition is input convertion
of the incoming signal
S1(t) = S(t)/[A + S(t)*S(t)]
where S(t) if signal from RX. Parameter A shold be controlled to get
best reciption. Than signal S1(t) is processed by conventional procedure
(say FFT).
It is very interesting to try such an algorithm.
73 de RA9MB/Alex
http://www.qsl.net/ra9mb
| 