[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Linrad

Hi Alex and all,

> Well, linrad knows what is pulse shape.  If pulse shape is determined
> by prefiltering we can input test signal before recieving real signal and
> linrad can put pulse function into memory. 
In other words, a pulse that has much larger bandwidth than the bandwidth
we bring into the computer will have it's shape determined solely
by the frequency response of our hardware.

> But how linrad can know  QRN pulse realy occures in the small "time window"? 
Oooh! Linrad has removed all strong signals first (selective limiter)

Since all pulses that have very large bandwidth will have nearly all energy in a 
single data point (or in a pair of data points) the pulse amplitude of even a rather weak
pulse will become very high compared to the noise floor. Any single data point that
is much larger than the average means we have a noise pulse at this particular time.

> Or at least how it can know number of pulses in a large time window? 
By looking at each and every sample and deciding that anything above the threshold
(that the user may adjust) is a noise pulse.

> If this is known though it is posible to determine pulses parameters by some minimisation
> procedure and substuct pulses then. 

But it will fail if the pulse is not wideband - a distant thunderstorm may produce
pulses that are filtered through a filter with rapidly varying phase and amplitude vs
frequency (multipath propagation) Sometimes we may resolve the different propagation
paths as different pulses, but generally the number of pulses is too large and
the pulses are too close in time if it is nighttime on a HF band (I think)

There are many other interesting problems with the pulse removal procedures and I am
sure much better algorithms will be developed in the future. Already the Linrad
procedure works well in many cases and it allows reception of signals you can not possibly
hear with any conventional radio:)


Leif  /  SM5BSZ