One thing you might look at is the "Rosenfell" algorithm used in some HP
spectrum analyzers (like the 8566). As I recall, they alternately take
the maximum and minimum values in the bins -- ie, if you're collapsing
three bins into one and the original values are
010 050 010|010 050 010|120 130 120|120 130 120|010 050 010|010 050 010
You'd end up with
050 010 130 120 050 010
The idea is that this looks more like the "grass" that you would see on
a non-digital analyzer and accentuates the difference between noise and
stable signal, where averaging the bins would lose that distinction --
it has the effect of removing indicators that you're looking at noise.
Not sure if this might be applicable to the application, and not sure I
got the details just right from memory, but I think searching
"Rosenfell" would find more details.
John
_______________________________________________
Discuss-gnuradio mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
