I think the best approach is just to include every
possible method in GNURadio. This can only make the platform
more versatile. I make use of overlapping a lot because
computation times are a pain. The trade-off in resolution is
acceptable to me because it has limits that I can work around.
This allows me to work with weak signals that would otherwise go
unnoticed without some heavy computation. I typically work at
resolutions of over 1 million to examine narrow-band signals.
The other method is to have a system setup to record the
spectrum to a file, then allow another computer to do the FFT at
different resolutions. SETI do the same thing with the BOINC
platform, only on a far greater scale.
Regards,
Mark McCarron
There are two distinct issues here: