Forum: GNU Radio USRP2 - Many Narrowband

Announcement (2017-05-07): www.ruby-forum.com is now read-only since I unfortunately do not have the time to support and maintain the forum any more. Please see rubyonrails.org/community and ruby-lang.org/en/community for other Rails- und Ruby-related community platforms.
Ryan P. (Guest)
on 2009-01-05 22:55
(Received via mailing list)
All:

I'm currently trying to extract a relatively small number of narrowband
channels out of about 10mhz.  With USRP1, I can get two channels easily
by
using both sets of DDCs.  In some instances, I can successfully use a
filterbank, however at 12.5khz channels, I quickly run out of horsepower
if
I don't keep the decimation high enough, which does not always let me
cover
the swath of RF I need.

Are there any other possibilities with the new USRP2 hardware to
implement
more DDCs in the firmware to do more of this work on the USRP side?  Any
other thoughts in general?
Matt E. (Guest)
on 2009-01-06 00:00
(Received via mailing list)
Ryan P. wrote:
> implement more DDCs in the firmware to do more of this work on the
> USRP side?  Any other thoughts in general?

You could implement more DDCs in the USRP2, but it makes more sense to
use filterbanks in software.  It may take something more complicated
than a straight forward filterbank across the whole bandwidth.  You
might want to do it in a few steps.  For example, do a filterbank to get
you 16 channels of ~500 kHz apiece, and then do smaller filterbanks on
only those 500kHz bands have signals of interest in them.

Matt
Eric B. (Guest)
on 2009-01-06 02:49
(Received via mailing list)
On Mon, Jan 05, 2009 at 01:44:53PM -0800, Matt E. wrote:
>> other thoughts in general?
>
> You could implement more DDCs in the USRP2, but it makes more sense to
> use filterbanks in software.  It may take something more complicated
> than a straight forward filterbank across the whole bandwidth.  You
> might want to do it in a few steps.  For example, do a filterbank to get
> you 16 channels of ~500 kHz apiece, and then do smaller filterbanks on
> only those 500kHz bands have signals of interest in them.
>
> Matt

The filterbank in the tree was done quick and dirty.
There's lots of room for improvement. (e.g.,  coalescing many of the
intermediate blocks.)

Eric
This topic is locked and can not be replied to.