Re: [Discuss-gnuradio] (no subject)

[Tom Rondeau]

On Mon, Apr 1, 2013 at 7:26 PM, vamshi krishna dodla
<address@hidden> wrote:
> Sir
>
> Thanks for your efforts and your reply, yeah i understand your concern
> regarding the information lost (also effects of multipath  and adjacent
> channel interference, which is good to some extent) by only taking band edge
> filter's output for purpose of timing recovery, but how about taking the
> sample rate as 2 sps for entire (timing and carrier recovery) loops(cascaded
> structure) and upsampling to 4 sps only at band edge input part, but rest of
> receiver operating at 2 sps only. (although the detector viz costas loop
> takes 1sps for detection purpose which should not be a problem).
>
> Thanks & Regards

So while I think that this discussion is really interesting, I find
email to be a terrible way to go about communicating these ideas. In
fact, we just had this discussion recently on the mailing list:

http://lists.gnu.org/archive/html/discuss-gnuradio/2013-03/msg00510.html


Would you be willing to set up a wiki page to convey your ideas here
better? I've started a dummy page for people to start discussions
here:

http://gnuradio.org/redmine/projects/gnuradio/wiki/SignalProcessing

(Also, I think you're new to the community, so you'll probably have to
create an account on gnuradio.org and then send either myself or
Martin Braun an email to add you to the GNU Radio project so you can
then edit the wiki.)

Thanks!
Tom



> On Mon, Apr 1, 2013 at 3:59 PM, Tom Rondeau <address@hidden> wrote:
>>
>> On Mon, Apr 1, 2013 at 2:46 PM, vamshi krishna dodla
>> <address@hidden> wrote:
>> > I have some questions regarding the coastas loop implemented, does it
>> > also
>> > consider the snr estimate too? like a Maximum Likelihood loop(tanh*
>> > snr).
>>
>> Hi Vamshi,
>>
>> No, the Costas loop does not account for SNR. This is among a series
>> of improvements that I'd like to see happen. We have an SNR estimator
>> block that sends tags and one that we can easily outfit to send
>> messages. We would then use this to send a message to blocks like this
>> so they can adjust their behavior based on current SNR estimates. I
>> wouldn't want to recalculate the SNR every time its needed.
>>
>> > Also can you suggest me any interpolator for timing recovery and also
>> > polyphase bandedge filters implementation. This would be really helpful
>> > to
>> > me.As there is very little material available online, also as you are
>> > experts in these fields your guidance will be very useful
>>
>> Take a look at gr-digital/lib/fll_band_edge_cc for a frequency locked
>> loop.
>>
>> There's a lot of debate over the merits of using just the band edge
>> information to do timing recovery, since you'd be throwing away
>> information by not taking the information in the entire symbol (also
>> effects of multipath and adjacent channel interference).
>> Implementations are very welcome, though. One thing we can do well in
>> GNU Radio is pit algorithms against each other under real signal
>> conditions to see how they work.
>>
>> For timing recovery, look at gr-digital/lib/pfb_clock_sync_ccf.
>>
>> Tom
>
>