discuss-gnuradio
[Top][All Lists]
Advanced

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

Re: [Discuss-gnuradio] Decimation


From: Prateek Dayal
Subject: Re: [Discuss-gnuradio] Decimation
Date: Sat, 20 Nov 2004 11:21:53 +0000

On Fri, 19 Nov 2004 15:07:54 -0700, Dan Gisselquist
<address@hidden> wrote:
> Any comments?  Yeah.
> 
> I had a similar conversation with my instructor two years ago.  He
> basically said that wavelets were the ideal way of breaking the world
> into time and frequency.  Be aware, however, wavelets do not divide the
> world into equal time and frequency "bins".  They are biased towards
> high frequency resolution for low frequencies, while at the same time
> achieving low time resolution for the same low frequencies.  Towards
> the higher frequency end of the scale, wavelets are often used to
> provide very precise time resolution at the expense of frequency
> resolution.  Applications for wavelets are numerous.
> 
> However (my two cents), wavelets are not as appropriate for examining
> the electromagnetic spectrum.  In particular, the assumption that
> higher frequencies need more time resolution is rendered invalid by a
> frequency agile communications system (something where the carrier
> frequency can be placed anywhere).
> 
> IMHO, decimation always requires an antialiasing low-pass filter
> coupled with the actual decimator.  Two primary implementations exist.
> The first filters and then decimates, the second couples the two
> operations together so that the only filter outputs computed are the
> valid outputs of the decimator.  The differences between these two
> approaches, in terms of computational cost, can be very dramatic.

Which approach does GNU Radio take ???


> 
> While you could use the wavelet transform to perform the same
> operation, it would only make sense if you wanted multiple outputs as a
> result.  For example, the wavelet transform might make sense if you
> wanted to separate the input into multiple different streams at
> (likely) different rates.
> 
> Other approaches, such as those using a DFT filter banks, make a lot
> more sense when you want equally spaced frequency bins for reasons
> discussed above.  If you'd like, I can provide some really good
> references for how to do this well ...
> 
> Just my two cents,
> 
> Dan
> 
> 
> 
> 
> On 11/19/2004 10:54:21 AM, Prateek Dayal wrote:
> > Hi ,
> >
> > I saw that in GNU Radio, decimation is carried out after using an
> > antialiasing low pass filter. In a discussion a professor at my
> > institute suggested that we can use wavelets for decimation and
> > interpolation ....
> >
> > any comments ???
> >
> > --
> > Prateek Dayal
> > B.Tech 4th Year
> > ECE
> > IIT Guwahati
> >
> > www.geocities.com/pmd_iitgw
> >
> > 
> > _______________________________________________
> > Discuss-gnuradio mailing list
> > address@hidden
> > http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
> >
> 


-- 
Prateek Dayal
B.Tech 4th Year
ECE
IIT Guwahati

www.geocities.com/pmd_iitgw




reply via email to

[Prev in Thread] Current Thread [Next in Thread]