Re: ListCorrelate on lists with different sampling rates
- To: mathgroup at smc.vnet.net
- Subject: [mg87366] Re: ListCorrelate on lists with different sampling rates
- From: dh <dh at metrohm.ch>
- Date: Wed, 9 Apr 2008 05:51:32 -0400 (EDT)
- References: <ftfell$bv7$1@smc.vnet.net>
Hi Curtis, downsampling will lead to aliasing (folding of high frequencies to low frequencies) if your original signal contains high frequencies. Therefore, either you are sure that your signal contains no high frequencies or you should apply a low pass filter before downsampling. Upsampling by an integer factor can be done by introducing zeros and applying a low pass filter. Downsampling by a integer factor can be done by low pass filtering and decimation. Sample rate changes by a rational factor can be done by first upsampling, then downsampling. All this stuff can be found in many books about digital signal processing. hope this helps, Daniel Curtis Osterhoudt wrote: > Hi, all, > > I've been beating my head against this problem (only half an hour at a > time, though, which may well be my mistake) for a little while now, and was > wondering if people had some input. > > I'm basically looking at the cross-correlation (or sliding dot-product) of > two sampled (discrete) signals---one a drive signal and one a received > signal---to pick out various features. I'm doing this using Mathematica's > ListCorrelate, which is working admirably, and I can easily pick out echoes > and so on. > > However, my question concerns what to do when the sampled signals have > different time scales. As I understand it, the ListCorrelate function really > only works (for this application) when the input lists are one-dimensional. > If the sampling rates are different, I can go ahead and (for example) > resample the higher-sample-rate signal (via interpolation) to the same rate > as the lower-sample-rate signal, and then perform the ListCorrelation. It > seems to me, though, that I'm sort of throwing away some information that I > don't need to in the resampling procedure. > > Is that the recommended procedure, or is there something more clever I can > do? > > ========================================================== > Curtis Osterhoudt > cfo at remove_this.lanl.and_this.gov > PGP Key ID: 0x4DCA2A10 > Please avoid sending me Word or PowerPoint attachments > See http://www.gnu.org/philosophy/no-word-attachments.html > ========================================================== >