what can I do/use after cross correlation ?

[leny]

Hi  !

I want to use in Labview PRBS (or MLS) to characterize my system.                          MLS signal------->|system|-----> out response signal (mls+noise)

so I try to demodulate my output signal using a cross correlation between my output and my input but I do not understand the result. What can I do after the cross correlation to be able to see a clean  time domain signal.

finally I would like to use several mls input signal.  

Thanks 

[CoastalMaineBird]

The purpose of cross-correlation is usually to define the time-lag between two signals. Basically if the SYSTEM introduces a TIME LAG of 0.1 sec, then the cross correlation will show a peak at 0.1, and a lesser value at other times.

If your output is really just (input + noise) then cross-correlation should always show a peak at T = 0.0. 

You say you're trying to "demodulate" a signal.  What sort of modulation is there? 

[leny]

Thank you for your response.

actually I use a MLS signal for my input, and I try to get the system response from the output, but I don't know how can I recover my signal !?

I get a peak at 0.1, then my response is 0.1 ms too late ?

 

Message Edité par leny le 03-22-2009 11:02 AM

[CoastalMaineBird]

The only thing you can tell from that graph is that your output is about 0.099 sec (whatever the X value is at the Y-peak) behind the input.

. 

That's assuming you have scaled the X-axis correctly.

That's the point where, if you slid the output back in time by that amount, they line up the best.

The other points on the graph are just other time-shift trials, and they don't match so well. 

That's the end of the story, using that graph.

If you want to recover the signal, you have to use other means.  Do you know the nature of the noise?

Do you know the nature of the signal?

I assume MLS = Max Length Sequence = pseudo-random signal. 

[leny]

showing below my output when I send a DC signal. in fact it is the blood pressure in a finger.

actually the interesting singnal is the "noise" here.

Then I want to recover this signal by the mls way .

Thanks for everything ! 

[CoastalMaineBird]

I'm confused - do you want to extract the BP signal, or the noise that's on top of it?

[leny]

hooo sorry for my English !

I want to extract the BP signal.

I called the BP signal "noise", because it's the perturbation in my system. 

I need to recover the BP signal like in the picture but using mls signal intput.

[CoastalMaineBird]

OK, you'll have to spend some words explaining what you need, and what you've tried.

What does the MLS have to do with anything?

What does cross-correlation have to do with anything.

Seems to me you pick out the maxima and minima of the signal and those are your BP points. 

What am I missing? 

[leny]

yes of course I can do that, or I can use a PB filter, but this method works only with one input signal.

I want to use something like CDMA to be able to use several input signals and after to be able to separate each signal. 

is it clear ?!

[leny]

up please !?