NaN in deconvolution

Hi guys,

I am stuck with a deconvolution part in my program. I have a filter o/p and the filtered o/p, but they both have zeros and numbers too. When I am trying to deconvolute, it returns "NaN" values. Is there any solution to it? I m stuck here for days. Please help me.

Thanks and regards,

Raj.

Answer 1

[Amith Kamath]

Since your data has zeros in them, deconvolution may return NaNs because you are dividing by zeros. What you could do is either of these:

try by adding 'eps' to all the zero values, so that it results in finite values, although not exactly accurate,

or, remove the NaN values by using ~isfinite or ~isnan.

[raj0689]

thanks Amith,

I will try that out and will let u know.. :)

[raj0689]

but Amith, how can i do it.. Since I haven't used eps, i am quite confused.

even_val=eps.*deconv(fil_sig_demod_even,fil_op);

Is above line the right format ?

Thanks again..

[Walter Roberson]

even_val=deconv(eps+fil_sig_demod_even,eps+fil_op);

[raj0689]

thanks Walter :) its working , but still i am not able to reverse my modulation.. getting lots of errors in getting +1s and -1s.. :(