Thread Subject: Noise cancelation

Hi to All,
I am a new member of matlab, and have some questions about noise cancelation. I have 2 sequences, which names are x(n) and y(n).
x(n)= e(n) + n(n) ,vector length is N=730
y(n)= e'(n) + n'(n) ,vector length is N=730
like this
x(n)=[

n(n) and n'(n) are the thermal noise and these are uncorelated, but e(n) and e'(n) are the errors and these are high corelated with each other. So, I want to separate the noise and error..How can I do this? Please give me some advice and instractions for this problem.
Thank you
Sincerely,

On 19 Nov, 22:25, "sdfr ewcs" <axiseyh...@gmail.com> wrote:
> Hi to All,
> I am a new member of matlab, and have some questions about noise cancelation. I have 2 sequences, which names are x(n) and y(n).
> x(n)= e(n) + n(n) ,vector length is N=730
> y(n)= e'(n) + n'(n) ,vector length is N=730
> like this
> x(n)=[
>
> n(n) and n'(n) are the thermal noise and these are uncorelated, but e(n) and e'(n) are the errors and these are high corelated with each other. So, I want to separate the noise and error..How can I do this?

You can't.

You say that e and e' are 'highly' correlated. This means
that they are not prefectly correlated, but can be separated
into a perfectly correlated part c and an uncorrelated part
u, as

e = c + u
e' = c' + u'

where

corr{e,e'} = corr{c+u,c'+u'} = corr{c,c'}

since, by definition, the correlation terms involving u
equal 0.

A similar argument works for the n and n' signals.

The best you can do is to match up the parts of the two
sequences that are correlated with each other, using
some sort of cross correlation. This might be useful
to you, but will *not* have any connection to your
classification of signal components as 'error' or 'noise'.

Rune