Thread Subject: How to separate a signal?

The observed signal

y(t) = x_1(t) + x_2(t) * exp(j A_1 (x^2)) + x_3(t) * exp(j A_2
(x^2)).

* stands for the convolution operation.

How can I separate the signal x_1(t) without influence of x_2(t) and
x_3(t)?

Thxs.

Tristan Imp <xin.zhang03@gmail.com> wrote in message
<aa11b5a4-4507-40dc-80fb-
686a6e32b1ba@l6g2000prm.googlegroups.com>...
> The observed signal
>
> y(t) = x_1(t) + x_2(t) * exp(j A_1 (x^2)) + x_3(t) *
exp(j A_2
> (x^2)).
>
> * stands for the convolution operation.
>
> How can I separate the signal x_1(t) without influence of
x_2(t) and
> x_3(t)?
>
> Thxs.
>

You have 3 sources x1, x2, x3 and only one mixed signal y?
SO what is the quantity 'x'? It is also one source or a
parameter of the mixing equation.

y(t) = x_1(t) + x_2(t) * exp(j A_1 (x^2)) + x_3(t) *
exp(j A_2 (x^2)).

I think it is not easy if you don't have any constraints
(such as sparsity, nonnegactive or independent
sources, ...) on the sources.

We have one toolbox called as ICALAB which allow you to
recover the sources from the linear mixed signals. I think
it maybe helpful for you.

Download ICALAB at this website

http://www.bsp.brain.riken.jp/ICALAB/ICALABSignalProc/

Anh Huy Phan
RIKEN - BSI