[Eeglablist] Signal Decomposition

[Arnaud Delorme]

Dear Santanu,

you should try the SOBI ICA algorithms which uses correlations across
times in performing
the signal separation (compiled version sobi or acsobiro Matlab
functions of the ICALAB
Matlab package; EEGLAB will detect these algorithms if they are in the
Matlab path and
you will be able to use them when calling the "Run ICA" menu).

Best,

Arno

Santanu Das wrote:

>HI:
>  I am Santanu Das and a graduate student in ASU. Presently I am trying to
>decompose an overlapping signal to it's components.
>Assume the model
>x(t) = sum_{i=1 to L} A_i *sin( omega t - theta_i)
>y(t)= x(t- t_1) + x(t- t_2) + x(t- t_3)
>
>So y(t) is the sum of the time shifted overlapping signals. The time shift
>is generally in micro range. In my case, y(t) is known and would like to
>extract the  time shifted components of x(t), which means in this case those
>3 components.
>