Page 66 - Algorithm Collections for Digital Signal Processing Applications using MATLAB

54 Chapter 2

respectively. Thus the ICA problem can be represented in the form of matrix
as described below.

y 11 y 21 x 11 x 21
y 12 y 22 x 12 x 22

y 13 y 23 x 13 x 23
y 14 y 24 x 14 x 24 a 11 a 21
y 15 y 25 x 15 x 25

y 16 y 26 = x 16 x 26 a 12 a 22
.
.
.

y 1n y 2n x 1n x 2n

T T T
[Y] = [X] [A]

Given the matrix [Y], obtaining the matrices [X] and [A] such that the
T
columns of the matrix [X] are independent to each other is the ICA problem.
Consider three independent signals and the corresponding mixed signals
along with the kurtosis values are displayed below.