Digital Signal Processing Reference
In-Depth Information
2
=1
m=0,
s
4
2
0
2
-
-
4
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
n
0.4
0.3
0.2
0.1
0
-
4
-
3
-
2
-
1
0
cells
1
2
3
4
Fig. 4.12
Normal random variable and its estimated PDF
Consider that the variable
X
(generated using the MATLAB file
randn.m
)isa
normal random variable with a mean value of 0 and a variance of 1. From (
4.94
), it
follows:
s
Y
¼ a
2
m
Y
¼ b
;
:
(4.95)
From (
4.93
) and (
4.95
), we get a linear transformation that has to be applied to
the normal variable
X
in order to obtain a normal variable with the desired mean
value and variance (
m
Y
and
s
Y
, respectively) of length
N
Y ¼ s
Y
X þ m
Y
:
(4.96)
Using (
4.93
) the MATLAB file
r ¼ fnorm
(
N,VA,ME
) generates the desired
normal variable
r
with a length of
N
and the given variance
VA
and mean value
ME
.
Example 4.4.2
Generate normal variables with the variance
s
2
¼
4 and four
different mean values:
6,
2, 3, 8. Estimate the corresponding PDFs.
Solution
The generated variables are shown in Fig.
4.13a
, and the corresponding
estimated densities are shown in Fig.
4.13b
. Note that the estimated densities have
the same shape and are only shifted along the
x
-axis. Similarly, according to the “3
s
rule” all signals have the same range around their mean values.
Search WWH ::
Custom Search