Digital Signal Processing Reference
In-Depth Information
Tutorial 27
Q: A random current signal x(t) has a two-sided, normalized (to unit resistance)
power spectral density (PSD) given by:
Amp
2
=
Hz
0
;
0\
jj
\1Hz
G
x
ð
f
Þ¼
0
:
01
jj
;
1\
jj
\
1
Hz
The signal is passed through a filter whose transfer function is H
ð
s
Þ¼
s
:
1. Determine the magnitude transfer function of the filter.
2. Determine the power transfer function of the filter.
3. Determine the average power (normalized to 1 X) of the output signal y(t).
Solution:
1.
H
ð
x
j j¼
H
ð
s
j j
s
¼
jx
¼
j
j j
¼
x
or
H
ð
f
j j¼
2pf
.
H
ð
f
j j
2
¼
1
2.
4p
2
f
2
:
3. P
¼
R
1
1
G
x
ð
f
Þ
H
ð
f
j j
2
df
¼
2
R
1
0
:
01
f
1
4p
2
f
2
df
¼
0
:
000253 W
=
X
:
1
Tutorial 28
Q: A signal x(t) = 4cos(x
0
t) is transmitted through two additive white Gaussian
noise (AWGN) channels and received as s
1
(t) = x(t)+n
1
(t) and s
2
(t) = x(t)+
n
2
(t), where SNR
1
= 10 dB and SNR
2
= 0 dB. Find:
1. The expected values of s
1
(t) and s
2
(t).
2. Noise power in both cases.
3. Roughly plot the pdf of n
1
(t) and n
2
(t). Which one has more spread around the
mean?
Solution:
1. Since we have AWGN, then
E
n
1
ð
t
f ¼E
n
2
ð
t
f ¼
0 [at any time t].
Note that
E
n
ð
t
fg¼
statistical mean of n(t) = m = $
-
?
np(n)dn, while the
time mean is m
t
¼
T
R
0
n
ð
t
Þ
dt
;
T being the total time. If t is discrete, then
m
t
¼
N
P
N
1
n
ð
k
Þ:
If n(t)isergodic, then m = m
t
.
k
¼
0
Search WWH ::
Custom Search