Beruflich Dokumente
Kultur Dokumente
Communication Systems
Spring 2013-14
Chapter 1
Communications
Is Signal = Message?
Digital
Examples
Analog Signals
Digital Signals
Binary Signals
1
0 0
1
0
t
Feb 03, 2014
Transmitter
Channel
Modulation
Coding
Attenuation
Noise
Distortion
Interference
Receiver
Detection (Demodulation+Decoding)
Filtering (Equalization)
b1b2 ...
Source
Encoder
m (t )
Transmitter
b1b2 ...
m (t )
s(t )
s (t )
Channel
Receiver
Source
Decoder
n(t)
Analog or Digital
Modulator
Transmitter
s (t )
h(t)
s(t ) n(t )
Channel
Analog or Digital
Demodulator
b1b2 ...
m ( t )
Receiver
10
Fundamental Limitations
Noise
11
Block Diagram
n(t)
m(t) (Modulator)
Analog
or Digital
Transmitter
s(t)
h(t)
Demodulator
Channel
m
(t)
Receiver
12
Performance Criterion
13
14
Example: PSTN
Characteristics
Circuit-switched network
Designed for voice communications (analog???)
Faxes and modems use PSTN for transmission of
digital data in analog form
15
Example:PSTN
Long distance line
Local
exchange
Local line
International
exchange
International line
Long distance
exchange
Long distance line
Local
exchange
16
Example: Cellular
Islamabad
PSTN
MTSO
MTSO
17
Example: Cellular
18
AM
Amplitude modulation
535-1605kHz
10kHz channels
FM
Frequency modulation
88-108MHz
Channels centered at 200kHz intervals starting at
88.1MHz
19
Various standards
IEEE 802.11a/b/g popular
IEEE 802.11b
IEEE 802.11a
20
21
Sensor Networks
Mesh Networks
22
n(t)
m(t) (Modulator)
Analog
or Digital
Transmitter
s(t)
h(t)
Demodulator
Channel
m
(t)
Receiver
23
Transmitter
Channel
Modulation
Coding
Attenuation
Noise
Distortion
Interference
Receiver
Detection (Demodulation+Decoding)
Filtering (Equalization)
24
Transmitter
Coding
25
Channel
Noise
Distortion
Inter-symbol interference
Constant attenuation
Variable attenuation
Interference
Crosstalk
EE-380: Communication Systems
26
Receiver
27
Performance Criterion
28
29
w0 = 2f0
Phasor representation
v(t) = A cos(w0t + ) = <[A ej(w0t+)]
Rotating phasors
Frequency plots
Amplitude
Phase
f0
A
w 0t +
Amplitude
Phase
A
f0
f0
Feb 03, 2014
f
30
A=2
f0
f0
f0
Amplitude
f0
Phase
31
Periodic Signals
T0
Feb 03, 2014
32
Average signal
hx(t)i = Tlim
!1
x(t)dt
T=2
Average power
hjx(t)j2i = Tlim
!1
1
T
T=2
R
1
T
T=2
R
jx(t)j2dt
T=2
33
34
Fourier Series
v(t) =
1
P
cnej2nf 0t
cn =
c0 =
1
T0
1
T0
f0 = T10
T0
v(t)dt = hv(t)i
T0
(fundamental frequency)
35
Sinc Function
1
sinc(x) = sinxx
0.8
0.6
0.4
0.2
-0.2
-0.4
-5
1
T
T=2
R
ej2ftdt =
T=2
Feb 03, 2014
1
j2fT
-4
-3
-2
-1
sin fT
(ejfT ejfT) = fT = sinc(fT)
36
= T10
T0
=2
R
Aej2nf 0tdt
A
n
sinc(
)
4
4
A
sinc(nf 0)
T0
for =
T0
4
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-3
-2
-1
nf0 =
Feb 03, 2014
T0
0.9
A
jnf 0
jnf 0
= j2nf
(e
e
)
0T 0
sin nf 0
nf 0
=2
A
T0
T0
cn =
1
T0
v(t)
n
4
37
1
P
cnej2nf 0t = c0 +
= c0 +
1
P
jcnje
n=1
= c0 + 2
= a0 + 2
1
P
n=1
j(2nf 0t+6 c n)
1
P
6 c n)
jc njej(2nf 0t+
n=1
1
P
n=1
1
P
n=1
38
Parsevals theorem
T=2
R
T=2
T=2
R
T=2
1
P
jx(t)j2dt
x(t)x(t)dt
c nc n
1
P
jc nj 2
39
Fourier Transform
1
v(t)
A
T0
T0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-3
-2
-1
nf0 =
n
4
40
Fourier Transform
R1
1
R1
x(t)e j2ftdt
X(f)e j2ftdt
x(t) $ X(f)
Feb 03, 2014
41
x(t=b)
ac
x(t
k )
a=b
x( t)
a+c
x(d t)
da
ab
x(t c)
x(t + c)
A
x(tb)
+ ak
t
42
x(t) = rect(t)
12
X(f) = sinc(f)
x(t) =
Arect(t)
X(f) = Asinc(f)
1
2
43
Properties of FT
44
Properties of FT
Modulation
x(t) cos(2f ct + ) $
e j
X(f
2
f c) +
e j
X(f
2
+ f c)
45
Properties of FT
Differentiation
d
x(t)
dt
$ j2fX(f)
dn
x(t)
dt n
$ (j2f)nX(f)
Integration
R
R1
1
1
x()d $ j2f
X(f)
x()d = X(0)
R1
X(f)df = x(0)
46
A generalized function
R1
x(t)(t)dt = x(0)
(defined by sifting)
R1
1
R1
1
R1
(t)dt = 1
x(t)(t t 1)dt = x(t 1)
x(t t 2)(t t 1)dt = x(t 1 t 2)
x(t) (t) =
Feb 03, 2014
R1
1
x()(t )d = x(t)
47
FT of 1 and exponentials
R1
ej2f0t $ (f f0)
1 $ (f)
and
(t) $ 1
f
)
+
sinc(f + f c)
c
2
2
48
Other Functions
Signum Function:
n
t>0
sgn(t) = +1
1 t<0
1
$ jf
1
t
49
FT of Periodic Signals
xp(t) =
1
P
cne
j2nf 0t
cn =
n
Let x(t) =
cn = f0
x p(t)
0
R1
jtj<T 0=2
else
1
T0
T0
1
P
) xp(t) =
f0 = T10
x(t mT 0)
m=1
1
P
xp(t) = f 0
X(nf 0)ej2nf 0t
n=1
1
P
xp(t) $ f 0
X(nf 0)(f nf 0)
n=1
Feb 03, 2014
50
FT of Periodic Signals
1
P
(t mT 0)
m=1n
0=2
Let x(t) = x p0(t) jtj<T
else
xp(t) =
Xp(f) = f 0
(f nf 0)
n=1
1
1
P
P
(t mT 0) $ f 0
(f nf 0)
m=1
Feb 03, 2014
T0
f0
f
f0 = T10
n=1
EE-380: Communication Systems
51
LTI Systems
Linearity
x(t)
If
x(t) =
then
y(t) =
P
n
P
cnxn(t)
cnG[xn(t)]
y(t)
y(t) = G[x(t)]
Time invariance
G[x(t )] = y(t )
cnG[xn(t )]
n
Feb 03, 2014
52
=
=
R1
1
R1
x()G[(t )]d
x()h(t )d
= x(t) h(t)
Feb 03, 2014
53
=
=
R1
1
R1
h()x(t )d
h()e j2f c(t)d
= e j2f ct
R1
h()e j2f cd
= ej2f ctH(f c)
Feb 03, 2014
Exponentials are
eigenfunctions of
LTI systems!
LTI Systems can
not generate new
frequencies!
54
Hilbert Transformer
j
f>0
HQ(f) = jsgn(f) =
+j
f<0
1
xQ(t) = t
1
x
(t) = x(t) hQ(t) = x(t) t
x
(t) $ jsgn(f)X(f)
Feb 03, 2014
55
56
Bandpass Signals
How does it look like in frequency domain?
Vbp(f)
fc W
V bp(f);
fc
fc + W
jfj > f c W;
fc W
fc
fc + W
jfj < f c + W
57
Bandpass Signals
vbp(t) = A(t) cos(2f ct + )
vbp(t) = A(t) cos 2f ct cos A(t) sin 2f ct sin
vbp(t) = vi(t) cos 2f ct vq(t) sin 2f ct
vbp(t) = vi(t) cos 2f ct + vq(t) cos(2f ct + =2)
Vbp(f) = 0:5[Vi(f fc) + Vi(f + fc)] + 0:5j[Vq(f fc) Vq(f + fc)]
V+(f) = Vbp(f)u(f) = 0:5Vi(f fc) + 0:5jVq(f fc)
Vlp(f) V+(f + fc) = 0:5Vi(f) + 0:5jVq(f)
Feb 03, 2014
58
Lowpass-Bandpass
vbp(t) = <fA(t)ej(2fct+)g
vbp(t) = 2<f12A(t)ejej2f ctg
vbp(t) = 2<fvlp(t)ej2fctg
Vbp(f) = Vlp(f fc) + Vlp( f fc)
Why this all lowpass-bandpass stuff needed?
Y lp(f) = H lp(f)Xlp(f)
Y bp(f) = H bp(f)Xbp(f)
Feb 03, 2014
59
Amplitude Modulation
Double-sideband (DSB)
60
Amplitude Modulation
Double-sideband+carrier (DSB+C) or simple AM
A max
A min
61
Amplitude Modulation
We can easily demodulate the signal if 1 + x(t) 0
Envelope Detection
> 1 ) min(1 + x(t)) < 0
Envelope Distortion
Phase Reversal
62
AM in Frequency domain
X(f)
X c(f)
USB
LSB
fc
LSB
fc B
+B f
USB
fc
fc + B
Key Points:
1. Power wasted in carrier
2. Requires twice the baseband bandwidth
Feb 03, 2014
63
AM Power Calculations
ST hx2c (t)i
xc(t) = Ac[1 + x(t)] cos 2f ct
) S T = 12A2c (1 + 2S x);
S x = hx2(t)i
Pc = ST =0 = 12A2c
Psb = 12(S T Pc) = 14A2c 2S x = 122S xPc (in one sideband)
Feb 03, 2014
64
AM Power Calculations
jx(t)j 1 ) 2S x 1
) Psb < 12Pc
2Psb < Pc
At least 50% power is wasted in carrier
65
Xc+(f) =
Ac
X(f
2
f c)
X c(f)
USB
LSB
fc
LSB
fc B
USB
fc
fc + B
f
66
Psb = 14A2c S x
P sb
?
A 2max
67
Power Calculations
What is
P sb
?
A 2max
= S4x
(100% modulation)
) AP2sb = S16x
max
68
Tone Modulation
Information signal is x(t) = Am cos 2f mt
For DSB-SC or DSB:
xc(t) = AcAm cos 2f mt cos 2f ct
A cA m
xc(t) = 2 cos 2(fc fm)t + cos 2(fc + fm)t
For DSB+C or AM:
m
xc(t) = Ac cos 2fct + A cA
cos 2(fc fm)t
2
m
+ A cA
cos 2(fc + fm)t
2
69
Tone Modulation
For DSB-SC or DSB:
A cA m
2
fc
fc + fm
Ac
AcAm
2
fc
fc + fm
70
Modulators
Product modulators
Hard to build
Generally limited by low power and low frequencies
Switching modulator
71
SSB-AM
First consider DSB-SC
USB
LSB
X c(f)
fc
LSB
USB
fc B
fc
fc + B
fc B
fc
fc + B
fc
72
BT = 2W
For SSB:
BT = W
S T = 1 Psb = 14A2c S x
73
A cA m
xc(t) = 2 cos 2(fc fm)t + cos 2(fc + fm)t
For USSB:
74
Tone Modulation
For DSB-SC or DSB:
A cA m
2
fc
fc + fm
For USSB:
A cA m
2
fc
fc + fm
75
SSB Filter
SSB Output
ybp(t) = xbp(t) ? hbp(t) = xc(t)
hbp(t)
76
xi(t) = A cx(t)
xq(t) = 0
77
fc
fc W
fc
fc + W
fc W
fc
fc + W
H bp+(f)
H lp(f) = H bp+(f + f c)
H lp(f)
(For USSB)
f
78
H lp(f)
(For LSSB)
f
jfj W
79
ybp(t) = 2<[ylp(t)ej2fct]
y bp(t) = 12Ac[x(t) cos 2f ct x
(t) sin 2f ct]
Feb 03, 2014
80
81
SSB Generation
DSB Input
SSB Filter
SSB Output
Hard to realize
Rely on the absence of spectral
components at low frequencies
82
VSB Signals
83
Demodulation
Frequency conversion
Synchronous detection
Envelope detection
84
Chapter 5
85
Exponential Modulation
86
Exponential Modulation
Transmitted/Modulated signal:
xc(t) = Ac cos(2f ct + (t))
87
Instantaneous Frequency
Instantaneous frequency:
1 d c(t)
1
f(t) 2
dt = 2
c(t)
88
180
89
f fc
90
(For FM)
R
Input
PM Modulator
(For PM)
FM Output
= 2f
91
92
93
Spectral Analysis of FM
94
2(t)
2!
+ ]
3(t)
3!
+ ]
xcq(t) Ac(t)
EE-380: Communication Systems
95
x(t)
PM
R
(t) =
2f x(t)
FM
X(f)
(f) =
jf X(f)=f
f>0
PM
FM
96
Tone Modulation
Am sin 2f mt
x(t) =
Am cos 2f mt
x(t)
R
(t) =
2f x(t)
) (t) = sin 2f mt
A m
= Am
( f m )f
PM
FM
PM
FM
PM
FM
97
f>0
98
fc fm
A c
4
fc
fc + fm
99
NOTE: cos( sin 2f mt) and sin( sin 2f mt) are periodic
Try to write their Fourier series expansion!
Feb 03, 2014
100
n even
n odd
where n > 0
and
J n()
1
2
xc(t) = Ac
e j( sin n)d
1
P
n=1
Feb 03, 2014
101
xc(t) = Ac
n=1
J 0()
J 1()
J 2()
J 2()
J 3()
fc fm
fc
J 3()
fc + fm
J 1()
102
Bessel Function
1.0
n=0
n=1
n=2
n=3
J ( )
n
n = 10
0.0
0 1 2 3
Feb 03, 2014
10
EE-380: Communication Systems
15
103
Bessel Function
1
Jn( )
=1
0.5
0
-0.5
0
Feb 03, 2014
=2
=5
= 10
1
n/
EE-380: Communication Systems
2
104
Bessel Function
105
106
FM Bandwidth
Consider Tone Modulation
Observation for 1
jJ n()j falls off quickly for jn=j > 1
Consider components only for jnj
Significant frequency components lie within fc fm
Observation for 1
jJ n()j is insignificant for jnj > 1
Consider components only for jnj = 1
Significant frequency components lie within f c f m
Feb 03, 2014
107
M() 1
B = 2( + 2) fm = 2(Amf4 + 2fm)
For worst case, use: Am = 1; fm = W
BT = 2(f + 2W);
>2
( = fW)
108
Approximations:
2DW = 2f ; D 1
BT =
2W;
D1
Carsons rule: BT 2(f + W) = 2(D + 1)W
Carsons rule is applied for extreme values of deviation
ratio. It underestimates the BW for practical systems.
For practical systems (2 < D < 10) :
D>2
Use BT 2(f + 2W) = 2(D + 2)W;
(Do Example 5.2-1 and Exercise 5.2-1 in text)
Feb 03, 2014
109
Single Tone
Generic Signal
D 1; 1
2fm
2DW
D 1; 1
2fm
2W
Carsons Rule
D; extreme
2( + 1)f m
2(D + 1)W
Practical designs
2( + 2)f m
2(D + 2)W
110
Direct method
111
(t) = x(t)
900
R
1
Input
Feb 03, 2014
Ac cos 2fct
NBPM Modulator
= 2Tf
EE-380: Communication Systems
NBFM Output
112
Non-linear
Device
Bandpass
Filter
Frequency Multiplier
Frequency
Converter
113
FM Detection
Phase-shift discriminator
FM to AM conversion
Approximates the differentiator
Zero-crossing detector
Frequency feedback
114
FM Detection: FM to AM
Differentiate the FM signal to get an AM signal!
xc(t) = Ac cos(2f ct + (t)) = Ac cos c(t)
R
xc(t) = Ac cos(2f ct + 2f x()d)
dx c(t)
dt
Signal
Differentiator
Envelope
Detector
DC
Block
115
FM Detection: Discriminator
(t) (t t1) t1 (t) = 2ft 1x(t)
Can we produce (t) (t t 1)?
Amplitude
Limiter+BPF
cos(2fct + (t))
Phase-shift
Network
LPF
yD(t) K Dfx(t)
sin(2fct + (t t 1))
116
117
e(t)
LPF
Ka
VCO
118
Used in
119
e(t)
LPF
= Av sin(2fct + )
e(t) = AcAv cos(2fct + ) sin(2fct + )
= AcAv[sin( ) sin(4fct + + )]
2
120
h(t)
Ka
y(t)
(t) = 2Kv
(t)
t y()d
Integrator
121
H(f)
(f)
Ka
Y(f)
K v=jf
aH(f)
Y(f) = 1+KaKH(f)(K
(f)
v=jf)
jfK aH(f)
= jf+K
(f)
aK vH(f)
jfKH(f)
= K1v jf+KH(f)
(f)
(K = K aK v)
122
(for FM input)
jfKH(f) f
f
Y(f) = K1v jf+KH(f)
jf X(f) = K
H L(f)X(f)
v
jfj < W
K>W
f
y(t) K
x(t)
v
First order PLL requires loop gain to be greater than
the frequency deviation K f
Thus second order PLLs are used in practice.
Feb 03, 2014
123
Non-linearities in FM
FM transmission is robust against channel variations!
vo(t) = a1vi(t) + a2v2i (t) + a3v3i (t) (Non-linear channel)
vi(t) = xc(t) = Ac cos(2f ct + (t))
vo(t) = K0 + K1(cos 2f ct + (t)) + K2(cos 4f ct + 2(t))
+ K3(cos 6f ct + 3(t))
We need a BPF; what are the requirements?
f c + f + W 2f c (2f + W) (Carsons rule)
fc 3f + 2W
With this BPF, output is: vo(t) = K1(cos 2f ct + (t))
Unlike AM, FM can combat amplitude non-linearities!
Feb 03, 2014
124