SIGNALS AND SYSTEMS USING MATLAB

Chapter 7 Fourier Analysis in Communications and Filtering

Luis F. Chaparro

Modulation systems

Given lowpass nature of most message signals, it is necessary to shift in frequency the

spectrum of the message to avoid using a very large antenna

This is attained by modulation: changing either the magnitude A(t) or the phase (t)

of a carrier
A(t) cos(2fc + (t)).
giving

Amplitude Modulation (AM): A(t) proportional to message, for constant phase

Frequency Modulation (FM), Phase modulation (PM): (t) changes with the
message
Communication system: cascade of transmitter, channel and receiver none LTI

2 / 21

AM modulation systems

AM Suppressed Carrier

message m(t), carrier cos(c t)

modulated signal s(t) = m(t) cos(c t)
c >> 2f0, f0 maximum frequency in m(t)
Fourier spectrum
1
S() = [M( c ) + M( + c )]
2
M() spectrum of m(t)
2 cos(c t)

cos(c t)
m(t)

Channel

Band-pass
filter

Transmitter

Receiver

AM-SC transmitter, channel and receiver

3 / 21

Low-pass
filter

m(t)

 Demodulation: assuming output of bandpass filter is s(t)

R() = S( c ) + S( + c )
1
= M() + [M( 2c ) + M( + 2c )]
2
Output of LPF is M() or m(t)
Demodulation requires exact carrier frequency:

if demodulator uses c + , > 0:

r (t) = s(t) cos((c + )t)

R()
= S( c ) + S( + c + )
1
= [M( + ) + M( )]
2
1
+ [M( 2(c + /2)) + M( + 2(c + /2)] .
2
Output of LPF is distorted message

4 / 21

 Commercial AM

s(t) = [K + m(t)] cos(c t) modulated signal

K chosen so that K + m(t) > 0
1
S() = K [( c ) + ( + c )] + [M( c ) + M( + c )] .
2
Example: Commercial AM modulation
1
0.8
0.6
0.4
y(t)

0.2
0
0.2
0.4
0.6
0.8
1
0

0.2

0.4

0.6

0.8
t (sec)

1.6

100

|Y()|

y(t)

1.4

150

0.5
0
0.5
1
0

1.2

0.005

0.01

0.015

0.02

0.025

0.03

0.035

50

0
2.5

0.04

1.5

0.5

1.5

0.5

t (sec)

0.5

1.5

2.5
4
x 10

0.5

1.5

2.5
4
x 10

f (Hz)

2
500
400

|Z()|

z(t)

1
0
1
2
0

300
200
100

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0
2.5

0.04

f (Hz)

t (sec)

Message (top), part of original signal and corresponding AM modulated signal

(bottom left), spectrum of the original signal and of the modulated signal (top and
bottom right).

5 / 21

 Single sideband modulation efficient use of spectrum by reducing bandwidth of

modulate signal
cos(c t)
m(t)

s(t)

BPF
H(j)

c + B

Upper side-band AM transmitter. c is the carrier frequency and B the bandwidth of

the message
Quadrature AM efficient use of spectrum by sending two messages on the same band
m1 (t)

m
1 (t)

cos(c t)

cos(c t)
+

s(t)

r(t)

shift
2
m2 (t)

LPF

2 shift

Transmitter

LPF

m
2 (t)

Receiver

QAM transmitter and receiver: s(t) is the transmitted signal and r (t) the received
signal

6 / 21

 Frequency division multiplexing sharing the spectrum

cos(1 t)
m1 (t)

cos(1 t)

BPF

cos(2 t)
m2 (t)

LPF

m
1 (t)

cos(2 t)

BPF

Channel

cos(3 t)

m
2 (t)

LPF

cos(3 t)

m3 (t)

BPF

LPF

m
3 (t)

Frequency division multiplexing (FDM) system

Frequency modulation (FM) nonlinear, timevarying modulation system

sFM (t) = cos(c t +

m( )d )

Narrowband FM: angle (t) small so that cos((t)) 1, sin((t)) (t)

d [c t + (t)]
d (t)
= m(t), IF (t) =
= c + m(t)
dt
dt
S() = F [cos(c t + (t))] = F [cos(c t) cos((t)) sin(c t) sin((t))]
1
[( c ) + ( + c )] + [( c ) ( + c )]
2j

() =
M()
j

7 / 21

Example: Narrowband FM
10
|M()|

m(t)

50
0

50
0
1

0.05

0.1

0.15

0
500

0.2

500

0
f

500

0.1
|X()|

x(t)

0.5
0

0.05

0.5
0

0.05

0.1
t (sec)

0.15

0
500

0.2

500
450
400

Frequency

350
300
250
200
150
100
50
0
0.6/7

0.7/7

0.8/7

0.9/7

1/7

1.1/7

1.2/7

1.3/7

0.2

Time

Top left message m(t) R= 80 sin(20t)u(t) and narrow-band FM signal

t
x(t) = cos(2fc t + 0.1 m( )d ); top-right magnitude spectra of m(t) and x(t).
Spectrogram of x(t) displaying evolution of its Fourier transform with respect to time.

8 / 21

 Wideband FM

messages: m1(t) = 80 sin(20t)u(t), m2(t) = 2000tu(t)

instantaneous frequencies IFi (t) = 2fci + 50mi (t) i = 1, 2
fc1 = 2500, fc2 = 25Hz
200
150
m(t)

m(t)

50
0

100
50

50
0

0.02

0.04

0.06

0.08

0
0

0.1

0.04

0.02

0.04

0.06

0.08

0.1

0.06

0.08

0.1

0.5
x(t)

0.02

0.5
x(t)

0
0.5

0
0.5

1
0

0.02

0.04

0.06

0.08

0.1

t (sec)

t (sec)

10
|M()|

|M()|

10
5

0
200
0.015

150

100

50

50

100

150

0
1000

200

3
0.01

500

1000

2.5
|X()|

|X()|

500
3

x 10

0.005

2
1.5

0
5000

1
5000

5000

f
5000

5000

5000

4500

4500

4000

4000

3500

3500

3000

Frequency

Frequency

0
f

2500
2000
1500

3000
2500
2000
1500

1000

1000

500

500

0
0.02 0.04 0.06 0.08

0.10 0.12 0.14 0.16 0.18

Time

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

Time

Left sinusoidal message and right ramp message: messages, FM modulated signals,
spectra of messages, spectra of FM signals, and spectrograms of FM signals

9 / 21

Analog filtering

Use of eigenfunction property of LTI systems periodic and aperiodic signals have

Fourier representations consisting of sinusoids of different frequencies, the frequency

components of any signal can be modified by appropriately choosing the frequency
response of the LTI system or filter
LTI system
Aej(0 t+)

A|H(j0 )|ej(0 t++

! H(j0 )

H(s)

Eigenfunction property of continuous LTI systems

Appropriate filter for a certain application is specified using the spectral

characterization of the input and the desired spectral characteristics of the output
Classical approach in filter design is to consider lowpass prototypes, with normalized

frequency and magnitude responses, which may be transformed into other filters with
the desired frequency response

10 / 21

Filtering basics

B(s)
Filter transfer function H(s) =
(LTI system with specific frequency response)
A(s)
filter output Y () = X ()H(j)
Lowpass filter design

1
1 + f (2)
such that for low frequencies f (2) << 1 |H(j)|2 1,
for high frequencies f (2) >> 1 |H(j)|2 0
Choose magnitude squared function |H(j)|2 =

Issues to consider:

selection of the appropriate function f (.),

the factorization needed to get H(s) from the magnitude squared function

frequency transformation to convert LPF into other filters

Magnitude specifications
1 2 |H(j)| 1
0 |H(j)| 1

0 p (passband)
s (stopband)

11 / 21

()

|H(j)|
1

min

1 2
1
p

max

p s

Equivalent magnitude specifications for a lowpass filter

Loss specifications

Loss function () = 10 log10 |H(j)|2 = 20 log10 |H(j)| dBs

0 () max
0 p (passband)
() min
s (stopband)
max = 20 log10(1 2), min = 20 log10(1)
General case: (0) = 1 , 2 in the passband and 3 in the stopband

(0) = 1 dc loss
max = 2 1 maximum attenuation in passband
min = 3 1 minimum attenuation in stopband

12 / 21

Butterworth lowpass filter design

Magnitude response

N th order lowpass Butterworth filter

1

0 2
0
|HN (j )| =
=
1 + [0]2N
hp
hp half-power or 3dB frequency
Factorization

S = s/hp S/j = 0 = /hp

1
HN (S)HN (S) =
1 + (S 2)N
D(S)D(S) = 1 + (S 2)N HN (S) = 1/D(S)
e j(2k1)
N 2N
j(2k1)
2N
Poles: (1) Sk = e
Sk = jN = e j(2k1+N)
e
Sk = e j(2k1+N)/(2N)

k = 1, , 2N

Poles in circle of radius 1

No poles on jaxis
Consecutive poles separated by /N radians

13 / 21

 Filter design

() = 10 log10 |HN (/hp )|2 = 10 log10(1 + (/hp )2N )

0 () max
0 p
min () <
s
= p
p
hp
= s
s
hp

(p ) = 10 log10(1 + (p /hp )2N ) max so that

1/2N
0.1max
10
1
(s ) = 10 log10(1 + (s /hp )2N ) min so that
(100.1min 1)1/2N

half-power frequency:
s
p

hp
(100.1max 1)1/2N
(100.1min 1)1/2N
minimum order:
log10[(100.1min 1)/(100.1max 1)]
N
2 log10(s /p )

14 / 21

Chebyshev lowpass filter design

Normalized magnitude squared function

0
,

=
p
1 + 2CN2 (0)
N order of filter, ripple factor, CN (.) Chebyshev polynomials

|HN (j0)|2 =

Chebyshev polynomials
0

CN ( ) =

cos(N cos1(0))
|0| 1
cosh(N cosh1(0)) |0| > 1.

Three term difference equation:

CN+1(0) + CN1(0) = 20CN (0), N 0
initial conditions
C0(0) = cos(0) = 1
C1(0) = cos(cos1(0)) = 0
C0(0) = 1,
C1(0) = 0,
C2(0) = 1 + 202,
C3(0) = 30 + 403,

15 / 21

 Filter design



0
2 2
( ) = 10 log10 1 + CN ( )
0

Ripple factor
p
= 100.1max 1,

=
p
0

RW = 1

1
1 + 2

Minimum order 
i0.5
h 0.1
10 min 1
cosh1 10
0.1max 1
N
cosh1( ps )
Half-power frequency :
0
(hp ) = 10 log10(1 + 2CN2 (hp )) = 3 dB, then
2

1+

0
2
CN (hp )

= 100.3 2

1
0
= cosh(N cosh1(hp ))


 
1
1
= p cosh
cosh1
N

CN (hp ) =
hp

16 / 21

 Factorization

0 = S/j, S = s/p
1
1
H(S)H(S) =
=
1 + 2CN2 (S/j) D(S)D(S)
Guillemins
 
1
1
a = sinh1
N

k = sinh(a) cos(k )

real part of pole

0k = cosh(a) sin(k )

imaginary part of pole

where 0 k < /2 are the angles corresponding to the Butterworth filters

(measured with respect to the negative real axis of the S plane)

17 / 21

Example: Lowpass filtering

x(t) = [2 cos(5t) + cos(10t) + 4 sin(20t)]u(t)
Design third-order lowpass Butterworth filter with a half-power frequency hp = 5
rad/sec, to attenuate frequency components of frequency 10 and 20
cos(10 t)2 cos(5 t)+4 sin(20 t)
1

0.7

0.6

x(t)

0.8

0
5

|H(j)|

10
t

15

20

10
t

15

20

0.5

0.4

0.3

0.2

y(t)

0.1

5
6

0.9

0
5

10

20

30

Signal x(t), top right figure; lowpass Butterworth filter with poles and magnitude
response shown on the left. Filtered signal, bottom right, is approximately the
lowfrequency component of x(t)

18 / 21

Example: Butterworth vs Chebyshev lowpass filters with hp = 5 rad/sec

Filtering x(t) = [2 cos(5t) + cos(10t) + 4 sin(20t)]u(t)
Specifications
(0) = 0 dB
max = 0.1 dB, p = 5 rad/sec
min = 15 dB, s = 10 rad/sec
cos(10 t)2 cos(5 t)+4 sin(20 t)
1
0.9

0.8

0.7

Chebyshev

Butterworth
Chebyshev

5
x(t)

0
5

0.6
|H(j)|

Butterworth

10
t

15

20

10
t

15

20

0.5
0.4
5
0.3

0.2

0.1
4

y(t)

0
5

10

20

30

19 / 21

Frequency transformations
s
Lowpass-Lowpass
S=
0
0
Lowpass-Highpass
S=
s
s 2 + 0
Lowpass-Bandpass
S=
s BW
s BW
Lowpass-Bandstop
S= 2
s + 0
S is the normalized and s the final variables
0 is a desired cut-off frequency and BW a desired bandwidth

0.8

0.8

|H1(j)|

|H(j)|

Example: Lowpass prototype filter (Butterworth)

0 = 40 and BW = 10
LP prototype
0.6
0.4
0.2

LPHP

0.6
0.4
0.2

10

20

0.8

0.8

0.6
0.4
0.2
0
0

60

80

30

40

LPBE

|H3(j)|

|H2(j)|

LPBP

40

0.6
0.4
0.2
0

10

20

30

40

10

20

Frequency transformations: (top-left) prototype lowpass filter, (top-right) lowpass to

highpass transformation, (bottom-left) lowpass to bandpass transformation,
(bottom-right) lowpass to band-eliminating transformation

20 / 21

Example: General filter design

Specifications
(0) = 0, max = 0.1, min = 60 dB
p = 10, s = 15 rad/sec

0.6
0.4

6
8

0.2
5

10

15

20

25

10
0

0.8

|H()|

<H()

|H()|

0.8

0
0

0
1

<H()

0.6
0.4

10

15

20

25

10
0

10

15

20

25

80

()

60

()

10

40
50

0
0

10

10

15

20

25

15

20

25

60
40
20

20

20

80

50

10
0

100

100
20

6
8

0.2

10

15

20

25

15

10

Elliptic (left) and Chebyshev2 (right) lowpass filter designs using analogfil function.
Clockwise, magnitude, phase, loss function and poles and zeros are shown for each design.

21 / 21