LECTURE 9-2: PULSE-

AMPLITUDE MODULATION
DR. TRINH VAN CHIEN
 INTRODUCTION TO LINE CODING

Techniques used for improving baseband transmission.

Focus on a 2-PAM modulation. Let us suppose to modify the TX filter p(t) for

• Spectral shaping: shaping of the transmitted waveform spectrum for

matching the channel response characteristics

• Expedite the recovery operation at the receiver side.

Typical examples are:

in the first case, spectral efficiency improvement and/or reduction of
transmitted power around DC (where many wired connections show bad
response)
in the second case, symbol synchronization

2
 BIPOLAR NRZ (NON RETURN TO ZERO)

Signal set M = {s1 (t ) = + APT (t ) , s2 (t ) = − APT (t ) }

s1 (t )

s2 (t )
A

0 T t 0 T t

1 −A
Versor b1 (t ) = PT (t )
T
Vector set M = {s1 = (+ ) , s2 = ( − ) }

(it coincides with a 2-PAM with rectangular pulse)

3
 BIPOLAR NRZ
Transmitted waveform uT

1 0 1 0
0 T 2T 3T 4T t

s(t )

s(t ) =  a[n] p(t − nT )

n

T 2T 3T 4T t
a[n] {+ , − }

4
 BIPOLAR NRZ
Signal spectrum | P( f ) |2
Gs ( f ) =  2
a = A2Tsinc2 ( fT )
T

1.0

Gs ( f )
0.8

0.6

0.4

0.2

0.0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 fT
5
 BIPOLAR NRZ

Signal spectrum | P( f ) |2
Gs ( f ) =  2
a = A2Tsinc2 ( fT )
T
0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50

-55

-60
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

6
 BIPOLAR RZ (RETURN TO ZERO)

Signal set M = {s1 (t ) = + APT / 2 (t ) , s2 (t ) = − APT / 2 (t ) }

s1 (t )

A1 s2 (t )

T /2 T
0 T /2 T t 0 t

2 −−A
1
Versor b1 (t ) = PT / 2 (t )
T
Vector set M = {s1 = (+ ) , s2 = ( − ) }

7
 BIPOLAR RZ
Transmitted waveform uT

1 0 1 0
0 T 2T 3T 4T t

s(t )

s(t ) =  a[n] p(t − nT )

n

T t
a[n] {+ , − }
2T 3T 4T

8
 BIPOLAR RZ
| P( f ) |2 A2T
Signal spectrum Gs ( f ) =  2
a = sinc2 ( fT / 2)
T 4
1.0

0.8

0.6

Gs ( f )
0.4

0.2

0.0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
fT

9
 BIPOLAR RZ
Signal spectrum | P( f ) |2 A2T
Gs ( f ) =  2
a = sinc2 ( fT / 2)
T 4
0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50

-55

-60
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

10
 EXAMPLE: BIPOLAR TRIANGULAR

Signal set M = {s1 (t ) = + AT (t ) , s2 (t ) = − AT (t ) }

s1 (t )
s2 (t )
A

0 T t
0 T t

Versor
3 −A
b1 (t ) = T (t )
T
Vector set M = {s1 = (+ ) , s2 = ( − ) }

11
 EXAMPLE: BIPOLAR TRIANGULAR
Transmitted waveform
uT

1 0 1 0
0 T 2T 3T 4T t

s(t )
s(t ) =  a[n] p(t − nT )
n

a[n] {+ , − } T 2T 3T 4T t

12
 EXAMPLE: BIPOLAR TRIANGULAR
| P( f ) |2 A2T
Signal spectrum Gs ( f ) =  2
a = sinc4 ( fT / 2)
T 4
1.0

0.8

Gs ( f )
0.6

0.4

0.2

0.0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
fT
13
 EXAMPLE: BIPOLAR TRIANGULAR
2 2
Signal spectrum
| P ( f ) | A T
Gs ( f ) =  a2 = sinc4 ( fT / 2)
T 4
0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

14
 MANCHESTER (BIPHASE)
Signal set M = {s1 (t ) = + Ax(t ) , s2 (t ) = − Ax(t ) }
x(t ) =  + PT / 2 (t ) − PT / 2 (t − T / 2) 
s1 (t ) s2 (t )

+A +A

T T t T t
T
2 2
−A −A

1
Versor b1 (t ) =  + PT / 2 (t ) − PT / 2 (t − T / 2)
T
Vector set M = {s1 = (+ ) , s2 = ( − ) }

15
 MANCHESTER (BIPHASE)
Transmitted waveform uT

1 0 1 0
0 T 2T 3T 4T t

s(t )

s(t ) =  a[n] p(t − nT )

n

a[n] {+ , − } T 2T 3T 4T t

16
 MANCHESTER (BIPHASE)
| P( f ) |2 2 sin ( fT / 2)
4
Signal spectrum Gs ( f ) =  2
=AT
( fT / 2)2
a
T
(maximum at f ≈ 0.74/T)
1.0

0.8

0.6
Gs ( f )

0.4

0.2

0.0 fT
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

17
 MANCHESTER (BIPHASE)

Signal spectrum
| P ( f ) |2
sin 4
( fT / 2)
Gs ( f ) =  a2 = A2T
T ( fT / 2)2
(maximum at f ≈ 0.74/T)
0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
18
 MANCHESTER (BIPHASE)
1   T 
p(t ) = b1 (t ) = +
 T /2
P (t ) − PT /2 t − 
T   2 
1 T  T  T T  T  3T  
P( f ) = +
 2 sinc  f  exp  − j 2 f  − sinc  f  exp  − j 2 f  =
T  2  4 2  2  4 
 T  T  T 
= + sinc  f  exp  − j 2 f   1 − exp ( − j fT ) 
 2  2  4 
T  T
P ( f ) = sinc 2  f  1 − cos ( − fT ) − j sin ( − fT ) =
2 2

4  2
T  T
= sinc 2  f  1 − cos ( fT ) + j sin ( fT ) =
2

4  2
T  T
= sinc 2  f  1 + cos 2 ( fT ) − 2 cos ( fT ) + sin 2 ( fT )  =  A 1 − cos A
4  2  sin   =
2 2
T  T  T  T
= sinc 2  f  1 − cos ( fT )  = Tsinc 2  f  sin 2   f 
2  2  2  2
19
 UNIPOLAR NRZ

Signal set M = {s1 (t ) = + APT (t ) , s2 (t ) = 0 }

s1 (t )

s2 (t )
A

0 T 0 T t
t

1
Versor b1 (t ) = PT (t )
T
Vector set M = {s1 = (+ ) , s2 = ( 0 ) }

20
 UNIPOLAR NRZ
Transmitted waveform
uT

1 0 1 0
0 T 2T 3T 4T t

s(t )
s(t ) =  a[n] p(t − nT )
n

a[n] {+ , 0} T 2T 3T 4T t

21
 UNIPOLAR NRZ

Signal spectrum
2
2
| P( f ) | a2 +
n  n
Gs ( f ) =  a2
T
+ 2
T

n =−
P    f − 
T   T

| P( f ) |2 = x sinc 2 ( fT ) xR

A Dirac delta at zero frequency

A2 A2
Gs ( f ) = Tsinc ( f ) +  ( f )
2

4 4

22
 UNIPOLAR NRZ
Signal spectrum A2 A2
Gs ( f ) = Tsinc ( f ) +  ( f )
2

4 4

Gs ( f )

Gv(f)

fT
23
 UNIPOLAR RZ

Signal set M = {s1 (t ) = + APT / 2 (t ) , s2 (t ) = 0 }

s1 (t )

s2 (t )
A1

0 T /2 T t 0 T t

Versor
2
b1 (t ) = PT / 2 (t )
T
Vector set M = {s1 = (+ ) , s2 = ( 0 ) }

24
 UNIPOLAR RZ
Transmitted waveform
uT

1 0 1 0
0 T 2T 3T 4T t

s(t )
s(t ) =  a[n] p(t − nT )
n

a[n] {+ , 0} T 2T 3T 4T t

25
 UNIPOLAR RZ
Signal spectrum | P( f ) | 
2 2 +
2
n  n
G( f ) =  a2
T
+
T
a
2 
n =−
P    f − 
T   T

2
 sin( fT / 2) 
P( f ) = z  ( z R)
2

 
 ( fT / 2) 

Dirac deltas at zero frequency and at odd multiples of 1/T

A2 A2 + 2  (2i + 1)   (2i + 1) 
Gs ( f ) =
16
Tsinc ( fT / 2) +
2

16 i =−
sinc 
 2  

  f −
T 


26
 UNIPOLAR RZ

Signal spectrum A2 A2 + 2  (2i + 1)   (2i + 1) 

Gs ( f ) =
16
Tsinc ( fT / 2) +
2

16 i =−
sinc 
 2  

  f −
T 


1.0
Gs ( f )
0.8

0.6

0.4

0.2

0.0
-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8
fT
t/T

27
 M-PAM CONSTELLATION:
CHARACTERISTICS

1. Base-band modulation
2. One-dimensional signal space
3. m signals, symmetrical with respect to the origin
4. Information associated to the impulse amplitude
PAM=Pulse Amplitude Modulation

28
 M-PAM CONSTELLATION:
CONSTELLATION

SIGNAL SET M = {si (t ) =  i p(t )}im=1

Versor b1(t)=p(t) (d=1)

VECTOR SET

M = {s1 = (−(m − 1) ) , s 2 = (−(m − 3) ),..., s m−1 = ( +( m − 3) ) , s m = ( +( m − 1) )}  R

Rb
k = log 2 (m) T = kTb R=
k

29
 M-PAM CONSTELLATION:
CONSTELLATION
Example: 4-PAM constellation

M = {s1 = (-3α) , s 2 = (-α), s 3 = (+α) , s 4 = ( +3α)}  R

b1 (t )

s1 = ( −3 ) s 2 = ( − ) 0 s 3 = ( + ) s 4 = ( +3 )

30
 M-PAM CONSTELLATION:
CONSTELLATION
Example: 8-PAM constellation

M = {s1 = (-7α ) , s 2 = (-5α ), s 3 = (-3α ) , s 4 = (-α ), s 5 = (+α) , s 6 = (+3α), s 7 = (+5α) , s8 = (+7α)}  R

b1 (t )
s1 = ( −7 ) s 2 = ( −5 ) s 3 = ( −3 ) s 4 = ( − ) s = ( + ) s 6 = ( +3 ) s 7 = ( +5 ) s8 = ( +7 )
05

31
 M-PAM CONSTELLATION:
BINARY LABELLING
e : Hk  M

It is always possible to build a Gray labeling

4-PAM:
00 / s1 01/ s 2 11/ s 3 10 / s 4
b1 (t )

( −3 ) ( − ) 0 ( + ) ( +3 )

8-PAM:
110 / s1 111/ s 2 101/ s 3 100 / s 4 000 / s 5 001/ s 6 011/ s 7 010 / s 8
b1 (t )
( −7 ) ( −5 ) ( −3 ) ( − ) ( + ) ( +3 ) ( +5 ) ( +7 )
0
32
 M-PAM CONSTELLATION:
TRANSMITTED WAVEFORM
1
Example: 4-PAM p(t ) = PT (t )
uT T

1 0 1 1 0 0 1 1
Tb 2Tb 3Tb 4Tb 5Tb 6Tb 7Tb 8Tb

sT (t )

3 T

 T

T 2T 3T 4T
− T

33
−3 T
M-PAM CONSTELLATION:
TRANSMITTED WAVEFORM
Example: 4-PAM p(t ) = RRC  = 0.5

34
 M-PAM CONSTELLATION:
BANDWIDTH AND SPECTRAL
EFFICIENCY

Case 1: p(t) = ideal low pass filter

Total bandwidth Bid =

R Rb / k
=
(ideal case) 2 2

Spectral efficiency Rb
(ideal case)
id = = 2k bps / Hz
Bid

35
 M-PAM CONSTELLATION:
BANDWIDTH AND SPECTRAL
EFFICIENCY

Case 2: p(t) = RRC filter roll off 

R R /k R (1 +  )
Total bandwidth B = (1 +  ) = b (1 +  )
2 2

Spectral efficiency =
Rb
=
2k
bps / Hz
B (1 +  )

36
 EXERCIZE

Given a baseband channel with bandwidth B up to 4000 Hz, compute the

maximum bit rate Rb we can transmit over it with a 256-PAM
constellation in the two cases:

• Ideal low pass filter

• RRC filter with =0.25

37
 M-PAM CONSTELLATION:
MODULATOR

p (t )

e( )

u T = (vT [n]) + ( a[ n]) + +

 a[n]p(t − nT )
− −
s(t ) =
n =−

Equal to 2-PAM, but we have m possible levels:

a[n] {−(m − 1) , − (m − 3) ,..., + (m − 3) , + (m − 1) }

38
 M-PAM CONSTELLATION:
DEMODULATOR

q(t )
r (t ) y (t ) 1[n] ML s R [ n] v R [ n]
e( )
CRITERION

t0 + nT
Symbol
synchronization
R = 1/ T

Equal to 2-PAM, but we have m possible levels:

a[n] {−(m − 1) , − (m − 3) ,..., + (m − 3) , + (m − 1) }

39
M-PAM CONSTELLATION: EYE
DIAGRAM
4-PAM, p(t) = RRC with  =0.5

40
M-PAM CONSTELLATION: EYE
DIAGRAM
8-PAM, p(t) = RRC with  =0.5

41
 M-PAM CONSTELLATION:
ERROR PROBABILITY

By applying the asymptotic approximation we can obtain

m −1  3k Eb 
Pb (e)  erfc  
mk  m2 − 1 N
 0 

42
 M-PAM CONSTELLATION:
ERROR PROBABILITY

Comparison: 2-PAM vs. 4-PAM

1  Eb 
2 − PAM: Pb (e) = erfc  
2 
 N0 

3  2 Eb 
4 − PAM: Pb (e)  erfc  
8 
 5 N0 

The 2-PAM constellation has better performance

The constellation gain is in the order of 10 log(5/2) = 4 dB

43
 M-PAM CONSTELLATION:
ERROR PROBABILITY
Comparison: 2-PAM vs. 4-PAM
2-PAM
1 4-PAM
0.1

0.01

1E-3

1E-4

1E-5

1E-6
BER

1E-7

1E-8

1E-9

1E-10

1E-11

1E-12
-5 0 5 10 15 20 25
Eb/N0 [dB]
44
 M-PAM CONSTELLATION:
ERROR PROBABILITY

2-PAM
1 4-PAM
0.1
8-PAM
16-PAM
0.01 32-PAM
1E-3 64-PAM
128-PAM
1E-4 256-PAM
1E-5

1E-6
BER

1E-7

1E-8
The performance decrease

for increasing m
1E-9

1E-10

1E-11

1E-12
-5 0 5 10 15 20 25 30 35 40 45 50

Eb/N0 [dB]

45
 M-PAM CONSTELLATION:
PERFORMANCE/SPECTRAL
EFFICIENCY TRADE-OFF
Given a baseband channel with bandwidth B and an m-PAM constellation,
by increasing the number of signals m=2k we increase the spectral efficiency

id = Rb / B = 2k bps / Hz

then we can transmit a higher bit rate Rb.

Unfortunately, the performance decrease:

fixed a BER value, the signal-to-noise ratio Eb/N0 necessary to achieve it
increases with m.

46
 EXAMPLE

Suppose B=4kHz.

With a (ideal) 2-PAM we transmit Rb = 8 kbps

With a (ideal) 256-PAM we transmit Rb = 64 kbps

However, fixed a target BER (e.g. BER=1e-10), a 256-PAM requires a larger

ratio Eb/N0 (34 dB of difference!).

As an example, at the parity of transmitted power, the link distance is very

lower (by a factor of 50!)

47
 LINEAR MODULATION
An m-PAM constellation is a base-band modulation characterized by a low
pass TX filter p(t).

Let us suppose to change this TX filter from p(t) to p(t)cos (2 f0 t)

➢ The constellation stays unchanged →

the BER performance are the same

➢ The signal spectrum changes

48
 LINEAR MODULATION

2
P( f )
s(t ) =  a[n] p(t − nT ) G( f ) =  a2
T
n

s '(t ) =  a[n] p '(t − nT )

1
n
G '( f ) = [G ( f − f 0 ) + G ( f + f 0 )]
4
p '(t ) = p (t ) cos(2 f 0t )

The signal spectrum is translated around frequency f0

49
 LINEAR MODULATION

A linear modulation simply translates the spectrum around frequency f0

(carrier frequency or Intermediate Frequency IF )

The modulation formats obtained by applying a linear modulation to

m-PAM modulations are called m-ASK (Amplitude Shift Keying).

The only one really important is 2-ASK, which is always called 2-PSK
(Phase Shift Keying).

50
 M-ASK CONSTELLATION:
CHARACTERISTICS

1. One-dimensional constellation identical to m-PAM

2. Versor b1 (t ) = p '(t ) = p(t ) cos(2 f 0t )

3. Signal spectrum centred around f0 → bandpass modulations

4. ASK (Amplitude Shift Keying)

51
 M-ASK CONSTELLATION:
SIGNAL SPECTRUM

Gs ( f ) = x  P( f − f 0 ) + P( f + f 0 )  xR
2 2
 

Example: p(t) = ideal low pass filter

Rb
Bid = R =
k

Rb
id = = k bps / Hz
− f0 f0 Bid

R R

52
 M-ASK CONSTELLATION:
PROPERTIES

Properties

➢ Spectral efficiency halved with respect to m-PAM

➢ BER performance identical to m-PAM

➢ No practical applications
(only exception 2-ASK which is always called 2-PSK)

53