Power Spectral Density of Digitally Modulated

Signals
Saravanan Vijayakumaran
sarva@ee.iitb.ac.in
Department of Electrical Engineering
Indian Institute of Technology Bombay

August 22, 2013

1 / 22

PSD Definition for Digitally Modulated Signals

Consider a real binary PAM signal
u(t) =

bn g(t nT )

n=

where bn = 1 with equal probability and g(t) is a baseband pulse of

duration T

PSD = F [Ru ( )] Neither SSS nor WSS

2 / 22

Cyclostationary Random Process

Definition (Cyclostationary RP)
A random process X (t) is cyclostationary with respect to time interval T if it is
statistically indistinguishable from X (t kT ) for any integer k .

Definition (Wide Sense Cyclostationary RP)

A random process X (t) is wide sense cyclostationary with respect to time
interval T if the mean and autocorrelation functions satisfy
mX (t)

mX (t T )

RX (t1 , t2 )

RX (t1 T , t2 T )

for all t,
for all t1 , t2 .

3 / 22

Power Spectral Density of a Cyclostationary Process

To obtain the PSD of a cyclostationary process with period T

Calculate autocorrelation of cyclostationary process RX (t, t )

Average autocorrelation between 0 and T , RX ( ) = T1 0T RX (t, t ) dt
Calculate Fourier transform of averaged autocorrelation RX ( )
R

4 / 22

Power Spectral Density of a Realization

Time windowed realizations have finite energy
xTo (t)

x(t)I[ To , To ] (t)

STo (f )

F(xTo (t))

x (f )
S

|STo (f )|2
To

(PSD Estimate)

PSD of a realization
2
x (f ) = lim |STo (f )|
S
To
To

|STo (f )|2
1

*
)

To
To

To
2

T2o

x ( )
xTo (u)xTo (u ) du = R

5 / 22

Power Spectral Density of a Cyclostationary Process

X (t)X (t ) X (t + T )X (t + T ) for cyclostationary X (t)
x ( )
R

To
2

1
To

1
KT

T2o

KT
2

x(t)x (t ) dt
x(t)x (t ) dt

for To = KT

KT
2
K

1
T

1
T

1
T

T
0

2
1 X
x(t + kT )x (t + kT ) dt
K
K

k = 2

E [X (t)X (t )] dt

0
T

RX (t, t ) dt = RX ( )
0

PSD of a cyclostationary process = F[RX ( )]

6 / 22

PSD of a Linearly Modulated Signal

Consider
u(t) =

bn p(t nT )

n=

u(t) is cyclostationary wrt to T if {bn } is stationary

u(t) is wide sense cyclostationary wrt to T if {bn } is WSS

Suppose Rb [k ] = E[bn bnk

]
P
Let Sb (z) = k = Rb [k ]z k

The PSD of u(t) is given by


 |P(f )|2
Su (f ) = Sb e j2fT
T

7 / 22

PSD of a Linearly Modulated Signal

Ru ( )
=
=

1
T

1
T

Ru (t + , t) dt
0

n= m=

1 X
T

k = m=

E [bn bm
p(t nT + )p (t mT )] dt
(m1)T

E [bm+k bm
p( kT + )p ()] d

mT

Z
1 X

E [bm+k bm
p( kT + )p ()] d
T

k =
Z

1 X
Rb [k ]
p( kT + )p () d
T

k =

8 / 22

PSD of a Linearly Modulated Signal

Ru ( ) =

1 X
Rb [k ]
p( kT + )p () d
T

k =

p( + )p () d

*
)

|P(f )|2

p( kT + )p () d

*
)

|P(f )|2 e j2fkT

Su (f ) = F [Ru ( )]

|P(f )|2 X
Rb [k ]e j2fkT
T


 |P(f )|2
Sb e j2fT
T

k =

where Sb (z) =

k =

Rb [k ]z k .

9 / 22

Power Spectral Density of Line Codes

Line Codes
0

Unipolar NRZ

Polar NRZ

Bipolar NRZ

Manchester

Further reading: Digital Communications, Simon Haykin, Chapter 6

11 / 22

Unipolar NRZ
Symbols independent and equally likely to be 0 or A
P (b[n] = 0) = P (b[n] = A) =

1
2

Autocorrelation of b[n] sequence

Rb [k ] =

A2
2

k =0

A2
4

k 6= 0

p(t) = I[0,T ) (t) P(f ) = T sinc(fT )e jfT

Power Spectral Density
Su (f ) =

|P(f )|2 X
Rb [k ]e j2kfT
T
k =

12 / 22

Unipolar NRZ
Su (f )

X
A2 T
A2 T
sinc2 (fT ) +
sinc2 (fT )
e j2kfT
4
4
k =


A T
A
n
sinc2 (fT ) +
sinc2 (fT )
f
4
4
T
n=

A2 T
A2
sinc2 (fT ) +
(f )
4
4

13 / 22

Normalized PSD plot

1

Su (f )
A2 T

Unipolar NRZ

0.5

0
0.5

1.5

fT

14 / 22

Polar NRZ
Symbols independent and equally likely to be A or A
P (b[n] = A) = P (b[n] = A) =

1
2

Autocorrelation of b[n] sequence

Rb [k ] =

2
A

k =0

k 6= 0

P(f ) = T sinc(fT )e jfT

Power Spectral Density
Su (f ) = A2 T sinc2 (fT )

15 / 22

Normalized PSD plots

1

Su (f )
A2 T

Unipolar NRZ
Polar NRZ

0.5

0
0.5

1.5

fT

16 / 22

Manchester
Symbols independent and equally likely to be A or A
P (b[n] = A) = P (b[n] = A) =

1
2

Autocorrelation of b[n] sequence

Rb [k ] =

2
A

k =0

k 6= 0

P(f ) = jT sinc fT2 sin fT

ejfT
2
Power Spectral Density

Su (f ) = A2 T sinc2

fT
2

sin2

fT
2

17 / 22

Normalized PSD plots

1

Su (f )
A2 T

Unipolar NRZ
Polar NRZ
Manchester

0.5

0
0.5

1.5

fT

18 / 22

Bipolar NRZ
Successive 1s have alternating polarity
0

Zero amplitude

+A or A

Probability mass function of b[n]

P (b[n] = 0)

P (b[n] = A)

P (b[n] = A)

1
2
1
4
1
4

Symbols are identically distributed but they are not independent

19 / 22

Bipolar NRZ
Autocorrelation of b[n] sequence
Rb [k ] =

A2 /2
A2 /4

k =0
k = 1
otherwise

Power Spectral Density

Su (f )

=
=
=

T sinc2 (fT )


A2
A2  j2fT
e
+ e j2fT

2
4

A2 T
sinc2 (fT ) [1 cos(2fT )]
2
A2 T sinc2 (fT ) sin2 (fT )

20 / 22

Normalized PSD plots

1

Su (f )
A2 T

Unipolar NRZ
Polar NRZ
Manchester
Bipolar NRZ

0.5

0
0.5

1.5

fT

21 / 22

Thanks for your attention

22 / 22