Random Processes – Spectral Characteristics

Module-4

1 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Synopsis

 Introduction
 Power Density Spectrum and its Properties
 Cross PSD and its properties,
 Relationship between Correlation and Power Spectrum
 Power Spectral density of a WSS discrete Time random
processes and Sequences.
 Power Spectrum of Complex Processes.

2 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Introduction
 In the previous module we studied the characterization of a random
process in time domain by its mean, autocorrelation function and
covariance functions.

 In this module, we will study the spectral description of the random

process using the Power Spectral Density (PSD).

 For a deterministic signal, it is well-known that the spectral

information can be obtained using Fourier transform.

 However, the Fourier transform cannot be obtained for random

signals because for such type of signals Fourier transform does not
exist.

3 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Introduction
 Therefore, the frequency domain analysis can be carried out on a
random signal by transforming the autocorrelation function into
frequency domain.

 That is, the Fourier transforms of autocorrelation function known as

power density spectrum can be used to obtain the frequency
domain representation of random process.

 Since it requires the knowledge of Fourier transform, a brief

introduction to Fourier transform is given in the next section.

4 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Fourier Transform
 Fourier transform of a continuous-time deterministic signal x(t) is
given by

X ( )   x(t )e  jt dt


 The function 𝑋(𝑗𝜔) is called the spectrum of x(𝑡).

 Since 𝑋(𝑗𝜔) is a complex quantity, it has magnitude 𝑋(𝑗𝜔) and

phase ∠𝑋(𝑗𝜔).

 The plot between 𝑋(𝑗𝜔) versus 𝜔 is known as magnitude

spectrum and

 The plot between ∠𝑋(𝑗𝜔) versus 𝜔 is known as phase spectrum.

5 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Fourier Transform
 The Fourier transform for a signal exists only if it is absolutely
integrable.That is

 The inverse Fourier transform of 𝑋(𝑗𝜔) is given by

1 
x(t ) 
2  
X ( )e jt d

6 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)

 We studied that Fourier transform exists for a signal if it is

absolutely integrable.

 The Fourier transform does not exist for a random process, since
we assume the ensemble of a random process exists for all time.

 Therefore, frequency-domain representation of a random process

cannot be obtained using the Fourier transform of x(t).

 Hence, we have to find some other way of obtaining frequency-

domain representation of random process.

7 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)
 For that purpose first we will consider a portion of a sample
function x(t) that exists between –T and T for our analysis. Let

 x(t ), T  t  T
xT (t )  
0, o/w
T
Assume T
xT (t ) dt  , for all finite T .
 and Fourier transform can be obtained using the formula

 T
X T ( )   xT (t )e  j t
dt   x(t )e  jt dt
 T

8 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)
 Using Parseval’s theorem, we can write the energy of x(t) as

 T 1 
E (T )   xT (t ) dt   x(t ) dt   X T ( ) d
2 2 2
 T 2 

 By dividing the above expression by 2T, we obtain the average power

P(T).That is,

average power in x(t ) over the interval (T , T )

X T ( )
2
1 T 1 
P (T )  T x(t ) dt  2  d
2

2T  2T

9 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)
 Note that the above equation does not represent the power in an
entire sample function.

 The power in the entire sample function can be obtained by letting T

to infinity.

 Then it represents the power in one sample function but does not
represent the average power.

x(t )  X (t ), take expectation, let T  .

 Therefore, to get the average power for the random process we take
expected value of the above expression as,

10 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)
E[ X T ( ) ]
2
1 T 1 
 lim T E[ X (t ) ]dt  2  lim d
2
PXX

T  2T T  2T

1 
PXX  A{E[ X (t ) ]} 2
PXX 
2  
S XX ( )d

E[ X T ( ) ]
2

S XX  lim
T  2T
 is known as power spectral density.

 SXX(f) represents the average power per hertz. Therefore, the term
Power Spectral Density (PSD).

11 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Power Spectral Density (PSD)
Definition:

 For a random process X(t), the Power Spectral Density is defined as

 where 𝑋𝑇 (𝑗𝜔) is Fourier transform of the truncated version of the

process.
w.s.s.  PXX  RXX (0)

E[ X T ( ) ]
2
1 T 1 
 lim T E[ X (t ) ]dt  2  lim d
2
PXX

T  2T T  2T
The average power of X(t)
12 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
Tutorial example
 Consider a random process, X (t )  A0 cos(0t  )

 -- uniformly distributed on (0, )
2
 Find the average power of X(t)

SOLUTION:
2 2
A A
E[ X (t ) 2 ]  E[ A02 cos 2 ( 0t  )]  E[ 0  0 cos(2 0t  2)]
2 2
A02 A02 2 2 A02 A02 

2
 
2 0 
cos(2 0t  2 )d  
2 2
sin(2 0t  2 ) 20

A02 A02
  sin(2 0t )
2 

13 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial example
 The average power of X(t)

1 T A02 A02 A02

 A{E[ X (t ) ]}  lim T [ 2   sin(20t )]dt  2
2
PXX
T  2T

14 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Tutorial example
 For the above numerical find the power spectral density of X(t).
X (t )  A0 cos(0t  )

SOLUTION:
E[ X T ( ) ]
2

S XX  lim
T  2T
T 1 j j0t
T
X T ( )   A0 cos( 0t  )e  j t
dt   A0 [e e  e  j e  j0t ]e  jt dt
T T 2
A0 j T j (0  ) t A0  j T  j (0  ) t
 e  e dt  e  e dt
2  T 2  T

T    0 )T  e T  )T ]
 j 
j sin[( 1 ] j  t T  je j sin[(
T
0 sin(  T )
 A0Te (T   )T j  t0T
e j t
dt  e A Te  2T
0
j(
   0 )T  T
15 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
 Tutorial example
T 1 j j0t
T
X T ( )   A0 cos( 0t  )e  j t
dt   A0 [e e  e  j e  j0t ]e  jt dt
T T 2
A0 j T j (0  ) t A0  j T  j (0  ) t
 e  e dt  e  e dt
2  T 2  T

j sin[(   0 )T ]  j sin[(   0 )T ]
 A0Te  A0Te
(   0 )T (   0 )T

sin[(  0 )T ]  j sin[(   0 )T ]
X T ( )  A0Te j
 A0Te
(  0 )T (  0 )T
sin[(  0 )T ] j sin[(   0 )T ]
X T ( )  A0Te
*  j
 A0Te
(  0 )T (  0 )T

16 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial example

sin 2
[(   )T ] sin 2
[(   0 )T ]
X T ( )  X T ( ) X T ( )  A0 [T T
2 * 2 2 0 2
]
(   0 ) T
2 2
(   0 ) T
2 2

j 2  j 2 sin[(   0 )T ] sin[(   0 )T ]
 A T (e
2 2
e )
(   0 )T (   0 )T
0


2 2
E[e j 2
e  j 2
]  E[2cos 2]   2
2cos 2 d  sin 2  /2
0
0   0

E[ X T ( ) ]
2
A02 T sin 2 [(   0 )T ] T sin 2 [(   0 )T ]
 [  ]
2T 2  (   0 ) T 2 2
 (   0 ) T
2 2

17 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial example
 In obtaining the result we have neglected the cross terms each
having factors at widely separated frequencies 𝜔0 and −𝜔0 such that
their product becomes very small.
sin 2 x
 x 2 dx   (C-54)

 T sin 2 ( T ) 2
 T sin x 1
  (T )2 d    x 2 T dx  1 (a)

T sin 2 ( T ) , if   0
lim  (b)
T   ( T ) 2
0, if   0

18 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Tutorial example
E[ X T ( ) ]
2
A02 T sin 2 [(   0 )T ] T sin 2 [(   0 )T ]
 [  ]
2T 2  (   0 ) T 2 2
 (   0 ) T
2 2

E[ X T ( ) ]
2
A02
S XX ( )  lim  [ (   0 )   (   0 )]
T  2T 2

 Finally we use the above result to obtain the average power as,

1  1 A02
 A02
PXX 
2  S XX ( )d  2  2 [ (  0 )   (  0 )]d  2

19 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Properties of Power Spectral Density (PSD)
 Property 1: The power density spectrum 𝑺𝑿𝑿(𝝎) is a real value and
non-negative function.
(1) S XX ( )  0

 Property 2: The power density spectrum is an even function in 𝝎.

(2) X (t ) real  S XX ( )  S XX ( )

 Property 3: The PSD is a real function if X(t) is real.
(3) S XX ( ) is real
 Property 4: Average power is equal to the integral of 𝑺𝑿𝑿(𝝎) .
1 
(4)
2  
S XX ( )d  A{E[ X (t ) 2 ]}
20 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
Properties of Power Spectral Density (PSD)
 Average power in the frequency band [𝝎𝟏, 𝝎𝟐] is given by

 Property 5: The power density spectrum of the derivation 𝑋(𝑡) =

𝑑𝑋(𝑡)
is 𝝎2 times the power spectrum of X(t).
𝑑𝑡
(5) S XX ( )   2 S XX ( )

 Property 6: If then 𝑺𝑿𝑿(𝝎) is a continuous

function of 𝝎

21 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Properties of Power Spectral Density (PSD)
 Property 8: If 𝑺𝑿𝑿(𝝎) is the power density spectrum of X(t) then
the power density spectrum of

𝜶 𝑿(𝒕) is 𝜶 𝟐 𝑺𝑿𝑿(𝝎).

 Property 9: If 𝑺𝑿𝑿(𝝎) is the power density spectrum of X(t) then

the power density spectrum of X(t) + b with E[X(t)] = 0 is

1 2
 Property 10: If 𝑆𝑋𝑋 (𝜔) and 𝑆𝑋𝑋 (𝜔) are the PSD of 𝑋1 (𝑡) and
𝑋2 (𝑡) respectively.
1 2
 If 𝑋1 (𝑡) ⊥ 𝑋2 (𝑡) then PSD of 𝑋1 𝑡 + 𝑋2 (𝑡) is 𝑆𝑋𝑋 𝜔 + 𝑆𝑋𝑋 (𝜔)

22 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Some common Fourier-transform pairs

23 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Some common Fourier-transform pairs

24 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Determine which of the following functions are valid PSDs.

 SOLUTION

1 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

2 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Determine which of the following functions are power spectrum and,
which are not.

 SOLUTION

 But SXX(w) is not always non-negative.

 the value cos 2w is –1.

3 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

4 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF
 Let 𝑥(𝑡) be an ensemble of the random process 𝑋(𝑡). Let us define
𝑥𝑇(𝑡) as

 The Fourier transform of 𝑥𝑇(𝑡) is given by

 Using Parseval’s theorem, we can write the energy of the process as,

5 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF
 The average power is given by

 To find the average power of the random process, we take the

expected value with T tending to infinity in the above equation.

6 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF
 We know

 Using the above equation, we can obtain

 We know 𝐸[𝑋(𝑡1) 𝑋(𝑡2)] = 𝑅𝑋𝑋(𝑡1, 𝑡2)

7 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF

 Now the inverse Fourier transform of SXX(w) is

8 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF

which is the time average of the

process’ autocorrection function.
9 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
Relationship between PSD and ACF
 The above equation says that the inverse Fourier transform of the
power density spectrum is the time average of the autocorrelation
function of the process.

 Therefore, we can write

 For a wide-sense stationary process,

10 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Relationship between PSD and ACF
 Therefore,

 And

 That is, power spectral density is Fourier transform of

autocorrelation function.

 The above equations are known as Wiener–Khintchine relations.

11 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 If 𝑅𝑋𝑋 (𝜏) = 𝑎𝑒 −𝑏 𝜏 , find the spectral density function, where ‘a’ and
‘b’ are constants.

 SOLUTION

12 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

13 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Find the power spectrum for random process
X (t )  A cos(0t  )
 whose autocorrelation function is given as,

A02
by Ex 6.2-1, RXX ( )  cos( 0 )
 SOLUTION 2

A02 j0
RXX ( )  (e  e  j0 )
4
A02 j0
RXX ( )  (e  e  j0 )
4

14 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
x(t )e jt 
FT
X (   )

1 
FT
2 ( )

A02
S XX ( )  [ (   0 )   (   0 )]
2

15 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 A WSS random process X(t) has PSD

 Find the autocorrelation function

 SOLUTION

16 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

17 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Cross-Power Density Spectrum
Definition 1:

 Consider the jointly wide-sense stationary process X(t) and Y(t),

their sum W(t) = X(t) + Y(t).

 The autocorrelation function of W(t) is

RWW (t , t   )  E[W (t )W (t   )]  E{[ X (t )  Y (t )][ X (t   )  Y (t   )]}
 RXX (t , t   )  RYY (t , t   )  RXY (t , t   )  RYX (t , t   )

 Applying Fourier transform on both sides, we get

1 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Cross-Power Density Spectrum
 The second and third terms in the above equation are known as
cross-power density spectrum defined as

 By taking inverse Fourier transform, we can obtain cross-correlation

function

2 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Cross-Power Density Spectrum
Definition 2:

 Consider two real random processes X(t) and Y(t). Let 𝑥𝑇(𝑡) and
𝑦𝑇(𝑡) be truncated ensembles of the random processes defined as

 x(t ), T  t  T  y (t ), T  t  T
xT (t )   yT (t )  
0, o/w 0, o/w
 The Fourier transforms of 𝑥𝑇(𝑡) and 𝑦𝑇(𝑡) ) are given by

3 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Cross-Power Density Spectrum
 The cross power between two processes is given by
1  1 T 1  X T ( )* YT ( )
PXY (T ) 
2T  xT (t ) yT (t )dt  2T T x(t ) y(t )dt  2  2T
d

 The cross power between two random processes can be obtained

by taking expected value of the above equations

1 T 1 E[ X T ( )* YT ( )]

PXY (T ) 
2T T RXY (t , t )dt  2  2T
d

total average Cross Power contained in X (t ), Y (t )

1 T 1  E[ X T ( )* YT ( )]
PXY  lim
T  2T T RXY (t , t )dt  2  Tlim
 2T
d

4 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Cross-Power Density Spectrum
 From the above expression the cross density power spectrum can
be written as
E[ X T ( )* YT ( )]
S XY ( )  lim
T  2T

1 
PXY 
2 

S XY ( )d

1 
PYX   SYX ( )d  PXY
*

2 

Total cross power = PXY  PYX

X (t ), Y (t ) orthogonal  PXY  PYX  0
5 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
Relationship between cross-power spectrum
and cross-correlation function

 That is, the cross-power density spectrum can be defined as the

Fourier transform of the time average of the cross-correlation
function.

6 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Properties of Cross Power Density Spectrum
 Property 1: For a real valued random process X(t) and Y(t),
(1) S XY ( )  SYX (  )  SYX ( )*
 Property 2: The real part of 𝑆𝑋𝑌(𝜔) and real part of 𝑆𝑌𝑋(𝜔) are
even functions of 𝜔
(2) Re[S XY ( )] & Re[SYX ( )] -- even
 Property 3: Imaginary parts of 𝑆𝑋𝑌(𝜔) and 𝑆𝑌𝑋(𝜔) are odd
functions of 𝜔
(3) Im[S XY ( )] & Im[SYX ( )] -- odd

7 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Properties of Cross Power Density Spectrum
 Property 4: If X(t) and Y(t) are orthogonal
(4) X (t ) & Y (t ) orthogonal  S XY ( )  SYX ( )  0
X (t ) & Y (t ) orthogonal  RXY (t , t   )  0  A[RXY (t , t   )]  0

 Property 5: If X(t) and Y(t) are uncorrelated and have constant

means, then
(5) X (t ) & Y (t ) uncorrelated & have constant mean X , Y
 S XY ( )  SYX ( )  2 XY  ( )

8 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

 Given the cross power spectrum of a random process as,

a  jb / W ,  W    W
S XY ( )  
 0, o/w

X (t ), Y (t ) -- jointly w.s.s.
 Find the cross correlation

9 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Tutorial Example
 SOLUTION

j W
1 W b j 1 b e 1 W b j
RXY ( ) 
2 W (a  j W )e d  2 (a  j W ) j 
2 W W e d
W

jW  jW j W

1 (a  jb)e (a  jb)e 1 be
 [  ]
2 j j 2 jW 2 W

1 a e jW  e  jW b jW  jW 1 b e j  e  j

 [  (e  e )] 
2  j  2 W 2 j
a b b
 sin(W )  cos(W )  sin(W )
   W 2

10 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Given the cross power spectrum of a random process as,
8
S XY ( )   0
(  j )3
X (t ), Y (t ) -- jointly w.s.s.
 Find the cross correlation

 SOLUTION

2  2
u ( ) e  FT

(  j )3

RXY ( )  4u ( ) 2 e 

11 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

 Let the cross correlation of two random process X(t) and Y(t) be

AB
RXY (t , t   )  {sin(0 )  cos[0 (2t   )]}
2
 Where A, B and 𝜔0 are constants.

 Find the cross power spectrum

12 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Tutorial Example
 SOLUTION RXY (t , t   ) 
AB
{sin(0 )  cos[0 (2t   )]}
2
1 T
T  2T T
A[ RXY (t , t   )]  lim RXY (t , t   )dt

AB AB 1 T

2
sin( 0 )  lim
2 T  2T   T
cos[ 0 (2t   )]dt

AB AB j0
 sin( 0 )  [e  e  j0 ]
2 4j
AB  j AB
S XY ( )  [2 (  0 )  2 (  0 )]  [ (   0 )   (   0 )]
4j 2

13 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Find the cross-spectral density function of two random processes
𝑋 𝑡 = 𝐴 𝐶𝑜𝑠 𝜔0 𝑡 + 𝐵 𝑠𝑖𝑛 𝜔0 𝑡
Y 𝑡 = −𝐴 𝐶𝑜𝑠 𝜔0 𝑡 + 𝐵 𝑠𝑖𝑛(𝜔0 𝑡)

 Both of them are ZERO mean and have same variance as 𝜎 2

 SOLUTION

14 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

15 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Bandwidth of the Power Spectral Density
 The power spectral density (PSD) of a signal quantifies the
distribution of power across different frequencies in the signal.

 It provides information about how much power is present at each

frequency component.

 The bandwidth of the PSD represents the frequency range within

which a significant portion of the signal's power is concentrated.

 It indicates the width of the spectrum occupied by the signal.

 In other words, it delineates the range of frequencies where the

signal carries meaningful information.

 The bandwidth of a random process is measured by considering the

positive frequency range over which the PSD is non-zero.
16 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
Bandwidth of the Power Spectral Density
Low Pass Bandwidth

 A random process is said to be a low-pass process, if its spectral

components are concentrated around 𝜔 = 0.

 That is, for a lowpass process most of the power is near dc

 The bandwidth of such a process is the lowest frequency for which

the PSD is non-zero.

17 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Bandwidth of the Power Spectral Density
Band Pass Bandwidth

 A process is said to be bandpass process for which the power is

centred around some non-zero frequency 𝜔 = 𝜔0 .

 For a bandpass process, the bandwidth is given by 𝑩 = 𝝎𝑯 − 𝝎𝑳 ,

 where 𝝎𝑳 is the lowest value of B such that PSD 𝑺𝑿𝑿 𝝎 = 𝟎 for all
value of 𝟎 < 𝝎 < 𝑩

 where 𝝎𝑯 is the highest value of B such that PSD 𝑺𝑿𝑿 𝝎 = 𝟎 for

all value of 𝝎 > 𝑩

18 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Bandwidth of the Power Spectral Density

𝝎𝑳 𝝎𝑯

19 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

RMS (Root Mean Square) Bandwidth
 The rms bandwidth is the spread of normalized power spectral
density.
X (t ) real  S XX ( ) even

 The spread of the power spectral density is equal to

 The area enclosed by the PSD is given by

20 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

RMS (Root Mean Square) Bandwidth
 Therefore, the rms bandwidth is equal to


S XX ( ) lowpass form  2
Wrms 
 
 2 S XX ( )d


S XX ( )d

 The above concept can be extended to Band pass form of power

spectrum.

 That is significant spectral components cluster near some frequencis

𝝎𝟎 and −𝝎𝟎 .

 If we assume, X (t ) real  S XX ( ) even

21 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

RMS (Root Mean Square) Bandwidth


4  (   0 ) 2 S XX ( ) d
S XX ( ) bandpass form  Wrms
2
 0

rms BW  0
S XX ( )d

mean frequency 0 
 S
0 XX ( )d

 S 0 XX ( )d

22 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example
 Find low pass rms bandwidth for the PSD

 SOLUTION

1 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

2 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

3 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

𝑾𝟑 𝝅 𝟐
𝑾 𝑾𝟐
𝑾𝟐𝒓𝒎𝒔 = 𝟖 = ⟹ 𝑾𝒓𝒎𝒔 =
𝟑𝑾𝝅 𝟑 𝟑
𝟖

4 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

 Tutorial Example
 Given the power spectrum S XX ( ) lowpass form

10
S XX ( ) 
[1  ( /10) 2 ]2
 Find the rms bandwidth, 3dB BW, 10dB BW
 SOLUTION

  10  /2 10
 XX     [1  ( /10)2 ]2    / 2 [1  tan 2  ]2  d
2
S ( ) d d 10sec

 / 2 100  /2  /2 1  cos 2
 d   100cos  d   100
2
d  50
 / 2 sec 2   / 2  / 2 2

  10 tan   d  10sec2  d
5 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT
 Tutorial Example
  10 2
 / 2 10 tan 
3 2

 XX
     [1  ( /10)2 ]2    / 2 [1  tan 2  ]2  d
2 2
S ( ) d d 10sec

 /2104 tan 2   /2  /2
4 1  cos 2
 d   10 sin  d   10
4 2
d  5000
 / 2 sec 
2   / 2   / 2 2


2
Wrms 
 
 2 S XX ( )d
 100


S XX ( )d

rms BW Wrms  10 rad/sec

6 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

3𝑑𝐵 𝐵𝑊 = 2 × 1.178 × 𝑅𝑀𝑆 𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑖𝑛 𝐻𝑒𝑟𝑡𝑧

10
3𝑑𝐵 𝐵𝑊 = 2 × 1.178 ×
2𝜋
𝟑𝒅𝑩 𝑩𝑾 = 𝟑. 𝟕𝟒𝟗𝑯𝒛

10𝑑𝐵 𝐵𝑊 = 2 × 4.798 × 𝑅𝑀𝑆 𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑖𝑛 𝐻𝑒𝑟𝑡𝑧

10
10𝑑𝐵 𝐵𝑊 = 2 × 4.798 ×
2𝜋
𝟏𝟎𝒅𝑩 𝑩𝑾 = 𝟏𝟓. 𝟐𝟔𝟑 𝑯𝒛

7 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

8 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

9 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

10 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

11 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

12 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

13 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

14 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

15 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

16 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

17 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Tutorial Example

18 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT

Summary

 Introduction
 Power Density Spectrum and its Properties
 Cross PSD and its properties,
 Relationship between Correlation and Power Spectrum
 Power Spectral density of a WSS discrete Time random
processes and Sequences.
 Power Spectrum of Complex Processes.

19 Dr. R.K.Mugelan, Associate Professor, SENSE, VIT