Cross Power Spectral Density: Z T Which Is Sum of Two Real Jointly WSS Random Processes

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 8

Cross power spectral density

Consider a random process Z (t ) which is sum of two real jointly WSS random processes
X(t) and Y(t). As we have seen earlier,

RZ (t ) = RX (t ) + RY (t ) + RXY (t ) + RYX (t )
If we take the Fourier transform of both sides,
S Z (w ) = S X (w ) + SY (w ) + FT ( RXY (t )) + FT ( RYX (t ))
where FT (.) stands for the Fourier transform.
Thus we see that S Z (w ) includes contribution from the Fourier transform of the cross-

correlation functions RXY (t ) and RYX (t ). These Fourier transforms represent cross power
spectral densities.
Definition of Cross Power Spectral Density
Given two real jointly WSS random processes X(t) and Y(t), the cross power spectral

density (CPSD) S XY (w ) is defined as

FTX T* (w ) FTYT (w )
S XY (w ) = lim E
T �� 2T

where FTX T (w ) and FTYT (w ) are the Fourier transform of the truncated processes

t t
X T (t ) = X(t)rect ( ) and YT (t ) = Y(t)rect ( ) respectively and *
denotes the complex
2T 2T
conjugate operation.
We can similarly define SYX (w ) by

FTYT* (w ) FTX T (w )
SYX (w ) = lim E
T �� 2T
Proceeding in the same way as the derivation of the Wiener-Khinchin-Einstein theorem
for the WSS process, it can be shown that

S XY (w ) = �RXY (t )e - jwt dt
-�

and

SYX (w ) = �RYX (t )e - jwt dt
-�
The cross-correlation function and the cross-power spectral density form a Fourier
transform pair and we can write

RXY (t ) = �S XY (w )e jwt dw
-�

and

RYX (t ) = �SYX (w )e jwt dw
-�

Properties of the CPSD


The CPSD is a complex function of the frequency w. Some properties of the CPSD of
two jointly WSS processes X(t) and Y(t) are listed below:
(1) S XY (w ) = SYX
*
(w )

Note that RXY (t ) = RYX ( -t )



\ S XY (w ) = �RXY (t )e - jwt dt
-�

= �RYX ( -t )e - jwt dt
-�

= �RYX (t )e jwt dt
-�
*
= SYX (w )

(2) Re( S XY (w )) is an even function of w and Im( S XY (w )) is an odd function of w


We have

S XY (w ) = �RXY (t )(cos wt + j sin wt )dt
-�
� �
= �RXY (t ) cos wt dt + j �RXY (t )sin wt ) dt
-� -�

= Re( S XY (w )) + j Im( S XY (w ))
where

Re( S XY (w )) = �RXY (t ) cos wt dt is an even function of w and
-�

Im( S XY (w )) = �RXY (t )sin wt dt is an odd function of w and
-�

(3) X(t) and Y(t) are uncorrelated and have constant means, then
S XY (w ) = SYX (w ) = m X mY d (w )

Observe that
RXY (t ) = EX (t + t )Y (t )
= EX (t + t ) EY (t )
= m X mY
= mY m X
= RXY (t )
\ S XY (w ) = SYX (w ) = m X mY d (w )

(4) If X(t) and Y(t) are orthogonal, then


S XY (w ) = SYX (w ) = 0

If X(t) and Y(t) are orthogonal,


RXY (t ) = EX (t + t )Y (t )
=0
= RXY (t )
\ S XY (w ) = SYX (w ) = 0

(5) The cross power PXY between X(t) and Y(t) is defined by
1 T
PXY = lim E �X (t )Y (t )dt
T �� 2T -T

Applying Parseval’s theorem, we get


1 T
PXY = lim E �X (t )Y (t ) dt
T �� 2T -T

1 �
= lim E �X T (t )YT (t )dt
T �� 2T -�

1 1 �
= lim �FTX T (w ) FTYT (w )dw
*
E
T �� 2T 2p -�
1 � EFTX T* (w ) FTYT (w )
= �lim dw
2p -�T �� 2T
1 �
= �S XY (w ) dw
2p -�
1 �
\ PXY = �S XY (w )d w
2p -�
Similarly,
1 �
PYX = �SYX (w )dw
2p -�
1 � *
= �S XY (w ) dw
2p -�
= PXY
*
Example Consider the random process Z (t ) = X (t ) + Y (t ) discussed in the beginning of
the lecture. Here Z (t ) is the sum of two jointly WSS orthogonal random processes
X(t) and Y(t).
We have,
RZ (t ) = RX (t ) + RY (t ) + RXY (t ) + RYX (t )
Taking the Fourier transform of both sides,

S Z (w ) = S X (w ) + SY (w ) + S XY (w ) + SYX (w )
1 � 1 � 1 � 1 � 1 �
\ �S Z (w )dw = �S X (w )dw + �SY (w )dw + �S XY (w )dw + �SYX (w )dw
2p -� 2p -� 2p -� 2p -� 2p -�
Therefore,
PZ (w ) = PX (w ) + PY (w ) + PXY (w ) + PYX (w )
Remark
 PXY (w ) + PYX (w ) is the additional power contributed by X (t ) and Y (t ) to the
resulting power of X (t ) + Y (t )
 If X(t) and Y(t) are orthogonal, then

S Z (w ) = S X (w ) + SY (w ) + 0 + 0
= S X (w ) + SY (w )
Consequently
PZ (w ) = PX (w ) + PY (w )

Thus in the case of two jointly WSS orthogonal processes, the power of the sum of
the processes is equal to the sum of respective powers.

Power spectral density of a discrete-time WSS random process


Suppose g[ n] is a discrete-time real signal. Assume g[n] to be obtained by sampling a

continuous-time signal g (t ) at an uniform interval T such that


g[n] = g (nT ), n = 0, �1, �2,...
The discrete-time Fourier transform (DTFT) of the signal g[n] is defined by

G (w ) = � g[ n]e - jw n
n =-�

G (w ) exists if { g[n]} is absolutely summable, that is, � g[n] < �. The signal g[n] is
n =-�

obtained from G (w ) by the inverse discrete-time Fourier transform


p
1 jw n
g[n]) =
2p �g (w )e
-p
dw

Following observations about the DTFT are important:


 w is a frequency variable representing the frequency of a discrete sinusoid.

Thus the signal g[n] = A cos(w0 n) has a frequency of w0 radian/samples.


 G (w ) is always periodic in w with a period of 2p . Thus G (w ) is uniquely
defined in the interval -p �w �p .
 Suppose { g[n]} is obtained by sampling a continuous-time signal g a (t ) at a
uniform interval T such that
g[n] = g a ( nT ), n = 0, �1, �2,...

The frequency w of the discrete-time signal is related to the frequency W of the

w
continuous time signal by the relation W =
T
where T is the uniform sampling interval. The symbol W for frequency of a
continuous signal is used in the signal-processing literature just to distinguish it from
the corresponding frequency of the discrete-time signal. This is illustrated in the Fig.
below.

 We can define the Z - transform of the discrete-time signal by the relation



G ( z ) = � g [ n] z - n
n =-�

where z is a complex variable. G (w ) is related to G ( z ) by


G (w ) = G ( z ) z = e jw

Power spectrum of a discrete-time real WSS process { X [n]}


Consider a discrete-time real WSS process { X [n]}. The very notion of stationarity

poses problem in frequency-domain representation of { X [n]} through the Discrete-time


Fourier transform. The difficulty is avoided similar to the case of the continuous-time
WSS process by defining the truncated process
�X [n] for n N
X N [n ] = �
�0 otherwise

The power spectral density S X (w ) of the process { X [n]} is defined as


1 2
S X (w ) = lim E DTFTX N (w )
N �� 2 N + 1

where
� N
DTFTX N (w ) = � X N [n]e - jwn = � X [n]e - jwn
n =-� n =- N

Note that the average power of { X [n]} is RX [0] = E X 2 [ n ] and the power spectral

density S X (w ) indicates the contribution to the average power of the sinusoidal


component of frequency w.

Wiener-Einstein-Khinchin theorem
The Wiener-Einstein-Khinchin theorem is also valid for discrete-time random processes.
The power spectral density S X (w ) of the WSS process { X [n]} is the discrete-time
Fourier transform of autocorrelation sequence.

S X (w ) = � RX [ m ] e - jwm -p �w �p
m =-�

RX [m] is related to S X (w ) by the inverse discrete-time Fourier transform and given by


p
1
RX [m]) = �S (w )e jwm d w
2p
X
-p
Thus RX [m] and S X (w ) forms a discrete-time Fourier transform pair. A generalized

PSD can be defined in terms of z - transform as follows



S X ( z) = �R [ m ] z
m =-�
x
-m

Clearly, S X (w ) = S X ( z ) z =e jw
-m
Example Suppose RX [m] = 2 m = 0, �1, �2, �3.... Then

S X (w ) = � RX [ m ] e - jw m
m =-�
m
� � 1�
= 1 + � � � e- jw m
m =-��2�
m �0

3
=
5 - 4cos w

The plot of the autocorrelation sequence and the power spectral density is shown in Fig.
below.

Example
Properties of the PSD of a discrete-time WSS process
 For the real discrete-time process { X [n]}, the autocorrelation function RX [m] is
real and even. Therefore, S X (w ) is real and even.
 S X (w ) �0.
 The average power of { X [ n]} is given by
1 p
EX 2 [n] = RX [0] = �S X (w )d w
2p -p
Similarly the average power in the frequency band [w1 , w2 ] is given by

w2
2 �S X (w )d w
w1

 S X (w ) is periodic in w with a period of 2p .

Interpretation of the power spectrum of a discrete-time WSS process


Assume that the discrete-time WSS process { X [n]} is obtained by sampling a continuous-

time random process { X a (t )} at an uniform interval, that is,


X [n] = X a (nT ), n = 0, �1, �2,...

The autocorrelation function RX [m] is defined by


R X [ m] = E X [ n + m ] X [ n ]
= E X a (n T + mT ) X a (n T )
= RX a ( mT )

\ RX [ m] = RX a ( mT ) m = 0, �1, �2,...

Thus the sequence { RX [m] } is obtained by sampling the autocorrelation function


RX a ( t ) at a uniform interval T .

The frequency w of the discrete-time WSS process is related to the frequency W of the

w
continuous time process by the relation W =
T

You might also like