Random-Process

1.1 Introduction
Random Processes:

• Any physical quantity that varies with time is a signal.

• Examples of signals are an electrocardiogram (ECG), and an electroencephalogram (EEG)signals.

• There are two types signals which are

1. Continuous time signals

2. Discrete time signals

• Continuous time signals have the time variable t takes values from −∞ to ∞ or in an interval between
t1 to t2 .

• Continuous time signals are indicated as x(t), y(t), z(t) and so on.

• A discrete-time signal is a set of measurements taken sequentially in time (e.g., at every millisecond).

• Each measurement point is usually called a sample, and a discrete-time signal is indicated by by
x(n), y(n), z(n) , where the index n is an integer that points to the order of the measurements in
the sequence.

• A random process is a time-varying function that contains the outcome of a random experiment for
each time instant, X(t).

• A random process is a time varying function, e.g., a signal.

• A random process consists of infinite number of random variables.

• Random Process are of two types

1. Continuous random process

2. Discrete random process

• Real random process also called stochastic process

• The collection of all possible sample functions of X(t) is called an ensemble of X(t).

1
1.2. Probability Distribution and Density Functions Random-Process
RP: Discrete and Continuous

The set of all possible sample functions

(a) Ensemble of continuous signals (b) Ensemble of discrete signals

{v(t, E i)} is called the ensemble and

defines the random process v(t) that
Figure 1.1: An Ensemble of Signals
describes the noise source.
Sample functions of a binary random
1.2 Probability Distribution and Density Functions
process.
The cdf of a random process is defined as

FX (x, t) = P {x(t) ≤ x}

The pdf of a random process is defined as

dFX (x, t)
fX (x, t) =
dx
The bivariate cdf of a random process is defined as

FX(t1 )X(t2 )(x, 1x2 ) = P {X(t1 ) ≤ x1 , X(t2 ) ≤ x2 }

The bivariate pdf of a random process is defined as

∂2
fX(x1 ,t1 )(x2 ,t2 ) (x, 1x2 ) = F
∂x1 ∂x2 X(t1 )X(t2 )
The bivariate cdf of a nth order multivariate random process is defined as

FX(t1 )X(t2 )(x, 1x2 )...(xn ,tn ) = P {X(t1 ) ≤ x1 , X(t2 ) ≤ x2 ...X(tn ) ≤ xn }

Similarly nth order multivariate density functions of order n exist

∂2
fX(x1 ,t1 )(x2 ,t2 )...(xn ,tn ) (x, 1x2 xn ) = F
∂x1 ∂x2 X(t1 )X(t2 )
The mean of a random process is defined as
Z ∞
E[X(t)] = xfX (x, t)dx
−∞

The mean square µX of a random process is defined as

Z ∞
2
X 2 (t) = E[X (t) = x2 fX (x, t)dx
−∞

2
1.3. Stationary Random-Process

2 of a random process is defined as

The variance σX
2
σX = E[(X − µX )2 ]
= E[X 2 − 2µX X + µ2X ]
= E[X 2 ] − 2µX E[X] + µ2X
= E[X 2 ] − 2µ2X + µ2X
= E[X 2 ] − µ2X

1.3 Stationary
A random process {X(t)} is stationary if the pdfs and its statistical properties are invariant with changes
in time. For example, for a stationary process, X(t) and X(t + ∆) have the same probability distributions.

FX(t) (x) = FX(t+∆) (x) f or all t, t + ∆

Stationary of order one: Consider a pdf of random process X(t) is

fX(t+ts ) (x) = fX(t) (x) = fX (x)

The pdf is independent of any time shift hence its mean of the random process X(t) is also a constant

µX(t) (x) = µX

The variance of the random process X(t) is also a constant

2 2
σX(t) = σX

Stationary of order two: Consider a pdf of random process X(t) is

fX(t1 +ts )X(t2 +ts ) (x1 , x2 ) = fX(t1 )X(t2 ) (x1 , x2 )

Let

t1 + ts = t
ts = t − t1
t2 + ts = t2 + t − t1
t2 + ts = t + (t2 − t1 )
= t+τ

The mean of the random process X(t) is also a constant

E[X(t)] = µX
E[X(t1 )X(t2 )] = E[X(t)X(t + τ )]

The correlation of a wide sense stationer is independent of time, it depends upon τ . When the time
difference is 0 i.e., τ = t2 − t1 = 0 ⇒ t1 = t2 = t

E[X(t)X(t)] = E[X 2 (t)] = σX

2
+ µ2X

The above relation is a autocorrelation RX (τ ) of a random process X(t)

RX (τ ) = E[X 2 (t)] = σX
2
+ µ2X

If a random process X(t) has first and second order stationary, then it called Wide-Sense
Stationary Random Processes
Wide-Sense Stationary Random Processes A continuous-time random process X(t) is wide-sense
stationary (WSS) if it follows the following properties

3
1.3. Stationary Random-Process

1. The mean is independent of time t

E[X(t)] = µX(t) = µX = constant

The variance of the random process X(t) is also a constant

2 2
σX(t) = σX

2. The autocorrelation function only depends on time difference

RX (τ ) = E[X(t)X(t)] = E[X 2 (t)] = σX

2
+ µ2X

A discrete-time random process {X(n), n ∈ Z} is weak-sense stationary or wide-sense stationary (WSS) if

1. µX (n) = µX f or all n ∈ Z

2. RX (n1 , n2 ) = RX (n1 − n2 ) f or all n1 , n2 ∈ Z

——————————-

4
1.4. Correlation Functions: Random-Process

1.4 Correlation Functions:

The correlation functions between two random variables is a measure of the similarity
between the variables.
There are two types of correlation functions
1. Auto-Correlation Functions
2. Cross-Correlation Functions

1.4.1 Auto-Correlation Functions:

The correlation of a signal X(t) with itself is called as autocorrelation. This is denoted as
RX (τ ) = E[X(t)X(t + τ )]
where t and τ are arbitrary. When τ = 0 the autocorrelation function is the average of the random process
squared. It is also called average power.
RX (0) = E[X(t)X(t)] = E[X 2 (t)] = σX
2
+ µ2X
For −τ
RX (−τ ) = E[X(t)X(t − τ )]
Let t0 = t − τ
RX (−τ ) = E[X(t0 + τ )X(t0 )]
= E[X(t0 )X(t0 + τ )]

RX (−τ ) = RX (τ )
The autocorrelation function is an even function of τ .
E[Xi2 ] = E[X 2 ] = µ2X + σX 2

j=i
E[Xi Xj ] =
E[Xi Xi ] = µ2X j 6= i
The bounds on auto-correlation function is

E[{X(t) ± X(t + τ )}2 ] ≥ = 0

E[{X 2 (t) ± 2X(t)X(t + τ ) + X 2 (t + τ )}2 ] ≥ = 0

By performing expectations
RX (0) ± 2RX (τ ) + RX (0) ≥ 0

|RX (τ )| ≤ RX (0)
If X(t) is a random process with non zero mean, then it is defined as auto-covariance function
CX (τ )
CX (τ )RX (0) = E[{X(t) − µX }{X(t + τ ) − µX }]
= RX (τ ) − µ2X
The power spectral density of is defined as
Z ∞
RX (τ )e−jωτ dτ ≥ 0 f or all ω
−∞

5
1.4. Correlation Functions: Random-Process

Properties of Auto Correlation Functions

• RX (τ ) is bounded as

|RX (τ )| ≤ RX (0)

• Auto Corrrelation function is even symmetry

RX (−τ ) = RX (τ )

• The mean value of the random process is obtained using autocorrelation using the following relation

E[X 2 (t)] = X 2 (t) = σX

2
+ µ2X = RX (0)

1.4.2 Cross-Correlation Functions:

Consider a two random process X(t) and Y (t) are wide sense stationary. When we consider both the
random process it is called as jointly wide sense stationary. Then their cross-correlation function is defined
as

RXY (τ ) = E[X(t)Y (t + τ )]
RY X (τ ) = E[Y (t)X(t + τ )]

where t and τ are arbitrary. The order of the subscript is

RXY (−τ ) = RY X (τ )

If Y (t) is periodic with period T , then

RXY (τ + T ) = RXY (τ )

If X(t) is periodic with T, then

RY X (τ + T ) = RY X (τ )

The bounds on cross-correlation function is

E[{X(t) ± kY (t + τ )}2 ] ≥ = 0
E[{X 2 (t) ± 2kX(t)Y (t + τ ) + k 2 Y 2 (t + τ )}2 ] ≥ = 0

By performing expectations

RX (0) ± 2kRXY (τ ) + k 2 RY (0) ≥ 0

If k = 1

RX (0) ± 2RXY (τ ) + RY (0) ≥ 0

Then it becomes
1
|RXY (τ )| ≤ = [RX (0) + RY (0)]
2
If k is a positive and real constant then

k 2 RY (0) + 2kRXY (τ ) + RX (0) ≥ 0

The quadratic will be never negative if it does not have real roots. If its discriminant is
2
4RXY (τ ) − 4RX (0)RY (0) ≤ 0

6
1.4. Correlation Functions: Random-Process

p
|RXY (τ )| ≤ RX (0)RY (0)

———————————————
For example for a quadratic equation

ax2 + bx + c2 = = 0

then its discriminant is b2 − 4ac

———————————————
The geometric mean is
p 1
RX (0)RY (0) ≤ [RX (0) + RY (0)]
2
A cross correlation function is

CXY (τ ) = E[(X(t) − µX )(Y (t + τ ) − µY )]

= RXY (τ ) − µX µY

CY X (τ ) = E[(Y (t) − µY )(X(t + τ ) − µX )]

= RY X (τ ) − µY µX

Addition and Subtraction

sin(A + B ) = sinAcosB − cosAsinB

sin(A − B ) = sinAcosB + cosAsinB
cos(A + B ) = cosAcosB − sinAsinB
cos(A − B ) = cosAcosB + sinAsinB
tanA + tanB
tan(A + B ) =
1 − tanAtanB
tanA − tanB
tan(A − B ) =
1 + tanAtanB

Product Identities
1
sinAcosB = (sin(A + B) + sin(A − B))
2
1
cosAsinB = (sin(A + B) − sin(A − B))
2
1
cosAcosB = (cos(A + B) + cos(A − B))
2
1
sinAsinB = (cos(A − B) − cos(A + B))
2

Example 4.2 The random process described by

Y (t) = Acos(ωc t + Θ)

where A and ωc t are constants and Θ is a random variable distributed uniformly between ±π
Find the pdf of random variable Θ, mean and autocorrelation function of Y

Solution:

7
1.4. Correlation Functions: Random-Process

The pdf of random variable Θ

1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable Y

Z π
1
µY = Acos(ωc t + θ) dθ
−π 2π
A π
Z
= cosωc tcosθ − sinωc tsinθdθ
2π −π
A
= [cosωc tsinθ + sinωc tcosθ]π−π
2π
A
= [0 + sinωc t(−1 − (−1))]
2π
= 0
Autocorrelation function RY (τ )
Z π  
1 1
E[Y (t)Y (t + τ )] = Acos(ωc t + θ) Acos(ωc (t + τ ) + θ) dθ
−π 2π 2π
A2 π
Z
= cos(ωc t + θ)cos(ωc (t + τ ) + θ)dθ
2π −π
A2 π
Z
= [cos(2ωc t + ωc τ + 2θ) + cosωc τ ]dθ
2π −π
A2 π
Z
= [cos(2ωc t + ωc τ + 2θ) + cosωc τ ]dθ
2π −π

x = 2ωc t + ωc τ + 2θ
dx
= 0+0+2
dθ
dx
dθ =
2
Limits, when θ = π
x = 2ωc t + ωc τ + 2π
when θ = −π
x = 2ωc t + ωc τ − 2π

Z π Z c τ −2π
2ωc t+ω
dx
[cos(2ωc t + ωc τ + 2θ)]dθ = (cosx)
−π 2
2ωc t+ωc τ +2π
1 c t+ωc τ −2π
= [sinx]2ω
2ωc t+ωc τ +2π
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ ) − sin(2ωc t + ωc τ )]
2
= 0

8
1.4. Correlation Functions: Random-Process

Z π
[cosωc τ ]dθ = cosωc τ [θ]π−π
−π
= 2πcosωc τ

A2
E[Y (t)Y (t + τ )] = 2πcosωc τ
2π
A2
= cosωc τ
2

The ensemble variance is constant

A2
σY2 = RY (0) =
2
From the above discussions it is observed that

1. E[Y (t)] = 0=Constant and

2. E[Y (t)Y (t + τ )] = RY (τ ), it is independent of absolute time hence

Y (t) is a wide sense stationary.

Example 4.6 Consider two random process X(t) and Y (t) which are independent, jointly wide
sense stationary random process described by

X(t) = Acos(ω1 t + Θ1 )
Y (t) = Bcos(ω2 t + Θ2 )

where A, B and ω1 t ω2 tare constants and Θ1 and Θ2 are random variable distributed uniformly
between ±π Let X(t) and Y (t) are related by

W (t)X(t) = X(t)Y (t)

Find the autocorrelation function of W

Solution:
The autocorrelation function of X(t) and Y (t) are

A2
RX (τ ) = cosω1 τ
2
B2
RY (τ ) = cosω2 τ
2

A2 B2
RW (τ ) = cosω1 τ cosω2 τ
2 2
A2 B 2
= cosω1 τ cosω2 τ
4
A2 B 2
= [cos(ω1 + ω2 )τ + (ω1 − ω2 )τ ]
8

9
1.4. Correlation Functions: Random-Process

6. The random process described by

X(t) = Asin(ωc t + Θ)

where A and ωc are constants and Θ is a random variable uniformly distributed between ±π.
Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean and
autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable X

Z π
1
µX = Asin(ωc t + θ) dθ
−π 2π
A π
Z
= sinωc t cosθ − cosωc t sinθdθ
2π −π
A
= [sinωc tsinθ + cosωc tcosθ]π−π
2π
A
= [0 + sinωc t(−1 − (−1))]
2π
= 0

Autocorrelation function RX (τ )

Z π
1
E[X(t)X(t + τ )] = (Asin(ωc t + θ)) (Asin(ωc (t + τ ) + θ)) dθ
−π 2π
A2 π
Z
= sin(ωc t + θ)sin(ωc (t + τ ) + θ)dθ
2π −π
A2 π 1
Z
= [cosωc τ − cos(2ωc t + ωc τ + 2θ)]dθ
2π −π 2
A2 π
Z
= [cosωc τ − cos(2ωc t + ωc τ + 2θ)]dθ
4π −π

x = 2ωc t + ωc τ + 2θ
dx
dθ =0+0+2
dθ = dx
2
Limits, when θ = π
x = 2ωc t + ωc τ + 2π
when θ = −π
x = 2ωc t + ωc τ − 2π

10
1.4. Correlation Functions: Random-Process

2ωc t+ω
Z c τ +2π
Z π
dx
[cos(2ωc t + ωc τ + 2θ)]dθ = (cosx)
−π 2
2ωc t+ωc τ −2π
1
= [sinx]2ωc t+ωc τ +2π
2ωc t+ωc τ −2π
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ ) − sin(2ωc t + ωc τ )]
2
= 0

Z π
[cosωc τ ]dθ = cosωc τ [θ]π−π
−π
= 2πcosωc τ

A2
E[X(t)X(t + τ )] = 2πcosωc τ
4π
A2
RX (τ ) = cosωc τ
2
From the above discussions it is observed that

1. E[X(t)] = 0=Constant and

2. E[X(t)X(t + τ )] = RX (τ ), it is independent of absolute time hence

X(t) is a wide sense stationary.

——————————————————————–
7. The random process described by

X(t) = Acos(ωc t + φ + Θ)

where A ωc and φ are constants and Θ is a random variable uniformly distributed between
±π. Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean and
autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable X

11
1.4. Correlation Functions: Random-Process

Z π
1
µX = cos(ωc t + φ + Θ) dθ
−π 2π

x = ωc t + φ + θ
dx
= 0 + 0 + dθ
dθ
dθ = dx

Limits, when θ = π

x = ωc t + φ + π

when θ = −π

x = ωc t + φ − π

Z π
1
µX = cos(ωc t + φ + Θ) dθ
−π 2π
A ωc t+φ+π
Z
= cosxdx
2π ωc t+φ−π
A
= [sinx]ωωcc t+φ+π
t+φ−π
2π
A
= [sin(ωc t + φ + π) − sin(ωc t + φ − π)]
2π
A
= [−sin(ωc t + φ) + sin(ωc t + φ)]
2π
= 0

Autocorrelation function RX (τ )

Z π
1
E[X(t)X(t + τ )] = (Acos(ωc t + φ + θ)) (Acos(ωc (t + τ ) + φ + θ)) dθ
−π 2π
A2 π
Z
= cos(ωc t + φ + θ)cos(ωc (t + τ ) + φ + θ)dθ
2π −π
A2 π 1
Z
= [cosωc τ − cos(2ωc t + ωc τ + 2θ)]dθ
2π −π 2
A2 π
Z
= [cos(2ωc t + ωc τ + 2φ + 2θ) + cosωc τ ]dθ
4π −π

x = 2ωc t + ωc τ + 2θ
dx
= 0+0+2
dθ
dx
dθ =
2
Limits, when θ = π

x = 2ωc t + ωc τ + 2π

when θ = −π

x = 2ωc t + ωc τ − 2π

12
1.4. Correlation Functions: Random-Process

2ωc t+ω
Z c τ +2π
Z π
dx
[cos(2ωc t + ωc τ + 2θ)]dθ = (cosx)
−π 2
2ωc t+ωc τ −2π
1
= [sinx]2ωc t+ωc τ +2π
2ωc t+ωc τ −2π
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ + 2π) − sin(2ωc t + ωc τ − 2π)]
2
1
= [sin(2ωc t + ωc τ ) − sin(2ωc t + ωc τ )]
2
= 0

Z π
[cosωc τ ]dθ = cosωc τ [θ]π−π
−π
= 2πcosωc τ

A2 A2
E[X(t)X(t + τ )] = RX (τ ) = 2πcosωc τ = cosωc τ
4π 2
From the above discussions it is observed that
1. E[X(t)] = 0=Constant and

2. E[X(t)X(t + τ )] = RX (τ ), it is independent of absolute time hence

X(t) is a wide sense stationary.

——————————————————————–
8. The random process described by

X(t) = Acos(ωc t + Θ) + B

where A, B and ωc are constants and Θ is a random variable uniformly distributed between
±π. Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean and
autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable X

µX = E[X(t)] = E[Acos(ωc t + Θ) + B]
Z π Z π
1
= Acos(ωc t + θ) dθ + B dθ
−π 2π −π 2π

13
1.4. Correlation Functions: Random-Process

x = ωc t + θ
dx = 0 + dθ
dθ = dx

Limits, when θ = π

x = ωc t + π

when θ = −π

x = ωc t − π

Z π
1
µX = Acos(ωc t + θ) dθ
−π 2π
A ωc t+π
Z
= cosxdx
2π ωc t−π
A
= [sinx]ωωcc t+π
t−π
2π
A
= [sin(ωc t + π) − sin(ωc t − π)]
2π
A
= [−sin(ωc t) + sin(ωc t)]
2π
= 0

Z π
1 B π
B dθ = [θ]
−π 2π 2π −π
B
= [2π]
2π
= B

Autocorrelation function RX (τ )

1
cosAcosB = (cos(A + B) + cos(A − B))
2

E[X(t)X(t + τ )] = E[(Acos(ωc t + φ + B)) (Acos(ωc (t + τ ) + B))]

= E[ A2 cos(ωc t + φ) (cos(ωc (t + τ ) + φ)) + AB (cos(ωc (t + τ ) + φ)) + AB (cos(ωc t + φ)) + B 2

A2 π 1
Z
= [cos(2ωc t + ωc τ + 2θ) + cos(ωc τ )] dθ + 0 + 0 + B 2
2π −π 2
A2
= cos(ωc τ ) + B 2
2

A2
E[X(t)X(t + τ )] = cos(ωc τ ) + B 2
2
A2
RX (τ ) = cos(ωc τ ) + B 2
2
From the above discussions it is observed that

1. E[X(t)] = B=Constant and

2. E[X(t)X(t + τ )] = RX (τ ), it is independent of absolute time hence

14
1.4. Correlation Functions: Random-Process

X(t) is a wide sense stationary.

—————————————————————————-
10. The random process described by
X(t) = A2 cos2 (ωc t + Θ)
where A and ωc are constants and Θ is a random variable uniformly distributed between ±π.
Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean and
autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable X

µX = E[X(t)] = E[A2 cos2 (ωc t + Θ)]

A2
= E[1 + cos(2ωc t + 2Θ)]
2
A2 A2 π
Z
= + cos(2ωc t + 2θ)dθ
2 2 −π

x = 2ωc t + 2θ
dx = 0 + 2dθ
dx
dθ =
2
Limits, when θ = π
x = 2ωc t + 2π
when θ = −π
x = 2ωc t − 2π
Z π
1
= cos(2ωc t + 2θ) dθ
−π 2π
2ωc t+2π
A2
Z
= cosxdx
4π 2ωc t−2π
A2
= [sinx]2ω c t+2π
2ωc t−2π
4π
A2
= [sin(2ωc t + 2π) − sin(2ωc t − 2π)]
4π
A2
= [sin(2ωc t) − sin(2ωc t)] = 0
4π
A2
µX =
2

15
1.4. Correlation Functions: Random-Process

Autocorrelation function RX (τ )

E[X(t)X(t + τ )] = E[ A2 cos2 (ωc t + Θ) Acos(ωc (t + τ ) + B)A2 cos2 (ωc t + Θ) ]

A2
= E[{1 + cos(2ωc t + 2Θ)}{1 + cos(2ωc (t + τ ) + 2Θ)}]
4
Z π Z π
A2 1 1
= {1 + cos(2ωc t + 2θ) dθ + cos(2ωc (t + τ ) + 2θ) dθ}
4 −π 2π −π 2π
Z π
1
+ cos(2ωc t + 2θ)cos(2ωc (t + τ ) + 2θ) dθ
−π 2π
2 Z π
A 1
= {1 + 0 + 0 + [cos(4ωc t + 2ωc τ + 4θ) + cos(2ωc τ )] dθ}
4 4π −π
A2 1
= {1 + cos(2ωc τ )}
4 2

A2 1
E[X(t)X(t + τ )] = {1 + cos(2ωc τ )}
4 2
A2 1
RX (τ ) = {1 + cos(2ωc τ )}
4 2
From the above discussions it is observed that
A2
1. E[X(t)] = 2 =Constant and

2. E[X(t)X(t + τ )] = RX (τ ), it is independent of absolute time hence

X(t) is a wide sense stationary.

—————————————————————————-
12. The random process described by

X(t) = Acos(ωc t + Θ) + Bcos(ωs t)

where A, B ωc , and ωs ωc 6= ωs are constants and Θ is a random variable uniformly distributed

between ±π. Is X(t) wide-sense stationary? If not, then why not? If so, then what are the
mean and autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (−π)
1
=
2π

1

 2π −π <θ <π
fΘ (θ) =
0 otherwise


The mean of random variable X

µX = E[X(t)] = E[Acos(ωc t + Θ) + Bcos(ωs t)]

Z π
= A cos(ωc t + 2θ) dθ + Bcos(ωs t)
−π 2π
= Bcos(ωs t)

16
1.4. Correlation Functions: Random-Process

From the above discussions it is observed that X(t) is not a wide sense stationary, because its mean is not
a constant.

—————————————————————————-
13. The random process described by

X(t) = Acos(ωc t + Θ)

where A and ωc , are constants and Θ is a random variable uniformly distributed between
± π2 . Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean and
autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π/2 − (−π/2)
1
=
π

1

 π − π/2 < θ < π/2
fΘ (θ) =
0 otherwise


The mean of random variable X

1
µX = E[X(t)] = E[Acos(ωc t + Θ) ]
π
Z π/2
1
= A cos(ωc t + θ) dθ
−π/2 π

x = ωc t + θ
dθ = dx

Limits, when θ = π/2

x = ωc t + π/2

when θ = −π/2

x = ωc t − π/2

Z π/2
1
µX = A cos(ωc t + θ) dθ
−π/2 π
Z ωc t+π/2
A
= cos(x)dx
π ωc t−π/2
A
= cos(ωc t + π/2) − cos(ωc t − π/2)
π
2A
= cos(ωc t)
π
From the above discussions it is observed that X(t) is not a wide sense stationary, because its mean is not
a constant.

17
1.4. Correlation Functions: Random-Process

—————————————————————————-
14. The random process described by

X(t) = Acos(ωc t + Θ)

where A and ωc , are constants and Θ is a random variable uniformly distributed between 0
and π. Is X(t) wide-sense stationary? If not, then why not? If so, then what are the mean
and autocorrelation function for the random process? [?]

Solution:
The pdf of random variable Θ
1 1
fΘ (θ) = =
b−a π − (0)
1
=
π

1

 π 0<θ<π
fΘ (θ) =
0 otherwise


The mean of random variable X

1
µX = E[X(t)] = E[Acos(ωc t + Θ) ]
Z π π
1
= A cos(ωc t + θ) dθ
0 π

x = ωc t + θ
dθ = dx

Limits, when θ = 0

x = ωc t

when θ = π

x = ωc t + π

Z π
1
µX = A cos(ωc t + θ) dθ
0 π
A ωc t+π
Z
= sin(x)dx
π ωc t
A
= sin(ωc t + π) − sin(ωc t)
π
−2A
= sin(ωc t)
π
From the above discussions it is observed that X(t) is not a wide sense stationary, because its mean is not
a constant.

—————————————————————————-
15. A random process described by

X(t) = V

18
1.4. Correlation Functions: Random-Process

where V is a random variable uniformly distributed between 0 and 4. That means that each
realization of X(t) = V is constant v, that the constant varies from one realization to the next,
and that the variation is described as uniformly distributed between 0 and 4. Is X(t) wide-
sense stationary? If not, then why not? If so, then what are the mean and autocorrelation
function for the random process? [?]

Solution:
The pdf of random variable X
1 1
fV (v) = =
b−a 4 − (0)
1
=
4

1

 4 0<θ<4
fX (x) =
0 otherwise


The mean of random variable X

Z 4  4
1 1 1 2 
µX = E[X(t)] = E[V ] = vdv = = 4 −0
4 0 4 2 0 8
= 2

 4
1 4 2
Z
2 1
E[X(t)X(t + τ )] = E[V ] = v dv =
4 0 4 3 0
1 3 
= 4 −0
12
16
=
3
From the above discussions it is observed that
1. E[X(t)] = 2=Constant and
16
2. E[X(t)X(t + τ )] = RX (τ ) = 3 , it is independent of absolute time hence
X(t) is a wide-sense stationary.

—————————————————————————-
16. A random process described by
X(t) = V
where V is a random variable uniformly distributed between 0 and 2. That means that each
realization of X(t) = V is constant v, that the constant varies from one realization to the next,
and that the variation is described as uniformly distributed between 0 and 2. Is X(t) wide-
sense stationary? If not, then why not? If so, then what are the mean and autocorrelation
function for the random process? [?]

Solution:
The pdf of random variable X
1 1
fV (v) = =
b−a 2 − (0)
1
=
2

19
1.4. Correlation Functions: Random-Process

1

 2 0<θ<2
fX (x) =
0 otherwise


The mean of random variable X

Z 2  2
1 1 1 2 
µX = E[X(t)] = E[V ] = vdv = = 2 −0
2 0 2 2 0 4
= 1

Z 4  4
1 1
E[X(t)X(t + τ )] = E[V 2 ] = v 2 dv =
2 0 2 3 0
1 3 
= 2 −0
6
4
=
3
From the above discussions it is observed that
1. E[X(t)] = 1=Constant and

2. E[X(t)X(t + τ )] = RX (τ ) = 43 , it is independent of absolute time hence

X(t) is a wide-sense stationary.

—————————————————————————-
17. A random process described by

X(t) = V

where V is a random variable uniformly distributed between 0 and 3. That means that each
realization of X(t) = V is constant v, that the constant varies from one realization to the next,
and that the variation is described as uniformly distributed between 0 and 3. Is X(t) wide-
sense stationary? If not, then why not? If so, then what are the mean and autocorrelation
function for the random process? [?]

Solution:
The pdf of random variable X
1 1
fV (v) = =
b−a 3 − (0)
1
=
2

1

 3 0<θ<3
fX (x) =
0 otherwise


The mean of random variable X

Z 3  3
1 1 1 2 
µX = E[X(t)] = E[V ] = vdv = = 3 −0
3 0 3 2 0 6
3
=
2

20
1.4. Correlation Functions: Random-Process

Z 3  3
2 1 2 1
E[X(t)X(t + τ )] = E[V ] = v dv =
3 0 3 3 0
1 3
 
= 3 −0
9
= 3

From the above discussions it is observed that

1. E[X(t)] = 32 =Constant and

2. E[X(t)X(t + τ )] = RX (τ ) = 3, it is independent of absolute time hence

X(t) is a wide-sense stationary.

—————————————————————————-

21
1.4. Correlation Functions: Random-Process

1.4.3 Addition of Random Processes:

Consider the addition of two random independent, jointly wide-sense stationary random precess X(t) and
Y (t) is

W (t) = X(t) + Y (t)

The autocorrelation function of the sum W (t) is

RW (τ ) = E[{X(t) + Y (t)}{X(t + τ ) + Y (t + τ )}]

= E[X(t)X(t + τ ) + X(t)Y (t + τ ) + Y (t)X(t + τ ) + Y (t)Y (t + τ )}]
= RX (τ ) + RXY (τ ) + RY X (τ ) + RY (τ )

If X(t) and Y (t) are independent, then the cross-correlation function is

RXY (τ ) = RY X (τ ) = µX µY

If X(t) or Y (t) has a zero mean then

RW (τ ) = RX (τ ) + RY (τ )

18. The autocorrelation function for random process Z(t) is


50(1 − |τ |/T ) − T ≤ t ≤ T
RZ (τ ) =
0 otherwise

A random process X(t) is the sum

X(t) = Z(t) + U

where U is a random variable with a mean µU = 4, a variance σU2 = 25, and is independent of
Z(t). Find the autocorrelation function of X(t) [?]

Solution:

E[Z(t)] = 0

Autocorrelation function RX (τ )

RX (τ ) = E[{Z(t) + U }{Z(t + τ ) + U }]
= E[Z(t)Z(t + τ ) + U Z(t + τ ) + U Z(t) + U 2 ]
= RZ (τ ) + 0 + 0 + E[U 2 ]
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + 25 + 42
= RZ (τ ) + 41

——————————-

19. The autocorrelation function for random process Z(t) is


40(1 − |τ |/T ) − T ≤ t ≤ T
RZ (τ ) =
0 otherwise

22
1.4. Correlation Functions: Random-Process

A random process X(t) is the sum

X(t) = Z(t) + U

where U is a random variable with a mean µU = 5, a variance σU2 = 15, and is independent of
Z(t). Find the autocorrelation function of X(t) [?]

Solution:

E[Z(t)] = 0

Autocorrelation function RX (τ )

RX (τ ) = E[{Z(t) + U }{Z(t + τ ) + U }]
= E[Z(t)Z(t + τ ) + U Z(t + τ ) + U Z(t) + U 2 ]
= RZ (τ ) + 0 + 0 + E[U 2 ]
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + 15 + 52
= RZ (τ ) + 40

——————————-

20. The autocorrelation function for random process Z(t) is


60(1 − |τ |/T ) − T ≤ t ≤ T
RZ (τ ) =
0 otherwise

A random process X(t) is the sum

X(t) = Z(t) + U

where U is a random variable with a mean µU = 4, a variance σU2 = 20, and is independent of
Z(t). Find the autocorrelation function of X(t) [?]

Solution:

E[Z(t)] = 0

Autocorrelation function RX (τ )

RX (τ ) = E[{Z(t) + U }{Z(t + τ ) + U }]
= E[Z(t)Z(t + τ ) + U Z(t + τ ) + U Z(t) + U 2 ]
= RZ (τ ) + 0 + 0 + E[U 2 ]
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + σU2 + µ2U
= RZ (τ ) + 20 + 42
= RZ (τ ) + 36

——————————-

23
1.4. Correlation Functions: Random-Process

21. The random process X(t) has the autocorrelation function


10(1 − |τ |/τN ) − τN ≤ τ ≤ τN
RX (τ ) =
0 otherwise

The random process Y (t) is independent of X(t) and has the autocorrelation function

15(1 − |τ |/T ) − T ≤ t ≤ T
RY (τ ) =
0 otherwise

X(t) = Z(t) + U

where T  τN . The random process Z(t) = X(t) + Y (t). For Z(t), find the autocorrelation
function, its total power, its dc power, and its ac power. Is Z(t) wide-sense stationary? [?]

Solution:
2 = 10, µ = 0, σ 2 = 15, µ = 0, Autocorrelation function R (τ )
σX X Y Y Z

RZ (τ ) = RX (τ ) + RY (τ )

It is mentioned that T  τN , RZ (τ ) is summation of the autocorrelation function of RX (τ ) and RY (τ ),

RX (τ ) is within the limits of RY (τ ), hence we have to consider the limits for RZ (τ ) from 0 ≤ |τ | ≤ τN
and τN ≤ |τ | ≤ T
From 0 ≤ |τ | ≤ τN the RZ (τ ) is

RZ (τ ) = 15(1 − |τ |/T ) + 10(1 − |τ |/τN )

From τN ≤ |τ | ≤ T the RZ (τ ) is

RZ (τ ) = 15(1 − |τ |/T )



 15(1 − |τ |/T ) + 10(1 − |τ |/τN ) 0 ≤ |τ | ≤ τN



RZ (τ ) = 15(1 − |τ |/T ) τN ≤ |τ | ≤ T




0 otherwise


From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is

2
= σX + σY2 = 10 + 15
= 25

24
1.4. Correlation Functions: Random-Process

22. The random process X(t) and Y (t) are jointly wide-sense stationary and they are
independent. Given that W (t) = X(t) + Y (t) and
 
|τ |
RX (τ ) = 10 exp −
3

(  
3−|τ |
10 3 −3≤τ ≤3
RY (τ ) =
0 otherwise

For W (t), find its autocorrelation function, its total power, its dc power, and its ac power.
Is W (t) wide-sense stationary? [?]

Solution:
2 =
It is mentioned that X(t) and Y (t) are jointly wide-sense stationary and they are independent. σX
10, µX = 0, σY2 = 10, µY = 0, Autocorrelation function RZ (τ )

RW (τ ) = RX (τ ) + RY (τ )

RZ (τ ) is summation of the autocorrelation function of RX (τ ) and RY (τ ), RX (τ ) is within the limits

of RY (τ ), hence we to consider the limits for RZ (τ ) from 0 ≤ |τ | ≤ 3 and other than this limits
From 0 ≤ |τ | ≤ 3 the RZ (τ ) is

   
|τ | 3 − |τ |
RZ (τ ) = 10 exp − + 10
3 3

Otherwise

 
3 − |τ |
RZ (τ ) = 10
3

    
|τ | 3−|τ |


 10 exp − 3 + 10 3 0 ≤ |τ | ≤ 3
RZ (τ ) =  
 10 3−|τ |

otherwise

3

From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is

2
= σX + σY2 = 10 + 10
= 20

25
1.4. Correlation Functions: Random-Process

23. The random process X(t) has the autocorrelation function


10 exp(1 − |τ |/τN ) − τN ≤ τ ≤ τN
RX (τ ) =
0 otherwise

The random process Y (t) is independent of X(t) and has the autocorrelation function
 
RY (τ ) = 13
ωBτ

where (2π)/ωB  τN . The random process Z(t) = X(t)+Y (t). For Z(t), find the autocorrelation
function, its total power, its dc power, and its ac power. Is Z(t) wide-sense stationary? [?]

Solution:
2 =
It is mentioned that X(t) and Y (t) are jointly wide-sense stationary and they are independent. σX
10, µX = 0, σY2 = 13, µY = 0, Autocorrelation function RZ (τ )

RW (τ ) = RX (τ ) + RY (τ )

RZ (τ ) is summation of the autocorrelation function of RX (τ ) and RY (τ ), RX (τ ) is within the limits

of RY (τ ), hence we have to consider the limits for RZ (τ ) from 0 ≤ |τ | ≤ τN and other than this limits
From 0 ≤ |τ | ≤ τN the RZ (τ ) is

 
RZ (τ ) = 10 exp(1 − |τ |/τN ) + 13
ωBτ

Otherwise

 
RZ (τ ) = 13
ωBτ

  
sin(ωBτ )


 10(1 − |τ |/τ N ) + 13 ωBτ 0 ≤ |τ | ≤ 3
RZ (τ ) =  
 13 sin(ωBτ )

otherwise

ωBτ

From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is

2
= σX + σY2 = 10 + 13
= 23

26
1.4. Correlation Functions: Random-Process

24. A random process X(t) has the autocorrelation function

RX (τ ) = 15exp (−2|τ |) −∞<τ <∞

The random process Y (t) is Y (t) = X(t) − 3

a. What is the autocorrelation function of Y (t)?

b. What are the total power, the dc power, and the ac power of Y (t) ?

c. What is the cross-correlation RXY (τ )? [?]

Solution:
2 = 15 and µ = 0.
It is given that RX (τ ) has σX X
It is given that Y (t) = X(t) − 3 σY2 = 15 and µX = −3.
a. The autocorrelation function of Y (t) is

RY (τ ) = E[Y (t)Y (t + τ )] = E[{X(t) − 3}{X(t + τ ) − 3}]

= E[X(t)X(t + τ ) − 3X(t + τ ) − 3X(t) + 9]
= RXY (τ ) + 9

b. The total power, the dc power, and the ac power of Y (t) are
The total power

E[Y 2 (t)] = σY2 + µ2Y = 15 + 9 = 24

The DC power is

µ2Y = 9

The AC power is

σY2 = 15

c. The cross-correlation RXY (τ ) is

RXY (τ ) = E[X(t)Y (t + τ )] = E[{X(t)}{X(t + τ ) − 3}]

= E[X(t)X(t + τ ) − 3X(t)]
= RX (τ )

25. A random process X(t) has the autocorrelation function

RX (τ ) = 10cos(ωc τ ) −∞<τ <∞

The random process Y (t) is Y (t) = X(t) − 4

a. What is the autocorrelation function of Y (t)?

b. What are the total power, the dc power, and the ac power of Y (t) ?

c. What is the cross-correlation RXY (τ )? [?]

27
1.4. Correlation Functions: Random-Process

Solution:
2 = 10 and µ = 0.
It is given that RX (τ ) has σX X
It is given that Y (t) = X(t) − 4 σY2 = 10 and µX = −4.
a. The autocorrelation function of Y (t) is

RY (τ ) = E[Y (t)Y (t + τ )] = E[{X(t) − 4}{X(t + τ ) − 4}]

= E[X(t)X(t + τ ) − 4X(t + τ ) − 4X(t) + 16]
= RXY (τ ) + 16

b. The total power, the dc power, and the ac power of Y (t) are
The total power

E[Y 2 (t)] = σY2 + µ2Y = 10 + 16 = 26

The DC power is

µ2Y = 16

The AC power is

σY2 = 10

c. The cross-correlation RXY (τ ) is

RXY (τ ) = E[X(t)Y (t + τ )] = E[{X(t)}{X(t + τ ) − 4}]

= E[X(t)X(t + τ ) − 4X(t)]
= RX (τ )

26. A random process X(t) has the autocorrelation function


13(1 − |τ |/T ) − T ≤ τ ≤ T
RX (τ ) =
0 otherwise

The random process Y (t) is Y (t) = X(t) − 2

a. What is the autocorrelation function of Y (t)?

b. What are the total power, the dc power, and the ac power of Y (t) ?

c. What is the cross-correlation RXY (τ )? [?]

Solution:
2 = 13 and µ = 0.
It is given that RX (τ ) has σX X
It is given that Y (t) = X(t) − 2 σY2 = 13 and µX = −2.
a. The autocorrelation function of Y (t) is

RY (τ ) = E[Y (t)Y (t + τ )] = E[{X(t) − 2}{X(t + τ ) − 2}]

= E[X(t)X(t + τ ) − 2X(t + τ ) − 2X(t) + 4]
= RXY (τ ) + 4

b. The total power, the dc power, and the ac power of Y (t) are

28
1.4. Correlation Functions: Random-Process

The total power

E[Y 2 (t)] = σY2 + µ2Y = 13 + 4 = 17

The DC power is

µ2Y = 4

The AC power is

σY2 = 13

c. The cross-correlation RXY (τ ) is

RXY (τ ) = E[X(t)Y (t + τ )] = E[{X(t)}{X(t + τ ) − 2}]

= E[X(t)X(t + τ ) − 2X(t)]
= RX (τ )

27. X(t) and Y (t) are zero mean jointly wide-sense stationary random process. The random
process Z(t) is

Z(t) = 3X(t) + Y (t)

Find the correlations RZ (τ ), RZX (τ ), RXZ (τ ), RZY (τ ) and RY Z (τ ) [?]

Solution:

RZ (τ ) = E[Z(t)Z(t + τ )] = E[{3X(t) + Y (t)}{3X(t + τ ) + Y (t + τ )}]

= E[{9X(t)X(t + τ )} + {3X(t)Y (t + τ )} + {3X(t + τ )Y (t)} + {Y (t)Y (t + τ )}]
= 9RX (τ ) + 3RXY (τ ) + 3RY X (τ ) + RY (τ )

RZX (τ ) = E[Z(t)X(t + τ )] = E[{3X(t) + Y (t)}X(t + τ )]

= E[{3X(t)X(t + τ )} + {Y (t)X(t + τ )}]
= 3RX (τ ) + RY X (τ )

RXZ (τ ) = E[X(t)Z(t + τ )] = E[X(t){3X(t + τ ) + Y (t + τ )}]

= E[{3X(t)X(t + τ )} + {X(t)Y (t + τ )}]
= 3RX (τ ) + RXY (τ )

RZY (τ ) = E[Z(t)Y (t + τ )] = E[{3X(t) + Y (t)}Y (t + τ )]

= E[{3X(t)Y (t + τ )} + {Y (t)Y (t + τ )}]
= 3RXY (τ ) + RY (τ )

RY Z (τ ) = E[Y (t)Z(t + τ )] = E[Y (t){3X(t + τ ) + Y (t + τ )}]

= E[{3Y (t)X(t + τ )} + {Y (t)Y (t + τ )}]
= 3RY X (τ ) + RY (τ )

29
1.4. Correlation Functions: Random-Process

28. X(t) and Y (t) are zero mean jointly wide-sense stationary random process. The random
process Z(t) is

Z(t) = 3X(t) + 2Y (t)

Find the correlations RZ (τ ), RZX (τ ), RXZ (τ ), RZY (τ ) and RY Z (τ ) [?]

Solution:

RZ (τ ) = E[Z(t)Z(t + τ )] = E[{3X(t) + 2Y (t)}{3X(t + τ ) + 2Y (t + τ )}]

= E[{9X(t)X(t + τ )} + {6X(t)Y (t + τ )} + {6X(t + τ )Y (t)} + 4{Y (t)Y (t + τ )}]
= 9RX (τ ) + 6RXY (τ ) + 6RY X (τ ) + 4RY (τ )

RZX (τ ) = E[Z(t)X(t + τ )] = E[{3X(t) + 2Y (t)}X(t + τ )]

= E[{3X(t)X(t + τ )} + 2{Y (t)X(t + τ )}]
= 3RX (τ ) + 2RY X (τ )

RXZ (τ ) = E[X(t)Z(t + τ )] = E[X(t){3X(t + τ ) + 2Y (t + τ )}]

= E[{3X(t)X(t + τ )} + 2{X(t)Y (t + τ )}]
= 3RX (τ ) + 2RXY (τ )

RZY (τ ) = E[Z(t)Y (t + τ )] = E[{3X(t) + 2Y (t)}Y (t + τ )]

= E[{3X(t)Y (t + τ )} + 2{Y (t)Y (t + τ )}]
= 3RXY (τ ) + 2RY (τ )

RY Z (τ ) = E[Y (t)Z(t + τ )] = E[Y (t){3X(t + τ ) + 2Y (t + τ )}]

= E[{6Y (t)X(t + τ )} + 2{Y (t)Y (t + τ )}]
= 6RY X (τ ) + 2RY (τ )

29. X(t) and Y (t) are zero mean jointly wide-sense stationary random process. The random
process Z(t) is

Z(t) = X(t) + 2Y (t)

Find the correlations RZ (τ ), RZX (τ ), RXZ (τ ), RZY (τ ) and RY Z (τ ) [?]

Solution:

RZ (τ ) = E[Z(t)Z(t + τ )] = E[{X(t) + 2Y (t)}{X(t + τ ) + 2Y (t + τ )}]

= E[{X(t)X(t + τ )} + {2X(t)Y (t + τ )} + {2X(t + τ )Y (t)} + 4{Y (t)Y (t + τ )}]
= RX (τ ) + 2RXY (τ ) + 2RY X (τ ) + 4RY (τ )

RZX (τ ) = E[Z(t)X(t + τ )] = E[{X(t) + 2Y (t)}X(t + τ )]

= E[{X(t)X(t + τ )} + 2{Y (t)X(t + τ )}]
= RX (τ ) + 2RY X (τ )

30
1.4. Correlation Functions: Random-Process

RXZ (τ ) = E[X(t)Z(t + τ )] = E[X(t){X(t + τ ) + 2Y (t + τ )}]

= E[{X(t)X(t + τ )} + 2{X(t)Y (t + τ )}]
= RX (τ ) + 2RXY (τ )

RZY (τ ) = E[Z(t)Y (t + τ )] = E[{X(t) + 2Y (t)}Y (t + τ )]

= E[{X(t)Y (t + τ )} + 2{Y (t)Y (t + τ )}]
= RXY (τ ) + 2RY (τ )

RY Z (τ ) = E[Y (t)Z(t + τ )] = E[Y (t){X(t + τ ) + 2Y (t + τ )}]

= E[{Y (t)X(t + τ )} + 2{Y (t)Y (t + τ )}]
= RY X (τ ) + 2RY (τ )

1.4.4 Multiplication of Random Process:

Consider the multiplication of two random independent, jointly wide-sense stationary random precess X(t)
and Y (t) is

W (t) = X(t)Y (t)

The mean of W (t) is

E[W (t)] = µW = E[X(t)]E[Y (t)] = µX µY

The variance of W (t) is

2
σW = E[X(t)]E[Y (t) − µX µY ]2
2
= (σX + µ2X )(σY2 + µ2Y )µ2X µ2Y

The autocorrelation function of the sum W (t) is

RW (τ ) = E[{X(t)Y (t)}{X(t + τ ) + Y (t + τ )}]

= E[X(t)X(t + τ )]E[Y (t)Y (t + τ )]
= RX (τ )RY (τ )

30. The random process X(t) is noise with the autocorrelation function

10(1 − |τ |/τN ) − τN ≤ τ ≤ τN
RX (τ ) =
0 otherwise

The random process Y (t) is independent of X(t) and has the autocorrelation function

15(1 − |τ |/T ) − T ≤ t ≤ T
RY (τ ) =
0 otherwise

where T  τN . The random process Z(t) = X(t) × Y (t). For Z(t), find the autocorrelation
function, its total power, its dc power, and its ac power. Is Z(t) wide-sense stationary? [?]

Solution:
2 = 10, µ = 0, σ 2 = 15, µ = 0, Autocorrelation function R (τ )
σX X Y Y Z

RZ (τ ) = RX (τ ) × RY (τ )

31
1.4. Correlation Functions: Random-Process

RZ (τ ) = 15(1 − |τ |/T ) × 10(1 − |τ |/τN )

= 150(1 − |τ |/T )(1 − |τ |/τN )


 150(1 − |τ |/T )(1 − |τ |/τN ) |τ | ≤ τN
RZ (τ ) =
0 otherwise


From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is 150

31. The random process X(t) is noise with the autocorrelation function

10(1 − |τ |/τN ) − τN ≤ τ ≤ τN
RX (τ ) =
0 otherwise

The random process Y (t) is independent of X(t) and has the autocorrelation function

RY (τ ) = 18cos(ωc τ ) −∞≤τ ≤∞

where 2π/ωc  τN . The random process Z(t) = X(t) × Y (t). For Z(t), find the autocorrelation
function, its total power, its dc power, and its ac power. Is Z(t) wide-sense stationary? [?]

Solution:
2 = 10, µ = 0, σ 2 = 18, µ = 0, Autocorrelation function R (τ )
σX X Y Y Z

RZ (τ ) = RX (τ ) × RY (τ )

RZ (τ ) = 15(1 − |τ |/T ) × 18cos(ωc τ )

= 180(1 − |τ |/T )(cos(ωc τ ))


 180(cos(ωc τ ))(1 − |τ |/T ) |τ | ≤ τN
RZ (τ ) =
0 otherwise


From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is 180

32. The random process X(t) is noise with the autocorrelation function

10(1 − |τ |/τN ) − τN ≤ τ ≤ τN
RX (τ ) =
0 otherwise

32
1.4. Correlation Functions: Random-Process

The random process Y (t) is independent of X(t) and has the autocorrelation function

sin(ωB τ )
RY (τ ) = 13 −∞≤τ ≤∞
(ωB τ )

where 2π/ωB  τN . The random process Z(t) = X(t) × Y (t). For Z(t), find the autocorrelation
function, its total power, its dc power, and its ac power. Is Z(t) wide-sense stationary? [?]

Solution:
2 = 10, µ = 0, σ 2 = 13, µ = 0, Autocorrelation function R (τ )
σX X Y Y Z

RZ (τ ) = RX (τ ) × RY (τ )

sin(ωB τ )
RZ (τ ) = 10(1 − |τ |/T ) × 13
(ωB τ )
sin(ωB τ )
= 130(1 − |τ |/T )
(ωB τ )


sin(ω τ )
 130 (ωB Bτ ) (1 − |τ |/T ) |τ | ≤ τN

RZ (τ ) =

 0 otherwise

From the above discussions it is observed that

1. The mean of RZ (τ ) = 0

2. RZ (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RZ (τ ) is 130

33. The random process X(t) and Y (t) are jointly wide-sense stationary and they are
independent. Given that W (t) = X(t) × Y (t) and
 
|τ |
RX (τ ) = 10exp − −∞≤τ ≤∞
3

(  
3−|τ |
11 3 −3≤τ ≤3
RY (τ ) =
0 otherwise

For W (t) find its autocorrelation function, its total power, its dc power, and its ac power. Is
W (t) wide-sense stationary? [?]

Solution:
2 = 10, µ = 0, σ 2 = 11, µ = 0, Autocorrelation function R (τ )
σX X Y Y W

RW (τ ) = RX (τ ) × RY (τ )

   
|τ | 3 − |τ |
RW (τ ) = 10exp − × 11
3 3
  
|τ | 3 − |τ |
= 110exp −
3 3

33
1.4. Correlation Functions: Random-Process

   
|τ | 3−|τ |
 110exp − 3

3 |τ | ≤ τN
RW (τ ) =

0 otherwise


From the above discussions it is observed that

1. The mean of RW (τ ) = 0

2. RW (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RW (τ ) is 110

34. The random process X(t) and Y (t) are jointly wide-sense stationary and they are
independent. Given that W (t) = X(t) × Y (t) and
 
|τ |
RX (τ ) = 12exp − −∞≤τ ≤∞
4

(  
4−|τ |
10 4 −4≤τ ≤4
RY (τ ) =
0 otherwise

For W (t) find its autocorrelation function, its total power, its dc power, and its ac power. Is
W (t) wide-sense stationary? [?]

Solution:
2 = 12, µ = 0, σ 2 = 10, µ = 0, Autocorrelation function R (τ )
σX X Y Y W

RW (τ ) = RX (τ ) × RY (τ )

   
|τ | 4 − |τ |
RW (τ ) = 12exp − × 10
4 4
  
|τ | 4 − |τ |
= 120exp −
4 4

   
|τ | 4−|τ |

 120exp − 4 4 |τ | ≤ 4
RW (τ ) =

0 otherwise


From the above discussions it is observed that

1. The mean of RW (τ ) = 0

2. RW (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RW (τ ) is 120

35. The random process X(t) and Y (t) are jointly wide-sense stationary and they are
independent. Given that W (t) = X(t) × Y (t) and
 
|τ |
RX (τ ) = 11exp − −∞≤τ ≤∞
5

34
1.4. Correlation Functions: Random-Process

(  
5−|τ |
12 5 −5≤τ ≤5
RY (τ ) =
0 otherwise

For W (t) find its autocorrelation function, its total power, its dc power, and its ac power. Is
W (t) wide-sense stationary? [?]

Solution:
2 = 11, µ = 0, σ 2 = 12, µ = 0, Autocorrelation function R (τ )
σX X Y Y W

RW (τ ) = RX (τ ) × RY (τ )

   
|τ | 5 − |τ |
RW (τ ) = 11exp − × 12
5 5
  
|τ | 5 − |τ |
= 132exp −
5 5

   
|τ | 5−|τ |

 132exp − 5 5 |τ | ≤ 5
RW (τ ) =

0 otherwise


From the above discussions it is observed that

1. The mean of RW (τ ) = 0

2. RW (τ ) is independent of absolute time, hence it is wide-sense stationary.

Total power or AC power of RW (τ ) is 132

35
1.5. Ergodic Random Processes Random-Process

1.5 Ergodic Random Processes

Any random process is wide-sense stationarity if it satisfies the conditions like mean and and
autocorrelation function are independent of time. To calculate mean and autocorrelation function of
a random process, it requires an ensemble of sample functions (data records). It is difficult to collect the
data in real time situations. In many real-life applications Its convenient to calculate the averages from a
single data record. This is possible in certain random processes called ergodic processes.
A random process is ergodic if time averages with a sample function equal to ensemble. The avaragieng
of random process is defined as

1 T /2
Z
hx(t)i = lim x(t)dt
T →∞ T −T /2

The autocorrelation function for random process x(t) is defined as

Z T /2
1
hx(t)x(t + τ )i = lim x(t)x(t + τ )dt
T →∞ T −T /2

——————————-

36. Assume that the data in the following table are obtained from a windowed sample
function obtained from an ergodic random process. Use(4.73) to estimate the autocorrelation
for τ = 0, 2 and 4 ms, where ∆t = 2ms

x(t) 1.5 2.1 1.0 2.2 -1.6 -2.0 -2.5 2.5 1.6 -1.8
k 0 1 2 3 4 5 6 7 8 9

Solution:
Autocorrelation function for discrete sequences is
k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

It is given that τ = 0, 2 and 4 ms, where ∆t = 2ms , when τ = 0 it is an autocorrelation

when τ = 2 and ∆t = 2ms i = 2/2 = 1, τ = 4 and ∆t = 2ms i = 4/2 = 2

k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

1 X−1
k=N
1
x(k)2

rX (0) = x(k)x(k) =
n n
k=0
1 
(1.5)2 + (2.1)2 + (1.0)2 + (2.2)2 + (−1.6)2 + (−2.0)2 + (−2.5)2 + (2.5)2 + (1.6)2 + (−1.8)2

=
10
= 3.736
k=N −2
1 X
rX (2 ms) = x(k)x(k + 1)
n
k=0
1
= [(1.5)(2.1) + (2.1)(1.0) + (1.0)(2.2) + (2.2)(−1.6) + (−1.6)(−2.0) + (−2.0)(−2.5)
10
+ (−2.5)(2.5) + (2.5)(1.6) + (1.6)(−1.8)]
1
= [3.15 + 2.1 + 2.2 − 3.52 + 3.2 + 5 − 6.25 + 4 − 2.88]
10
= −0.700

36
1.5. Ergodic Random Processes Random-Process

k=N −3
1 X
rX (4 ms) = x(k)x(k + 2)
n
k=0
1
= [(1.5)(1.0) + (2.1)(2.2) + (1.0)(−1.6) + (2.2)(−2.0) + (−1.6)(−2.5) + (−2.0)(2.5)
10
+ (−2.5)(1.6) + (2.5)(−1.8)]
= −0.938

37. Assume that the data in the following table are obtained from a windowed sample
function obtained from an ergodic random process. Use(4.73) to estimate the autocorrelation
for τ = 0, 3 and 6 ms, where ∆t = 3ms

x(t) 1.0 2.2 1.5 -3.0 -0.5 1.7 -3.5 -1.5 1.6 -1.3
k 0 1 2 3 4 5 6 7 8 9

Solution:
Autocorrelation function for discrete sequences is
k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

It is given that τ = 0, 3 and 6 ms, where ∆t = 3ms , when τ = 0 it is an autocorrelation

when τ = 3 and ∆t = 3ms i = 3/3 = 1, τ = 6 and ∆t = 23ms i = 6/3 = 2

k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

1 X−1
k=N
1
x(k)2

rX (0) = x(k)x(k) =
n n
k=0
1 
(1.0)2 + (2.2)2 + (1.5)2 + (−3.0)2 + (−0.5)2 + (1.7)2 + (−3.5)2 + (−1.5)2 + (1.6)2 + (−1.3)2

=
10
= 3.898

k=N −2
1 X
rX (3 ms) = x(k)x(k + 1)
n
k=0
1
= [(1.0)(2.2) + (2.2)(1.5) + (1.5)(−3.0) + (−3.0)(−0.5) + (−0.5)(1.7) + (1.7)(−3.5)
10
+ (−3.5)(−1.5) + (−1.5)(1.6) + (1.6)(−1.3)]
= −0.353

k=N −3
1 X
rX (6 ms) = x(k)x(k + 2)
n
k=0
1
= [(1.0)(1.5) + (2.2)(−3.0) + (1.5)(−0.5) + (−3.0)(1.7) + (−0.5)(−3.5) + (1.7)(−1.5)
10
+ (−3.5)(1.6) + (−1.5)(−1.3)]
= −1.540

37
1.5. Ergodic Random Processes Random-Process

38. Assume that the data in the following table are obtained from a windowed sample
function obtained from an ergodic random process. Use(4.73) to estimate the autocorrelation
for τ = 0, 7 and 14 ms, where ∆t = 7ms

x(t) 1.5 0.4 0.8 0.3 -0.4 -1.7 2.0 -2.0 0.8 -0.2
k 0 1 2 3 4 5 6 7 8 9

Solution:
Autocorrelation function for discrete sequences is
k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

It is given that τ = 0, 2 and 4 ms, where ∆t = 2ms , when τ = 0 it is an autocorrelation

when τ = 7 and ∆t = 7ms i = 7/7 = 1, τ = 14 and ∆t = 7ms i = 14/7 = 2

k=N −1−i
1 X
rX (i) = x(k)x(k + i)
n
k=0

1 X−1
k=N
1
x(k)2

rX (0) = x(k)x(k) =
n n
k=0
1 
(1.5)2 + (0.4)2 + (0.8)2 + (0.3)2 + (−0.4)2 + (−1.7)2 + (2.0)2 + (−2.0)2 + (0.8)2 + (−0.2)2

=
10
= 1.487

k=N −1
1 X
rX (7 ms) = x(k)x(k + 1)
n
k=0
1
= [(1.5)(0.4) + (0.4)(0.8) + (0.8)(0.3) + (0.3)(−0.4) + (−0.4)(−1.7) + (−1.7)(2.0)
10
+ (2.0)(−2.0) + (−2.0)(0.8) + (0.8)(−0.2)]
= −0.744
k=N −2
1 X
rX (14 ms) = x(k)x(k + 2)
n
k=0
1
= [(1.5)(0.8) + (0.4)(0.3) + (0.8)(−0.4) + (0.3)(−1.7) + (−0.4)(2.0) + (−1.7)(−2.0)
10
+ (2.0)(0.8) + (−2.0)(−0.2)]
= 0.509

38
1.6. Power Spectral Densities Random-Process

1.6 Power Spectral Densities

The power spectral density of a wide-sense stationary random process is the Fourier transform of the
autocorrelation function. The distribution of power over range frequencies is described by power power
density spectrum.
Fourier transform pair is
Z ∞
XT (jω) = xT (t)e−jωt dt
−∞
Z ∞
1
xT (t) = XT (jω)ejωt dω
2π −∞
The power of signal is its square. Averaging the power of the signal x(t) within a window of width T
is

1 T /2 2
Z
PT = x (t)dt
T −T /2
1 ∞ 2
Z
= x (t)dt
T −∞ T
Z ∞ Z ∞
1 1
= xT (t) xT (t)ejω dωdt
T −∞ 2π −∞
Z ∞ Z ∞
1 1
= xT (jω) xT (t)ejωt dtdω
2πT −∞ 2π −∞
Z ∞
1
= xT (jω)xT (−jω)dω
2πT −∞

xT (−jω) = XT∗ (jω)

Z T /2 Z ∞
1 2 1
PT = X (t)dt = |XT (jω)|2 dω
T −T /2 2π −∞
Z T /2 Z ∞
1
X 2 (t)dt = |XT (jω)|2 dω
−T /2 2π −∞

Expectation of the above equation is

"Z # Z ∞
T /2 
2 1 2
E X (t)dt = E |XT (jω)| dω
−T /2 2π −∞
Z T /2 Z ∞
1
E[X 2 (t)]dt = E[|XT (jω)|2 ]dω
−T /2 2π −∞
Z ∞
1
T E[X 2 (t)] = E[|XT (jω)|2 ]dω
2π −∞
Z ∞
1
T RX (0) = E[|XT (jω)|2 ]dω
2π −∞
Z ∞
1
RX (0) = lim E[|XT (jω)|2 ]dω
2π −∞ T →∞
The above equation can be split into two pars as
Z ∞
1
RX (0) = SX (ω)dω
2π −∞
SX (ω ) = lim E[|XT (jω)|2 ]
T →∞

39
1.6. Power Spectral Densities Random-Process

39. A PSD is as shown in Figure 1.2 where the constants are a = 55, b = 5, ωo = 1000 and
ω1 = 100. Calculate values for E[X 2 (t)], the σX
2 and |µ |.
X

S X ( ) a


0 0  1 0 0  1

Figure 1.2

Solution:

a
S X ( ) a

b b


(0  1 ) 0 (0  1 ) 0 0  1 0 0  1

Figure 1.3

ωo − ω1 = 1000 − 100 = 900

ωo + ω1 = 1000 + 100 = 1100
(ωo + ω1 ) − (ωo − ω1 ) = 1100 − 1000 = 200

The given spectrum is in the triangular form, its base is 200 and its height is 5 and it also has a = 55.
The modified two sided spectrum is redrawn and is as shown in Figure 1.3. The x axis is ω radians/sec.
The Autocorrelation function is

RX (0) = E[X 2 (t)] = Area under P SD = Area of impulse + Area of T riangle

1
= 2× [Area of impulse + Area of T riangle]
2π Z
∞
2 1
= σX = SX (ω)dω
2π −∞
 
1
= 55 + (5)(200)
2π 2
= 176.6620
|µX | = 0

40. A PSD is as shown in Figure 1.2 where the constants are a = 450, b = 6, ωo = 10, 000 and

ω1 = 1000. Calculate values for E[X 2 (t)], the σX

2 and |µ |.
X

Solution:
The given spectrum has the following details constant a = 450

ωo − ω1 = 10, 000 − 1000 = 9900

ωo + ω1 = 10000 + 1000 = 11000
(ωo + ω1 ) − (ωo − ω1 ) = 11000 − 9000 = 2000

The signal magnitude is b = 6

40
1.6. Power Spectral Densities Random-Process

Autocorrelation function is
Z ∞
1
RX (0) = E[X 2 (t(] = σX2
= SX (ω)dω
2π −∞
 
1
= 450 + (6)(2000)
2π 2
= 2053.0988
|µX | = 0

41. A PSD is as shown in Figure 1.2 where the constants are a = 72, b = 4, ωo = 1000 and
ω1 = 50. Calculate values for E[X 2 (t)], the σX
2 and |µ |.
X

Solution:
The given spectrum has the following details constant a = 72

ωo − ω1 = 1000 − 50 = 950
ωo + ω1 = 1000 + 50 = 1050
(ωo + ω1 ) − (ωo − ω1 ) = 1050 − 950 = 100

The signal magnitude is b = 4

Autocorrelation function is
Z ∞
2 1 2
RX (0) = E[X (t(] = = σX
SX (ω)dω
2π −∞
 
1
= 72 + (4)(100)
2π 2
= 86.5803
|µX | = 0

42. A PSD is as shown in Figure 1.4 where the constants are a = 3, b = 5, ω1 = 8 and ω2 = 12.
Calculate values for RX (0), the variance and µ2X .

S X ( )


2  1 0 1 2

Figure 1.4

Solution:

ω1 = 8
ω2 = 12
ω2 − ω1 = 12 − 8 = 4

41
1.6. Power Spectral Densities Random-Process

The given spectrum is in the rectangular form, its base is 4 and its height is 5 and it also has a = 3 at
ω = 0. The x axis is ω radians/sec. The Autocorrelation function is

RX (0) = E[X 2 (t)] = Area under P SD = Area of impulse + Area of Rectangle

1
= [Area of impulse + Area of Rectangle]
2π Z ∞
2 2 1
= µX + σ X = SX (ω)dω
2π −∞
1
= (3 + 2[(5)(12 − 8)])
2π
= 6.8437
2 3
µX = = 0.4775
2π
2
σX = RX (0) − µ2X = 6.8437 − 0.4775
= 6.3662

43. A PSD is as shown in Figure 1.4 where the constants are a = 5, b = 3, ω1 = 7 and ω2 = 13.
Calculate values for RX (0), the variance and µ2X .

Solution:
The given spectrum has the following details constant a = 5

ω1 = 7
ω2 = 13
ω2 − ω1 = 13 − 7 = 6

The signal magnitude is b = 3

Autocorrelation function is
Z ∞
1
RX (0) = µ2X + 2
σX = SX (ω)dω
2π −∞
1
= (5 + 2(3)(13 − 7))
2π
= 6.5254
3
µ2X = = 0.7958
2π
2
σX = RX (0) − µ2X = 6.5254 − 0.7958
= 5.7296

44. A PSD is as shown in Figure 1.4 where the constants are a = 4, b = 3, ω1 = 9 and ω2 = 14.
Calculate values for RX (0), the variance and µ2X .

Solution:
The given spectrum has the following details constant a = 4

ω1 = 9
ω2 = 14
ω2 − ω1 = 14 − 9 = 5

The signal magnitude is b = 3

42
1.6. Power Spectral Densities Random-Process

Autocorrelation function is
Z ∞
1
RX (0) = µ2X + σX
2
= SX (ω)dω
2π −∞
1
= (4 + 2(3)(14 − 9))
2π
= 5.4113
3
µ2X = = 0.6366
2π
2
σX = RX (0) − µ2X = 5.4113 − 0.6366
= 4.7746

45. A PSD is as shown in Figure 1.5 where the constants are a = 5, ωo = 100 and w = 8.
Calculate values for RX (0), the variance and |µX |.

S X ( )

a 2w


0 0 0

Figure 1.5

Solution:
The given spectrum has the following details

w = 8
2w = 16

The signal magnitude is a = 5

Autocorrelation function is
Z ∞
2 1
RX (0) = σX = SX (ω)dω
2π −∞
2
= (5(16))
2π
= 25.4648
µX = 0

46. A PSD is as shown in Figure 1.5 where the constants are a = 3, ωo = 150 and w = 7.
Calculate values for RX (0), the variance and |µX |.

Solution:
The given spectrum has the following details

w = 7
2w = 14

The signal magnitude is a = 3

43
1.6. Power Spectral Densities Random-Process

Autocorrelation function is
Z ∞
2 1
RX (0) = σX = SX (ω)dω
2π −∞
2
= (3(14))
2π
= 13.3690
µX = 0

47. A PSD is as shown in Figure 1.6 where the constants are a = 4, ωo = 125 and w = 6.
Calculate values for RX (0), the variance and |µX |.

Solution:
The given spectrum has the following details

w = 6
2w = 12

The signal magnitude is a = 3

Autocorrelation function is
Z ∞
2 1
RX (0) = σX = SX (ω)dω
2π −∞
2
= (4(12))
2π
= 15.2789
µX = 0

48. A PSD is as shown in Figure 1.6 where the constants are a = 300, b = 10, ωM = 100.
Calculate values for RX (0), the variance and |µX |.

S X ( )


M 0 M

Figure 1.6

Solution:
The given spectrum has the following details

−ωM + ωM = 100 + 100 = 200

The signal magnitude is a = 300, b = 10

44
1.6. Power Spectral Densities Random-Process

Autocorrelation function is
Z ∞
1
RX (0) = µ2X + 2
σX = SX (ω)dω
2π −∞
1
= (300 + (10)(200))
2π
2 2000
σX = = 318.3099
2π
300
µ2X = = 47.7465
2π
|µX | = 6.9099
2 2
σX = RX (0) − σX = 318.3099 − 47.7465 = 366.0564

49. A PSD is as shown in Figure 1.6 where the constants are a = 200, b = 20, ωM = 80.
Calculate values for RX (0), the variance and |µX |.

Solution:
The given spectrum has the following details
−ωM + ωM = 80 + 80 = 160
The signal magnitude is a = 200, b = 20
Autocorrelation function is
Z ∞
1
RX (0) = µ2X + σX
2
= SX (ω)dω
2π −∞
1
= (200 + (20)(160))
2π
= 541.1268
300
µ2X = = 47.7465
2π
|µX | = 6.9099
2
σX = RX (0) − µ2X = 541.1268 − 31.8310 = 509.2958

50. A PSD is as shown in Figure 1.6 where the constants are a = 300, b = 15, ωM = 75.
Calculate values for RX (0), the variance and |µX |.

Solution:
The given spectrum has the following details
−ωM + ωM = 75 + 75 = 150
The signal magnitude is a = 300, b = 15
Autocorrelation function is
Z ∞
1
RX (0) = µ2X + σX
2
= SX (ω)dω
2π −∞
1
= (300 + (15)(150))
2π
= 405.8451
300
µ2X = = 47.7465
2π
|µX | = 6.9099
2
σX = RX (0) − µ2X = 405.8451 − 47.7465 = 358.0986

45
1.7. Weiner-Khinchin Relations Random-Process

RX (τ ) SX (ω)
1 cosωc τ π[δ(ω − ωc ) + δ(ω + ωc )]
2 δ(τ ) 1
2a
3 exp(−a|τ |) a > 0 2 +a2

T (1 − |τ |/T ) − T ≤ τ ≤ T ω 2
sin(ωT /2)
4 T2 ωT /2
0 otherwise 
1 − ωB ≤ ω ≤ ωB
5 (ωB /π) sin(ωBτ)
ωB τ 0 otherwise

1.7 Weiner-Khinchin Relations

Weiner-Khinchin Relation states that the PSD and the autocorrelation function for wide-sense stationary
random process are Fourier transform pair:
Z ∞
SX (ω) = RX (τ )e−jωτ dτ
−∞
Z ∞
1
RX (τ ) = SX (ω)ejωτ dω
2π −∞

Let a rectangular function of time τ be


T (1 − |τ |/T ) − T ≤ τ ≤ T
0 otherwise

51. Use the property of an integrated unit-impulse function. (E.3; see Appendix E) to verify
the property item 1 in table 4.1.

Solution:

SX (ω) = π[δ(ω − ωc ) + δ(ω + ωc )]

Show that

RX (τ ) = cosωc τ

It is given that
Z ∞
g(x)δ(x − x0 )dx = g(x0 )
−∞

Based on the above relation

Z ∞ Z ∞
1 1
SX (ω)ejωt dω = π[δ(ω − ωc ) + δ(ω + ωc )]ejωτ dω
2π −∞ 2π −∞
1  jωc τ
+ e−jωc τ

= e
2
= cosωc τ

52. Use the property of an integrated unit-impulse function. (E.3; see Appendix E) to verify
the property item 2 in table 4.1.

Solution:

SX (ω) = 1

46
1.7. Weiner-Khinchin Relations Random-Process

Show that

RX (τ ) = δ(τ )

It is given that
Z ∞
g(x)δ(x − x0 )dx = g(x0 )
−∞

Based on the above relation

Z ∞ Z ∞
−jωt
RX (τ )e dτ = δ(τ )e−jωτ dω = e0
−∞ −∞
= 1

53. Use the property of an integrated unit-impulse function. (E.3; see Appendix E) to verify
the property item 3 in table 4.1.

RX (τ ) = exp(−a|τ |) a > 0

Show that
2a
SX (ω ) =
ω2 + a2
Solution: It is given that

Z ∞
g(x)δ(x − x0 )dx = g(x0 )
−∞

Based on the above relation

Z 0 Z ∞
−aτ −jωt
SX (ω) = e e dτ + eaτ e−jωt dτ
−∞ −∞
Z 0 Z ∞
−(a+jω)τ
= e(a−jω)τ dτ + e dτ
−∞ 0
" #0 " #∞
e(a−jω)τ e(a+jω)τ
= +
(a − jω) (a + jω)
−∞ 0
1 1 2a
= − = 2
(a − jω) (a + jω) (a + ω 2 )

47
1.7. Weiner-Khinchin Relations Random-Process

Multiplication of Two Random Process

Consider a two random process X(t) and Y (t) have the autocorrelation functions RX (τ ) and RY (τ ),
multiplication of two random process is

W (t) = X(t)Y (t)

The multiplication of autocorrelation functions is

RW (τ ) = RX (τ )RY (τ )

Z ∞
SW (ω) = RX (τ )RY (τ )dτ
−∞
Z ∞  Z ∞ 
1
= RX (τ ) SY (u)e du e−jωτ dτ
juτ
−∞ 2π
Z ∞ Z ∞−∞
1
= SY (u) RX (τ )e−jωτ dτ
2π −∞ −∞
Z ∞
1 1
= SY (u)SX (ω − u)du = SX (ω) ∗ SY (ω)
2π −∞ 2π

56. The random process X(t) and Y (t) have the autocorrelation functions

RX (τ ) = e(−10|τ |) and RY (τ ) = 5cos(600τ )

If Z(t) = X(t)Y (t), and if X(t) and Y (t) are independent , what is the PSD for Z(t) ?

Solution:

RX (τ ) = e(−10|τ |)

20
SX (ω ) =
ω2 + 100

RY (τ ) = 5cos(600τ )

SY (ω) = 5π[δ(ω − 600) + δ(ω + 600)]

1
SZ (ω ) = SX (ω)SY (ω)
2π Z
∞
1 20
= 5π[δ(ω − u − 600) + δ(ω − u + 600)] 2 du
2π −∞ u + 100
 
1 1
= 50 +
(ω − 600)2 + 100 (ω + 600)2 + 100

57. The random process X(t) and Y (t) have the autocorrelation functions

RX (τ ) = e(−10|τ |) and RY (τ ) = 6cos(400τ )

If Z(t) = X(t)Y (t), and if X(t) and Y (t) are independent , what is the PSD for Z(t) ?

Solution:

48
1.7. Weiner-Khinchin Relations Random-Process

RX (τ ) = e(−10|τ |)

20
SX (ω ) =
ω2 + 100

RY (τ ) = 6cos(400τ )

SY (ω) = 6π[δ(ω − 400) + δ(ω + 400)]

1
SZ (ω ) = SX (ω)SY (ω)
2π Z
∞
1 20
= 6π[δ(ω − u − 400) + δ(ω − u + 400)] 2 du
2π −∞ u + 100
 
1 1
= 60 +
(ω − 400)2 + 100 (ω + 400)2 + 100

58. The random process X(t) and Y (t) have the autocorrelation functions

RX (τ ) = e(−10|τ |) and RY (τ ) = 7cos(500τ )

If Z(t) = X(t)Y (t), and if X(t) and Y (t) are independent , what is the PSD for Z(t) ?

Solution:

RX (τ ) = e(−10|τ |)

20
SX (ω ) =
ω 2 + 100

RY (τ ) = 7cos(500τ )

SY (ω) = 7π[δ(ω − 500) + δ(ω + 500)]

1
SZ (ω ) = SX (ω)SY (ω)
2π Z
∞
1 20
= 7π[δ(ω − u − 500) + δ(ω − u + 500)] 2 du
2π −∞ u + 100
 
1 1
= 70 +
(ω − 500) + 100 (ω + 500)2 + 100
2

59. A bandlimited wide-sense stationary random process X(t) has the PSD

7cos(πω/2ωM ) − ωM ≤ ω ≤ ωM
SX (ω) =
0 otherwise

A carrier random process C(t) which is independent of X(t), is

√
C(t) = 60cos(ωC t + Θ)

49
1.7. Weiner-Khinchin Relations Random-Process

where ωC  ωM , and where Θ is a random variole uniformly distributed between ±π. If

Y (t) = X(t)C(t), what is the PSD for Y (t) ?

Solution:

√
C(t) = 60cos(ωC t + Θ)
RC (τ ) = 30cos(ωC τ )
SC (ω) = 30π[δ(ω − ωC ) + δ(ω + ωC )]
SX (ω) = 7cos(πuω/2ωM )

1
SY (ω ) = SX (ω) ? SC (ω)
2π Z
∞
1
= 7cos(πu/2ωM )30π[δ(ω − u − ωC ) + δ(ω − u + ωC )]du
2π −∞
= 105cos(π(ω − ωC )/2ωM ) + 105cos(π(ω + ωC )/2ωM )

60. A bandlimited wide-sense stationary random process X(t) has the PSD

5cos(πω/2ωM ) − ωM ≤ ω ≤ ωM
SX (ω) =
0 otherwise

A carrier random process C(t) which is independent of X(t), is

√
C(t) = 200cos(ωC t + Θ)

where ωC  ωM , and where Θ is a random variole uniformly distributed between ±π. If

Y (t) = X(t)C(t), what is the PSD for Y (t) ?

Solution:

√
C(t) = 200cos(ωC t + Θ)
RC (τ ) = 100cos(ωC τ )
SC (ω) = 100π[δ(ω − ωC ) + δ(ω + ωC )]
SX (ω) = 5cos(πuω/2ωM )

1
SY (ω ) = SX (ω) ? SC (ω)
2π Z
∞
1
= 5cos(πu/2ωM )100π[δ(ω − u − ωC ) + δ(ω − u + ωC )]du
2π −∞
= 250cos(π(ω − ωC )/2ωM ) + 250cos(π(ω + ωC )/2ωM )

61. A bandlimited wide-sense stationary random process X(t) has the PSD

6cos(πω/2ωM ) − ωM ≤ ω ≤ ωM
SX (ω) =
0 otherwise

A carrier random process C(t) which is independent of X(t), is

√
C(t) = 34cos(ωC t + Θ)

50
1.7. Weiner-Khinchin Relations Random-Process

where ωC  ωM , and where Θ is a random variole uniformly distributed between ±π. If

Y (t) = X(t)C(t), what is the PSD for Y (t) ?

Solution:

√
C(t) = 34cos(ωC t + Θ)
RC (τ ) = 17cos(ωC τ )
SC (ω) = 17π[δ(ω − ωC ) + δ(ω + ωC )]
SX (ω) = 6cos(πuω/2ωM )

1
SY (ω ) = SX (ω) ? SC (ω)
2π Z
∞
1
= 6cos(πu/2ωM )17π[δ(ω − u − ωC ) + δ(ω − u + ωC )]du
2π −∞
= 51cos(π(ω − ωC )/2ωM ) + 51cos(π(ω + ωC )/2ωM )

62. The wide sense stationary random process X(t) has a PSD that is a constant 5 × 10−6 V 2
when |f | < 1kHz and is 0 otherwise. The random process Y (t) = 10cos(ωo t + Θ) V where
fO = 100KHz and Θ is uniformly distributed between ±π. X(t) and Y (t) are independent
Z(t) = X(t)Y (t). What is the PSD for Z(t)?

Solution:

Y (t) = 10cos(ωo t + Θ)
RY (τ ) = 50cos(ωo τ )
SY (ω) = 50π[δ(ω − ωo ) + δ(ω + ωo )]

Z ∞
1
SZ (ω) = SX (u) ? SY (ω)
2π −∞
Z ∞
1
= SX (u)SY (ω − u)du
2π −∞
Z ∞
1
= 5 × 10−6 × 50π[δ(ω − u − ωo ) + δ(ω − u + ωo )]du
2π −∞
= 125 × 10−6 V 4 | ± ωo − ω| < 2π × 103

63. The wide sense stationary random process X(t) has a PSD that is a constant 7 × 10−6 V 2
when |f | < 0.9kHz and is 0 otherwise. The random process Y (t) = 8cos(ωo t + Θ) V where
fO = 100KHz and Θ is uniformly distributed between ±π. X(t) and Y (t) are independent
Z(t) = X(t)Y (t). What is the PSD for Z(t)?

Solution:

Y (t) = 8cos(ωo t + Θ)
RY (τ ) = 32cos(ωo τ )
SY (ω) = 32π[δ(ω − ωo ) + δ(ω + ωo )]

51
1.7. Weiner-Khinchin Relations Random-Process

Z ∞
1
SZ (ω ) = SX (u) ? SY (ω)
2π −∞
Z ∞
1
= SX (u)SY (ω − u)du
2π −∞
Z ∞
1
= 7 × 10−6 × 32π[δ(ω − u − ωo ) + δ(ω − u + ωo )]du
2π −∞
= 112 × 10−6 V 4 | ± ωo − ω| < 2π × 103

64. The wide sense stationary random process X(t) has a PSD that is a constant 6 × 10−6 V 2
when |f | < 1.1kHz and is 0 otherwise. The random process Y (t) = 9cos(ωo t + Θ) V where
fO = 100KHz and Θ is uniformly distributed between ±π. X(t) and Y (t) are independent
Z(t) = X(t)Y (t). What is the PSD for Z(t) ?

Solution:

Y (t) = 9cos(ωo t + Θ)
RY (τ ) = 40.5cos(ωo τ )
SY (ω) = 40.5π[δ(ω − ωo ) + δ(ω + ωo )]

Z ∞
1
SZ (ω ) = SX (u) ? SY (ω)
2π −∞
Z ∞
1
= SX (u)SY (ω − u)du
2π −∞
Z ∞
1
= 6 × 10−6 × 40.5π[δ(ω − u − ωo ) + δ(ω − u + ωo )]du
2π −∞
= 121.5 × 10−6 V 4 | ± ωo − ω| < 2π × 103

52
1.8. Linear Systems: Random-Process

1.8 Linear Systems:

Consider a signal x(t) is passed through a filter with h(t), then the output of the filter y(t) is expressed as
Z ∞
y(t) = h(u)x(t − u)du
0

where h(t) is the impulse response of the system. Also above mentioned equation represents the convolution
operation between input signal x(t) with h(t). The laplace transform the above system is

Y (s) = H(s)X(s)

where X(s) and Y (s) are the Laplace transforms of the input signal x(t) and output t(t) signals.
The above relation is applied for the random process X(t) and which expressed as
Z ∞
y(t) = h(u)X(t − u)du
0

1.8.1 The mean of Y (t):

The mean of the output random process is
Z ∞
µY = E[Y (t] = h(u)E[X(t − u)]du
0

Z ∞
µY = E[Y (t] = µX h(u)du
0

The Laplace transform of h(t)

Z ∞
H(s) = h(t)e−st dt
0

µY = E[Y (t] = µX H(0)

1.8.2 Cross-Correlating Y (t) and X(t):

Consider a two random process X(t) and Y (t) are wide sense stationary. When we consider both the
random process it is called as jointly wide sense stationary. Then their cross-correlation function is defined
as
Z ∞
RXY (τ ) = E[X(t)Y (t + τ )] = h(u)E[X(t)X(t + τ − u)]du
0

Z ∞
RXY (τ ) = h(u)RX (τ − u)du
0

Z ∞ Z ∞
SXY (jω) = h(u) RX (τ − u)e−jωτ dτ du
0 −∞

Let v = τ − u
Z ∞ Z ∞
−jωu
SXY (jω) = h(u)e RX (v)e−jωv dv
0 −∞
= H(jω)SX (ω)

53
1.8. Linear Systems: Random-Process

1.8.3 Autocorrelation of Y (t):

RY (τ ) = E[Y (t)Y (t + τ )]
Z ∞ Z ∞
= h(v) h(u)E[X(t − v)X(t + τ − u)dudv
0 0
Z ∞ Z ∞
= h(v) h(u)RX (τ + v − u)dudv
0 0

Z ∞ Z ∞ Z ∞
SY (ω) = h(v) h(u) RX (τ + v − u)e−jωτ dτ dudv
0 0 −∞

Let w = τ + v − u
Z ∞ Z ∞ Z ∞
−jωu
SY (ω) = h(v)e jωv
dv h(u)e RX (w)e−jωw dw
0 0 −∞
= H(−jω)H(jω)SX (ω)
= |H(jω)|2 SX (ω)

54
1.8. Linear Systems: Random-Process

65. Suppose in a given application

 aKe−aτ τ ≥ 0


RXY (τ ) =
0 τ ≤0


aK
SXY (jω ) =
jω + a
What are RY X (τ ) and SY X (jω) in this case? Interpret your results. [?]

Solution:
We Know that

RY X (τ ) = RXY (−τ )

 aKeaτ τ ≥ 0


RY X (τ ) =
0 τ ≤0


And also
aK
SY X (jω) = SY∗ X (jω) =
−jω + a

66. Suppose that the PSD input to a linear system is SX (ω) = K. The cross-correlation of
the input X(t) with the output Y (t) of the linear system is found to be
 −τ
 e + 3e−2τ τ ≥ 0
RXY (τ ) = K
0 τ ≤0


What is the power filter function |H(jω)|2 ? . [?]

Solution:
If SX (ω) = K then

RX (τ ) = Kδ(τ )

Z ∞
RXY (τ ) = h(u)RX (τ − u)du
0

Z ∞
RXY (τ ) = K h(u)δ(τ − u)du
0
= Kh(τ )

Similarly
 −t
 e + 3e−2t t ≥ 0
h(t) = K
0 t≤0


1 3 (s + 2) + 3(s + 1) 4s + 5
H(s) = + = = 2
s+1 s+2 (s + 1)(s + 2) s + 3s + 2

55
1.8. Linear Systems: Random-Process

j4ω + 5
H(jω ) =
−ω 2 + 3jω + 2

16ω 2 + 25
|H(jω)|2 =
ω 4 + 5ω 2 + 4

67. Suppose that the PSD input to a linear system is SX (ω) = K. The cross-correlation of
the input X(t) with the output Y (t) of the linear system is found to be
 −τ
 3e + e−2τ τ ≥ 0
RXY (τ ) = K
0 τ ≤0


What is the power filter function |H(jω)|2 ? . [?]

Solution:
If SX (ω) = K then

RX (τ ) = Kδ(τ )

Z ∞
RXY (τ ) = h(u)RX (τ − u)du
0

Z ∞
RXY (τ ) = K h(u)δ(τ − u)du
0
= Kh(τ )

Similarly
 −t
 3e + e−2t t ≥ 0
h(t) = K
0 t≤0


3 1 3(s + 2) + (s + 1) 4s + 7
H(s) = + = = 2
s+1 s+2 (s + 1)(s + 2) s + 3s + 2

j4ω + 7
H(jω ) =
−ω 2 + 3jω + 2

16ω 2 + 49
|H(jω)|2 =
ω 4 + 5ω 2 + 4

68. Suppose that the PSD input to a linear system is SX (ω) = K. The cross-correlation of
the input X(t) with the output Y (t) of the linear system is found to be
 −2τ
 2e + 3e−τ τ ≥ 0
RXY (τ ) = K
0 τ ≤0


What is the power filter function |H(jω)|2 ? . [?]

56
1.8. Linear Systems: Random-Process

Solution:
If SX (ω) = K then

RX (τ ) = Kδ(τ )

Z ∞
RXY (τ ) = h(u)RX (τ − u)du
0

Z ∞
RXY (τ ) = K h(u)δ(τ − u)du
0
= Kh(τ )

Similarly
 −t
 2e + 3e−t t ≥ 0
h(t) = K
0 t≤0


2 3 2(s + 1) + 3(s + 2) 5s + 8
H(s) = + = = 2
s+2 s+1 (s + 1)(s + 2) s + 3s + 2

j5ω + 8
H(jω ) =
−ω 2 + 3jω + 2

25ω 2 + 64
|H(jω)|2 =
ω 4 + 5ω 2 + 4

69. The PSD of the random process X(t) and the transfer function of a network are
1 s
SX (ω) = and H(s) =
ω2 + (100) 2 (s + 10)(s + 9000)

Y (s) = H(s)X(s). Find µY , SXY (jω) and SY (ω) . [?]

Solution:
SX (ω) doesn’t have any dc component (unit impulse function) when ω = 0, hence

µX = 0

Also the relation is

µY = µX H(0) = 0

SXY (jω ) = H(jω)SX (ω)

jω
=
(ω 2 + (100)2 )(jω + 10)(jω + 9000)

SY (ω ) = |H(jω)|2 SX (ω)
ω2
=
(ω 2 + (100)2 )(ω 2 + (10)2 )(ω 2 + (9000)2 )

57
1.8. Linear Systems: Random-Process

70. The PSD of the random process X(t) and the transfer function of a network are
1 s
SX (ω) = and H(s) =
ω2 + (80) 2 (s + 9)(s + 10000)

Y (s) = H(s)X(s). Find µY , SXY (jω) and SY (ω) . [?]

Solution:
SX (ω) doesn’t have any dc component (unit impulse function) when ω = 0, hence

µX = 0

Also the relation is

µY = µX H(0) = 0

SXY (jω ) = H(jω)SX (ω)

jω
=
(ω 2 + (80)2 )(jω + 9)(jω + 10000)

SY (ω ) = |H(jω)|2 SX (ω)
ω2
=
(ω 2 + (80)2 )(ω 2 + (9)2 )(ω 2 + (10000)2 )

71. The PSD of the random process X(t) and the transfer function of a network are
1 s
SX (ω) = and H(s) =
ω2 + (70) 2 (s + 10)(s + 11000)

Y (s) = H(s)X(s). Find µY , SXY (jω) and SY (ω) . [?]

Solution:
SX (ω) doesn’t have any dc component (unit impulse function) when ω = 0, hence

µX = 0

Also the relation is

µY = µX H(0) = 0

SXY (jω ) = H(jω)SX (ω)

jω
=
(ω 2 + (70)2 )(jω + 10)(jω + 11000)

SY (ω ) = |H(jω)|2 SX (ω)
ω2
=
(ω 2 + (70)2 )(ω 2 + (10)2 )(ω 2 + (11000)2 )

58