# 1

Overview of Random Processes
2
GOAL
This fundamental course is concerned with the statistical
characterization of random signals.
EE 301
(deterministic signals)
(no randomness)
Signals & Systems
EE 306
(random signals)
• Deterministic processes :
physical process is represented by explicit mathematical relation
• Random processes :
result of a large number of separate causes. Described in
probabilistic terms and by properties which are averages.
3
Let denote the random outcome of an experiment.
To every such outcome, suppose a waveform
is assigned.
The collection of such
waveforms form a
stochastic process.
For fixed (the set of
all experimental outcomes),
is a specific time function.
For fixed t,
is a random variable.
The ensemble of all such realizations
over time represents the stochastic process X(t).
ξ
) , ( ξ t X
S
i
∈ ξ
) , (
1 1 i
t X X ξ =
) , ( ξ t X
t
1
t
2
t
) , (
n
t X ξ
) , (
k
t X ξ
) , (
2
ξ t X
) , (
1
ξ t X
M
M
M
) , ( ξ t X
0
) , ( ξ t X
Random (stochastic) Processes
4
Random (stochastic) Processes
For example
where is a uniformly distributed random variable in
represents a stochastic process.
Stochastic processes are everywhere:
Noise, detection and classification problems, pattern recognition,
stock market fluctuations, various queuing systems
all represent stochastic phenomena.
), cos( ) (
0
ϕ ω + = t a t X
ϕ
(0, 2 ), π
5
Random (stochastic) Processes
If X(t) is a stochastic process, then for fixed t, X(t) represents
a random variable. Its distribution function (cdf) is given by
Notice that depends on t, since for a different t, we obtain
a different random variable. Further
represents the first-order probability density function (pdf) of the
process X(t).
} ) ( { ) , ( x t X P t x F
X
≤ =
) , ( t x F
X
dx
t x dF
t x f
X
X
) , (
) , ( =

6
Random (stochastic) Processes
For t = t
1
and t = t
2
, X(t) represents two different random variables
X
1
= X(t
1
) and X
2
= X(t
2
) respectively. Their joint distribution is
given by
and
represents the second-order density function of the process X(t).
Similarly represents the n
th
order density
function of the process X(t). Complete specification of the stochastic
process X(t) requires the knowledge of
for all and for all n. (an almost impossible task
in reality).
} ) ( , ) ( { ) , , , (
2 2 1 1 2 1 2 1
x t X x t X P t t x x F
X
≤ ≤ =
) , , , , , (
2 1 2 1 n n
t t t x x x f
X
L L
) , , , , , (
2 1 2 1 n n
t t t x x x f
X
L L
n i t
i
, , 2 , 1 , L =
2
1 2 1 2
1 2 1 2
1 2
( , , , )
( , , , )

X
X
F x x t t
f x x t t
x x

=
∂ ∂

7
Random (stochastic) Processes
Mean of a Stochastic Process:
represents the mean value of a process X(t). In general, the mean of
a process can depend on the time index t.
Autocorrelation function of a process X(t) is defined as
and it represents the interrelationship between the random variables
X
1
= X(t
1
) and X
2
= X(t
2
) generated from the process X(t).
Properties:
1.
2.
*
1 2 2 1
( , ) ( , )
XX XX
R t t R t t =
. 0 } | ) ( {| ) , (
2
> = t X E t t R
XX
(Average instantaneous power)

( ) { ( )} ( , )
X
t E X t x f x t dx µ
+∞
−∞
= =

* *
1 2 1 2 1 2 1 2 1 2 1 2
( , ) { ( ) ( )} ( , , , )
XX X
R t t E X t X t x x f x x t t dx dx = =
∫ ∫

8
Random (stochastic) Processes
The function
represents the autocovariance function of the process X(t).
) ( ) ( ) , ( ) , (
2
*
1 2 1 2 1
t t t t R t t C
X X XX XX
µ µ − =
Similarly
0
2
0
0
( ) { ( )} {cos( )}
1
cos( ) 0.
2
X
t E X t aE t
t d
π
µ ω ϕ
ω ϕ ϕ
π
= = +
= + =

). ( cos
2
)} 2 ) ( cos( ) ( {cos
2
)} cos( ) {cos( ) , (
2 1 0
2
2 1 0 2 1 0
2
2 0 1 0
2
2 1
t t
a
t t t t E
a
t t E a t t R
XX
− =
+ + + − =
+ + =
ω
ϕ ω ω
ϕ ω ϕ ω
Example ). 2 , 0 ( ~ ), cos( ) (
0
π ϕ ϕ ω U t a t X + =
9
Stationary Random Processes
Stationary processes exhibit statistical properties that are
invariant to shift in the time index.
Thus, for example, second-order stationarity implies that
the statistical properties of the pairs
{X(t
1
) , X(t
2
) } and {X(t
1
+c) , X(t
2
+c)} are the same for any c.
Similarly first-order stationarity implies that the statistical
properties of X(t
i
) and X(t
i
+c) are the same for any c.
10
Stationary Random Processes
In strict terms, the statistical properties are governed by the
joint probability density function. Hence a process is n
th
-order
Strict-Sense Stationary (S.S.S) if
for any c, where the left side represents the joint density function of
the random variables and
the right side corresponds to the joint density function of the random
variables
A process X(t) is said to be strict-sense stationary if above eqn. is
true for all
) , , , , , ( ) , , , , , (
2 1 2 1 2 1 2 1
c t c t c t x x x f t t t x x x f
n n n n X X
+ + + ≡ L L L L
) ( , ), ( ), (
2 2 1 1 n n
t X X t X X t X X = = = L
). ( , ), ( ), (
2 2 1 1
c t X X c t X X c t X X
n n
+ =

+ =

+ =

L
. and , 2 , 1 , , , 2 , 1 , c any n n i t
i
L L = =
11
Stationary Random Processes
For a first-order strict sense stationary process,
for any c.
In particular c = – t gives
i.e., the first-order density of X(t) is independent of t.
In that case
) , ( ) , ( c t x f t x f
X X
+ ≡
) ( ) , ( x f t x f
X X
=

[ ( )] ( ) , E X t x f x dx a constant. µ
+∞
−∞
= =

12
Stationary Random Processes
i.e., the second order density function of a SSS process depends only
on the difference of the time indices
In that case, the autocorrelation function is given by
i.e., it depends only on the difference of the time indices.
2 1
. t t τ − =
*
1 2 1 2
*
1 2 1 2 2 1 1 2
*
2 1
( , ) { ( ) ( )}
( , , )
( ) ( ) ( ),
XX
X
XX XX XX
R t t E X t X t
x x f x x t t dx dx
R t t R R
τ
τ τ
=
= = −
= − = = −
∫ ∫

Similarly, for a second-order strict-sense stationary process
for any c. For c = – t
1
we get
) , , , ( ) , , , (
2 1 2 1 2 1 2 1
c t c t x x f t t x x f
X X
+ + ≡
1 2 1 2 1 2 2 1
( , , , ) ( , , )
X X
f x x t t f x x t t ≡ −
13
Stationary Random Processes
The basic conditions for the first and second order stationarity are
usually difficult to verify.
In that case, we often resort to a looser definition of stationarity,
known as Wide-Sense Stationarity (W.S.S).
A process X(t) is said to be Wide-Sense Stationary if
(i)
and
(ii)
i.e., the mean is a constant and the autocorrelation function
depends only on the difference between the time indices.
µ = )} ( { t X E
*
1 2 2 1
{ ( ) ( )} ( ),
XX
E X t X t R t t = −
14
Stationary Random Processes
Remarks:
1. Notice that these conditions do not say anything about the
nature of the probability density functions, and instead deal
with the average behavior of the process.
2. Strict-sense stationarity always implies wide-sense
stationarity.
SSS WSS
However, the converse is not true in general, the only exception
being the Gaussian process.
If X(t) is a Gaussian process, then wide-sense stationarity (w.s.s)
strict-sense stationarity (s.s.s).
15
Ergodic Random Processes
If almost every member of the ensemble shows the same statistical
behavior as the whole ensemble, then it is possible to determine the
statistical behavior by examining only one typical sample function.
⇒ Ergodic process
For ergodic process, the mean values and autocorrelation function
can be determined by time averages as well as by ensemble
averages, that is,
Ergodic in the mean process:
Ergodic in the autocorrelation process:
These conditions can exist if the process is stationary.
→ ergodic stationary (not vice versa)
{ }
1
( ) ( )
2
lim
T
T
T
E X t X t dt
T

→∞
=

( )
{ }
* *
1
( ) ( ) ( ) ( )
2
lim
T
XX
T
T
R E X t X t X t X t dt
T
τ τ τ

→∞
= + = +

16
Power Spectral Density
• Power spectrum of X(t)
• Autocorrelation of X(t)
• Power spectrum and Autocorrelation function
are Fourier transform pair
• Total average power=
2
( ) [ ( )] ( )
j f
x x x
S f F R R e d
π τ
τ τ τ

−∞
= =

1 2
( ) [ ( )] ( )
j f
x x x
R F S f S f e df
π τ
τ

−∞
= =

(0) ( )
x x
R S f df

−∞
=

17
Random Processes as Inputs/Outputs to LTI Sytems
A deterministic system transforms each input waveform into
an output waveform by operating only on the
time variable t.
Thus a set of realizations at the input corresponding to a process X(t)
generates a new set of realizations at the
output associated with a new process Y(t).
) , (
i
t X ξ
)] , ( [ ) , (
i i
t X T t Y ξ ξ =
)} , ( { ξ t Y
Our goal is to study the output process statistics in terms of the input
process statistics and the system function.
] [⋅ T
 → 
) (t X
 → 
) (t Y
t t
) , (
i
t X ξ
) , (
i
t Y ξ
18
Random Processes as Inputs/Outputs to LTI Sytems
Linear Systems: represents a linear system if
Let represent the output of a linear system.
Time-Invariant System: represents a time-invariant system if
i.e., shift in the input results in the same shift in the output also.
If satisfies both, then it corresponds to a linear time-invariant (LTI)
system.
] [⋅ L
)} ( { ) ( t X L t Y =
)}. ( { )} ( { )} ( ) ( {
2 2 1 1 2 2 1 1
t X L a t X L a t X a t X a L + = +
] [⋅ L
) ( )} ( { )} ( { ) (
0 0
t t Y t t X L t X L t Y − = − ⇒ =
] [⋅ L
19
Random Processes as Inputs/Outputs to LTI Sytems
LTI

∞ +
∞ −
∞ +
∞ −
− =
− =

) ( ) (
) ( ) ( ) (
τ τ τ
τ τ τ
d t X h
d X t h t Y
arbitrary
input
t
) (t X
t
) (t Y
) (t X ) (t Y
LTI systems can be uniquely represented in terms of their output
to a delta function
LTI ) (t δ ) (t h
Impulse
Impulse
response of
the system
t
) (t h
Impulse
response
then
20
Random Processes as Inputs/Outputs to LTI Sytems
Output Statistics: The mean of the output process is given by
). ( ) ( ) ( ) (
} ) ( ) ( { )} ( { ) (

t h t d t h
d t h X E t Y E t
X X
Y
∗ = − =
− = =

∞ +
∞ −
∞ +
∞ −
µ τ τ τ µ
τ τ τ µ
h(t)
) (t
X
µ ) (t
Y
µ
In particular if X(t) is wide-sense stationary, then we have
so that
X X
t µ µ = ) (
constant. a c d h t
X X Y
, ) ( ) (

µ τ τ µ µ = =

∞ +
∞ −
21
Random Processes as Inputs/Outputs to LTI Sytems
Output Statistics: The autocorrelation function of the output is given by
*
( ) ( ) ( ) ( ).
YY XX
R R h h τ τ τ τ = ∗ − ∗
h(τ) h*(-τ)
( )
XY
R τ
→
→  → 
( )
YY
R τ ( )
XX
R τ
h*(-τ) h(τ)
( )
YX
R τ
→
→  → 
( )
YY
R τ
( )
XX
R τ
Thus, Y(t) is w.s.s process.
X(t) and Y(t) are jointly w.s.s.
LTI system
h(t)
wide-sense
stationary process
wide-sense
stationary process.
) (t X
) (t Y
22
Random Processes as Inputs/Outputs to LTI Sytems
Output Statistics: The power spectral density function of the output is
2
*
( ) ( ) ( ) ( ) ( ) ( )
YY XX XX
S f S f H f S f H f H f = =
H(f) H*(f)
( )
XY
S f
→
→  → 
( )
YY
S f ( )
XX
S f
H*(f) H(f)
( )
YX
S f
→
→  → 
( )
YY
S f
( )
XX
S f
{ }
{ }
*
( ) ( ) ( ) ( )
XY XY
S f F R F E X t Y t τ τ

= = +

{ }
{ }
*
( ) ( ) ( ) ( )
YX YX
S f F R F E Y t X t τ τ

= = +

23
White Noise
• A random process X(t) is called a white process if it
has a flat power spectrum.
– If S
x
(f) is constant for all f
• It closely represents thermal noise
f
Sx(f)
The area is infinite
(Infinite power !)
0
( )
2
n
N
S f =
24
White Noise
• Autocorrelation
2 2 0 0
( ) ( ) ( )
2 2
j f j f
x x
N N
R S f e df e df
π τ π τ
τ δ τ
∞ ∞
−∞ −∞
= = =
∫ ∫
Sn(f)
N
0
/2
f
N
0
/2
Rx(τ)
τ
Rx(τ)=0 if τ=t
2
-t
1
≠0
X(t
1
) and X(t
2
) are uncorrelated if t
1
≠ t
2
25
White Gaussian Noise
• The sampled random variables will be statistically
independent Gaussian random variables
Sn(f)
N
0
/2
f
N
0
/2
Rx(τ)
τ
τ=0
τ≠0
26
Poisson Random Process
Poisson random variable:
L , 2 , 1 , 0 ,
! " duration of interval
an in occur arrivals "
= =
)
`
¹
¹
´
¦

k
k
e
k
P
k
λ
λ
∆ =

⋅ = = µ µ λ
T
T np

0 T
43 42 1
arrivals k
27
Poisson Random Process
Definition: X(t) = n(0, t) represents a Poisson process if
the number of arrivals n(t
1
, t
2
) in an interval (t
1
, t
2
) of length
t = t
2
– t
1
is a Poisson random variable with parameter
Thus
. t λ
1 2 2 1
, , 2 , 1 , 0 ,
!
) (
} ) , ( { t t t k
k
t
e k t t n P
k
t
− = = = =

L
λ
λ
t t n E t X E λ = = )] , 0 ( [ )] ( [
). , min( ) , (
2 1 2 1
2
2 1
t t t t t t R
XX
λ λ + =
X(t) : not WSS
28
Poisson Impulse Process
Although X(t) does not represent a wide sense stationary
process, its derivative does represent a wide sense
stationary process (called Poisson Impulse Process).
) (t X

) (t X ) (t X

dt
d ) (⋅
constant a
dt
t d
dt
t d
t
X
X
,
) (
) ( λ
λ µ
µ = = =

2
1 2 1 2
( ) ( ).
X X
R t , t t t λ λ δ
′ ′
= + −
29
Gaussian Random Process
• A random process X(t) is a Gaussian process if for all
n and for all , the random variables
has a jointly Gaussian density function, which may be
expressed as
– where
1 2
( , , , )
n
t t t K
2
{ ( ), ( ), , ( )}
i n
X t X t X t K
1
/ 2 1/ 2
1 1
( ) exp[ ( ) ( )]
2 (2 ) [det( )]
T
n
f x x m C x m
C π

= − − −
2
[ ( ), ( ), , ( )]
T
i n
x X t X t X t = K
( ) m E X =
{ } (( )( ))
ij i i j j
C c E x m x m = = − −
: n random variables
: mean value vector
: nxn covariance matrix
30
Gaussian Random Process
• Property 1
– For Gaussian process, knowledge of the mean(m)
and covariance(C) provides a complete statistical
description of process
• Property 2
– If a Gaussian process X(t) is passed through a LTI
system, the output of the system is also a
Gaussian process. The effect of the system on
X(t) is simply reflected by the change in mean(m)
and covariance(C) of X(t)