Continuity and Differentiation of Random Processes

We know that the dynamic behavior of a system is described by differential

equations involving input to and output of the system. For example, the behavior
of an RC network is described by a first order linear differential equation with the
source voltage as the input and the capacitor voltage as the output. What happens
if the input voltage to the network is a random process?

Each realization of the random process is a deterministic function and the

concepts of differentiation and integration can be applied to it. Our aim is to
extend these concepts to the ensemble of realizations and apply calculus to
random process.

We discussed the convergence and the limit of a random sequence. The continuity of the
random process can be defined with the help of convergence and limits of a random
process. We can define continuity with probability 1, mean-square continuity, and
continuity in probability etc. We shall discuss the mean-square continuity and the
elementary concepts of corresponding mean-square calculus.

Mean-square continuity of a random process

Recall that a sequence of random variables { X n } converges to a random variable X if
lim E [ X n - X ] = 0
2

and we write
l.i.m. X n = X
n

A random process { X ( t ) } is said to be continuous at a point t = t0 in the mean-square

sense if l.i.m. X ( t ) = X ( t0 ) or equivalently
t t0

X ( t ) - X ( t0 )
2
lim E
t t0 = 0

Mean-square continuity and autocorrelation function

(1) A random process { X ( t)} is MS continuous at t0 if its auto correlation function

RX ( t1 , t2 ) is continuous at (t0 , t0 ).

Proof:

= E ( X ( t ) - 2 X ( t ) X ( t0 ) + X ( t0 ) )
X ( t ) - X ( t0 )
2 2 2
E

= RX ( t , t ) - 2 RX ( t , t0 ) + RX ( t0 , t0 )

If RX ( t1 , t 2 ) is continuous at (t0 , t0 ), then

 X ( t ) - X ( t0 )
= lim RX ( t , t ) - 2 R X ( t , t0 ) + R X ( t 0 , t0 )
2
lim E

t t0 t t0

= RX ( t0 , t0 ) - 2 RX ( t0 , t0 ) + RX ( t0 , t0 )
=0

(2) If { X ( t)} is MS continuous at t0 its mean is continuous at t0 .

This follows from the fact that

(E ) E
( X (t) - X (t ) )
2 2
X ( t) - X ( t )
0 0

2
X ( t ) - X ( t0 ) X ( t ) - X ( t0 )
2
\ lim
E

lim E
= 0
t t0 t t0

\ EX ( t ) is continuous at t0 .

Example
Consider the random binary wave { X (t )} discussed in Example . A typical realization
of the process is shown in Fig. below. The realization is a discontinuous function.

X (t )

K 1
Tp

0
-1
Tp
K
t

The process has the autocorrelation function given by

t
1-
t Tp
RX (t ) = Tp

0
otherwise

We observe that RX (t ) is continuous at t = 0. Therefore, RX (t ) is continuous at all t .

Example For a Wiener process { X ( t ) } ,
RX ( t1 , t2 ) = a min ( t1 , t2 )
where a is a constant.
\ RX ( t , t ) = a min ( t , t ) = a t
Thus the autocorrelation function of a Wiener process is continuous everywhere implying
that a Wiener process is m.s. continuous everywhere. We can similarly show that the
Poisson process is m.s. continuous everywhere.

Mean-square differentiability

The random process { X ( t ) } is said to have the mean-square derivative X ' ( t ) at a point
X ( t + Dt ) - X ( t )
t G, provided approaches X ' ( t ) in the mean square sense as
Dt
Dt 0 . In other words, the random process { X ( t ) } has a m-s derivative X ' ( t ) if
X ( t + Dt ) - X ( t )
2

lim E - X ' ( t ) = 0
Dt 0
Dt

Remark
(1) If all the sample functions of a random process X ( t ) are differentiable, then the
above condition is satisfied and the m-s derivative exists.
Example
Consider the random-phase sinusoid { X (t )} given by

X (t ) = A cos( w0 t + f ) where A and w0 are constants and f ~ U [0, 2p ]. Then for each
f , X (t ) is differentiable. Therefore, the m.s. derivative is

(t ) = - Aw0 sin( w0t + f )

X

M.S. Derivative and Autocorrelation functions

2 RX ( t1 , t2 )
The m-s derivative of a random process X ( t ) at a point t G exists if

t1 t2
exists at the point (t , t ).

Applying the Cauchy criterion, the condition for existence of m-s derivative is

X ( t + Dt1 ) - X ( t ) X ( t + Dt 2 ) - X ( t )
2

lim E - = 0
Dt1 , Dt2 0
Dt1 Dt2
Expanding the square and taking expectation results,

X ( t + Dt1 ) - X ( t ) X ( t + Dt 2 ) - X ( t )
2

lim E -
Dt1 , Dt2 0
Dt1 Dt 2
R ( t + Dt1 , t + Dt1 ) + RX ( t , t ) - 2 RX ( t + Dt1 , t ) R ( t + Dt 2 , t + Dt 2 ) + RX ( t , t ) - 2RX ( t + Dt 2 , t )
= lim X + X
Dt1 ,Dt2 0
Dt12
Dt 2 2
R ( t + Dt1 , t + Dt2 ) - RX ( t + Dt1 , t ) - RX ( t , t + Dt 2 ) + RX ( t , t )
-2X
Dt1Dt2

2 RX ( t1 , t2 )
Each of the above terms within square bracket converges to if the

t1 t2 t =t ,t 1 2 =t

second partial derivative exists.

X ( t + Dt1 ) - X ( t ) X ( t + Dt2 ) - X ( t ) 2 RX ( t1 , t 2 ) 2 R X ( t1 , t 2 ) 2 R X ( t1 , t2 )
2

\ lim E - = + -2
Dt1 , Dt2 0
Dt1 Dt2
t1 t2 t1 t2
t1 t2 t =t1 ,t = t2

=0

2 RX ( t1 , t2 )
Thus, { X ( t ) } is m-s differentiable at t G if exists at (t , t ) G G.

t1 t2

Particularly, if X ( t ) is WSS,

RX ( t1 , t2 ) = RX ( t1 - t2 )

Substituting t1 - t2 = t , we get

2 RX ( t1 , t2 ) 2 RX ( t1 - t2 )
=

t1 t2
t1 t2
dRX ( t ) ( t1 - t2 )
= .
t1 dt
t2
d 2 RX ( t )
t
=-
dt 2

t1
d 2 RX ( t )
=-
dt 2
Therefore, a WSS process X ( t ) is m-s differentiable if RX ( t ) has second derivative at
t = 0.

Example
Consider a WSS process { X ( t ) } with autocorrelation function
RX ( t ) = exp ( - a t )
RX ( t ) does not have the first and second derivative at t = 0. { X ( t ) } is not mean-square
differentiable.
Example The random binary wave { X ( t ) } has the autocorrelation function
t
1-
t Tp
RX (t ) = Tp

0
otherwise

RX ( t ) does not have the first and second derivative at t = 0. Therefore, { X ( t ) } is not
mean-square differentiable.
Example For a Wiener process { X ( t ) } ,
RX ( t1 , t2 ) = a min ( t1 , t2 )
where a is a constant.
a t if t2 < 0

\ RX ( 0, t2 ) = 2
0 other wise

a if t2 < 0

RX ( 0, t2 )
\ = 0 if t2 > 0
t1
does not exist if if t2 = 0
2 RX ( t1 , t2 )
\ does not exist at (t1 = 0, t2 = 0)
t1 t2

Thus a Wiener process is m.s. differentiable nowhere.

Mean and Autocorrelation of the Derivative process

We have,
 X ( t + Dt ) - X ( t )
EX ' ( t ) = E lim
Dt 0 Dt
EX ( t + Dt ) - EX ( t )
= lim
Dt 0 Dt
m ( t + Dt ) - m X ( t )
= lim X
Dt 0 Dt
= mX '( t )

For a WSS process EX ' ( t ) = m X ' ( t ) = 0 as m X ( t ) = constant.

EX ( t1 ) X ' ( t2 ) = RXX ' ( t1 , t2 )

X ( t2 + Dt2 ) - X ( t2 )
= EX ( t1 ) lim
Dt2 0 Dt2
X ( t1 ) X ( t2 + Dt2 ) - X ( t1 ) X ( t 2 )
E

= lim
Dt2 0 Dt2
RX ( t1 , t2 + Dt2 ) - RX ( t1 , t2 )
= lim
Dt2 0 Dt2

RX
= ( t1 , t2 )
t2

Similarly we can show that

2 RXX ( t1 , t2 )
EX ' ( t1 ) X ' ( t2 ) =

t1 t2

For a WSS process

dR ( t )
EX ( t1 ) X ' ( t2 ) = RX ( t1 - t 2 ) = X

t dt

and
2 RX ( t1 - t2 )
( t1 ) X
EX ( t2 ) =

t1 t2
d 2 RX ( t )
=
dt 2
d 2 RX ( t )
\ var ( X
( t) ) =
dt 2 t =0
Mean Square Integral
Recall that the definite integral (Riemannian integral) of a function x ( t ) over the interval
[ t0 , t ] is defined as the limiting sum given by

t n -1
x ( t ) dt =
lim x ( t ) D
k k
n Dk 0
t0 k =0

Where t0 < t1 < ................ < tn -1 < t n = t are partitions on the interval [ t0 , t ] and
D k = tk +1 - tk and t k [ tk -1 , tk ] .

For a random process { X ( t ) } , the m-s integral can be similarly defined as the process
{Y ( t ) } given by
t n -1
Y ( t) =
X ( t ) dt = l.i.m. X ( t ) D
k k
n Dk 0
t0 k =0

Existence of M.S. Integral

t

It can be shown that the sufficient condition for the m-s integral X ( t ) dt
to
t0
t t

exist is that the double integral

R ( t ,t ) dt dt
X 1 2 1 2 exists.
t0 t0

If { X ( t ) } is M.S. continuous, then the above condition is satisfied and the

process is M.S. integrable.
Mean and Autocorrelation of the Integral of a WSS process
We have
t
EY ( t ) = E
X ( t ) dt
t0
t
EX ( t ) dt
=
t0
t
m X dt
=
t0

= m X (t - t 0 )
Therefore, if m X 0, { Y ( t ) } is necessarily non-stationary.
RY ( t1 , t2 ) = EY (t1 )Y (t2 )
t1 t2

= E
X ( t 1 ) X ( t 2 ) dt1dt 2
t0 t0
t1 t2

=
EX ( t 1 ) X ( t 2 ) dt 1dt 2
t0 t0
t1 t2

=
RX ( t 1 - t 2 ) dt 1dt 2
t0 t0

which is a function of t1 and t2 .

Thus the integral of a WSS process is always non-stationary.

Fig. (a) Realization of a WSS process X (t )

(b) corresponding integral Y (t )

Remark The nonstatinarity of the M.S. integral of a random process has physical
importance the output of an integrator due to stationary noise rises unboundedly.
Example The random binary wave { X ( t ) } has the autocorrelation function
 t
1-
t Tp
RX (t ) = Tp

0
otherwise

RX ( t ) is continuous at t = 0 implying that { X ( t ) } is M.S. continuous. Therefore,

{ X ( t)} is mean-square integrable.