Uncorrelated and Independent Increments

If the increments X (t 2 ) − X (t1 ) and X (t 4 ) − X (t 3 ) of a process X (t ) are

uncorrelated (or independent) for any t1 < t 2 ≤ t 3 < t 4 , then X (t ) is a process with
uncorrelated (or independent) increments. The Poisson and the Wiener processes are
independent increment processes.

Cross-Correlation and Cross-Covariance

Given two stochastic processes X (t ) and Y (t ) , we define

R XY (t1 , t 2 ) = E{X (t1 )Y (t 2 )} = RYX (t 2 , t1 )

as their cross-correlation and

C XY (t1 , t 2 ) = E{[ X (t1 ) − η X (t1 )][Y (t 2 ) − η Y (t 2 )]} = R XY (t1 , t 2 ) − η X (t1 )η Y (t 2 )

as their cross-covariance.

Two processes are orthogonal if

R XY (t1 , t 2 ) = 0 for every t1 and t 2 .

They are uncorrelated if

C XY (t1 , t 2 ) = 0 for every t1 and t 2 .

Two processes are independent if the group of random variables X (t1 ) ,…, X (t n ) are
independent of the group Y (t1′ ) ,…, Y (t m′ ) for any t1 ,…, t n , t1′ ,…, t m′ , i.e.

f ( x1 ,..., x n , y1 ,..., y m ) = f ( x1 ,..., x n ) f ( y1 ,..., y m ) .

ME529 1 G. Ahmadi
Stationary Processes
Definition: Strict-Sense Stationary (SSS)

A random process X (t ) is SSS if its statistics are not affected by a shift in the
time origin. That is, the two processes X (t ) and X (t + τ ) have the same statistics.

A random process X (t ) is SSS if its n th order density satisfies the condition

f ( x1 ,..., x n , t1 ,..., t n ) = f ( x1 ,..., x n , t1 + τ ,..., t n + τ ) for any τ and any n .

In particular,

f ( x; t ) = f ( x; t + τ ) for any τ .

This implies that

f ( x; t ) = f ( x ) ,

that is, the first-order density is independent of time. Similarly, one finds

f ( x1 , x 2 ; t1 , t 2 ) = f ( x1 , x 2 ; t1 − t 2 ) .

Hence, it follows that

E{X (t )} = η = const ,

R (t1 , t 2 ) = E{X (t1 )X (t 2 )} = R (t1 − t 2 ) ,

{ }
E X 2 (t ) = σ 2 = const .

Definition: Two processes X (t ) and Y (t ) are jointly SSS if the joint statistics of X (t )
and Y (t ) are the same as those of X (t + τ ) and Y (t + τ ) . This implies that

f XY ( x, y; t1 , t 2 ) = f XY ( x, y, t1 − t 2 ) ,

and

E{X (t1 )Y (t 2 )} = R XY (t1 − t 2 ) .

ME529 2 G. Ahmadi
Definition: Wide-Sense Stationary (WSS)

A process X (t ) is WSS (or weakly stationary) if its mean is constant and its
autocorrelation depends only on τ = t1 − t 2 . i.e.,

E{X (t )} = η = const

E{X (t + τ )X (t )} = R(τ ) .

Definition: Two processes X (t ) and Y (t ) are jointly WSS (weakly stationary) if each
satisfy the conditions for WSS and their cross-correlation depends only on the time
difference. i.e.,

E{X (t + τ )Y (t )} = R XY (τ ) .

ME529 3 G. Ahmadi