Random Process Characteristic Equation NOTES

Unit.3.
Classification of Random Processes
UNIT – III RANDOM PROCESSES
PART - A
Problem 1. Define I & II order stationary Process

Solution:
I Order Stationary Process:
A random process is said to be stationary to order one if is first order density function
does not change with a shift in time origin.
i.e., f X ( x1 : t1 ) = f X ( x1, t1 + δ ) for any time t1 and any real number .
i.e., E ⎣ ⎡ X (t )⎦ ⎤ = X =
Constant. II Order Stationary
Process:
A random process is said to be stationary to order two if its second-order density
functions does not change with a shift in time origin.
i.e., f X ( x1, x2 : t1, t2 ) =f X ( x1, x2 : t1 + δ , t2 + δ ) for all t1, t2 and .
Problem 2. Define wide-sense stationary process

Solution:
A random process X (t ) is said to be wide sense stationary (WSS) process if the
following conditions are satisfied
(i). E ⎡ ⎣ X (t )⎤ ⎦ = i.e., mean is a constant
μ
(ii). R (τ ) = E ⎣ ⎡ X (t1 ) X (t2 )⎤ ⎦ i.e., autocorrelation function depends only on the time
difference.
Problem 3. Define a strict sense stationary process with an example

Solution:
A random process is called a strongly stationary process (SSS) or strict sense
stationary if all its statistical properties are invariant to a shift of time origin.
This means that X (t) and X (t + ) have the same statistics for any and any t
Example: Bernoulli process is a SSS process
Problem 4. Define nth order stationary process, when will it become a SSS process?
Solution:
A random process X (t) is said to be stationary to order n or nth order stationary
if its nth order density function is invariant to a shift of time origin.
i.e., f X ( x1, x2 ,..., xn , t1, t2 ,..., tn ) = f X ( x1, x2 ,..., xn , t1 + δ , t2 + δ ,..., tn + δ ) for all
t1, t2 ,..., tn & h .
1
Unit.3. Classification of Random Processes
A nth order stationary process becomes a SSS process when n → ∞ .
2
Problem 5. When are two random process said to be orthogonal?

Solution:
Two process {X (t )} & {Y (t )} are said to be orthogonal, if
E { X ( t1 ) Y ( t2 )} = 0.
Problem 6. When are the process {X (t )} & {Y (t )} said to be jointly stationary in the
wide sense?
Solution;
Two random process {X (t )} & {Y (t )} are said to be jointly stationary in the
wide sense, if each process is individually a WSS process and RXY ( t1 , t2 ) is a function
of (t2 − t1 ) only.
7. Write the postulates of a poisson process?

Solution:
If { X (t )} represents the number of occurrences of a certain event in (0, t) then
the discrete random process {X (t )} is called the poisson process, provided the
following postualates are satisfied
(i) P ⎡⎣ 1 occumence in ( t, t + Δt ) ⎦⎤ = λΔt + o ( Δt )
(ii) P ⎡⎣ no occurrence in ( t, t + Δt ) ⎤⎦ = 1− λΔt + o ( Δt )
(iii) P ⎡⎣ 2 or more occurrences in ( t, t + Δt ) ⎤⎦ = λΔt + o ( Δt )
(iv) X (t) is independent of the number of occurrences of the event in any interval prior
and after the interval (0, t) .
(v) The probability that the event occurs in a specified number of times (t0 , t0 + t )
depends only on t , but not on t0 .
8. When is a poisson process said to be homogenous?

Solution:
The rate of occurrence of the event λ is a constant, then the process is called a
homogenous poisson process.
9. If the customers arrive at a bank according to a poisson process with a mean rate of 2
per minute, find the probability that, during an 1- minute interval no customer arrives.
Solution:
Here λ = 2 , t =1
e −λt ( λt )
n
∴ P { X (t ) = n} = n!
, n = 0,1, 2,...
Probability during 1-min interval, no customer arrives = P { X (t ) = 0} = e−2 .
3
10. Define ergodic process.

Solution:
A random process {X (t )} is said to be ergodic, if its ensemble average are equal to
appropriate time averages.
11. Define a Gaussian process.

Solution:
A real valued random process {X (t )} is called a Gaussian process or normal
process, if the random variables X (t1 ) , X (t2 ),..., X (tn ) are jointly normal for every
n =1, 2,... and for any set of t1, t2 ,...
The nth order density of a Gaussian process is given by
1 ⎡ 1 n n ⎤
f ( x1, x2 ,..., xn ;t1, t2 ,..., tn ) = exp ⎢ − ∑∑ Λ ij ( xi − μi ) ( xj − μ j )⎥
⎣ 2 Λ ⎦
( 2π ) Λ
n 2 1/ 2 i=1 j =1
Where μi = E { X ( ti )} and Λ is the n

th
order square matrix (λ ) , where
ij
ij }
= C X (ti ) , X (t j ) and Λ ij = Cofactor of λij in Λ.
12. Define a Markov process with an example.

Solution:
If for t1 < t2 < t3 < ... < tn < t ,
P { X ( t ) ≤ x / X ( t1 ) = x1 , X ( t2 ) = x2 ,..., X ( tn ) = X n } = P { X ( t ) ≤ x / X ( tn ) = xn }
then the process {X (t )} is called a markov process.
Example: The Poisson process is a Markov Process.
13. Define a Markov chain and give an example.

Solution:
If for all n ,
P { X n = an / Xn−1 = an−1, X n− 2 = an−2 ,...X 0 = a0} = P { X n = an / Xn−1 = an−1} ,
then the process xn } , n = 0,1,... is called a Markov chain.
Example: Poisson Process is a continuous time Markov chain.
Problem 14. What is a stochastic matrix? When is it said to be regular?

Solution: A sequence matrix, in which the sum of all the elements of each row is 1, is
called a stochastic matrix. A stochastic matrix P is said to be regular if all the entries of
Pm (for some positive integer m ) are positive.
⎛0 1⎞
Problem 15. If the transition probability matrix of a markov chain is ⎜ 1 1⎟ find the
⎜ ⎟
⎝2 2⎠
steady-state distribution of the chain.
4
Solution:
Let π = (π 1 , π 2 ) be the limiting form of the state probability distribution on
stationary state distribution of the markov chain.
By the property of π , π P = π
⎛0 1 ⎞
i.e., ( π 1 , π 2 ) ⎜ 1 1 ⎟⎟ = ( π 1 , π 2 )
⎜
⎝2 2 ⎠
1
π 2 = π1 ------------------------------ (1)
2
1
π + π 2 = π 2 ------ (2)
1
2
Equation (1) & (2) are one and the same.
Consider (1) or (2) with π1 + π 2 = 1 , since π is a probability distribution.
π1 + π 2 = 1
1
Using (1) , π 2 + π 2 = 1
2
3π 2
=1
2
2
π=
2
3
2 1
π = 1− π = 1− =
1 2
3 3
1 2
π 2 = 1− π1 = 1− =
3 3
1 2
∴π = & π = .
1 2
3 3
PART-B
Problem 16. a). Define a random (stochastic) process. Explain the classification of
random process. Give an example to each class.
Solution:
RANDOM PROCESS
A random process is a collection (orensemble) of random variables {X (s, t )} that are
functions of a real variable, namely time t where s S (sample space) and t T
(Parameter set or index set).
CLASSIFICATION OF RANDOM PROCESS
Depending on the continuous on discrete nature of the state space S and parameter set T ,
a random process can be classified into four types:
(i). It both T & S are discrete, the random process is called a discrete random sequence.
Example: If X n represents the outcome of the nth toss of a fair dice, then { X n, n ≥ 1} is a
discrete random sequence, since T = 1, 2, 3,...} and S = 1, 2, 3, 4, 5, 6} .
5
(ii). If T is discrete and S is continuous, the random process is called a continuous

random sequence.
Example: If X n represents the temperature at the end nth hour of a day, then
{ X n ,1 ≤ n ≤ 24} is a continuous random sequence since temperature can take any value is
an interval and hence continuous.
(iii). If T is continuous and S is discrete, the random process is called a discrete random
process.
Example: If X (t) represents the number of telephone calls received in the interval (0, t)
then X ( t )} random process, since S = 0,1, 2, 3,...} .
(iv). If both T and S are continuous, the random process is called a continuous random
process f
Example: If X (t) represents the maximum temperature at a place in the interval (0, t)
X (t )} is a continuous random process.
b). Consider the random process X (t ) = cos(t + φ ) , where φ is uniformly distributed in

π π
the interval − to . Check whether the process is stationary or not.
2 2
Solution: ⎛ π π⎞
Since 0 is uniformly distributed in − ⎜ , ,
2 2⎟
⎝ ⎠
1 π π
f (0)= , − < 0 <
π 2 2
π
2
E ⎡ ⎣ X (t )⎤ ⎦ = ∫ X (t ) f ( 0 )d 0
π
−
2
∫ cos (t + 0 ).
π2
1 d0
=
−
π π
2
π
2
= ∫ cos (t + 0 ) d 0
π
−
2
π
1
= ⎡⎣ sin ( t + 0 )⎤⎦ π 2
π −
2
2
= cost ≠ Constant.
π
Since E ⎡ ⎣ X (t is a function of t , the random process X (t )} is not a stationary
)⎤ ⎦
process.
6
Problem 17. a). Show that the process X (t )} whose probability distribution under
⎧ ( at ) n−1 ,
⎪ n−1 n = 1, 2,...
⎪ (1+ at )
certain conditions is given by P { X ( t ) = n} = ⎨ is evolutionary.
⎪ at
⎪ , n=0
⎩ 1+ at
Solution:
The probability distribution is given by
X (t) = n : 0 1 2 3 . . . .
(at )
2
at 1 at
P ( X (t) = n) : . . . .
(1+ at ) (1+ at ) (1+ at )
2 3 4
1+ at
∞
E ⎡ ⎣ X (t )⎤ ⎦ = ∑ npn
n=0
3 ( at )
2
1 2at
= 2 + 3 + 4 + ...
(1+ at ) (1+ at ) (1+ at ) 2
= ⎧
1 2 ⎪⎨1+ 2 ⎜⎛ at ⎟⎞+ 3 ⎜ ⎛ at ⎟ ⎞ ⎫⎪
1+ at 1+ at + ...⎬
(1+ at ) ⎪⎩ ⎝ ⎠ ⎝ ⎠ ⎪⎭
2
1 at
= 2 1
(1+ at) 1+ at
E ⎡ ⎣ X (t )⎦ ⎤ = 1 = Constant
∞
E ⎡ ⎣ X 2 (t )⎤ ⎦ = ∑ n2 pn
n=0
( at ) ( at )
n −1
=∑ = ∑∞ ⎡ n ( n +1) − n⎤
n−1
∞ n2
⎣ ⎦
(1+ at ) (1+ at )
n+1 n+1
n=1 n=1
⎡ ∞ ⎛ at n−1 −∑∞n ⎜ ⎛ at⎟ n−1 ⎤ ⎥

⎢∑
1
= 2 n ( n +1) ⎜ ⎟⎞ 1+ at ⎞
1+ at
(1+ at ) ⎢⎣ ⎝ ⎠ n=1 ⎝ ⎠ ⎥⎦
n=1
−3 −2
1 ⎡ ⎡ at ⎤ – ⎡⎢ 1− at ⎥⎤ ⎤⎥
= 2 ⎢ 2 ⎢ 1− 1+ at ⎦
1 1+ at ⎥⎦
( + at ) ⎢ ⎣ ⎣ ⎣ ⎥⎦
E ⎡ ⎣ X 2 (t )⎦ ⎤ = 1+ 2at ≠ Constant 2
Var X (t )} = E X 2 (t ) E ( X (t ))
Var X (t )} = 2at
∴The given process X (t )} is evolutionary
7
b). Examine whether n the Poisson process {X (t )} given by the probability law
e−λt ( λt )
P { X (t ) = n} = , n = 0,1, 2,... is evolutionary.
n!
Solution:
∞
E ⎡ ⎣ X (t )⎤ ⎦ = ∑ npn
e− λ t ( λ t )
n=0 n
∞
= ∑n
n=0 n!
e−λt ( λ t )
∞ n
=∑
n=1 ( n −1)!
( λt )
∞ n−1
= ( λt ) e −λt ∑
n=1 (n −1)! 2
⎡ λt ( λt ) ⎤
= ( λ t ) e−λt ⎢1+ + + ...⎥
⎢⎣ 1! 2! ⎥⎦
= ( λt ) e e
−λ t λt
E ⎡ ⎣ X (t )⎤ ⎦ = λt
E ⎡ ⎣ X (t )⎦ ⎤ ≠
Constant.
Hence the Poisson process { X (t )} is evolutionary.
Problem 18. a). Show that the random process X (t ) = A cos(ω t+ θ ) is WSS if A & ω
are constants and θ is uniformly distributed random variable in (0, 2π ) .
Solution:
Since is uniformly distributed random variable in (0, 2π )
⎧1
⎪ , 0 < 0 < 2π
f (θ ) = ⎨2π
⎪⎩0 , elsewhere
2π
E ⎡⎣ X ( t ) ⎤⎦ = ∫ X ( t ) f (θ ) dθ
0
2π
1
= ∫ 2π Acos (ωt + θ ) dθ
0
2π
= A cos ( ω t + θ ) dθ
2π 0∫
A 2π
= ⎡⎣ sin ( ωt + θ ) ⎤⎦
2π 0
8
A
= ⎡ sin ( ωt + 2π ) − sin ( ωt ) ⎤
⎣ ⎦
2π
A
= ⎡⎣ sin ( ωt ) − sin ( ωt ) ⎤⎦ ⎡⎣ sin ( 2π + θ ) = sinθ ⎤⎦
2π
E ⎡ ⎣ X (t )⎤ ⎦ = 0
RXX (t1, t2 ) = E X (t1 ) X (t2 )⎤ ⎦
= E ⎡ ⎣ A2cos (ωt + θ ) cos (ωt + θ )⎤ ⎦
1 2
⎡ cos ( ω ( t1 + t2 ) + 2θ ) + cos ( ω ( t1 − t2 ) ) ⎤
= A2 E
⎢
⎣⎢ 1 2 ⎦⎥
A2 2π ⎡ ⎤
cos ( ω ( t1 + t2 ) + 2θ ) + cos ( ω ( t1 − t2 ) ) ⎦ dθ
2 ∫0 2π ⎣
=
2π
A2 ⎡ sin ⎡⎣ ω ( t1 + t 2 ) + 2θ ⎤⎦ cos ω t ⎥
⎤
= ⎢ +θ
4π ⎢ ⎣ 2 ⎥⎦ 0
A2
=
4π
[2πcosωt]
A2
= cosωt = a function of time difference
2
Since E ⎡ ⎣ X (t )⎦ ⎤ = constant
RXX (t1, t2 ) =a function of time difference
∴{X (t )} is a WSS.
b). Given a random variable y with characteristic function φ (ω ) = E (eiωy ) and a random
process define by X (t ) = cos (λt + y ) , show that { X (t )} is stationary in the wide sense
if φ (1) = φ ( 2 ) = 0 .
Solution:
Given 0 (1) = 0
⇒ E [cosy + isiny] = 0
∴ E [cosy] = E [sin y] = 0
Also 0 ( 2 ) = 0
⇒ E [cos2 y + isin2 y] = 0
∴ E [cos2 y] = E [sin 2 y] = 0
E { X ( t )} = E ⎡⎣ cos ( λt + y ) ⎤⎦
= E [cosλtcosy − sinλtsiny]
9
= cosλt E [cosλt] − sinλt E [siny] = 0

RXX (t1, t2 ) = E X (t1 ) X (t2 )⎤ ⎦
= E ⎡⎣ cos ( λ t1 + y ) cos ( λt 2 + y ) ⎤⎦
⎡ cos ( λ ( t1 + t2 ) + 2 y ) + cos ( λ ( t1 − t2 ) ) ⎤
=E⎢ ⎥
⎢ 2 ⎥⎦
⎣ ) + 2 y ) + cos (λ (t − t ))⎤ ⎦
= E ⎣ cos ( λ ( t + t
1
⎡
1 2 1 2
2
= E ⎡⎣ cosλ ( t + t ) cos2 y − sinλ ( t + t ) sin2 y + cos ( λ ( t − t ))⎤ ⎦
1
1 2 1 2 1 2
2
1 1
= cosλ (t + t ) E (cos2 y ) − sinλ (t + t ) E (sin2 y ) + cos (λ (t − t ))
1
1 2 1 2 1 2
2 2 2
1
= cos (λ (t − t )) = a function of time difference.
1 2
2
Since E ⎡ ⎣ X (t )⎦ ⎤ = constant
RXX (t1, t2 ) =a function of time difference
∴{X (t )} is stationary in the wide sense .
Problem 19. a). If a random process { X (t )} is defined by {X (t )} = sin (ω t+ Y ) where

Y is uniformly distributed in (0, 2π ) . Show that { X (t )} is WSS.
Solution:
Since y is uniformly distributed in (0, 2π ) ,
1
f ( y) = , 0 < y < 2π
2π
2π
E ⎡ ⎣ X (t )⎤ ⎦ = ∫ X (t ) f ( y ) dy
0
2π
1
=
2π ∫ sin (ωt + y ) dy
0
1 2π
= ⎡⎣ −cos ( ωt + y ) ⎤⎦
2π 0
1
=− ⎡ cos ( ωt + 2π ) − cosωt ⎤ = 0
⎣ ⎦
2π
RXX ( t1 , t2 ) = E ⎡⎣ sin ( ωt1 + y ) sin (ωt2 + y )⎤⎦
⎡ cos ( ω ( t1 − t2 ) ) − cos ( ω ( t1 + t2 ) + 2 y ) ⎤
=E⎢ ⎥
⎢ 2
⎦⎥
) )⎤⎦ − E ⎡⎣ cos (ω ( t + t ) + 2 y )⎤ ⎦
1
⎣
= E ⎣ cos ( ω ( t − t
1
⎡
10
1 2 1 2
2 2
11
1 1 2π 1
= cos ( ω ( t1 − t2 ) ) − ∫ cos ( ω ( t1 + t2 ) + 2 y ) dy
2 2 0 2π2π
1 1 ⎡ sin ( ω ( t + t ) + 2 y ) ⎤
= cos (ω (t − t )) −
1 2
⎢ ⎥
4π ⎢ ⎣
1 2
2 2 ⎦⎥ 0
= cos ( ω ( t − t ) ) − ⎡⎣ sin ( ω ( t + t ) + 2π ) − sinω (t + t
1 1
)⎤ ⎦
8π
1 2 1 2 1 2
2
1
= cos (ω (t − t ) ) is a function of time difference.
1 2
2
∴{X (t )} is WSS.
b). Verify whether the sine wave random process X (t ) = Y sin ωt, Y is uniformly
distributed in the interval ( 1,1) is WSS or not
Solution:
Since y is uniformly distributed in( 1,1),
1
f ( y ) = , −1 < y < 1
2
1
E [ X (t)] = ∫ X (t) f ( y)dy

−1
1
1
= ∫ y sinωt
−1
2 dy
sinωt 1
=
2 ∫
−1
ydy
Sinωt
= (0) = 0
2
RXX (t1, t2 ) = E X (t1 ) X (t2 )⎤ ⎦
= E ⎡ ⎣ y2sinωt 1sinωt ⎤ 2⎦
⎡ 2 cosω (t1 − t2 ) − cosw (t1 + t2 ) ⎤
=
E ⎢⎣ y ⎥⎦
2
cosω (t1 − t2 ) − cosw (t1 + t2 )
= E ( y2 )
cosω (t − t ) −2cosw(t + t ) 1
=
∫−1 y2 f ( y ) dy
1 2 1 2
2
cosω (t − t ) − cosw(t + t ) 1 1
= 1 2 1 2
∫ y2dy
2 2 −1
12
cosω ( t − t ) − cosw ( t + t ) ⎛ y 3 ⎞
1
= 1 2 1 2
⎜ ⎟
4 ⎝ 3 ⎠−1
cosω (t1 − t2 ) − cosw (t1 + t2 ) ⎛ 2 ⎞
= ⎜ ⎟
4 ⎝3⎠
cosω (t1 − t2 ) − cosw(t1 + t2 )
=
6
RXX (t1, t2 ) a function of time difference alone.
Hence it is not a WSS Process.
Problem 20. a). Show that the process X (t ) = A cos λt + B sin λt (where A & B are
random variables) is WSS, if (i) E ( A) = E ( B) = 0 (ii) E ( A2 ) = E ( B 2 ) and (iii)
E ( AB ) = 0 .
Solution:
Given X (t ) = Acosλt + Bsinλt , E ( A) = E ( B) = 0 , E ( AB ) = 0 ,
E ( A2 ) = E ( B 2 ) = k (say)
E ⎡ ⎣ X (t )⎦ ⎤ = cosλt E ( A) + sinλt E ( B )
E ⎡ ⎣ X (t )⎦ ⎤ = 0 = is a constant. E ( A) = E ( B) = 0
R ( t1 , t2 ) = E { X ( t1 ) X ( t2 )}
= E {( Acosλt1 + Bsinλt1 )( Acosλt2 + Bsinλt2 )}
= E ( A2 )cosλt cos
1 λt +2 E ( B )sinλt sin1λt + E
2
2 ( AB )[sin λt cosλ1t 2 + cosλt sin
1 λt ]2
= E ( A )cosλt cosλt + E ( B ) sinλt sinλt + E ( AB ) sinλ (t + t )
2 2
1 2 1 2 1 2
= k (cosλt1cosλt2 + sinλt1sinλt2 )
= kcosλ (t1 − t2 ) = is a function of time difference.
∴{X (t )} is WSS.
b). If X (t ) = Ycost + Zsint for all t & where Y & Z are independent binary random
2 1
variables. Each of which assumes the values – 1 & 2 with probabilities &
3 3
respectively, prove that {X (t )} is WSS.
Solution:
Given
Y=y : 1 2
2 1
P( Y = y ) :
3 3
13
2 1
E ( Y ) = E ( Z ) = −1× + 2 × = 0
3 3
E (Y 2 ) = E ( Z 2 ) = ( −1) × + ( 2 ) ×
2 2 2 1
3 3
E (Y 2 ) = E ( Z 2 ) = + = = 2
2 4 6
3 3 3
Since Y & Z are independent
E (YZ ) = E (Y ) E (Z ) = 0 ---- (1)
Hence E ⎡ ⎣ X (t )⎦ ⎤ = E [ ycost + zsint ]
= E [ y]cost + E [z]sint
E ⎡ ⎣ X (t )⎦ ⎤ = 0 = is a ⎡⎣ E ( y ) = E ( z ) = 0⎤⎦
constant.
RXX (t1, t2 ) = E ⎡ ⎣ X (t1 ) X (t2
)⎤ ⎦
= E ⎣ ⎡ ( ycost1 + zsint1 )( ycost2 + zsint2 )⎤ ⎦
= E ⎡ ⎣ y2cost1 2
+ yzcost1 sint 2 + zysint1cost 2 + z2sint 1sint 2⎤ ⎦
cost
= E ( y2 )cost1 cost 2 + E [ yz]cost1 sint 2 + E[zy]sint1co2st 2 + E ⎡⎣ z ⎤ ⎦ sint1 sint2
= E ( y2 )cost1cost 2+ E ( z 2 ) sint 1sint 2
= 2 [ cost cost + sint sint ] ⎡⎣ E ( y 2 ) = E ( z 2 ) = 2⎤⎦
1 2 1 2
= 2 cos (t1 − t2 ) = is a function of time difference.

∴{X (t )} is WSS.
Problem 21. a). Check whether the two random process given by
X (t ) = A cosωt + B sin ωt & Y (t ) = B cosωt − Asin ωt . Show that X (t) & Y (t) are
jointly WSS if A & B are uncorrelated random variables with zero mean and equal
variance random variables are jointly WSS.
Solution:
Given E ( A) = E ( B) = 0
Var ( A) = Var ( B ) = σ 2
∴ E ( A2 ) = E ( B 2 ) = σ 2
As A & B uncorrelated are E (AB) = E ( A)E (B) = 0 .
E ⎡ ⎣ X (t )⎦ ⎤ = E [ Acosωt + Bsinωt ]
= E ( A) cosωt + E ( B ) sinωt = 0
E ⎡ ⎣ X (t )⎦ ⎤ = 0 = is a constant.
RXX (t1, t2 ) = E ⎣ ⎡ X (t1 ) X (t2 )⎤ ⎦
14
= E ( Acoswt1 + Bsinwt2 )( Acoswt 2 + Bsinwt2 )
15
= E A2coswt coswt + ABcoswt sinwt + BAsinwt c2oswt + B sinwt sinwt

1 2 1 2 1 2 1 2
= coswt coswt E A2 + coswt sinwt E [ AB ] + sinwt coswt E [ BA]2+ sinwt sinwt E B
1 2 1 2 1 2 1 2
= 2
[coswt coswt + sinwt sinwt ]
1 2 1 2
= 2
cosw (t 1 t )2 [ E ( A2 ) = E ( B 2 ) = σ 2 & E ( AB ) = E (BA) = 0]
RXX (t1, t2 ) =is a function of time difference.
E ⎡⎣ Y ( t ) ⎤⎦ = E [ Bcosωt − Asinωt ]
= E ( B) cosωt − E ( A) sinωt = 0
RYY (t1, t2 ) = E ⎡ ⎣ ( Bcosωt1 − Asinω t1 )( Bcosωt2 − Asinω t2 )⎤ ⎦
= E ⎡⎣ B 2 cosω t cos
1 ωt 2
– BAcosωt sin
1
ωt −2 AB sinωt c2os
1
ωt
+ A sinωt sin
1 ωt ⎤ 2⎦ 2
= E ( B 2 )cosωt 1cosωt 2 – E (BA)cosωt sin 1 ωt + 2E ( A )sinωt sin1ωt

1 ωt −2 E ( A B ) sinωt cos
2
2
= σ cosω ( t − t )
2
[ E ( A ) = E ( B ) = σ & E ( AB ) = E (BA) = 0]
2 2 2
1 2
RYY (t1, t2 ) =is a function of time difference.

RXY (t1, t2 ) = E ⎡ ⎣ X (t1 )Y (t2 )⎤ ⎦
= E ⎡ ⎣ ( Acosωt1 + Bsinωt1 )( Bcosωt2 + Asinω t2 )⎤ ⎦
= E ⎡ ⎣ ABcosωt1 cosωt2 − A2cosωt sin
1 ωt 2
+ B2sinωt 1cosωt 2 – BAsinωt sin
1 ωt 2⎦
⎤
= σ 2 [sinωt cosωt − cosωt sinωt ]
( ) = E (B2 ) = σ
1 2 1 2
= σ 2 sinω (t − t ) [ E A 2 2
& E ( AB) = E (BA) = 0]
1 2
RXY (t1, t2 ) = is a function of time difference.

Since { X (t )}& {Y (t )}are individually WSS & also RXY (t1, t2 ) is a function of time
difference.
The two random process {X (t )}& {Y (t )}are jointly WSS.
b). Write a note on Binomial process.

Solution:
Binomial Process can be defined as a sequence of partial sums S n n =1, 2,...} Where
Sn = X1 + X 2 +...+ X n Where X i denotes 1 if the trial is success or 0 if the trial is
failure.
As an example for a sample function of the binomial random process with
( x1 , x2 ,...) = (1,1, 0, 0,1, 0,1,...) is ( s1 , s2 , s3 ,...) = (1, 2, 2, 2, 3, 3, 4,...) , The process increments
by 1 only at the discrete times ti ti = iT , i =1, 2,...
Properties
(i). Binomial process is Markovian
16
(ii). Sn is a binomial random variable so, P(Sn = m) = nC mpmqn m

, E [ S n ] = np &
var [ Sn ] = np (1 p )
(iii) The distribution of the number of slots mi between ith and (i 1) arrival is
th
geometric with parameter p starts from 0. The random variables mi , i =1, 2,... are
mutually independent.
The geometric distribution is given by p (1 p ) , i = 1, 2,...
i 1
(iv) The binomial distribution of the process approaches poisson when n is large and p
is small.
Problem 22. a). Describe Poisson process & show that the Poisson process is Markovian.
Solution:
If { X (t )} represents the number of occurrences of a certain event in (0, t) then
the discrete random process {X (t )} is called the Poisson process, provided the
following postulates are satisfied
(i) P ⎡⎣ 1 occumence in ( t, t + Δt ) ⎤⎦ = λΔt + o ( Δt )
(ii) P ⎡⎣ no occurrence in ( t, t + Δt ) ⎤⎦ = 1− λΔt + o ( Δt )
(iii) P ⎡⎣ 2 or more occurrences in ( t, t + Δt ) ⎤⎦ = o ( Δt )
(iv) X (t) is independent of the number of occurrences of the event in any interval prior
and after the interval(0, t) .
(v) The probability that the event occurs in a specified number of times (t0 , t0 + t )
depends only on t , but not on t0 .
Consider
P⎡ ⎣ X = n / X (t P ⎣ ⎡ X (t1 ) = n1, X (t2 ) = n2 , X (t3 ) = n3 ⎤ ⎦
(t ) = n , X (t ) = n )⎤ ⎦ =
P ⎡ X (t ) = n , X (t ) = n ⎤
3 3 2 2 1 1
⎣ 1 1 2 2⎦
e− λt λ n t n1 ( t 2 1 ) (t – t )n −n
n2 −n1
3 3 1 –t 3 2
= 3 2
n1 !(n2 − n1 )!(n3 − n2 )!
e− λt λ n t n1 ( t 2 – t1 ) n −n
2 2 1 2 1
=
n1 !(n2 − n1 )!
− λ ( t 3 −t 2 ) n3 −n2
=e λ t( 3 − t2 n3 −n 2
)
(n 3 − n 2 )!
P ⎡⎣ X ( t3 = n3 / X ( t2 ) = n2 , X ( t1 ) = n1 ) ⎤⎦ = P ⎡⎣ X ( t3 ) = n3 / X ( t2 ) = n2 ⎤⎦
This means that the conditional probability distribution of X (t3 ) given all the past
values X (t1) = n1 , X (t2 ) = n2 depends only on the most recent values X (t2 ) = n2 .
17
i.e., The Poisson process possesses Markov property.
b). State and establish the properties of Poisson process.

Solution:
(i). Sum of two independent poisson process is a poisson process.
Proof:
The moment generating function of the Poisson process is
M (u ) = E euX (t )
X (t )
= eux P X (t ) = x
x=0
( t)
t x
e
= eux
x=0 x! 1
eu ( t ) e 2u ( t )
2
= e 1+ t
+ +...
1! 2!
=e e t teu
t (e 1) u
M X (t) (u ) = e
Let X1 (t) and X 2 (t) be two independent Poisson processes
Their moment generating functions are,
t (e 1) u
t (e 1) u
M X (t) (u ) = e and M X (t) (u ) = e

1 2
1 2
M X (t)+ X (t) (u) = M X (t) (u)M X

1 2 1 (t)2 (u)
(
t eu 1 ) (
t eu 1)
=e1 e2
( + )t (e 1) u
=e
1 2
By uniqueness of moment generating function, the process X 1 (t ) + X 2 (t )} is a

Poisson process with occurrence rate ( 1 + 2 ) per unit time.
(ii). Difference of two independent poisson process is not a poisson process.
Proof:
Let X (t ) = X1 (t ) − X 2 (t )
E { X ( t )} = E { X 1 ( t )} − E { X 2 ( t )}
= ( λ1 − λ2 ) t
E {X (t )} = E { X 2 (t )} + E {X 2 (t )} − 2E {X
2
(t )}E {X (t )}
1 2 1 2
= (λ t + λ t )1 + (λ t + λ t ) −2 2 (λ t )( λ1 t )
2 2
1
2 2
2 2
= (λ + λ ) t + (λ − λ ) t2
2
1 2 1 2
≠ ( λ – λ ) t + (λ − λ )
2
t2
1 2 1 2
Recall that E { X 2 (t )} for a poisson process {X (t )} with parameter λ is given by
18
E {X 2 (t )} = λt + λ 2t2
∴ X1 (t ) − X 2 (t ) is not a poisson process.
(iii). The inter arrival time of a poisson process i.e., with the interval between two
successive occurrences of a poisson process with parameter λ has an exponential
1
distribution with mean .
λ
Proof:
Let two consecutive occurrences of the event be Ei & Ei+1 .
Let Ei take place at time instant ti and T be the interval between the occurrences of Ei
Ei+1 .
Thus T is a continuous random variable.
P (T > t ) = P{Interval between occurrence of Ei and Ei+1 exceedst}
= P { Ei +1 does not occur upto the instant ( ti + t )}
= P No event occurs in the interval (ti ,ti + t )}
= P { X (t ) = 0} = P0 (t )
= e− λ t
∴The cumulative distribution function of T is given by
F ( t ) = P {T ≤ t} = 1− e−λt
∴The probability density function is given by
f ( t ) = λe −λt , ( t ≥ 0 )
1
Which is an exponential distribution with mean .
λ
Problem 23. a). If the process {N (t ) : t ≥ 0} is a Poisson process with parameter λ ,

obtain P ⎡⎣ N ( t ) = n⎤⎦ and E ⎡⎣ N ( t ) ⎤⎦ .
Solution:
Let λ be the number of occurrences of the event in unit time.
Let Pn (t ) represent the probability of n occurrences of the event in the interval (0, t) .
i.e., Pn (t ) = P { X (t ) = n}
∴ Pn (t + Δt ) = P { X (t + Δt ) = n}
= P{ n occurences in the time (0,t + Δt )}
⎧n occurences in the interval ( 0,t ) and no occurences in ( t, t + Δt ) or
⎪
= P ⎨n −1occurences in the interval ( 0, t ) and 1 occurences in ( t, t + Δt ) or
⎪
⎩ n − 2 occurences in the interval ( 0, t ) and 2 occurences in ( t, t + Δt ) or...
= Pn (t )(1− λΔt ) + Pn−1 (t ) λΔt + 0 + ...
19
Pn (t + Δt ) − Pn (t )
= λ {P n−1 ( ) n ( )}
t −P t
∴
Δt
Taking the limits as t 0
P (t ) = λ {P (t ) − P (t )} ------- (1)
d
dt n n−1 n
This is a linear differential equation.

t
P n( t ) eλt = ∫ λ P ( t )eλt
n−1
-------------------------
(2)
0
Now taking n =1 we get

t
eλt P1( t ) = λ ∫ P0 ( t )e λ dt ---------------- (3)

t
Now, we have,
P0 (t + t ) = P 0 occurences in ( 0, t + t )
= P 0 occurences in (0, t ) and 0 occurences in (t, t + t )
= P0 (t)[1 t]
P0 (t + t) P0 (t) = P0 (t)( t)
P0 (t + t ) P0 (t)
= P (t)0
t
Taking limit t 0
Lt P0 (t + t) P0 (t)
= P (t)0
t 0 t
dP0 (t)
= P ( t)
0
dt
dP0 (t)
= dt
P 0 ( t)
log P0 (t) = t +c
P0 (t ) = e t+c
P0 (t) = e tec
P0 (t) = e t A ----------------------(4)
Putting t = 0 we get
P0 (0) = e0 A = A
i.e., A =1
(4) we have
P0 (t) = e t
substituting in (3) we get
t
e t P 1(t) = e te t dt
0
20
= dt = t
0
P1 (t) = e t t
Similarly n = 2 in (2) we have,
t
P2 (t)e t = P1 (t)e tdt

0
t
= e t
te tdt
0
t2
= 2
2
P (t)e t = e ( t )
t 2
2
2!
Proceeding similarly we have inn general
e −λt ( λt )
Pn (t ) = P{ X (t ) = n} = , n = 0,1,...
n!
Thus the probability distribution of X (t ) is the Poisson distribution with parameter λt .
E ⎡ ⎣ X (t )⎤ ⎦ = λt .
.b). Find the mean and autocorrelation and auto covariance of the Poisson process.
Solution:
The probability law of the poisson process { X (t )} is the same as that of a poisson
distribution with parameter λt
∞
E ⎡ ⎣ X (t )⎦ ⎤ = ∑ nP ( X (t ) = n)
e− λ t ( λ t )
n=0 n
∞
= ∑n
n=0 n!
λt ( λt )
∞ n−1
= e −λt ∑n=1 ( n −1)!

= λte − λt eλt
E ⎡ ⎣ X (t )⎤ ⎦ = λt
Var ⎣⎡ X ( t ) ⎤⎦ = E ⎡⎣ X 2 ( t ) ⎤⎦ − E ⎡⎣ X ( t ) ⎤⎦
2
E X{ (t )} = ∑ n2 P ( X (t ) = n)
2
n=0
e−λt ( λt )
n
∞
n=0
(
= ∑ n −n+n 2
) n!
21
e−λt ( λ t ) e−λt ( λt )
∞ n ∞ n
= ∑ n ( n −1) +∑n
n=0 n! n=0 n!
e−λt ( λ t )
∞ n
= ∑ n ( n −1) + λt
n=0 n!
⎡ ( λt )2 ( )
λ t
3
( λt )
4
⎤
=e − λt
⎢ + + + ...⎥ + λt
⎣⎢ 1 1! 2! ⎥⎦
= ( λ t ) e−λt eλt + λt
2
= (λt ) + λt
2
E {X 2 (t )} = λt + λ 2t2
∴Var ⎡ ⎣ X (t )⎤ ⎦ = λt
RXX ( t1 , t2 ) = E { X ( t1 ) X ( t2 )}
{
= E X ( t1 ) ⎣⎡ X ( t2 ) − X ( t1 ) + X ( t1 )⎤⎦ }
{
= E X (t ) ⎡ X (t ) − X (t )⎤ + E ⎡ X 2 (t )⎤ }
1 ⎣ 2 1 ⎦ ⎣ 1 ⎦
= E ⎡ X (t )⎤ E ⎡ X (t ) − X (t )⎤ + E ⎡ X 2 (t )⎤
⎣ 1 ⎦ ⎣ 2 1 ⎦ ⎣ 1⎦
Since X (t )} is a process of independent increments.

= λt ⎡ λ (t − t )⎤ + λt + λt2 if t ≥ t (by − (1))
1⎣ 2 1 ⎦ 1 1 2 1
= λ 2t t + λt if t ≥ t
1 2 1 2 1
R ( t ,t ) = λ 2 t t + λ min ( t ,t )
XX 1 2 1 2 1 2
Auto Covariance
C XX ( t1 , t2 ) = RXX ( t1 , t2 ) − E { X ( t1 )} E { X ( t2 )}
= λ 2t , t + λt − λ 2t t
1 2 1 1 2
= λt1, if t2 ≥ t1
= λ min (t1, t2 )
Problem 24. a). Prove that the random process X (t ) = A cos(ω t+ θ ) . Where A, ω are
constants θ is uniformly distributed random variable in (0, 2π ) is ergodic.
Solution:
Since θ is uniformly distribution in (0, 2π ) .
1
f (θ ) = , 0 < θ < 2π
2π
Ensemble average same,
2π
1
E ⎡⎣ X ( t ) ⎤⎦ = ∫ cos ( ωt + θ ) dθ
2π 0
22
2π
1
=
2π ∫ ( cosωt cosθ − sinωt sinθ ) dθ
0
1
[cosωt sinθ + sinωt cosθ ]
2π
=
2π 0
= 1 [sinωt − sinωt] = 0
2π
E ⎡ ⎣ X (t )⎤ ⎦ = 0
RXX (t1, t2 ) = E X (t1 ) X (t2 )⎤ ⎦
= E ⎣⎡ cos ( ωt1 + θ ) cos ( ωt 2 + θ ) ⎤⎦
1
= E ⎡ cos ( ωt + ωt + 2θ ) + cos ( ωt − ωt ) ⎤
2 2π ⎣ 1 2 1 2⎦
1 1
= ∫ ⎡ cos ( ωt1 + ω t2 + 2θ ) + cos ( ωt1 − ωt 2 ) ⎤⎦ dθ
2 0 2π ⎣
1 ⎡ sin (ωt + ωt + 2θ ) + θ cos ( ω ( t − t ) ) ⎤ 2π
= 1 2 1 2
⎥
2 ⎥⎦ 0
⎢
4π ⎢ ⎣
2π
= cos (ω (t − t ))
4π 1 2
R (t , t ) = cos (ω (t − t ))
1
XX 1 2 1 2
2
The time average can be determined by
lim 1 T
X (t ) =
T → ∞ 2T −T ∫ cos (ωt + θ ) dt
= lim 1 ⎢ ⎡ sin ( ωt + θ ) ⎤ ⎥
T
T → ∞ 2T ω
⎣ ⎦ −T
lim 1
= ⎡ sin ( ωT + θ ) − sin ( −ωT + θ ) ⎤
T → ∞ 2Tω ⎣ ⎦
X (t ) = 0 [ As T − > ∞]
The time auto correlation function of the process,
Lt 1 T
T → ∞ 2T ∫ X ( t ) X ( t + τ ) dt
−T
T
lim 1
=
T → ∞ 2T ∫ cos (ωt + θ )cos (ωt + ωτ + θ ) dt
−T
lim 1 T ⎡ cos ( ωt + ωτ + 2θ ) + cosωτ ⎤
= T → ∞ 2T
∫ ⎢⎣
−T
2
⎥ dt
⎦
23
= lim 1 ⎢⎡ sin ( 2ωt + ωt + 2θ )

T
⎤
T → ∞ 4T 2 + tcosωt ⎥
⎣ ⎦ −T
lim 1 ⎢ ⎡ sin ( 2ω t + ωt + 2θ ) + sin ( −2ωT + ωτ + 2θ ) ⎤ ⎥
= T → ∞ 4T 2ω + 2Tcosωτ
⎣ ⎦
cosωτ
=
2
Since the ensemble average=time average the given process is ergodic.
b). If the WSS process { X (t )} is given by X (t ) = 10 cos(100t + θ ) where θ is uniformly

distributed over ( −π , π ) prove that { X ( t )} is correlation ergodic.
Solution:
To Prove { X (t )} is correlation ergodic it is enough to show that when
Lt 1 T
T → ∞ 2T ∫ X ( t )X ( t + τ ) dt = RXX (τ )
−T
RXX (τ ) = E ⎣⎡ X ( t ) X ( t + τ ) ⎤⎦
= E ⎡⎣ 10 cos (100t + θ ) 10 cos (100t +100τ + θ ) ⎤⎦
⎪⎧ ⎡ cost ( 200t +100τ + 2θ ) + cos (100τ ) ⎤ ⎫⎪
= E ⎨100 ⎢ ⎥⎬
2
⎪⎩ ⎣ ⎦ ⎪⎭
π
1
= 50 ∫ ⎡ cos ( 200t +100τ + 2θ ) + cos (100τ ) ⎤⎦ dθ
−π
2π ⎣ π
= 50 ⎢⎡ sin ( 200t +100τ + 2θ ) ⎤
2π 2 + θ cos (100τ ) ⎥
⎣ ⎦ −π
RXX (τ ) = 50 cos (100τ )
Lt 1 T
T → ∞ 2T −T∫ X ( tT) X ( t + τ ) dt
Lt 1
= (100t + θ )10cos (100t +100τ + θ )⎤⎦ dt
T → ∞ 2T T−T ∫ ⎡10cos
⎣
Lt 25
=
T → ∞ T −T ∫ ⎡⎣ cos ( 200t +100τ + 2θ ) + cos (100τ ) ⎤⎦ dtT
Lt 25 ⎡ sin ( 200t +100τ + 2θ ) ⎤
=T →∞ T ⎢ 200 + t cos (100τ ) ⎥
⎣ ⎦ −T
lim 50
lim
XT = Tcos (100τ )
T →∞ T →∞ T
= 50 cos (100τ ) = is a function of time difference
24
∴{X (t )} is correlation ergodic.
Problem 25. a). If the WSS process { X (t )} is given by X (t ) = cos (ωt + φ ) where φ is
uniformly distributed over ( −π , π ) prove that { X ( t )} is correlation ergodic.
Solution:
To Prove { X (t )} is correlation ergodic it is enough to show that when
Lt 1 T
∫ X ( t )X ( t + τ ) dt = RXX (τ )
T → ∞ 2T −T
RXX (τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
= E ⎡⎣ cos (ω t + 0 ) cos (ωt + ωτ + 0 )⎤⎦

⎡ cos ( 2ωt + ωτ + 20 ) + c os ( ωτ ) ⎤
=E ⎢ ⎥
⎢
⎣ 2 ⎥⎦
1 1
= π ⎡ cos ( 2ω ωτ ) ωτ ⎤
2 ∫−π 2π ⎣
t+ − 20 + cos ⎦ d 0
1 ⎡ sin (2ωt + ωτ + 20 )
π
⎤
= + 0 cosωτ ⎥
4π ⎢⎣ 2 ⎥⎦ −π
1
RXX (τ ) = cos (ωτ )
2
Consider, T
Lt 1
T → ∞ 2T ∫ X ( t ) X ( t + τ ) dt
−T T
Lt 1
=
T→
Lt∞ 2T1 ∫ cos (ωt + 0 )cos (ωt + ωτ + 0 ) dt
∫ ⎣⎡ cos ( 2ωt + ωτ + 20 ) + cos (ωτ ) ⎤ dt⎦
−T
=
T → ∞ 4T
T
−T
1 ⎡⎢ sin (2ωt + ωτ + 20 ) + tcos (ωτ )⎥ ⎤

T
Lt
=
T → ∞ 4T ⎢ ⎣ 2ω ⎥⎦ −T
1
= cos (ωτ )
2
lim 1 T 1
T → ∞ 2π ∫ X ( t ) X ( t + τ ) dt =
−T 2
cos ( ωτ ) = R (τ )
∴{X (t )} is correlation ergodic.
25
b). A random process{X (t )} defined as{X (t )} = A cosωt + B sin ωt , where A & B are
the random variables with E ( A) = E ( B) = 0 and E ( A2 ) = E ( B 2 ) & E ( AB ) = 0 . Prove
that the process is mean ergodic.
Solution:
To prove that the process is mean ergodic we have to shoe that the ensemble mean is
same as the mean in the time sense.
Given E ( A) = E (B) = 0 ------------------ 1
E (A2 ) = E (B2 ) = k (say) & E (AB) = 0 ----- 2.
Ensemble mean is
E ⎡ ⎣ X (t )⎦ ⎤ = E [ Acosωt + Bsinωt ]
= E ( A) cosωt + E ( B ) sinωt = 0 Using (1)
Time Average
lim 1 T
X (t ) =
T → ∞ 2T −T ∫ ( Acosωt + B sinωt ) dt
lim 1 ⎡ Asinω t Bcosω t ⎤
T
= –
T → ∞ 2T ⎢ ⎣ ω ω ⎥ −T
lim 1 ⎡ ⎛ Asinωt Bcosωt ⎞ ⎛⎟ Asin
= T → ∞ 2T ⎢ ⎜ ⎜ ω (−T ) − Bcosω (−T ) ⎞⎤ ⎟⎥
ω − ω − ω ω
⎝ ⎦ ⎠
⎢⎣ ⎝ ⎠⎥ ⎦
lim 1 ⎡ Asinωt Bcos ωt Asin ωT BcosωT ⎤
= − + +
⎢
T → ∞ 2T ⎣ ω ω ω ω ⎥⎦
lim 1 ⎡ 2 AsinωT ⎤
=
T → ∞ 2T ⎢ ⎣ ω ⎥⎦
A lim ⎡ sinωT ⎤ = 0 .
=
ω T → ∞⎢ ⎣ T ⎥ ⎦
The ensemble mean =Time Average
Hence the process is mean ergodic.
26.a). Prove that in a Gaussian process if the variables are uncorrelated, then they are
independent
Solution:
Let us consider the case of two variables X t & X t
1 2
If the are uncomelated then,

r12 = r21 = r = 0
Consequently the variance covariance matrix
⎛σ 2 0⎞
∑= ⎜ 1 2 ⎟
⎝0 σ 2 ⎠
26
⎛σ 2 0⎞
Matrix of co-factors = ∑ = ⎜
2
⎟
ij
⎝0 σ 21 ⎠
∴ ∑ = σ 2σ 12 − 02 = σ 2σ 2 1 2
Eij
Now ( X − μ ) ∑ −1 ( X − μ ) = ( X − μ ) ( X − μ)
1 1
∑
Let us consider
( X − μ ) Eij ( X − μ ) 2
1
⎡σ 0 ⎤ ⎡ x1 − μ ⎤1 ⎡ x1 − μ1 ⎤
[X − μ , X – μ ]⎢ 2 ⎥ = ⎡ ⎣ ( – μ )σ + 0 + 0 + ( x – μ )σ 2 ⎤ ⎦ ⎢
2
⎥⎢ ⎥
x
1 1 2 2 0 σ2 ⎣ x −μ 1 1 2 2 2 1 x −μ
⎣ 1 ⎦ 2 2⎦ 2 2 ⎦
⎣
= (x − μ ) σ 2 +(x − μ ) σ 2
2 2
1 1 2 2 2 1
Now the joint density of X t & X t is

1 2
1 ⎡ ( x − μ ) 2 σ 2 +( x − μ ) 2 σ 2 ⎤
−
1
f (x ,x )= e 2σ σ1 ⎢ 2⎣
2 2 1 1 1 2 2 2⎥⎦
( 2π )
1 2 2/2
σ σ12 2
2
⎡
−1 ⎛ x − μ ⎞
2
⎛ x −μ ⎞ ⎤
2
⎢ ⎜ 1 1 ⎟ +⎜ 2 2 ⎟ ⎥
1 2 ⎢ ⎝ σ1 ⎠ ⎝ σ2 ⎠ ⎥⎦
= e ⎣
( 2π ) σ 1σ 2 2 2
1 ⎛ x −μ ⎞ 1 ⎛ x −μ ⎞
− ⎜ 1 1⎟ – ⎜ 2 2⎟
= 1 2 ⎝ σ1 ⎠ 2⎝ σ2 ⎠
e e
2π σ1σ 2
2 2
1 ⎛ x −μ ⎞ 1⎛ x −μ ⎞
− 2⎜ 1 1 ⎟ – ⎜ 2 2⎟
= 1 e ⎝ σ1 ⎠ 1 2⎝ σ2 ⎠
e
σ1 2π σ2 2π
f ( x1, x2 ) = f ( x1 ) f ( x2 )
∴ X t& X t are independent.
1 2
b). If { X ( t )} is a Gaussian process with μ ( t ) = 10 & C ( t1 , t2 ) = 16 e− t −t1 2

. Find the
probability that (i) X (10) ≤ 8 and (ii) X (10) − X ( 6 ) ≤ 4 .
Solution:
Since {X (t )} is a Gaussion process, any member of the process is a normal random
variable.
∴ X (10) is a normal RV with mean μ (10) = 10 and variance C (10,10) = 16 .
⎧ X (10 ) −10 ⎫
P { X (10 ) ≤ 8} = P ≤ −0.5
⎨ ⎬
4
⎩ ⎭
= P{Z ≤ −0.5} (Where Z is the standard normal RV)
= 0.5 − P{Z ≤ 0.5}
= 0.5 0.1915 = 0.3085 .
27
(from normal tables)
28
X (10) X (6) is also a normal R V With mean μ (10) − μ (6) = 10 −10 = 0

Var { X (10) − X (6)} = Var {X (10)}+ Var {X (6)} − 2Cov{X (10), X (6)}
= Cov (10,10) + Cov (6, 6) − 2Cov (10, 6)
= 16 +16 − 2 ×16e−4 = 31.4139
P X (10) − X ( 6 ) ≤ 4 = P ⎨⎧⎪ X (10) − X (6) ≤
{ } 4 ⎫⎪
⎬
⎪⎩ 5.6048 5.6048 ⎪⎭
= P{ Z ≤ 0.7137}
= 2× 0.2611 = 0.5222 .
Problem 27. a). Define a Markov chain. Explain how you would clarify the states and
identify different classes of a Markov chain. Give example to each class.
Solution:
Markov Chain: If for all n,
P { X n = an / Xn−1 = an−1, X n−2 = an−2 ,..., X 0 = a0} = P { X n = an / Xn−1 = an−1} then the
process X n } , n = 0,1, 2,... is called a Markov Chain.
Classification of states of a Markov chain
Irreducible: A Markov chain is said to be irreducible if every state can be reacted from
every other state, where Pij(n) > 0 for some n and for all i & j.
⎡ 0.3 0.7 0 ⎤
Example: ⎢ 0.1 0.4 0.5⎥
⎢ ⎥
⎢⎣ 0 0.2 0.8⎥⎦
Period: The Period di of a return state i is defined as the greatest common division of all
m such that Pij(m) > 0
i.e., d i = GCD m : p(ijm) > 0 }
State i is said to be periodic with period di if di >1 and a periodic if di =1.
Example:
⎡ 0 1⎤
So states are with period 2.
⎢ 1 0⎥
⎣ ⎦
⎡ 1 3⎤
⎢ 4 4⎥
⎢ ⎥ The states are aperiodic as period of each state is 1.
⎢ 1 1⎥
⎢⎣ 2 2
⎥⎦
Ergodic: A non null persistent and aperiodic state is called ergodic.
Example:
29
⎡0 0 1 0⎤
⎢0 0 0 1⎥
⎢ ⎥
⎢0 1 0 0 ⎥ Here all the states are ergodic.
⎢ ⎥
⎢ 1 1 1 1⎥
⎢⎣ 2 8 8 4⎥ ⎦
b). The one-step T.P.M of a Markov chain { X n ; n = 0,1, 2,...} having state space
⎡ 0.1 0.5 0.4 ⎤
⎢ 0.2 0.2 ⎥ and the initial distribution is π 0 = (0.7, 0.2, 0.1) .
S = {1, 2, 3} is P = ⎢ 0.6 ⎥
⎢⎣ 0.3 0.4 0.3⎥⎦
Find (i) P ( X 2 = 3 X 0 = 1) (ii) P ( X 2 = 3) (iii) P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 1) .
Solution:
(i) P ( X 2 = 3 / X 0 = 1) = P ( X 2 = 3 / X1 = 3) P ( X1 = 3 / X 0 = 1) + P ( X 3 = 3 / X1 = 2) P ( X1 = 2 / X 0 = 1)
+P ( X 2 = 3 / X1 = 1) P ( X1 = 1/ X 0 = 1)
= (0.3)(0.4) + (0.2)(0.5) + (0.4)(0.1) = 0.26
⎛ 0.43 0.31 0.26 ⎞
⎜ ⎟
P = P.P = 0.24 0.42 0.34
2
⎜ ⎟
⎜ 0.36 0.35 0.29 ⎟
⎝ ⎠
3
(ii). P ( X 2 = 3) = ∑ P ( X 2 = 3 / X 0 = i ) P ( X 0 = i )
i=1
= P ( X 2 = 3 / X 0 = 1)P ( X 0 = 1) + P ( X 2 = 3 / X 0 = 2 ) P ( X 0 = 2)
+ P ( X 2 = 3 / X 0 = 3)P ( X 0 = 3)
= P 2P ( X = 1) + P2 P ( X = 2) + P2 P ( X = 3)
13 0 23 0 33 0
= 0.26 × 0.7 + 0.34 × 0.2 + 0.29 × 0.1 = 0.279

(iii). P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 1)
= P[ X 0 = 1, X1 = 3, X 2 = 3] P [ X3 = 2 / X 0 = 1, X1 = 3, X 2 = 3]
= P[ X 0 = 1, X1 = 3, X 2 = 3] P [ X3 = 2 / X 2 = 3]
= P [ X 0 = 1, X1 = 3] P [ X 2 = 3 / X 0 = 1, X1 = 3] P [ X 3 = 2 / X 2 = 3]
= P[ X 0 = 1, X1 = 3] P [ X 2 = 3 / X1 = 3]P [ X3 = 2 / X 2 = 3]
= P [ X 0 = 1] P [ X1 = 3 / X 0 = 1] P [ X 2 = 3 / X1 = 3] P [ X 3 = 2 / X 2 = 3]
= (0.4)(0.3)(0.4)(0.7) = 0.0336
30
Problem 28. a). Let { X n ; n = 1, 2, 3,.....} be a Markov chain with state space S = {0,1, 2}
⎡⎢ 0 1 0 ⎤⎥
1 1 1
and 1 – step Transition probability matrix P = ⎢ (i) Is the chain ergodic?
⎥
⎢ 4 2 4⎥
Explain (ii) Find the invariant probabilities. ⎢ 0 1 0⎥
Solution: ⎣ ⎦
⎡ 1 1 1⎤
⎢ ⎥
⎢⎡ 01 11 10 ⎥⎤ ⎢⎡ 01 11 01 ⎤⎥ ⎢ 14 23 14 ⎥
P2 = P.P = ⎢ ⎥⎢ ⎥ =⎢ ⎥
⎢⎢ 4 2 4 ⎥⎥ ⎢⎢ 4 2 4 ⎥⎥ ⎢⎢ 8 4 8 ⎥⎥
0 1 0 0 1 0 1 1 1
⎣ ⎦⎣ ⎦ ⎢ ⎥
⎢ ⎣ 4 2 4⎥ ⎦
⎛1 1 1 ⎞ ⎛1 3 1⎞
⎜ 4 2 4 ⎟⎛⎜ 0 1 0 ⎟⎞ ⎜ 8 4 8 ⎟
1 3 1 1 1 1 ⎜ 3 5 3⎟
P =P P=
3 2 ⎜ ⎟ ⎜ ⎟ = ⎜ ⎟
⎜ 8 4 8 ⎟⎜ 4 2 4 ⎟ ⎜ 16 8 16 ⎟
⎜ ⎟ ⎜ ⎟
1 1 1 ⎜⎝ 0 1 0 ⎟⎠ ⎜ 18 34 18 ⎟
⎜4 2 4 ⎟
⎝ ⎠ ⎝ ⎠
P11(3) > 0, P13(2) > 0, P21(2) > 0, P22(2) > 0, P33(2) > 0 and all other Pij(1) > 0
Therefore the chain is irreducible as the states are periodic with period 1
i.e., aperiodic since the chain is finite and irreducible, all are non null persistent
The states are ergodic.
⎡⎢ 0 1 0 ⎤⎥
1 1 1
[π 0 π 1 π 2 ] ⎢ 4 2 4 ⎥ = [π 0 π1 π2 ]
⎢ ⎥
⎢ 0 1 0⎥
⎣ ⎦
π1
= π – − − − − − − − − − − −(1)
0
4
π
π + 1 + π 2 = π – − − − − − − − − − − −(2)
0 1
2
π1
= π – − − − − − − − − − − −(3)
2
4
π 0 + π1 + π 2 = 1− − − − − − − − − − − −(4)
π π
From (2) π + π = π − 1 = 1
0 2 1
2 2
∴π 0 + π1 + π 2 = 1
π1
+π =1
1
2
31
3π1
=1
2
2
π=
1
3
π1
From (3) =π
2
4
1
π=
2
6
2 1
Using (4) π + + = 1
0
3 6
4 +1
π + =1
0
6
5 1
π + = 1⇒ π =
0 0
6 6
1 2 1
∴π 0 + , π 1 = & π 2 = .
6 3 6
⎛0 1 0⎞
⎜1 1 ⎟⎟ and
b). Find the nature of the states of the Markov chain with the TPM P = ⎜ 0
⎜2 2⎟
⎜1 0 0 ⎟⎠
⎝
the state space(1, 2, 3).
Solution:
⎛1 1⎞
0
⎜ 2 ⎟⎟
⎜2
P = ⎜0 1 0 ⎟
2
⎜1 ⎟
⎜ 0 1⎟
⎝2 2⎠
⎛ 0 1 0⎞
⎜1 ⎟
P3 = P2.P = ⎜ 0 1 ⎟= P
⎜2 2⎟
⎜0 1 0 ⎟
⎝ ⎠
⎛1 1⎞
0
⎜2 2 ⎟⎟
⎜
P = P .P = ⎜ 0 1 0 ⎟ = P2
4 2 2
⎜1 ⎟
⎜ 0 1 ⎟
⎝2 2⎠
∴P = P & P
2n 2 2 n+1
=P
Also P002 > 0, P011 > 0, P022 > 0
P101 > 0, P112 > 0, P121 > 0
32
P2 > 0, P1 > 0, P2 > 0

20 21 22
The Markov chain is irreducible

Also Pii2 = Pii4 = ... > 0 for all i
The states of the chain have period 2. Since the chain is finite irreducible, all states are
non null persistent. All states are not ergodic.
Problem 29. a). Three boys A, B and C are throwing a ball to each other. A always
throws to B and B always throws to C, but C is as likely to throw the ball to B as to A.
Find the TPM and classify the states.
Solution:
A B C
⎛ ⎞
A⎜ 0 1 0⎟
P= ⎜ ⎟
B⎜0 0 1⎟
C⎜1 1 ⎟
⎜ 0⎟
⎝2 2 ⎠
⎛ ⎞
⎜0 0 1⎟
⎜ ⎟
P = P× P = 1
2 ⎜ 1 0 ⎟
⎜2 2 ⎟
⎜ 1 1 ⎟
⎜0 ⎟
⎝ 2 2⎠
⎛1
⎜ 1 ⎞
2 2 0 ⎟
⎜ ⎟
1 1⎟
P3 = P2 × P = ⎜ 0 2 2⎟
⎜
⎜ ⎟
1 1 1
⎜ ⎟
⎝4 4 2⎠
For any i = 2, 3
Pii2 Pii3,... > 0
G.C.D of 2, 3, 5,... = 1
The period of 2 and 3 is 1. The state with period 1 is aperiodic all states are ergodic.
b). A man either drives a car or catches a train to go to office each day. He never goes 2
days in a row by train but he drives one day, then the next day he is just as likely to drive
again as he is to travel by train. Now suppose that one the first day of the week, the man
tossed a fair dice and drove to work iff a 6 appeared. Find the probability that he takes a
train on the third day and also the probability that on the third day and also the
probability that he drives to work in the long run.
Solution:
State Space = (train, car)
33
The TPM of the chain is

T C
⎛0 1⎞
P=T ⎜ ⎟
1 1⎟
C ⎜
⎝2 2⎠
1
P (traveling by car) = P (getting 6 in the toss of the die)=
6
5
& P (traveling by train) =
6
⎛5 1 ⎞
P(1) = ⎜ , ⎟
⎝6 6 ⎠
⎛0 1 ⎞ ⎛ 1 11 ⎞
⎛ 5 ,1 ⎞ ⎜ 1 1 ⎟ = , ⎟
P = P P=⎜
(2) (1) ⎟ ⎜
⎝ 6 6 ⎠⎜ ⎟ ⎝ 12 12 ⎠
⎝⎛20 2 ⎠1 ⎞
⎛ 1 11 ⎞ ⎛ 11 13 ⎞
() () , ⎜1 1 ⎟= , ⎟
P = P P=⎜
3 2 ⎟ ⎜
⎝ 12 12 ⎠⎜ ⎟ ⎝ 24 24 ⎠
⎝2 2 ⎠
11
P (the man travels by train on the third day) =
24
Let π = (π 1 , π 2 ) be the limiting form of the state probability distribution or stationary
state distribution of the Markov chain.
By the property of π , π P = π
⎛0 1 ⎞
(π 1 π 2 ) ⎜1 1 ⎟⎟ = (π 1 π 2 )
⎜
⎝2 2 ⎠
1
π 2 = π1
2
1
π + π 2 =π
1 2
2
& π1 + π 2 = 1
1 2
Solving π1 = & π 2 =
3 3
2
P{The man travels by car in the long run }= .
3
Problem 30 a). Three are 2 white marbles in urn A and 3 red marbles in urn B. At each
step of the process, a marble is selected from each urn and the 2 marbles selected are inter
changed. Let the state ai of the system be the number of red marbles in A after i
changes. What is the probability that there are 2 red marbles in A after 3 steps? In the
long run, what is the probability that there are 2 red marbles in urn A?
34
Solution:
State Space X n } = (0,1, 2) Since the number of ball in the urn A is always 2.
0 1 2
⎛ ⎞
⎜0 1 0 ⎟
0 ⎜ ⎟
P= 1 1 1
1 ⎜ ⎟
⎜6 2 3 ⎟
2 ⎜
2 1⎟
⎜0 ⎟
⎝ 3 3⎠
X n = 0 , A = 2W (Marbles) B = 3R (Marbles)
Xn+1 = 0 P00 = 0
Xn+1 = 1 P01 = 1
Xn+1 = 2 P02 = 0
Xn = 0 , A = 1W &1R (Marbles) B = 2R &1W (Marbles)
1
X =0 P =
n+1 10
6
1
X =1 P =
n+1 11
2
1
X =2 P =
n+1 12
3
Xn = 2 , A = 2R (Marbles) B =1R & 2W (Marbles)
Xn+1 = 0 P20 = 0
2
X =1 P =
n+1 21
3
1
X =2 P =
n+1 22
3
P = (1, 0, 0) as there is not red marble in A in the beginning.
( 0)
P(1) = P(0)P = (0, 1, 0)

⎛1 1 1⎞
P(2) = P(1) P =
⎜ , , ⎟
⎛⎝16 223 3 ⎠5 ⎞
P(3) = P(2)P = , ,
⎝⎜ 12 36 18 ⎟ ⎠ 5
= 2 } = P( 3 ) =
P (There are 2 red marbles in A after 3 steps) = P X 3 2
18
Let the stationary probability distribution of the chain be π = (π 0 , π 1 , π 2 ) .
By the property of π , π P = π & π 0 + π1 + π 2 = 1
35
⎛ ⎞
⎜0 1 0⎟
⎜ ⎟
1 1 1
( π0 π π ) ⎜
⎜
6 2 3⎟ ( 0 1
⎟ = ππ π )1
2 2
⎜
2 1⎟
⎜0 ⎟
⎝ 3 3⎠
1
π1 = π 0
6
1
π + π + 2 π 2 = π1
0
2 1 3
1 1
π 1 + π 2 = π2
3 3
& π 0 + π1 + π 2 = 1
1 6 3
Solving π 0 = , π 1 = , π 2 =
10 10 10
P {here are 2 red marbles in A in the long run} = 0.3.
b). A raining process is considered as a two state Markov chain. If it rains the state is 0
⎛ 0.6 0.4 ⎞
and if it does not rain the state is 1. The TPM is P = ⎜ ⎟ . If the Initial distribution
⎝ 0.2 0.8 ⎠
is (0.4, 0.6) . Find it chance that it will rain on third day assuming that it is raining today.
Solution:
⎛ 0.6 0.4 ⎞⎛ 0.6 0.4 ⎞
P2 = ⎜ ⎟⎜ ⎟
⎝0.2 0.8 ⎠⎝0.2 0.8 ⎠
⎛ 0.44 0.56 ⎞
=⎜ ⎟
⎝ 0.28 0.32 ⎠
P[rains on third day / it rains today = P[ X 3 = 0 / X1 = 0]
= P002 = 0.44
36
Unit.2. Two Dimensional Random Variables
UNIT – II TWO DIMENSIONAL RANDOM VARIABLES
Part.A
Problem 1. Let X and Y have joint density function f ( x, y ) = 2, 0 < x < y < 1.Find the marginal
density function. Find the conditional density function Y given X = x .
Solution:
Marginal density function of X is given by
∞
f X ( x) = f ( x) = ∫ f ( x, y ) dy
−∞
1 1
= ∫ f ( x, y ) dy = ∫ 2dy = 2 ( y )
1
x
x x
= 2 (1− x ) , 0 < x < 1.

Marginal density function of Y is given by
∞
fY ( y ) = f ( y ) = ∫ f ( x, y ) dx
−∞
y
= ∫ 2dx = 2 y, 0 < y < 1 .
( x ) = f ( x, y ) = 2 = 1 .
0
Conditional distribution function of Y given X = x is f y

f ( x) 2 (1− x ) 1− x
⎧0 , x < −a
⎪1 ⎛ x ⎞
Problem 2. Verify that the following is a distribution function. F ( x ) = ⎨ ⎜ +1⎟ , −a < x < a .
2 a
⎪ ⎝ ⎠
⎪⎩1 ,x > a
Solution:
F ( x) is a distribution function only if f ( x) is a density function.
d 1
f ( x ) = ⎡⎣ F ( x ) ⎤⎦ = , –a < x<a
dx 2a
∞
∫ f ( x) = 1
−∞
a 1 1 1
a
∴∫
2a
dx =
2a
[ ]−a 2a ⎡⎣ a − ( −a )⎤⎦
x =
−
a
1
1
=
.2a = 1 .
2a
Therefore, it is a distribution function.
x2
Problem 3. Prove that

∫ f X ( x) dx = p ( x1 < x < x2 )
x1
Solution:
( x ) dx = ⎡ F ( x ) ⎤
x2
x2 f
∫
x1
X ⎣ X ⎦ x1
= FX ( x2 ) − FX ( x1 )
= P [ X ≤ x 2 ] − P [ X ≤ x 1]
= P[x1 ≤ X ≤ x2 ]
Problem 4. A continuous random variable X has a probability density function f ( x ) = 3x2 ,

0 ≤ x ≤ 1. Find ' a ' such that P ( X ≤ a ) = P ( X > a ) .
Solution:
1
Since P ( X ≤ a ) = P ( X > a ) , each must be equal to because the probability is always 1.
2
1
∴ P ( X ≤ a) =
2
a
1
⇒ ∫ f ( x) dx =
0
2
a
a
1 ⎡ x3 ⎤ 1
∫ 3x 2
dx =
2
⇒ 3 ⎢ ⎥ = a 3
= .
0 ⎣ 3 ⎦0 2
1
⎛ 1 ⎞3
∴a = ⎜ ⎟
⎝2 ⎠
Problem5. Suppose that the joint density function

⎧ Ae−x− y , 0 ≤ x ≤ y, 0 ≤ y ≤ ∞
f ( x, y ) = ⎨ Determine A .
⎩0 , otherwise
Solution:
Since f ( x, y ) is a joint density function
∞ ∞
∫ ∫ f ( x, y ) dxdy = 1.
−∞ −∞
∞y
⇒ ∫ ∫ Ae−xe− ydxdy = 1
0 0
2
∞ y
⎛ e− x ⎞
⇒ A∫ e− y ⎜ ⎟ dy = 1
0 ⎝ −1 ⎠0
∞
⇒ A∫ ⎡⎣ e− y − e −2 y ⎤⎦ dy = 1
0
−2 y ∞
⎡ e− y e ⎤
⇒ A ⎢− ⎥ =1
⎣ −1 −2 ⎦ 0
⎡ 1⎤
⇒A =1⇒ A = 2
⎢ ⎣ 2⎥ ⎦
Problem 6. Examine whether the variables X and Y are independent, whose joint density
function is f ( x, y ) = xe−x( y+1) , 0 < x, y < ∞ .
Solution:
The marginal probability function of X is
∞ ∞
f X ( x) = f ( x ) = xe− x( y+1)dy
∫ f ( x, y ) dy = ∫
−∞ 0
∞
− x( y +1)
⎡ ⎤
= x⎢ e ⎥ = − ⎡⎣ 0 − e − ⎤⎦ = e − ,
x x
⎣ −x ⎦ 0
The marginal probability function of Y is
∞ ∞
fY ( y ) = f ( y ) = xe − x( y+1)dx
∫ f ( x, y ) dx = ∫
−∞ 0
⎪⎧ e − x( y +1) ⎤ ∞ ⎡ e− x( y +1) ⎤ ⎫ ⎪∞
=x ⎡ −⎢ ⎥
2 ⎬
⎨⎢ ⎥
⎩⎪ ⎣ − ( y +1) ⎦ 0 ⎢ ( y +1) ⎥ ⎪
⎣ ⎦⎭ 0
1
=
( y +1)
2
1
Here f ( x ) . f ( y ) = e−x × ≠ f ( x, y )
(1+ y )
2
∴ X and Y are not independent.
Problem 7. If X has an exponential distribution with parameter 1. Find the pdf of y = x

Solution:
Since y = x , x = y2
Since X has an exponential distribution with parameter 1, the pdf of X is given by
f X ( x) = e−x , x > 0 ⎡⎣ f ( x ) = λ e−λ x , λ = 1⎤⎦
dx
∴ fY ( y ) = f X ( x )
dy
2
= e − x 2 y = 2 ye − y
3
fY ( y ) = 2 ye − y , y > 0
2
⎛ π π⎞
Problem 8. If X is uniformly distributed random variable in −⎜ , , Find the probability
2 2⎟
⎝ ⎠
density function of Y = tanX .
Solution:
Given Y = tanX ⇒ x = tan−1 y
dx 1
∴ =
dy 1+ y2
⎛ π π⎞
Since X is uniformly distribution in − ⎜ , ,
2 2⎟
⎝ ⎠
1 1
f ( x) = =
b−a π +π
X
2 2 π
1 π
f ( x) = , − < x <
X
π 2 2 ⎞
Now f ( y ) = f ( x) dx = 1 ⎛ 1 , −∞< y <∞
⎜ ⎟
dy π ⎝ 1+ y2 ⎠
Y X
1
∴ f ( y) = , −∞ < y < ∞
π (1+ y2 )
Y
Problem 9. If the Joint probability density function of ( x, y ) is given by f ( x, y ) = 24 y (1− x ) ,

0 ≤ y ≤ x ≤ 1 Find E ( XY ) .
Solution:
1 1
E ( xy ) = ∫ ∫ xyf ( x, y ) dxdy y
0y
1 1
= 24 ∫ ∫ xy 2 (1− x ) dxdy x= y
0y
1
⎡ 1 y2 y3 ⎤ 4
= 24∫ y ⎢ − 2
+ ⎥ dy = . x
0 ⎣6 2 3⎦ 15
Problem 10. If X and Y are random Variables, Prove that Cov ( X ,Y ) = E ( XY ) − E ( X ) E (Y )

Solution:
cov ( X ,Y ) = E ⎣ ⎡ ( X − E ( X ) )( Y − E (Y ))⎤ ⎦
= E ( XY − XY − YX + XY )
= E ( XY ) − XE (Y ) − YE ( X ) + XY
= E ( XY ) − XY − XY + XY
4
= E ( XY ) − E ( X ) E (Y ) ⎡⎣ E ( X ) = X , E ( Y ) = Y ⎤⎦
Problem 11. If X and Y are independent random variables prove that cov ( x, y ) = 0
Proof:
cov ( x, y ) = E ( xy ) − E ( x) E ( y )
But if X and Y are independent then E ( xy ) = E ( x ) E ( y )
cov ( x, y ) = E ( x) E ( y ) − E ( x) E ( y )
cov ( x, y ) = 0.
Problem 12. Write any two properties of regression coefficients.

Solution:
1. Correction coefficients is the geometric mean of regression coefficients
2. If one of the regression coefficients is greater than unity then the other should be less
than 1.
σy σ
bxy = r and b = r x
yx
σx σy
If bxy > 1 then byx < 1 .
Problem 13. Write the angle between the regression lines.
Solution: The slopes of the regression lines are

σy
m1= r , m 2 = 1 σ y
σ rσ
x x
If θ is the angle between the lines, Then

σ xσ y ⎡ 1− r 2 ⎤
tan θ = 2 2 ⎢
σ x+ σ y ⎣ r ⎥⎦
π
When r = 0 , that is when there is no correlation between x and y, tan θ = ∞ (or) θ =
2
and so the regression lines are perpendicular
When r = 1or r = −1 , that is when there is a perfect correlation +ve or −ve , θ = 0 and so the
lines coincide.
Problem 14. State central limit theorem

Solution:
If X1, X 2....Xn is a sequence of independent random variable E ( X i ) = μi and
Var ( X i ) = σ i2 , i = 1, 2,..... n and if Sn = X 1 + X 2 +...... + Xn then under several conditions Sn
n n
follows a normal distribution with mean μ = ∑ μi and varianceσ = ∑σ 2

2
i as n → ∞ .
i =1 i =1
Problem 15. i). Two random variables are said to be orthogonal if correlation is zero.
5
ii). If X = Y then correlation coefficient between them is 1.
Part-B
Problem 16. a). The joint probability density function of a bivariate random variable ( X ,Y ) is
⎧⎪k ( x + y ) , 0 < x < 2, 0 < y < 2
f ( x, y ) = where ' k ' is a constant.
XY ⎨
⎩⎪0 , otherwise
i. Find k .
ii. Find the marginal density function of X and Y .
iii. Are X and Y independent?
iv. Find fY
X
( y x ) and f ( x y ) .
X
Y
Solution:
(i). Given the joint probability density function of a brivate random variable ( X ,Y ) is
⎪⎧ K ( x + y ) , 0 < x < 2, 0 < y < 2
f XY (x, y ) = ⎨
⎪⎩0 , otherwise
∞ ∞ ∞∞
Here ∫∫ f XY ( x, y ) dxdy = 1 ⇒ ∫ ∫ K ( x + y ) dxdy = 1

−∞ −∞ −∞ −∞
2 2 2 2
⎡ x2 ⎤
∫∫ K ( x + y ) dxdy = 1 ⇒ K ∫ ⎢ 2 + xy ⎥ dy = 1
0 0 0 ⎣ ⎦0
2
⇒ K ∫ ( 2 + 2 y ) dy = 1
0
2
⇒ K ⎡⎣ 2 y + y 2 ⎤⎦ 0 = 1
⇒ K [8 − 0] = 1
1
⇒K=
8
(ii). The marginal p.d.f of X is given by
∞
12
( ) ( x + y )dy
8 ∫0
f X ( x) = ∫ f x, y dy =
−∞
2
1⎡ 2
⎤
= 8 xy + y ⎥ = 1+4 x
⎢ 2
⎣ ⎦0
∴ The marginal p.d.f of X is ⎧ x +1
, 0< x<2
⎪
f X ( x) = ⎨ 4
⎪⎩0 , otherwise
The marginal p.d.f of Y is
∞
12
fY ( y ) = ∫ f ( x, y ) dx = ∫ ( x + y ) dx
−∞
80
6
2
1 ⎢⎡ x2 ⎤
= 8 2 + yx ⎥
⎣ ⎦0
1 y +1
= [2 + 2 y] =
8 4
∴The marginal p.d.f
⎧ y +1
of Y is
, 0< y<2
⎪
fY ( y) = ⎨ 4
⎪⎩0 , otherwise
(iii). To check whether X and Y are independent or not.
( x +1) ( y +1)
f X ( x ) fY ( y ) = ≠ f XY ( x, y )
4 4
Hence X and Y are not independent.
(iv). Conditional p.d.f fY

X
( y x1) is given by
f ( x, y ) ( x + y) ( x + y)
fY ( )
y
= = 8 =
1
f X ( x)
( x +1) 2 ( x +1)
X x 1
4
( )
⎜ ⎟
y 1⎛x+ y⎞
Y
f X x = , 0 < x < 2, 0 < y < 2
⎛ 1 2 ⎝ x +1 ⎠
0< y< ⎞
2
⎜
(v) P ⎜
⎜
2 ⎟
x = 1⎟ = ∫ fY
⎟ 0 X (
y = 1 dy
x
)
⎝ ⎠
1
1 2 1+ y 5
2 ∫0 2
= dy = .
32
Problem 17.a).⎧1 If X and Y are two random variables having joint probability density function
( 6 − x − y ) , 0 < x < 2, 2 < y < 4
⎪ Find (i) P ( X < 1∩Y < 3)
f ( x, y ) = ⎨ 8
⎪⎩0 , otherwise
(
(ii) P ( X + Y < 3) (iii) P X < 1
Y <3
. )
b). Three balls are drawn at random without replacement from a box containing 2 white, 3 red
and 4 black balls. If X denotes the number of white balls drawn and Y denotes the number of
red balls drawn find the joint probability distribution of ( X ,Y ) .
Solution:
a).
y =3 x=1
P ( X < 1∩ Y < 3) = ∫ ∫ f ( x, y ) dxdy

y =−∞ x =−∞
7
y =3 x=1
1
=∫ ∫ 8 (6 − x − y ) dxdy
y =2 x=0
3 1
1
=
8 ∫2 ∫0
(6 − x − y ) dxdy
1
13⎡ x2 ⎤
= ∫ ⎢ 6x − – xy ⎥ dy
82⎣ 2 ⎦0 3
1 3 ⎡ 11 ⎤ 1 ⎡ 11y y 2 ⎤
= ∫ ⎢ − y ⎥ dy = ⎢ − ⎥
8 ⎣ 2 ⎦ 8 ⎣2 2 ⎦
2 2
3
P ( X < 1∩ Y < 3) =
8
1 3− x
1
(ii). P ( X + Y < 3) = ∫ ∫ (6 − x − y ) dydx
0 2
8
11⎡ 3− x
y2 ⎤
= ∫ ⎢ 6 y − xy − ⎥ dx
8 ⎣ 2 ⎦
0 2
11 ( 3 − x) − [12 − 2x − 2]⎤⎥ dx
2
⎡ ( ) ( )
= 8 ∫ ⎢ ⎢6 3 − x − x 3 − x − 2 ⎥⎦
0⎣
1 1 ⎡ (
9 + x2 − 6x ) ⎤
∫
= ⎢18 − 6x − 3x + x − 2
(
− 10 − 2x ⎥ dx )
80 2 ⎥⎦
⎢⎣ 9 x 6x 2
⎤
11⎡
= ∫ ⎢ 18 − 9x + x 2 − − + −10 + 2x ⎥ dx
80⎣ 2 2 2 ⎦
11⎡ 7 x2 ⎤
= ∫ ⎢ − 4x + ⎥ dx
8 2 2
0⎣ ⎦
1
3
1 ⎢⎡ 7x 4x 2 x ⎤ 1⎡ 7 1⎤
= 8 2 – 2 + ⎥ = 8 ⎢ ⎣ 2− 2 + 6 ⎥ ⎦
⎣ 6 ⎦0
1 ⎡ 21−12 +1 ⎤ 1 ⎛ 10 ⎞ 5
= = = .
⎢ ⎥ 8 ⎜ 6 ⎟ 24
8 ⎣ 6 ⎦ ⎝ ⎠
P ( x < 1∩ y < 3)
(
(iii). P X < 1
Y <3 = ) P ( y < 3)
2
The Marginal density function of Y is fY ( y ) = f ( x, y )dx

∫ 0
2
1
=∫ ( 6 − x − y)dx
0 8
8
1⎡ x2 2
= 8 6x − 2 – yx ⎤⎥
⎢
⎣ ⎦0
1
= [12 − 2 − 2 y]
8
5− y
= , 2 < y < 4.
4
x =1 y =3
∫ ∫ (6 − x − y ) dxdy
1
(
8
P X <1
Y <3 = ) x=0 y =2
y =3
∫ fY ( y ) dy
y =2
3 3
= 3 ⎛ 5 −8y ⎞ = 1 ⎡ 8 y 2 ⎤ 3
∫⎝⎜ 4 ⎠⎟ dy ⎢ 5 y − 2 ⎥
2 4⎣ ⎦2
3 8 3
= × = .
8 5 5
b). Let X takes 0, 1, 2 and Y takes 0, 1, 2 and 3.

P ( X = 0,Y = 0) = P( drawing 3 balls none of which is white or red)
= P( all the 3 balls drawn are black)
4C 4 × 3× 2 ×1 1
= 3= = .
9C3 9 × 8× 7 21
P ( X = 0,Y = 1) = P( drawing 1 red ball and 2 black balls)
3C1 × 4C2 3
= =
9C3 14
P ( X = 0,Y = 2) = P( drawing 2 red balls and 1 black ball)
3C2 × 4C1 3× 2 × 4 × 3 1
= = = .
9C3 9 × 8× 7 7
P ( X = 0,Y = 3) = P( all the three balls drawn are red and no white ball)
3C3 1
= =
9C3 84
P ( X = 1,Y = 0) = P( drawing 1White and no red ball)
2 × 4× 3
2C1 × 4C2
= = 1× 2
9C3 9 × 8× 7
1× 2× 3
9
12 ×1× 2× 3 1
= = .
9 × 8× 7 7
P ( X = 1,Y = 1) = P( drawing 1White and 1 red ball)
2× 3
2C1 × 3C1 9× 8× 7 2
= = =
9C3 1× 2 × 3 7
P ( X = 1,Y = 2) = P( drawing 1White and 2 red ball)
2C1 × 3C2 2 × 3× 2 1
= = =
9C3 9 × 8× 7 14
1× 2 × 3
P ( X = 1,Y = 3) = 0 (Since only three balls are drawn)
P ( X = 2,Y = 0) = P( drawing 2 white balls and no red balls)
2C2 × 4C1 1
= =
9C3 21
P ( X = 2,Y = 1) = P( drawing 2 white balls and no red balls)
2C2 × 3C1 1
= =
9C3 28
P ( X = 2,Y = 2) = 0
P ( X = 2,Y = 3) = 0
The joint probability distribution of ( X ,Y ) may be represented as
Y
0 1 2 3
X
1 3 1 1
0
21 14 7 84
1 2 1
1 0
7 7 14
1 1
2 0 0
21 28
Problem 18.a). Two fair dice are tossed simultaneously. Let X denotes the number on the first
die and Y denotes the number on the second die. Find the following probabilities.
(i) P ( X + Y ) = 8 , (ii) P ( X + Y ≥ 8) , (iii) P ( X = Y ) and (iv)
( )
6
P X +Y = Y .
=4
b) The joint probability mass function of a bivariate discrete random variable ( X ,Y ) in given by
the table.
X
Y 1 2 3
1 0.1 0.1 0.2
2 0.2 0.3 0.1
10
Find
i. The marginal probability mass function of X and Y .
ii. The conditional distribution of X given Y = 1.
iii. P ( X + Y < 4)
Solution:
a). Two fair dice are thrown simultaneously
⎧(1,1)(1, 2 ) ... (1, 6 ) ⎫
⎪ 2,1 2, 2 ... 2, 6 ⎪
⎪( )( ) ( )⎪
S=⎨ ⎬ , n(S ) = 36
⎪ . . ... . ⎪
⎪( 6,1)( 6, 2 ) ... ( 6, 6 ) ⎪
⎩ ⎭
Let X denotes the number on the first die and Y denotes the number on the second die.
1
Joint probability density function of ( X ,Y ) is P ( X = x,Y = y ) = for
36
x = 1, 2, 3, 4, 5, 6 and y = 1, 2, 3, 4, 5, 6
(i) X + Y = { the events that the no is equal to 8 }
= {(2, 6),(3, 5 ) , ( 4, 4 ) , ( 5, 3 ) , ( 6, 2)}
P ( X + Y = 8) = P ( X = 2,Y = 6) + P ( X = 3,Y = 5) + P ( X = 4,Y = 4)
+ P ( X = 5,Y = 3) + P ( X = 6,Y = 2)
1 1 1 1 1 5
= + + + + =
36 36 36 36 36 36
(ii) P ( X + Y ≥ 8)
⎧( 2, 6 ) ⎫
⎪ ⎪
( 3, 5) , ( 3, 6 )
X + Y = ⎪( 4, 4 ) , ( 4, 5 ) , ( 4, 6 )
⎪
⎨⎪ ⎬⎪
( 5, 3) , ( 5, 4 )( 5, 5 ) , ( 5, 6 )
⎪ ⎪
⎪( 6, 2 ) , ( 6, 3) , ( 6, 4 ) , ( 6, 5 )( 6, 6 ) ⎪
⎩ ⎭
∴ P ( X + Y ≥ 8) = P ( X + Y = 8) + P ( X + Y = 9) + P ( X + Y = 10)
+ P ( X + Y = 11) + P ( X + Y = 12)
5
= + 4 + 3 + 2 + 1 = 15 = 5
36 36 36 36 36 36 12
(iii) P ( X = Y )
P ( X = Y ) = P ( X = 1,Y = 1) + P ( X = 2,Y = 2) +...... + P ( X = 6,Y = 6)
1 1 1
= + + .......... + =
6 1
=
36 36 36 36 6
P ( X + Y = 6 ∩Y = 4)
(
(iv) P X + Y = 6
Y = 4 = ) P (Y = 4)
11
1
Now P ( X + Y = 6 ∩ Y = 4) =
36
6
P (Y = 4) =
36
1
(
∴P X +Y = 6
Y =4
) = 366 = 61 .
36
b). The joint probability mass function of ( X ,Y ) is
X 1 2 3 Total
Y
1 0.1 0.1 0.2 0.4
2 0.2 0.3 0.1 0.6
Total 0.3 0.4 0.3 1
From the definition of marginal probability function
PX ( xi ) = ∑ PXY ( xi , y j )
yj
When X = 1 ,
PX ( xi ) = PXY (1,1) + PXY (1, 2)
= 0.1+ 0.2 = 0.3
When X = 2 ,
PX ( x = 2) = PXY (2,1) + PXY (2, 2)
= 0.2 + 0.3 = 0.4
When X = 3 ,
PX ( x = 3) = PXY (3,1) + PXY (3, 2)
= 0.2 + 0.1 = 0.3
∴ The marginal probability mass function of X is
⎧0.3 when x = 1
⎪
PX ( x ) = ⎨0.4 when x = 2
⎪0.3 when x = 3
⎩
The marginal probability mass function of Y is given by PY ( y j ) = ∑ PXY ( xi , y j )
xi
3
When Y =1, PY ( y = 1) = ∑ PXY ( xi ,1)

xi =1
= PXY (1,1) + PXY (2,1) + PXY (3,1)

= 0.1+ 0.1+ 0.2 = 0.4
3
When Y = 2 , PY ( y = 2) = ∑ PXY ( xi , 2)
xi =1
= PXY (1, 2) + PXY (2, 2) + PXY (3, 2)

= 0.2 + 0.3 + 0.1 = 0.6
12
∴ Marginal probability
⎧0.4 mass function of Y is
P ( y) = when y = 1
⎨
⎩0.6 when y = 2
Y
(ii) The conditional distribution of X given Y =1 is given by

(
P X=x
Y =1 )=
P ( X = x ∩ Y = 1)
P (Y = 1)
From the probability mass function of Y , P( y =1) = Py (1) = 0.4
(
When X =1 , P X = 1
Y =1) =
P ( X = 1∩ Y = 1)
P (Y = 1)
PXY (1,1) 0.1
= = = 0.25
PY (1) 0.4
(
When X = 2 , P X = 2 =
PXY (2,1) 0.1
) = = 0.25
Y =1 PY (1) 0.4
(
When X = 3 , P X = 3
Y =1 =
PXY (3,1) 0.2
) = = 0.5
PY (1) 0.4
(iii). P ( X + Y < 4) = P{( x, y ) / x + y < 4 Where x = 1, 2, 3; y = 1, 2}
= P{(1,1), (1, 2),(2,1)}
= PXY (1,1) + PXY (1, 2) + PXY (2,1)
= 0.1+ 0.1+ 0.2 = 0.4
Problem 19.a). If X and Y are two random variables having the joint density function
1
f ( x, y ) = ( x + 2 y ) where x and y can assume only integer values 0, 1 and 2, find the
27
conditional distribution of Y for X = x .
b). The joint probability density function of ( X ,Y ) is given by
⎧ 2 x2
⎪ xy + , 0 ≤ x ≤ 2, 0 ≤ y ≤ 1
f XY ( x, y ) = ⎨ 8 . Find (i) P( X >1) , (ii) P( X < Y ) and
⎪0 , otherwise
⎩
(iii) P ( X +Y 1)
Solution:
a). Given X and Y are two random variables having the joint density function
1
f ( x, y ) = ( x + 2 y ) − − − −(1)
27
Where x = 0,1, 2 and y = 0,1, 2
Then the joint probability distribution X and Y becomes as follows
13
Y
0 1 2 f1 ( x )
X
1 2 3
0 0
27 27 27
2 3 4 9
1
27 27 27 27
4 5 6 15
2
27 27 27 27
The marginal probability distribution of X is given by f1 ( X ) = ∑ P ( x, y ) and is calculated in
j
the above column of above table.

f ( x, y )
The conditional distribution of Y for X is given by f1 Y = ( y
X
=x )
= f 1 ( x)
and is obtained in
the following table.

X 0 1 2
Y
1 2
0 0
3 3
1 3 5
1
9 9 9
1 1 1
2
6 3 2
(
P Y =0
) = P (PX( X= 0,Y = 0)
=
0
=0
X =0 = 0) 6
27
2
(
P Y =1
) =
P ( X = 0,Y = 1) 27 1
= =
X =0 P ( X = 0) 6 3
27
4
(
P Y =2
X =0 ) =
P ( X = 0,Y = 2) 27 2
= =
P ( X = 0) 6 3
27
1
(
P Y =0
X =1 ) =
P ( X = 1,Y = 0) 27 1
= =
P ( X = 1) 9 9
27
3
(
P Y =1
X =1 ) =
P ( X = 1,Y = 1) 27 3 1
= = =
P ( X = 1) 9 9 3
27
14
5
(
P Y =2 P ( X = 1,Y = 2) 27 5
X =1 = ) = =
P ( X = 1) 9 9
27
2
(
P Y =0
X =2
=
P ( X = 2,Y = 0) 27 1
) = =
P ( X = 2) 12 6
27
4
(
P Y =1
X =2
=
P ( X = 2,Y = 1) 27 1
) = =
P ( X = 2) 12 3
27
6
(
P Y =2
X =2
=
P ( X = 2,Y = 2) 27 1
) = =
P ( X = 2) 12 2
27
x2
b). Given the joint probability density function of ( X ,Y ) is f XY ( x + y ) = xy 2 + ,
8
0 ≤ x ≤ 2, 0 ≤ y ≤ 1
∞
(i). P ( X > 1) = f X ( x ) dx
∫ 1
1
The Marginal density function of X is f X ( x) = f ( x, y ) dy

∫ 0
2
1
x ⎞
f X ( x) = ∫ ⎜⎛ xy2 + y =1
⎟ dy y
0⎝ 8 ⎠
1
⎡ xy 2 x 2 y ⎤ x x2
=⎢ + ⎥ = 3 + 8 , 1< x < 2
⎣ 3 8 ⎦0
2
⎛ x x2 ⎞
P ( X > 1) = ∫ ⎜ + ⎟ dx x
3 8
1⎝ ⎠
2
⎡ x2 + x ⎥⎤
3
19
= ⎢ 6 24 = 24 . y x =1 x=2
⎣ ⎦1
(ii) P ( X < Y ) =
∫ ∫ f XY ( x, y ) dxdy
R2
y =1
1 y
⎛ x2 ⎞
P( X < Y ) = y=x
∫y =0 x=0∫ ⎜⎝ xy2 + 8 ⎟⎠ dxdy
y
1
⎡ 2 2 x3 ⎤
= ∫⎢ x y
+ ⎥ dy x=0 x
2 24 ⎦ 0
0⎣
15
1
1
3
⎛ y4 y ⎞ ⎡ y5 y 4 ⎤
= ∫ ⎜ + ⎟ dy = ⎢ + ⎥
2 24 10 96 ⎦ 0
0 ⎝ ⎠ ⎣
1 1 96 +10 53
= + = =
10 96 960 480
(iii) P ( X + Y ≤ 1) = f XY ( x, y ) dxdy
∫∫ R3
Where R3 is the region

1 1− y
⎛ x2 ⎞
P ( X + Y ≤ 1) =
∫ ∫ ⎜ xy2 + 8 ⎟⎠ dxdy
y =0 x=0 ⎝
y
1− y
1
⎡ 2 2
= ∫ ⎢ ⎜⎛ x y + x ⎞⎤ dy
3
2 ⎟⎥ y =1
y=0 ⎣ ⎝
24 ⎠⎦
0
( )
1 ⎛ 1− y 2 y 2
(1− y ) ⎞
3
= ⎜
∫ ⎜ 2 + 24 ⎟dy ⎟ x+ y =1
y=0 ⎝ ⎠
1 ⎛
( )
1+ y2 − 2 y y2 (1− y )3 ⎞
= ∫⎜ + 24 ⎟⎟dy x
0⎜ 2
⎝ ⎠ 1
2 y ⎤⎥ 1 + (1− y ) ⎥⎤
4
⎡⎡ y 3
y5 2
=⎢⎢ + − 4 2 96 x =1
⎢⎣ ⎣ 3 5 ⎥⎦ 0
⎦
1 1 1 1
= + − + =
13
.
6 10 4 96 480
Problem20 a). If the joint distribution functions of X and Y is given by

⎧⎪(1− e x )(1− e− y ) , x > 0, y > 0
F ( x, y ) = ⎨
⎪⎩0 , otherwise
i. Find the marginal density of X and Y .
ii. Are X and Y independent.
iii. P(1< X < 3, 1< Y < 2).
b). The ⎧ 6 −joint
x − y probability distribution of X and Y is given by
, 0 < x < 2, 2 < y < 4
⎪
f ( x, y ) = ⎨ 8 (
. Find P 1< Y < 3
X =2 )
.
⎪⎩0 , otherwise
Solution:
( )(
a). Given F ( x, y ) = 1 − e − x 1− e − y )
= 1− e− x − e− y + e−( x+ y )
The joint probability density function is given by
16
∂ 2 F ( x, y )
f ( x, y ) =
∂x∂y
−( x+ y )
∂2 ⎡ –x −y ⎤
=
1− e −e +e
∂x∂y ⎣ ⎦
∂ ⎡ −y
= e − e −( x+ y ) ⎤
∂x ⎧⎣ −( x + y ) ⎦
⎪e , x ≥ 0, y ≥ 0
∴
f ( x, y ) = ⎨
⎪⎩0 , otherwise
(ii) The marginal probability function of X is given by
f ( x) = f X ( x )
∞ ∞
= ∫ f ( x, y ) dy = ∫ e−( x+ y )dy
−∞ 0
∞
−( x+ y )
⎡ ⎤
=⎢ e ⎥
−1
⎣ ⎦0
∞
= ⎡ −e−( x+ y ) ⎤
⎣ ⎦0
=e , x>0
−x
The marginal probability function of Y is

f ( y ) = fY ( y )
∞
= ∫ f ( x, y ) dx
−∞
∞
− ∞
( x+ y )
= ∫e dx = ⎡ −e−( x+ y ) ⎤
⎣ ⎦0
0
= e− y , y > 0
∴ f ( x) f ( y ) = e−xe− y = e−( ) = f ( x, y )
x+ y
∴ X and Y are independent.

(iii) P (1 < X < 3,1 < Y < 2) = P (1 < X < 3)× P (1 < Y < 2) [Since X and Y are independent]
3 2
= ∫ f ( x)dx × ∫ f ( y ) dy.
1 1
3 2
= ∫ e−xdx × ∫ e− ydy
1 1
3 2
⎡⎤ e− x
⎡ e− y ⎤
=⎢ ⎥ ⎢× ⎥
⎣ − ⎦1 ⎣ − ⎦1
1
( 1
= −e−3 + e−1 −e −2 + e −1 )( )
= e−5 − e−4 − e−3 + e−2
17
( )
3
(
b). P 1< Y < 3
X =2 )= f y x = 2 dy
1
4
f X ( x) = ∫ f ( x, y ) dy
2
4 6−x− y
⎛ ⎞
= ∫⎜ ⎟dy
⎝ 8 ⎠ 4
2
1⎛ y2 ⎞
= ⎜ 6 y − xy − ⎟
8⎝ 2 ⎠2
1
= (16 − 4x −10 + 2x )
8
6− x− y
( )
f y
x
=
f ( x, y )
f ( x)
= 8
6 − 2x
=
6− x− y
6 − 2x
3
8
(
P 1< Y < 3
X =2 )=
1
( )
f y x = 2 dy
3 4− y
⎛ ⎞
= ∫⎜ ⎟dy
⎝ 2 ⎠
2
3
1⎡ 2
⎤
= 2 4y − y ⎥
⎢ 2
⎣ ⎦2
3
1⎡ y 2
⎤ 1⎡
= 2 4 y − ⎥ = 2 14 − 17 ⎤⎥ 11 .
⎢ 2 ⎢ 2 = 4
⎣ ⎦2 ⎣ ⎦
Problem 21. a). Two random variables X and Y have the following joint probability density
⎧2 − x − y, 0 ≤ x ≤ 1, 0 ≤ y ≤ 1
function f ( x, y ) = ⎨ . Find the marginal probability density function
⎩0 , otherwise
of X and Y . Also find the covariance between X and Y .
6− x− y
b). If f ( x, y ) = , 0 ≤ x ≤ 2, 2 ≤ y ≤ 4 for a bivariate ( X ,Y ) , find the correlation
8
coefficient
Solution:
⎧2 − x − y, 0 ≤ x ≤ 1, 0 ≤ y ≤ 1
a) Given the joint probability density function f ( x, y ) = ⎨
⎩0 , otherwise
∞
Marginal density function of X is f X ( x) = ∫ f ( x, y ) dy

−∞
1
= ∫ ( 2 − x − y ) dy
0
18
1
⎡ 2
⎤
= 2 y − xy − y ⎥
⎢ 2
⎣ ⎦0
1
= 2− x−
2
⎧3
− x, 0 < x ≤ 1
f X ( x) = ⎪⎨2
⎪⎩0 , otherwise
1
Marginal density function of Y is fY ( y ) = ( 2 − x − y ) dx

∫ 0
1
⎡ x2
= 2x − 2 – xy ⎥⎤
⎢
⎣ ⎦0
3
= −y
2
⎧3
− y, 0 ≤ y ≤ 1
fY ( y ) = ⎪⎨2
⎪⎩0 , otherwise
Covariance of ( X ,Y ) = Cov ( X ,Y ) = E ( XY ) − E ( X ) E (Y )
1
1 1
⎛3 ⎞ ⎡ 3 x2 – x 3 ⎤ = 5
E ( X ) = ∫ xf X ( x ) dx = ∫ x ⎜ − x ⎟ dx = ⎢ ⎥
3 12
2 22
0 0 ⎝ ⎠ ⎣ ⎦ 0
1 1
⎛3 ⎞ 5
E (Y ) = ∫ yfY ( y ) dy = ∫ y ⎜ − y ⎟ dy =
2 12
0 0 ⎝ ⎠
Cov ( X ,Y ) = E ( XY ) − E ( X ) E (Y )
1 1
E ( XY ) = ∫ ∫ xy f ( x, y ) dxdy
0 0
1 1
= ∫ ∫ xy ( 2 − x − y ) dxdy
0 0
1 1
= ∫ ∫ (2xy − x2 y − xy2 )dxdy

0 0 1
1
⎡ 2x2 y x3 x2
⎤
=∫⎢ − y− y2 ⎥ dy
0 ⎣ 2 3 2 ⎦0
1
⎛ 1 y2 ⎞
= ∫ ⎜ y − − ⎟ dy
3 2
0⎝ ⎠1
2
⎡y y y3 ⎤ 1
=⎢ − − ⎥ =
⎣ 2 3 6 ⎦0 6
19
1 5 5
Cov ( X ,Y ) = × −
6 12 12
1 25 1
= − =− .
6 144 144
E ( XY ) − E ( X ) E (Y )
b). Correlation coefficient ρXY =
σ Xσ Y
Marginal density function of X is 4 6−x− y
∞ ⎛ ⎞ 6 − 2x
f X ( x ) = ∫ f ( x, y ) dy = ∫ ⎜ ⎟ dy =
8 8
−∞ 2⎝ ⎠
Marginal density function of Y is
2 6− x− y
∞ ⎛ ⎞ 10 − 2 y
fY ( y ) = ∫ f ( x, y ) dx = ∫ ⎜ ⎟ dx =
8 8
−∞ 0 ⎝ ⎠
2 2
⎛ 6 − 2x ⎞
Then E ( X ) = ∫ xf X ( x) dx = ∫ x ⎜ ⎟ dx
8
0 0 ⎝ ⎠
2
2 3
= 1 ⎢⎡ 6x − 2x ⎥⎤
8 2 3
⎣ ⎦0
1⎡ 16 ⎤ 1 20 5
= 12 − = × =
8 ⎢⎣ 13 ⎥ ⎦ 48 3 6
4
⎛ 10 − 2 y ⎞ 1 ⎡ 10 y 2 2 y 3 ⎤ 17
E (Y ) = ∫ y ⎜ dy = − =
8 ⎟ 8⎢ 2 3 ⎥ 6
2 ⎝ ⎠ ⎣ ⎦2 2
( )
E X 2 = x2 f ( x) dx
2 2
= x2
⎛ 6 − 2x ⎞
dx = 1 ⎡ 6x 3
−
2x 4 ⎤
=1
∫
40
x
−
∫0 1⎜⎝ ⎡ 108y 3 ⎟⎠ 2 y 4 8⎤ 4⎢⎣ 325 4 ⎥⎦ 0
⎛ 10 2 y ⎞
E (Y 2 ) = ∫ y2 ⎜ ⎟ dy = ⎢ – ⎥ =
2 ⎝ 8 ⎠ 8 ⎣ 3 4 ⎦2 3
Var ( X ) = σ 2 = E ( X 2 ) − ⎡ E ( X )⎤ = 1−
2 ⎛ 5 ⎞ 2 11
=
X ⎣ ⎦ ⎜ ⎟
⎝ 6⎠ 36
Var ( Y ) = σ = E ( Y ) − ⎡ E ( Y ) ⎤ =
2 25 ⎛ 17 ⎞ 2 11
2 2
− ⎜ ⎟=
Y ⎣
⎦ 3 ⎝ 6 ⎠ 36
4 2
⎛6−x− y ⎞
E ( XY ) = ∫ ∫ xy ⎜ ⎟ dxdy
⎝ 8 ⎠
2 0
2
1 4 ⎡ 6x 2 y x 3 y x 2 y 2 ⎤
= 8∫⎢ 2
– − ⎥ dy
2⎣
3 2 ⎦0
4
4 8 2⎞ ⎡ 2 3
⎤
= ∫ ⎜12 y − y − 2 y dy = ⎢ 2 y
1 12 y 2
1 ⎛ 8 y
⎟ – − ⎥
82⎝ 3 ⎠ 8⎣ 2 3 2 3 ⎦2
20
1⎡ 64 128 16 16 ⎤ 1 ⎡ 56 ⎤
= 96 − − − 24 + + =
8⎢⎣ 3 3 ⎥ ⎢
3 3 ⎦ 8 ⎣ 3 ⎦
⎥
7
E ( XY ) =
3
7 ⎛⎜5 ⎞⎛
⎟⎜17 ⎞ ⎟
−
E ( XY ) − E ( X ) E (Y )
6 6
ρXY 3 ⎝ ⎠⎝ ⎠
= =
σ XσY 11 11
6 6
1
ρ =− .
XY
11
Problem 22 a). Let the random variables X and Y have pdf
1
f ( x, y ) = , ( x, y ) = (0, 0),(1,1), (2, 0) . Compute the correlation coefficient.
3
b) Let X1 and X 2 be two independent random variables with means 5 and 10 and standard
devotions 2 and 3 respectively. Obtain the correlation coefficient of UV where U = 3X1 + 4 X 2
and V = 3X1 − X 2 .
Solution:
a). The probability distribution is
X 0 1 2 P (Y )
Y
1 1
0 3 0 0 3
1 1
1 0 3 0 3
1 1
0 0 0 3 3
P( X ) 1 1 1
3 3 3
⎛ 1⎞ ⎛ 1⎞ ⎛ 1⎞
E ( X ) = ∑ x p ( x ) = 0 × + 1× + 2 × = 1
i i i ⎜ 3⎟ ⎜ 3⎟ ⎜ 3 ⎟
i
⎛⎝ 1 ⎞ ⎠ ⎛ ⎝ 1 ⎞ ⎛⎠ ⎝1 ⎞ 1⎠
E ( Y ) = ∑ y p ( y ) = 0 × + 1× + 0 × = ⎟
i j j ⎜ 3⎟ ⎜ 3⎟ ⎜ 3 3
j ⎝ 1 ⎠ ⎝ 1 ⎠ ⎝1 5⎠
( ) ⎛ ⎞ ⎛
E X 2 = ∑ x 2 p ( x ) = 0 × + 1× + 4 × = ⎟
⎜
⎞ ⎛ ⎞
i i
3⎟ ⎜ 3⎟ ⎜ 3 3
i ⎝ ⎠ ⎝ ⎠ ⎝ ⎠
( )
Var( X ) = E X 2 − ⎡⎣ E ( X ) ⎤⎦ = −1 =
2 5
3
2
3
21
( )
E Y2 = ∑ y 2 p( y
j
) = ⎛⎜0 × 1 ⎞ +⎟ ⎛1×⎜ 1 ⎞ + ⎟⎛ 0 × 1⎜ ⎞ = 31⎟
j 3
3 3
j ⎠ ⎝ ⎠ ⎝⎝ ⎠
1 1 2
∴V (Y ) = E Y 2 − E ( Y )⎤⎦ = − =
2
( ) 3 9 9
E ( XY ) − E ( X ) E (Y )
Correlation coefficient ρXY =
V ( X )V (Y )
E ( XY ) = ∑∑ xi y j p ( xi , y j )
i j
1 1 1 1
= 0.0. + 0.1.0 +1.0.0 +1.1. +1.2.0 + 0.0.0 + 0.1.0 + 0.2. =
3 3 3 3
1 ⎛1⎞
− (1) ⎜ ⎟
ρXY = 3 ⎝ 3⎠ = 0
2 2
×
3 9
Correlation coefficient = 0 .
b). Given E ( X1 ) = 5, E ( X 2 ) = 10
V ( X1 ) = 4, V ( X 2 ) = 9
Since X and Y are independent E ( XY ) = E ( X ) E (Y )
E (UV ) − E (U ) E (V )
Correlation coefficient =
Var (U )Var (V )
E (U ) = E (3X1 + 4X 2 ) = 3E ( X1 ) + 4E ( X 2 )
= (3× 5) + ( 4 ×10) = 15 + 40 = 55.
E (V ) = E (3X1 − X 2 ) = 3E ( X1 ) − E ( X 2 )
= (3× 5) −10 = 15 −10 = 5
E (UV ) = E ⎡⎣ ( 3X 1 + 4 X 2 )( 3X 1 − X 2 )⎤⎦
= E ⎡⎣ 9X 1 − 3X 1 X 2 +12 X 1 X 2 − 4 X 2 ⎤⎦
2 2
( ) ) − 4E ( X )
= 9E X 12 − 3E ( X 1 X 2) +12E ( X X1 2
2
2
= 9E ( X ) + 9E ( X X ) − 4E ( X )
2
1 1 2
2
2
= 9E ( X ) + 9E ( X ) E ( X ) − 4E ( X )
2
1 1 2
2
2
= 9E ( X ) + 450 − 4E ( X )
2 2
V ( X ) = E ( X ) − ⎡ E ( X )⎤
1 2 2
2
1 ⎣ 1 ⎦ 1
E ( X ) = V ( X ) + ⎡⎣ E ( X ) ⎤ = 4 + 25 = 29
2
2
1 ⎦ 1 1
E ( X ) = V ( X ) + ⎡⎣ E ( X ) ⎤ = 9 +100 = 109
2
2
2
⎦
2 2
22
∴ E (UV ) = (9 × 29) + 450 − (4 ×109)

= 261+ 450 − 436 = 275
Cov(U ,V ) = E (UV ) − E (U ) E (V )
= 275 − ( 5× 55 ) = 0
Since Cov (U ,V ) = 0, Correlation coefficient = 0 .
Problem 23. a). Let the random variable X has the marginal density function
1 1
f ( x ) = 1, − < x < and let the conditional density of Y be
2 2
⎧
1, x < y < x +1, – 1 < x < 0
f( )
y
x =⎪ ⎨ 2
1
. Prove that the variables X and Y are uncorrelated.
⎪1, – x < y < 1− x, 0 < x <
⎩ 2
b). Given f ( x, y ) = xe− ( + ) , x ≥ 0, y ≥ 0 . Find the regression curve of Y on X .
x y 1
Solution:
1 1 1
2 2
⎡ x2 ⎤ 2
a). We have E ( X ) = ∫ xf ( x) dx = ∫ xdx = ⎢ ⎥ =0
−
1
−
1 ⎣2 ⎦ − 1
2 2 2
1
0 x+1 2 1−x
E ( XY ) = ∫ ∫ xydxdy + ∫ ∫ xydxdy
−
1 x 0 −x
2
1
0
⎡ x+1 ⎤ 2
⎡ 1− x ⎤
=∫x⎢ ∫ ydy ⎥ dx + ∫ x ⎢ ∫ ydy ⎥ dx
−1 ⎣ ⎦ 0 ⎣ −x ⎦
x
1
2
12
0
1
=
2 ∫ x ( 2x +1) dx + 2 ∫ x (1 − 2x ) dx
1 0
−
2
1
1 ⎡ 2x3 x2 ⎤ 0 1 ⎡ x2 2x3 ⎤ 2
= 2⎢ 3 + 2 ⎥ + 2⎢ 2 − 3 ⎥ =0
⎣ ⎦ −1 ⎣ ⎦0
2
Since Cov ( X ,Y ) = E ( XY ) − E ( X ) E (Y ) = 0 , the variables X and Y are uncorrelated.
y
b). Regression curve of Y on X is E x ( )
( x ) = ∫ yf ( y x ) dy
∞
E y
−∞
f ( x, y )
f ( y / x) =
fX ( X )
23
Marginal density function f X ( x) =

∫ f ( x, y ) dy
0
∞
= x ∫ e − x( y +1) dy
0
∞
− x( y+1)
⎡ e ⎤
⎥ = e−x x ≥ 0
= x⎢
−x ,
⎣ ⎦0
f
X ( x )( )
Conditional pdf of Y on X is f y x = f ( x, y ) = xe− − = xe− xy
e –x
xy x
The regression curve of Y on X is given by
( x ) =∫
∞
E y yxe−xydy
0
∞
⎡ e− xy e− xy ⎤
− x⎢ y
= x2 ⎥
⎣ −x ⎦0
( ) =x1 ⇒ y = x1 and hence xy⎧ x=+1.y

E yx
, 0 < x < 1, 0 < y < 2
⎪
Problem 24. a). Given f ( x, y ) = ⎨ 3 ,
⎪⎩0 , otherwise
obtain the regression of Y on X and X on Y .
b). Distinguish between correlation and regression Analysis
Solution:
a). Regression of Y on X is E Y X ( )
( X ) = ∫ yf ( y x ) dy
α
E Y
−α
f ( x, y )
f (Y ) =
X f ( x) X
∞
f X ( x) = ∫ f ( x, y ) dy
−∞ 2
⎛ x+ y⎞
2 1⎡ y2 ⎤
=∫⎜
⎟ dy = ⎢ xy + ⎥
3 3 2
0⎝ ⎠ ⎣ ⎦0
2 ( x +1)
=
3
( )
f Y =
f ( x, y ) = x + y
X f X ( x) 2(x +1)
24
(
2 y x+ y
)
Regression of Y on X = E Y (X ) = ∫ 2 ( x +1) dy
0
2
= 1 ⎢⎡ xy2 + y 3⎥ ⎤
2 ( x +1) 2 3
⎣ ⎦0
1 ⎡ 8⎤
= 2x + = 3x + 4
2 ( x +1) ⎢ ⎣ 3⎥ ⎦ 3 x +1
( )
( Y ) = ∫ xf ( x y ) dx
∞
E X
−∞
f ( x y ) = f f( x,( yy))
Y
∞
fY ( y ) = ∫ f ( x, y ) dx
−∞ 1
⎛ x+ y⎞
1 1 ⎡ x2 ⎤
=∫⎜
⎟ dx = ⎢ + xy ⎥
⎝ 3 ⎠ 3 2
0 ⎣ ⎦0
1⎡ 1 ⎤
= +y
3⎢⎣ 2 ⎥ ⎦
2( x + y)
y( )
f x =
2 y +1
( Y) x+ y
1
Regression of X on Y = E X =∫ dx
0 2 y +1
1
2
= 1 ⎡⎢ x ⎤
2 y +1 2 + xy ⎥
⎣ ⎦0
1
+y
2 1
= = .
2 y +1 2
b).
1. Correlation means relationship between two variables and Regression is a Mathematical
Measure of expressing the average relationship between the two variables.
2. Correlation need not imply cause and effect relationship between the variables. Regression
analysis clearly indicates the cause and effect relationship between Variables.
3. Correlation coefficient is symmetric i.e. rxy = ryx where regression coefficient is not symmetric
4. Correlation coefficient is the measure of the direction and degree of linear relationship
between two variables. In regression using the relationship between two variables we can predict
the dependent variable value for any given independent variable value.
25
Problem 25. a). X any Y are two random variables with variances σ 2 x and σ 2y respectively and
yσ x
r is the coefficient of correlation between them. If U = X + KY andV = X + , find the
σy
value of k so that U and V are uncorrelated.
b). Find the regression lines:
X 6 8 10 18 20 23
Y 40 36 20 14 10 2
Solution:
GivenU = X + KY
E (U ) = E ( X ) + KE (Y )
σ
V = X + XY
σY
σ
E (V ) = E ( X ) + X E (Y )
σY
If U and V are uncorrelated, Cov (U ,V ) = 0
E ⎡⎣ (U − E (U ) )(V − E (V ) ) ⎤⎦ = 0
⎡ ⎛ σX σX ⎞⎤
⇒
E ⎢ ( X + KY − E ( X ) − KE (Y ) ) × ⎜ X + Y − E( X )− E (Y ) ⎟⎥ = 0
⎣ ⎝ σ Y σ Y ⎠⎦
⎡ σX ⎫
⎤⎪
⇒ E ⎨ ⎡ ( X − E ( X )) + K (Y − E (Y ) ) ⎤ × ⎢( X − E ( X ) ) + (Y − E (Y ) ) =⎥ ⎬0
⎣ ⎦
⎩ ⎣ σ
Y ⎦⎭
⎧ σX σ ⎫
⇒ E ( X − E ( X ) ) + ( X − E ( X ))( Y − E (Y ) ) + K ( Y − E (Y ) )( X − E ( X ) ) + K (Y − E (Y ) ) = 0
2 2
X
⎨ ⎬
⎩ σY Y σ ⎭
σX σX
⇒V (X )+ Cov ( X ,Y ) + KCov ( X ,Y ) + K V (Y ) = 0
σY σY
⎡ σX ⎤ σX
K ⎢ Cov ( X ,Y ) + − (Y )⎥ = −V ( X )
V Cov ( x, y )
⎣ σY ⎦ σY
σ
−V ( X ) − X rσ σ
σY X Y
−σ 2X − rσ 2 X
K= σ =
rσ X σ Y + X
V (Y ) rσ X σ Y + σ X σ Y
σY
−σ 2X (1+ r ) = − σ X .
=
σ X σ Y (1+ r ) σY
26
b).
X Y X2 Y2 XY
6 40 36 1600 240
8 36 64 1296 288
10 20 100 400 200
18 14 324 196 252
20 10 400 100 200
23 2 529 4 46
X = 85 Y =122 X 2 =1453 Y 2 = 3596 XY =1226
∑x 85 ∑y 122
X = = = 14.17 , Y = = = 20.33
n 6 n 6
2 2
∑ x2 ⎛ ∑ x ⎞ 1453 ⎛ 85 ⎞
σx = –⎜ ⎟ = – ⎜ ⎟ = 6.44
n ⎝ n ⎠ 6 ⎝6 ⎠
2
∑ y2 ⎛ ∑ y ⎞
2
3596 ⎛ 122 ⎞
σy = –⎜ ⎟ = – = 13.63
n ⎝ n ⎠ 6 ⎜⎝ 6 ⎠⎟
1226
∑ xy
−xy − (14.17)(20.33)
r= n = 6 = −0.95
σ xσ y (6.44)(13.63)
σx 6.44
bxy = r = −0.95× = −0.45
σy 13.63
σy 13.63
byx = r = −0.95× = −2.01
σx 6.44
The regression line X on Y is
( )
x − x = bxy y − y ⇒ x −14.17 = −0.45 y − y ( )
⇒ x = −0.45 y + 23.32
The regression line Y on X is
( )
y − y = byx x − x ⇒ y − 20.33 = −2.01( x −14.17 )
⇒ y = −2.01x + 48.81
Problem 26. a) Using the given information given below compute x , y and r . Also compute
σ y when σ x = 2, 2x + 3y = 8 and 4x + y = 10 .
b) The joint pdf of X and Y is
27
X
Y -1 1
1 3
0 8 8
2 2
1 8 8
Find the correlation coefficient of X and Y .
Solution:
a). When the regression equation are Known the arithmetic means are computed by solving the
equation.
2x + 3y = 8------------- (1)
4x + y = 10 ------------ (2)
(1) × 2 ⇒ 4x + 6 y = 16 --------- (3)
( 2) − ( 3) ⇒ −5 y = −6
6
⇒y=
5
⎛6⎞
Equation (1) ⇒ 2x + 3 =8
⎜5 ⎟
⎝ ⎠
18
⇒ 2x = 8 −
5
11
⇒x=
5
11 6
i.e. x = & y =
5 5
To find r , Let 2x + 3y = 8 be the regression equation of X on Y .
3
2x = 8 − 3y ⇒ x = 4 − y
2
3
⇒ bxy = Coefficient of Y in the equation of X on Y = −
2
Let 4x + y = 10 be the regression equation of Y on X
⇒ y = 10 − 4x
⇒ byx = coefficient of X in the equation of Y on X = −4 .
r = ± bxybyx
= − ⎛ − 3 ⎞ −4
⎜ 2 ⎟( )
( b &b
xy yx
are negative )
⎝ ⎠
= −2.45
Since r is not in the range of (−1 ≤ r ≤ 1) the assumption is wrong.
Now let equation (1) be the equation of Y on X
28
8 2x
⇒ y= −
3 3
⇒ byx = Coefficient of X in the equation of Y on X
2
b =−
yx
3
from equation (2) be the equation of X on Y
1
b =−
xy
4
2 1
r = ± bxybyx = – × − = 0.4081
3 4
2
To compute σ from equation ( 4 ) b = −
y yx
3
σy
But we know that byx = r σ
x
2 σy
⇒− = 0.4081×
3 2
⇒ σ y = −3.26
b). Marginal probability mass function of X is
1 3 4
When X = 0, P ( X ) = + =
8 8 8
2 2 4
X = 1, P ( X ) = + =
8 8 8
Marginal probability mass function of Y is
1 2 3
When Y = −1, P (Y ) = + =
8 8 8
3 2 5
Y = 1, P (Y ) = + =
8 8 8
4 4 4
E ( X ) = ∑ x p ( x) = 0 × +1× =
x 8 8 8
3 5 3 5 2
E (Y ) = ∑ y p ( y ) = −1× +1× = − + =
y 8 8 8 8 8
( ) 4
E X 2 = ∑ x2 p ( x) = 02 × +12 × =
4 4
x 8 8 8
E (Y ) = ∑ y p ( y ) = ( −1) × +1 × = + = 1
2 2 2 3 2 5 3 5
y 8 8 8 8
( )
V ( X ) = E X 2 − ( E ( X ))
2
2
4 ⎛4⎞ 1
= −⎜ ⎟=
8 ⎝8 ⎠ 4
29
V ( Y ) = E (Y 2 ) − ( E (Y ) )
2
2
⎛ 1 ⎞ 15
= 1− ⎜ ⎟ =
⎝ 4 ⎠ 16
E ( XY ) = ∑∑ xy p ( x, y )
x
1y 3 2 ⎛2⎞
= 0× + 0 × + ( −1) +1× =0
⎜
⎝8⎟ ⎠
8 8 8
1 1 1
Cov ( X ,Y ) = E ( XY ) − E ( X ) E (Y ) = 0 − × = −
1 2 4 8
Cov ( X ,Y ) −
r= = 8 = −0.26 .
V ( X ) V (Y ) 1 15
4 16
Problem 27. a) Calculate the correlation coefficient for the following heights (in inches) of
fathers X and their sons Y .
X 65 66 67 67 68 69 70 72
Y 67 68 65 68 72 72 69 71
b) If X and Y are independent exponential variates with parameters 1, find the pdf of
U = X −Y .
Solution:
X Y XY X2 Y2
65 67 4355 4225 4489
66 68 4488 4359 4624
67 65 4355 4489 4285
68 72 4896 4624 5184
69 72 4968 4761 5184
70 69 4830 4900 4761
72 71 5112 5184 5041
X = 544 Y = 552 XY = 37560 X 2 = 37028 Y 2 = 38132
∑x 544
X= = = 68
n 8
∑y 552
Y= = = 69
n 8
XY = 68× 69 = 4692
30
1 2 1
σX = ∑ x2 − X = (37028) − 682 = 4628.5 − 4624 = 2.121
n 8
1 1
σY = ∑ y2 − y2 = (38132) − 692
= 4766.5 − 4761 = 2.345
n 8
1 1
Cov ( X ,Y ) = ∑ XY − X Y = (37650) − 68× 69
n 8
= 4695 − 4692 = 3
The correlation coefficient of X and Y is given by
Cov ( X ,Y ) 3
r ( X ,Y ) = =
σ XσY (2.121)(2.345)
3
=
= 0.6032 .
4.973
b). Given that X and Y are exponential variates with parameters 1
f X ( x) = e−x , x ≥ 0, f Y( y ) = e− y , y ≥ 0
Also f XY ( x, y ) = f X ( x) f y ( y ) since X and Y are independent
= e− xe− y
= e−( x+ y ) ; x ≥ 0, y ≥ 0
Consider the transformations u = x − y and v = y
⇒ x = u +v, y = v v
∂x ∂x
∂ ( x, y ) ∂v 1 −1
J = = ∂u = =1
∂ (u, v) ∂y ∂y 0 1
∂u ∂v
fUV (u, v) = f XY ( x, y ) J = e− xe− y = e ( )e−v
− u +v
= e−(u+2v) , u + v ≥ 0, v ≥ 0 RI R II
In Region I when u < 0 v= u
∞ ∞
f (u ) = ∫ f (u, v) dv = ∫ e−u .e−2vdv u

−u −u
∞
⎡ e−2v ⎤
=e ⎢ −u
⎥
⎣ −2 ⎦ −u
e−u e
u
= 0−e ⎦ = ⎤
2u
⎡
−2 ⎣ 2
In Region II when u > 0
∞
f (u ) = ∫ f (u, v)dv
0
∞ −( u +2v ) −u
= e dv = e
∫
0 2
31
⎧ eu
, u<0
∴ f (u ) = ⎨ ⎪ 2
−u
⎪e
u>0
,
⎩⎪ 2
Problem 28. a) The joint pdf of X and Y is given by f ( x, y ) = e−(x+ y) , x > 0, y > 0 . Find the
X +Y
pdf of U = .
2
b) If X and Y are independent random variables each following N (0, 2) , find the pdf of
Z = 2 X + 3Y . If X and Y are independent rectangular variates on (0,1) find the distribution of
X
.
Y
Solution:
x+ y
a). Consider the transformation u = &v = y
2
⇒ x = 2u − v and y = v
∂x ∂x
∂ ( x, y ) ∂v 2 −1
J= = ∂u = =2
∂ (u, v) ∂y ∂y 0 1
∂u ∂v
fUV (u, v) = f XY ( x, y ) J
= e−( x+ y ) 2 = 2e−( x+ y) = 2e−(2u −v+v)
= 2e−2u , 2u − v ≥ 0, v ≥ 0
u
fUV (u, v) = 2e−2u , u ≥ 0, 0 ≤ v ≤
2
u u
2 2
f (u ) = ∫ fUV (u, v ) dv = ∫ 2e−2udv

0 0
u
= ⎡⎣ 2e v⎤⎦ −2u 2
0
⎧ u −2u
⎪2 e , u ≥ 0
f (u ) = ⎨ 2
⎪⎩0 , otherwise
b).(i) Consider the transformations w = y ,

i.e. z = 2x + 3y and w = y
1
i.e. x = ( z − 3w), y = w
2
32
∂x ∂x 1
∂ –3 1
J =
( x, y ) =
∂z ∂w =
2
2 = .
∂ ( z, w)
∂y ∂y
0 1
2
∂z ∂w
Given that X and Y are independent random variables following N (0, 2)
(
− x2 + y2 )
∴ f XY ( x, y ) = 1 e 8 , −∞ < x, y < ∞
8π
The joint pdf of ( z, w) is given by
fZW ( z, w) = J f XY ( x, y )
⎡1 ⎤
( z −3w)
2
− + w2
1 1 ⎢ ⎥
= . e ⎣4 8 ⎦
2 8π 1
1 − ⎡ ( z −3w) +4 w ⎤ 2 2
= e 32 ⎥ ⎦ , −∞ < z, w < ∞ .
16π ⎢ ⎣
The pdf of z is the marginal pdf obtained by interchanging fZW ( z, w) w.r.to w over the range of
w.
1 ∞ ⎛ − 1 ( z −6 wz +13w ) ⎞ 2 2
∴ fZ ( z ) =
∫ ⎜e 32
16π −∞ ⎝
⎟dw
⎠
z ∞ ⎛ − 13 ⎜ w − 6 wz +⎛⎜3z ⎞⎟ −⎛⎜3 z ⎞⎟ ⎟ ⎞
2
⎛ ⎞
2
2 2
–
1
e 32 ∫ ⎜ e 32 ⎝ 13 ⎝ 13 ⎠ ⎝ 13 ⎠ ⎠ ⎟dw
⎜ ⎟
=
16π −∞ ⎜ ⎟
⎝ ⎠
∞ ⎛ 3z ⎞ ⎞
2
z 9z2 13 ⎛
2
z2 ∞ 13 2
– ⎟
1
∫ e dt
− + ⎜ w− ⎟ dw −32 t
= e 32 13×32 ∫ ⎜ e 32 ⎝ 13 ⎠ 1 − 8×13
16π −∞
⎜ ⎟ = 16π e −∞
⎝ ⎠
13 2 13 16 r32
r = t ⇒ dr = tdt ⇒ dr = dt ⇒ dr = dt
32 16 13t 13
16 13 4 –
1
dr = dt ⇒ r 2 dr = dt
13 r32 13 × 2
2 4 –
z ∞
−
1 2
2
dr
= 16π 13 × 2 e 8×13 ∫0 e−r r
z2
z2 ∞ 1 z2 −
1 − −
e 2( 2 )
1 – 1 2
= ∫e r π =
−r 13
e 8×13 2
dr = e 8×13
2π 13 × 2 0 2π 13 × 2 2 13 2π
i.e. Z (
N 0, 2 13 )
b).(ii) Given that X and Y are uniform Variants over (0,1)
⎧1, 0 < x < 1 ⎧1, 0 < y < 1
∴ fX ( x) = ⎨ and f Y ( y ) = ⎨
⎩0, otherwise ⎩ 0, otherwise
33
Since X and Y are independent,

f ( x, y ) = f ( x ) f ⎧1, 0 < x, y < 1
XY X y
( y) ⎨
⎩0, otherwise
x
Consider the transformation u = and v = y
y
i.e. x = uv and y = v
∂x ∂x
∂ ( x, y ) v 0 =v
J= = ∂u ∂v =
∂ (u, v) ∂y ∂y u 1
∂u ∂v
∴ fUV (u, v ) = f XY ( x, y ) J
= v, 0 < u < ∞, 0 < v < ∞
The range for u and v are identified as follows.
0 < x < 1 and 0 < y < 1.
⇒ 0 < uv < 1 and 0 < v < 1
⇒ uv > 0, uv < 1, v > 0 and v < 1
⇒ uv > 0 and v > 0 ⇒ u > 0
Now f (u ) = fUV (u, v ) dv
∫
The range for v differs in two regions
1
f (u ) = ∫ fUV (u, v)dv

0 1
1
⎡ v2 ⎤ 1
= ∫ vdv = ⎢ ⎥ = 2 , 0 < u < 1
0 ⎣2 ⎦ 0
1
1 1 2
⎡v ⎤ u
1
f (u ) = ∫ fUV (u, v )dv = ∫ vdv = ⎢ ⎥
u u
= 2u2 , 1 ≤ u ≤ ∞
0 0 ⎣2 ⎦ 0
⎧1
⎪2 , 0 ≤ u ≤1
∴ f (u ) =
⎨
1
⎪ , u >1
⎪⎩ 2u2
Problem 29. a) If X1, X 2 , .....Xn are Poisson variates with parameter λ = 2 . Use the central limit
theorem to estimate P (120 < Sn < 160) where sn = X1 + X 2 + ...... + Xn and n = 75 .
b) A random sample of size 100 is taken from a population whose mean is 60 and variance is
400. Using central limit theorem, with what probability can we assent that the mean of the
sample will not differ from μ = 60 by more than 4.
Solution:
a). Given that E ( X i ) = λ = 2 and Var ( X i ) = λ = 2
34
[Since in Poisson distribution mean and variance are equal to λ ]

i.e. μ = 2 and σ 2 = 2
(
By central limit theorem, Sn – N n μ , nσ 2 )
Sn N (150,150)
⎛ 120 −150 160 −150 ⎞
∴ P (120 < Sn < 160) = P ⎜ <z< ⎟
⎝ 150 150 ⎠
= P(−2.45 < z < 0.85)
= P(−2.45 < z < 0) + P(0 < z < 0.85)
= P(0 < z < 2.45) + P(0 < z < 0.85) = 0.4927 + 0.2939 = 0.7866
b). Given that n = 100 , μ = 60 ,σ 2 = 400

Since the probability statement is with respect to mean, we use the Linderberg-levy form of
central limit Theorem.
⎛ σ2 ⎞ σ2
X N ⎜ μ, ⎟ i.e. X follows normal distribution with mean ' μ ' and variance .
⎝ n ⎠ n
⎛ 400 ⎞
i.e. X N ⎜60, 100 ⎟
⎝ ⎠
X N (60, 4)
⎡ mean of the sample will not ⎤ ⎡ X will not differ from⎤
P ⎢ differ from 60 by more than 4 ⎥ = P ⎢ μ = 60 by morethan 4 ⎥
⎣ ⎦ ⎣ ⎦
(
= P X −μ ≤ 4 )
= P ⎡⎣ −4 ≤ X − μ ≤ 4 ⎤⎦
= P ⎡⎣ −4 ≤ X − 60 ≤ 4 ⎤⎦
⎡ 56 − 60 64 − 60 ⎤
= P ⎡ 56 ≤ X ≤ 64 ⎤ = P ≤z≤
⎣ ⎦ ⎣⎢ 2 2 ⎥⎦
= P[−2 ≤ Z ≤ 2]
= 2P [0 ≤ Z ≤ 2] = 2 × 0.4773 = 0.9446
Problem 30 a) If the variable X1, X 2 , X 3 , X 4 are independent uniform variates in the interval
(450, 550) , find P (1900 ≤ X1 + X 2 + X 3 + X 4 ≤ 2100) using central limit theorem.
b) A distribution with unknown mean μ has a variance equal to 1.5. Use central limit
theorem to find how large a sample should be taken from the distribution in order that the
probability will be at least 0.95 that the sample mean will be within 0.5 of the population mean.
Solution
a). Given that X follows a uniform distribution in the interval (450, 550)
35
b + a 450 + 550
Mean = = = 500
2 2
( b − a ) ( 550 − 450 )
2 2
Variance = = = 833.33
12 12
By CLT S n= X 1+ X 2+ X 3+ X follows
4 a normal distribution with N ( nμ , nσ 2 )
S − nμ
The standard normal variable is given by Z = n
nσ 2
1900 − 4× 500 100
when Sn = 1900 , Z = =− = −1.732
4× 833.33 57.73
2100 − 2000
when S = 2100 , Z = = 100 = 1.732
n 57.73
4× 833.33
∴ P (1900 ≤ Sn ≤ 2100) = P (−1.732 < z < 1.732)
= 2 × P (0 < z < 1.732) = 2× 0.4582 = 0.9164 .
b). Given E ( X i ) = μ and Var ( X i ) = 1.5

Let X denote the sample mean.
By C.L.T. X follows N⎛μ, 1.5 ⎞
⎪ n ⎪
⎝ ⎠
( )
We have to find ' n ' such that P μ − 0.5 < X < μ + 0.5 ≥ 0.95
( )
i.e. P −0.5 < X − μ < 0.5 ≥ .95
( )
P X − μ < 0.5 ≥ .95
⎡ ⎤
σ
P⎢ < 0.5⎥ ≥ 0.95
z n ⎦
⎣
⎡
n⎤
P ⎢ z < 0.5 ⎥ ≥ 0.95
⎣ σ ⎦
⎡ n ⎤
P ⎢ z < 0.5 ⎥ ≥ 0.95
⎣ 1.5 ⎦
( )
ie P Z < 0.4082 n ≥ 0.95
Where ' Z ' is the standard normal variable.
The Last value of ' n ' is obtained from P (Z )
< 0.4082 n = 0.95
( )
2P 0 < z < 0.4082 n = 0.95
⇒ 0.4082 n = 1.96 ⇒ n = 23.05
∴The size of the sample must be atleast 24.
36
Unit.4. Correlation and spectral densities
UNIT – IV CORRELATION AND SPECTRAL DENSITIES
PART-A
Problem1. Define autocorrelation function and prove that for a WSS process { X (t )},
RXX (−τ ) = RXX (τ ) 2.State any two properties of an autocorrelation function.
Solution:
Let {X (t )}be a random process. Then the auto correlation function of the process
{ X (t )} is the expected value of the product of any two members X (t1 ) and X (t2 ) of
the process and is given by RXX (t1, t2 ) = E ⎣ ⎡ X (t1 ) X (t2 )⎤ ⎦ or
RXX ( t, t + τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
For a WSS process{X (t )}, RXX (τ ) = E ⎡ ⎣ X (t ) X (t −τ )⎤ ⎦
∴ RXX ( −τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦ = E ⎡⎣ X ( t + τ ) X ( t ) ⎤⎦ = R ( t + τ − t ) = R (τ )
Problem 2. State any two properties of an autocorrelation function.
Solution:
The process { X ( t )} is stationary with autocorrelation function R (τ ) then
( i ) R (τ ) is an even function of τ
( ii ) R (τ ) is maximum at τ = 0 i.e., R (τ ) ≤ R ( 0 )
Problem 3. Given that the autocorrelation function for a stationary ergodic process with
4
no periodic components is R (τ ) = 25 + . Find the mean and variance of the
1+ 6τ 2
process{X (t )}.
Solution:
Lt Lt 4
μ 2x = R (τ ) = 25 + = 25
τ →∞ τ →∞ 1+ 6τ 2
∴μx = 5
( )
E X 2 (t ) = R XX ( 0) = 25 + 4 = 29
Var ( X ( t ) ) = E ⎡⎣ X 2 ( t ) ⎤⎦ − E ⎡⎣ X ( t ) ⎤⎦ = 29 − 25 = 4
2
Problem 4. Find the mean and variance of the stationary process { X (t )} whose
25τ 2 + 36
autocorrelation function R (τ ) = .
6.25τ 2 + 4
Solution:
1
36
25 +
25τ 2 + 36
R (τ ) = = τ2
6.25τ 2 + 4 6.25 + 2
4
τ
Lt
μ2 = R (τ ) = 25 2500
= =4
τ →∞
x
6.25 625
μx = 2
36
E ⎡ ⎣ X 2 (t )⎦ ⎤ = RXX ( 0 ) = = 9
4
{ }
Var X ( t )⎤⎦ = E ⎡⎣ X ( t ) ⎤⎦ − E ⎡⎣ X ( t ) ⎤⎦ = 9 − 4 = 5
2 2
Problem 5. Define cross-correlation function and mention two properties.

Solution:
The cross-correlation of the two process { X (t )} and {Y (t )} is defined by
RXY (t1, t2 ) = E X (t1 )Y (t2 )⎤ ⎦
Properties: The process { X (t )} and {Y (t )} are jointly wide-sense stationary with the
cross-correlation function RXY (τ ) then
(i) RYX (τ ) = RXY (−τ )
(ii) RXY (τ ) ≤ RXX ( 0 ) RYY (0)
Problem 6. Find the mean – square value of the process { X (t )} if its autocorrelation
−τ 2
function is given by R (τ ) = e 2
.
Solution:
⎛ τ2 ⎞
−
= RXX ( 0 ) = e
Mean-Square value = E ⎡ ⎣ X
⎜ ⎟
2
=1 (t )⎤ ⎦ 2
⎝ ⎠τ =0
Problem 7. Define the power spectral density function (or spectral density or power
spectrum) of a stationary process?
Solution:
If {X (t )} is a stationary process (either in the strict sense or wide sense with auto
correlation function R (τ ) , then the Fourier transform of R (τ ) is called the power
spectral density function of { X (t )} and denoted by SXX (ω ) or S (ω ) or S X (ω )
∞
Thus S ( ω ) = ∫ R (τ ) e −iωτ dτ
−∞
Problem 8. State any two properties of the power spectral density function.
Solution:
(i). The spectral density function of a real random process is an even function.
2
(ii). The spectral density of a process{X (t )}, real or complex, is a real function of ω and
non-negative.
Problem 9. State Wiener-Khinchine Theorem.
Solution:
If X T (ω ) is the Fourier transform of the truncated random process defined as
⎧⎪ X (t ) for t ≤ T
XT (t ) = ⎨
⎪⎩0 for t > T
Where { X (t )} is a real WSS Process with power spectral density function S (ω ) , then
S (ω ) = Lt ⎡ 1
E X
(ω ) ⎤
{
2
}
T → ∞ ⎢ ⎣ 2T T ⎥⎦
Problem 10. If R (τ ) = e −2 λ (τ ) is the auto Correlation function of a random
process{X (t )}, obtain the spectral density of { X (t )}.
Solution:
∞
S ( ω ) = ∫ R (τ ) e −iωτ dτ
−∞
∞
= ∫ e −2 λ τ ( cosωτ − isinωτ ) dτ
−∞
∞
= 2∫ e −2λτ cosωτ dτ
0
∞
⎡ 2e −2 λτ ⎤
2 (
=⎢ 2 −2λ cosωτ + ω sinωτ ) ⎥
⎣ 4λ + ω ⎦0
4λ
S (ω ) =
4λ 2 + ω 2
Problem 11. The Power spectral density of a random process { X (t )} is given by
⎧⎪π , ω < 1
SXX (ω ) = ⎨ Find its autocorrelation function.
⎪0,
⎩ elsewhere
Solution:
1∞
RXX (τ ) = ∫ SXX ( ω ) eiωτ d ω
2π −∞
1 1
= π eiωτ d ω
∫
2π −1 1
= ⎡⎢ e ωτ ⎤⎥
i
1
2 iτ
⎣ ⎦ −1
⎢1 ⎡ eiτ − e −iτ ⎤ ⎥
=2 iτ
⎣ ⎦
3
1 ⎢⎡ eiτ − 2i
e − iτ ⎤ ⎥ sinτ τ
=τ =
⎣ ⎦
Problem 12. Define cross-Spectral density.
Solution:
The process { X (t )} and {Y (t )} are jointly wide-sense stationary with the cross-
correlation function RXY (τ ) , then the Fourier transform of RXY (T ) is called the cross
spectral density function of { X (t )} and {Y (t )} denoted as SXY (ω )
∞
Thus SXY (ω ) = ∫ R (τ )e
XY
−iωτ
dτ
−∞
Problem 13. Find the auto correlation function of a stationary process whose power
ω ≤1
spectral density function is given by s (ω ) = ⎨⎪ω⎧ for
2
.
⎪⎩0 for ω > 1
Solution:
1∞
R (τ ) = S ( ω )eiωτ d ω
∫
2π −∞
1 1 2
= ∫ ω ( cos ωτ + i sin ωτ )dω
2π −1
1 ⎡ 2 ⎛ sin ωτ ⎞ ⎛ −cos ωτ ⎞ ⎛ − sin ωτ ⎞⎤ 1
+ 2⎜
1
= ∫ ω 2 .cos ωτ dω = ω
⎢ ⎜ ⎟ − 2ω ⎜ 2 ⎟ 3 ⎟⎥
−1 π⎣ ⎝ τ ⎠ ⎝ τ ⎠ ⎝ τ ⎠⎦ 0
1 ⎡ − sin τ 2 cos τ 2 sin τ ⎤
R (τ ) = + −
π⎢ ⎣ τ τ2 τ3 ⎥⎦
1
Problem 14. Given the power spectral density : Sxx (ω ) = , find the average power
4 + ω2
of the process.
Solution:
1∞
R (τ ) = S ( ω )eiωτ d ω
2π −∞ ∫
∞
1 ⎡ 1 ⎤ eiωτ d ω
= ∫ ⎢
2π −∞ ⎣ 4 + ω ⎦
2⎥
Hence the average power of the processes is given by

E ⎡ ⎣ X 2 (t )⎤ ⎦ = R (0)
1 ∞ dω
2π ∫−∞ 4 + ω 2
=
1 ∞ dω
= 2∫ 2
2π 02 + ω 2
4
1 ⎡ 1 −1 ⎛ ω⎜ ⎞⎤ ⎟ ∞
= tan
π ⎢⎣ 2 ⎝ ⎠⎥⎦ 0
1 ⎡ π ⎤ 12
= −0 = .
π ⎢⎣ 4 ⎦⎥ 4
Problem 15. Find the power spectral density of a random signal with autocorrelation
function e −λ τ .
Solution:
∞
S ( ω ) = ∫ R (τ )e −iωτ dτ
−∞
∞
= ∫ e−λ τ ( cos ωτ − i sin ωτ ) dτ

−∞
∞
= 2 ∫ e −λ τ cos ωτ dτ
0
∞
⎡ e− λτ ⎤
2(
= 2⎢ −λ cos ωτ + ω sin ωτ )⎥
⎣ λ +ω
2
⎦0
⎡
= 2 0− 1
(−λ )⎤ = 2 2λ 2
⎣ λ +ω
2 2
⎦ λ +ω
PART-B
Problem 16. a). If { X ( t )} is a W.S.S. process with autocorrelation function RXX (τ ) and
if Y (t ) = X (t + a ) − X (t − a ) .Show that RYY (τ ) = 2RXX (τ ) − RXX (τ + 2a ) − RXX (τ − 2a ) .
Solution:
Ryy (τ ) = E ⎡⎣ y ( t ) y ( t + τ ) ⎤⎦
{ }
= E ⎡⎣ X ( t + a ) − X ( t − a ) ⎤⎦ ⎡⎣ X ( t + τ + a ) − X ( t + τ − a ) ⎤⎦
= E ⎡⎣ X ( t + a ) X ( t + τ + a ) ⎤⎦ − E ⎡⎣ X ( t + a ) X ( t + τ − a )⎤⎦
– E ⎣⎡ X ( t − a ) X ( t + τ + a ) ⎤⎦ + E ⎣⎡ X ( t − a ) X ( t + τ − a ) ⎤⎦
= Rxx (τ ) − E ⎡⎣ X ( t + a ) X ( t + a + τ − 2a ) ⎤⎦
– E ⎡⎣ X ( t − a ) X ( t − a + τ + 2a ) ⎤⎦ + Rxx (τ )
= 2Rxx (τ ) − Rxx (τ − 2a) − Rxx (τ + 2a)
b). Assume a random signal Y (t ) = X (t ) + X (t − a ) where X (t ) is a random signal and
' a ' is a constant. Find RYY (τ ) .
Solution:
RYY (τ ) = E ⎡⎣ Y ( t ) Y ( t + τ ) ⎤⎦
{
= E ⎡⎣ X ( t ) + X ( t − a ) ⎤⎦ ⎡⎣ X ( t + τ ) + X ( t + τ − a ) ⎤⎦ }
5
= E ⎡⎣ X ( t ) X ( t + τ )⎤⎦ + E ⎡⎣ X ( t ) X ( t + τ − a ) ⎤⎦
+ E ⎡⎣ X ( t − a ) X ( t + τ ) ⎤⎦ + E ⎡⎣ X ( t − a ) X ( t + τ − a )⎤⎦
= Rxx (τ ) + Rxx (τ + a ) + Rxx (τ − a ) + Rxx (τ )
Ryy (τ ) = 2Rxx (τ ) + Rxx (τ + a ) + Rxx (τ − a )
Problem 17. a). If { X (t ) } and {Y (t )} are independent WSS Processes with zero means,
find the autocorrelation function of{Z (t )} , when (i) Z (t ) = a + bX (t ) + C Y (t ) ,
(ii) Z (t ) = aX (t )Y (t ) .
Solution:
Given E ⎡ ⎣ X (t )⎤ ⎦ = 0 E ⎡⎣ Y ( t )⎤⎦ = 0 -------- (1)
,
{X (t)} and {Y (t)} are independent
E { X (t )Y (t )} = E ⎣ ⎡ X (t )Y (t )⎦ ⎤ = 0 -------- (2)
(i). RZZ (τ ) = E ⎡⎣ Z ( t ) Z ( t + τ ) ⎤⎦
{
= E ⎡⎣ a + bX ( t ) + cY ( t ) ⎤⎦ ⎡⎣ a + bX ( t + τ ) + cY ( t + τ ) ⎤⎦
= E {a 2
+ ab X ( t + τ ) + acY ( t + τ ) + baX ( t ) + b 2 X ( t ) X ( t + τ )
+ bc X ( t ) Y ( t + τ ) + caY ( t ) + cbY ( t ) X ( t + τ ) + c 2 y ( t ) y ( t + τ ) }
( )
= E a 2 + abE ⎣⎡ X ( t + τ ) ⎤⎦ + acE ⎡⎣ Y ( t + τ )⎤⎦ + baE ⎡⎣ 2X ( t ) ⎤⎦ + b E ⎡⎣ X ( t ) X ( t + τ )⎤⎦
+ bc E ⎡⎣ X ( t ) y ( t + τ ) ⎤⎦ + caE ⎡⎣ Y ( t ) ⎤⎦ + cbE ⎡⎣ Y ( t ) X ( t + τ ) ⎤⎦ + c 2 E ⎡⎣ Y ( t ) Y ( t + τ ) ⎤⎦
= a2 + b2R XX (τ ) + c2 R (τ )
YY
RZZ (τ ) = E ⎡⎣ Z (t ) Z ( t + τ )⎤⎦
= E ⎡⎣ aX ( t ) Y ( t ) aX ( t + τ ) Y ( t + τ ) ⎤⎦
= E ⎡⎣ a 2 X ( t ) X ( t + τ ) Y ( t ) Y ( t + τ ) ⎤⎦
= a 2 E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦ E ⎡⎣ Y ( t ) Y ( t + τ ) ⎤⎦
R (τ ) = a 2 R (τ ) R (τ )
ZZ XX YY
b). If { X (t )} is a random process with mean 3 and autocorrelation Rxx (τ ) = 9 + 4e −0.2 τ .

Determine the mean, variance and covariance of the random variables Y = X (5) and
Z = X (8) .
Solution:
Given Y = X (5) & Z = X (8), E ⎡ ⎣ X (t )⎦ ⎤ = 3 -------------(1)
Mean of Y = E [Y ] = E ⎡⎣ X ( 5) ⎤⎦ = 3
Mean of Z = E [ Z ] = E ⎡⎣ X ( 8 ) ⎤⎦ = 3
We know that
6
Var ( Y ) = E ( Y 2 ) − ⎡⎣ E ( Y ) ⎤⎦
2
( ) (
E Y 2 = E X 2 ( 5) )
But E ⎡ ⎣ X (t )⎦ ⎤ = RXX (0)
2
= 9 + 4e−0.2101
= 9 + 4 = 13
Thus Var ( Y ) = 13 − ( 3) = 13 − 9 = 4
2
Var ( Z ) = E ( Z 2 ) − ⎡⎣ E ( Z ) ⎤⎦
2
E ⎡⎣ Z 2 ⎤⎦ = E ⎡ ⎣X 2 ( 8 ) ⎤ ⎦ ⎡⎣ Z = X ( 8) ⎤⎦
= RXX ( 0)
= 9 + 4 = 13
Hence Var ( Z ) = 13 − 32 = 13 − 9 = 4( )
E [YZ ] = R (5,8) = 9 + 4e−0.25−8 ( R (t , t
1 2 ) = 9 + 4e−0.2 t −t
1 2
)
= 9 + 4e −0.6
Covariance = R (t1, t2 ) − E ⎣ ⎡ X (t1 )⎤ ⎦ E ⎡ ⎣ X (t2 )⎤ ⎦

= R (5,8) − E [5] E [8]
= 9 + 4e−0.6 − ( 3× 3) = 4e−0.6 = 2.195
Problem 18. a). The autocorrelation function for a stationary process is given by
2
Rxx (τ ) = 9 + 2e −τ Find the mean value of the random variable Y = X (t ) dt and

.
∫ 0
variance of X (t ) .
Solution:
Given Rxx (τ ) = 9 + 2e τ
Mean of X (t ) is given by
Lt
X = E ⎡⎣ X ( t ) ⎤⎦ = Rxx (τ )
2 2
τ →∞
Lt
τ →∞ (
= 9 + 2e
) −τ
2
X =9
X =3
Also E ⎡ ⎣ X 2 (t )⎦ ⎤ = RXX (0) = 9 + 2e 0 = 9 + 2 = 11
Var { X ( t )} = E ⎡⎣ X 2 ( t ) ⎤ ⎦ − ⎡ ⎣E ( X ( t ) ) ⎤ ⎦
2
= 11− 32 = 11− 9 = 2
Mean of Y ( t ) = E ⎡⎣ Y ( t ) ⎤⎦
7
⎡2 ⎤
= E ⎢ ∫ X (t ) dt ⎥
⎣0 ⎦
2
= ∫ E ⎡ ⎣ X (t )⎤ ⎦ dt
0
2
= ∫ 3dt = 3 ( t ) 0 = 6
2
∴ E ⎡⎣ Y ( t ) ⎤⎦ = 6
b). Find the mean and autocorrelation function of a semi random telegraph signal process.
Solution:
Semi random telegraph signal process.
If N (t ) represents the number of occurrences of a specified event in (0, t )
and X ( t ) = ( −1) , then { X ( t )} is called the semi random signal process and N (t ) is a
N (t )
poisson process with rate λ .

By the above definition X (t ) can take the values = 1 and −1 only
P ⎡⎣ X ( t ) = 1⎤⎦ = P ⎡⎣ N ( t ) is even⎤⎦
e−λt ( λt )
k
= ∑
K =even K!
( λt )
2
⎡ ⎤
= e−λt ⎢1+ + ...⎥
⎣⎢ 2 ⎥⎦
P ⎡⎣ X ( t ) = 1⎤⎦ = e −λt coshλt
P ⎡⎣ X ( t ) = −1⎤⎦ = P ⎡⎣ N ( t ) is odd ⎤⎦
e−λt ( λt )
k
= ∑
K =odd K!
( λt )
3
⎡ ⎤
= e −λt ⎢ λ t + + ...⎥
⎢⎣ 3! ⎥⎦
= e sinhλ t
−λ t
Mean{X (t )} = ∑ K P ( X (t ) = K )
K =−1,1
= 1× e −λt coshλt + (−1) × e−λt sinhλt

= e−λt [coshλt − sinhλt ]
= e −2 λt ⎡⎣ coshλ t − sinhλt = e −λt ⎤⎦
R (τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
= 1× P ⎡⎣ X ( t ) X ( t + τ ) = 1⎤⎦ + −1× P ⎡⎣ X ( t ) X ( t + τ ) = −1⎤⎦
8
( λτ ) ( λτ )
n n
= ∑
n=even
eλt
∑ e− λ t
n! – n=odd n!
= e −λτ coshλτ − e λτ sinhλτ
= e −λτ [ coshλτ − sinhλτ ]
= e− λτ e− λτ
R (τ ) = e −2λτ
Problem 19. a). Find Given the power spectral density of a continuous process as
ω2 + 9
SXX (ω) = 4
ω + 5ω2 + 4
find the mean square value of the process.
Solution:
We know that mean square value of { X (t )}
1 ∞
= E X 2 (t ) =
{ } ∫ SXX (ω ) dω
2π −∞
1 ∞ ω2 + 9
2π ∫−∞ ω 4 + 5ω 2 + 4
= dω
( )
1 ∞ ω2 + 9
= .2∫ 4 dω
2π 0ω + 5ω 2 + 4
1∞ ω2 + 9
= ∫ 4 dω
π 0ω + ω 2 + 4ω2 + 4
1∞ ω2 + 9
= ∫ 2 2 dω
(
π 0ω ω +1 + 4 ω 2 +1 ) ( )
∞
ω2 + 9
{
i.e., E X (t ) = ∫
2 1
} dω
π 0 ω 2 + 4 ω 2 +1 ( )( )
let ω2 = u
∴ We have ω2 + 9 u+9
=
(
ω 2 + 4 ω 2 +1 )(
( u + 4 )( u +1) )
−4 + 9 −1+ 9
5 8
= −4 +1 + −1+ 4 = − + ….Partial fractions
u+4 u +1 3(u + 4) 3(u +1)
i.e., ω +9
2
5 8
=− +
(ω 2 + 4 ω 2 +1 )( 3 ω2 + 4 )
3 ω 2 +1 ( ) ( )
∴From (1),
1 ∞1⎡ −5 8 ⎤
{
E X 2 (t ) = ∫ ⎢ }+ ⎥ dω
π 0 (
3 ⎢⎣ ω 2 + 4 ) (ω 2
)
+1 ⎥⎦
9
∞
1 ⎡ 1 ω ⎤
= ⎢ −5. tan
−1 + 8 tan −1 ω ⎥
3π ⎣ 2 2 ⎦0
1 ⎡ 5 ⎛π ⎞ ⎛π ⎞ ⎤
= − ⎜ +⎟ 8 ⎜ −⎟0
3π ⎢ ⎣ 2 ⎝ 2 ⎠ ⎝ 2 ⎠ ⎥⎦
1 π⎛ 5 ⎞ 1 ⎛ 11 ⎞
= . − +8 =
⎜ ⎟ 6⎜ 2 ⎟
3π 2 2
⎝ 11 ⎠ ⎝ ⎠
E {X 2 (t ) } = .
12
b). If the 2n random variables Ar and Br are uncorrelated with zero mean and
( ) ( )
E Ar2 = E B 2r = σ 2r. Find the mean and autocorrelation of the process
n
X (t ) = ∑ Ar cosωrt + Br sin ωrt .

r =1
Solution:
Given E ( A )r = E ( B )r = 0 & E ( A2 )r = E ( B 2 r) = σ 2 r
n
⎡ ⎤
Mean: E ⎣ ⎡ X (t )⎤ ⎦ = ⎢ ∑ Arcosωrt + Br sinωrt ⎥
⎣ r =1 ⎦
n
= ∑ ⎣⎡ E ( Ar ) cosωr t + E ( Br ) sinωr t ⎤⎦
r =1
E ⎡ ⎣ X (t )⎤ ⎦ = E ( Ar ) = E ( Br ) = 0
0
Autocorrelation function:
R (τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
⎧n n
= E ⎨∑∑ ( Ar cosωr t + Br sinωr t ) ( As cosωs ( t + τ ) + Bs sinωs ( t + τ ) )
⎩ r =1 s=1
Given 2n random variables Ar and Br are uncorrelated
E [ Ar As ], E [ Ar Bs ], E [Br As ], E [Br , Bs ] are all zero for r ≠ s
n
= ∑ E ( A2r) cosω trcosω t (r t + τ ) + E ( B 2 ) rsinω tsin

r ω ( t r+ τ )
r =1
n
= ∑σ cos
r ω
2
r (t − t −τ )
r =1
n
= ∑σ cos
r ω
2
r ( −τ )
r =1
n
R (τ ) = ∑σ 2cos
r ω τ r
r =1
Problem 20. a). If { X (t )} is a WSS process with autocorrelation R (τ ) = Ae−α τ ,

determine the second – order moment of the random variable X (8) − X (5) .
Solution:
10
Given R (τ ) = Ae−α τ
R (t1, t2 ) = Ae−α t1 −t2
E ⎡ ⎣ X 2 (t )⎦ ⎤ = RXX (0) = Ae−α 0 = A
∴ E ⎡⎣ X ( 8 ) − X ( 5) ⎤⎦ = E ⎡⎣ X 2 ( 8 ) ⎤⎦ + E ⎡⎣ X 2 ( 5 ) − 2E ⎡⎣ X ( 8 ) X ( 5 ) ⎤⎦ ⎤⎦
2
E ⎡⎣ X 2 ( 8) ⎤⎦ = E ⎡⎣ X 2 ( 5) ⎤⎦ = A
Also E ⎡⎣ X ( 8) X ( 5 ) ⎤⎦ = R ( 8, 5) = Ae −α 8 −5 = Ae −3α
Substituting (2), (3) in (1) we obtain
E ⎡⎣ X ( 8 ) − X ( 5 ) ⎤⎦ = A + A − 2 Ae −3α
2
= 2 A − 2 Ae−3α
(
= 2 A 1− e −3α )
.
b). Two random process { X (t )} and {Y (t )} are given by
X (t ) = A cos(ω t+ θ ) ,Y (t ) = Asin (ωt + θ ) where A & ω are constants and θ is a
uniform random variable over 0 to 2 π . Find the cross-correlation function.
Solution:
RXY (τ ) = E ⎡⎣ X ( t ) Y ( t + τ ) ⎤⎦
= E ⎡⎣ Acos ( ωt + θ ) Asin ( ω ( t + τ ) + θ ) ⎤⎦
= A2 E ⎡⎣ sin ( ωt + ωτ + θ ) cos ( ωt + θ )⎤⎦
1
θ is a uniform random variables fθ (θ ) = , 0 ≤ θ ≤ 2π
2π
2π
A2
∴ RXY (τ ) =
2π ∫ sin (ωt + ωτ + θ ) cos (ωt + θ ) dθ
= A∫ ⎜ ⎛ sin ( 2ωt + ωτ + 2θ ) + sin ( ωt ) ⎞
0
2 2π
2π ⎟ dθ
0 ⎝
2 ⎠
2π
= 4π − cos ( 2ωt 2+ ωτ + 2θ ) + θ sin (ωt )⎥ ⎤
A ⎡
2
⎢
⎣ ⎦0
2
A
=
4π
[0 + 2π sinωτ ]
A2
= sinωτ
2
Problem 21. a). Find the cross-correlation function of
W (t ) = A (t ) + B (t ) & Z (t ) = A (t ) − B (t ) where A (t ) and B (t ) are statistically
independent random variables with zero means and autocorrelation function
RAA τ = e− τ , − ∞ < τ < ∞ , R BB τ = 3e − τ – ∞ < τ < ∞ respectively.
Solution:
11
Given E ⎡⎣ A ( t ) ⎤⎦ = 0, E ⎡⎣ B ( t ) ⎤⎦ = 0
A (t ) & B (t ) are independent
RAB (τ ) = E ⎡⎣ A ( t ) B ( t + τ ) ⎤⎦ = E ⎡⎣ A ( t ) ⎤⎦ E ⎡⎣ B ( t + τ ) ⎤⎦ = 0
Similarly RBA (τ ) = E ⎡⎣ B ( t ) A ( t + τ ) ⎤⎦ = E ⎡⎣ B ( t ) ⎤⎦ E ⎡⎣ A ( t + τ ) ⎤⎦ = 0
∴ RWZ (τ ) = E ⎡W
⎣ ( t ) Z ( t + τ ) ⎤⎦
{
= E ⎡⎣ A ( t ) + B ( t ) ⎤⎦ ⎡⎣ A ( t + τ ) − B ( t + τ ) ⎤⎦ }
= E ⎡⎣ A ( t ) A ( t + τ ) − A ( t ) B ( t + τ ) + B ( t ) A ( t + τ ) − B ( t ) B ( t + τ ) ⎤⎦
= E ⎡⎣ A ( t ) A ( t + τ ) ⎤⎦ − E ⎡⎣ A ( t ) B ( t + τ ) ⎤⎦ + E ⎡⎣ B ( t ) A ( t + τ ) ⎤⎦ − E ⎡⎣ B ( t ) B ( t + τ ) ⎤⎦
= RAA (τ ) − RAB (τ ) + RBA (τ ) − RBB (τ )
= RAA (τ ) − RBB (τ )
−τ
= e – 3e − τ
RWZ (τ ) = −2e − τ
b). The random processes { X (t )} and {Y (t )} defined by
X (t ) = A cos ωt + B sin ωt,Y (t ) = B cos ωt − A sin ωt where A & B are uncorrelated zero
mean random variables with same variance find its autocorrelation function.
Solution:
Given E ( A) = E ( B) = 0 & E A2 = E B 2 ( ) ( ) =σ 2
Also A and B are uncorrelated i.e., E ( AB ) = 0

RXY (τ ) = E ⎡⎣ X ( t ) Y ( t + τ ) ⎤⎦
{
= E [ Acosω t + B sinωt ] ⎣⎡ B cosω ( t + τ ) − Asinω ( t + τ ) ⎤⎦ }
⎡ ABcosωtcosω ( t + τ ) − A cosωtsinω ( t + τ ) ⎤
2
= E⎢ ⎥
⎢⎣ +B 2 sinω tcosω ( t + τ ) − BAsinωtsinω ( t + τ ) ⎥⎦
( )
= E ( AB ) cosωtcosω ( t + τ ) − E A2 cosω tsinω ( t + τ )
( )
+ E B 2 sinωtcosω ( t + τ ) − E ( BA ) sinωtsinω ( t + τ )
= E ( B 2 ) sinωtcosω ( t + τ ) − E ( A2 ) cosωtsinω ( t + τ ) ( E ( AB ) = 0 )
= σ ⎡⎣ sinωtcosω ( t + τ ) − cosωtsinω ( t + τ ) ⎤⎦ ⎡⎣ E2 ( A
2
) = E2 (B ) = σ 2
⎤
⎦
= σ 2 sinω ( t − t −τ )
= σ 2 sinω ( −τ )
RXY (τ ) = −σ 2 sinωt ⎣⎡ sin ( −θ ) = −sinθ ⎤⎦
12
Problem 22. a). Consider two random processes X (t ) = 3cos (ωt + θ ) and
⎛ π⎞
Y ( t ) = 2 cos ω t + θ − where θ is a random variable uniformly distributed in
⎜ 2⎟
⎝ ⎠
(θ , 2π ) . Prove that RXX ( 0 ) RYY ( 0 ) ≥ RXY (τ ) .
Solution:
Given X (t ) = 3cos (ωt + θ )
⎛ π⎞
Y ( t ) = 2cos ωt + θ −
⎜ 2⎟
⎝ ⎠
RXX (τ ) = E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
= E ⎣⎡ 3 cos ( ωt + θ ) 3cos ( ωt + ωτ + θ ) ⎤⎦
9
= E ⎡⎣ cos ( 2ωt + ωτ + 2θ ) + cos ( −ωτ ) ⎤⎦
2
1
θ is uniformly distributed in (0, 2π ) , f (θ ) =
2π
9 2π 1
= ∫ ⎡⎣ cos ( 2ωt + ωτ + 2θ ) + cos ωτ ⎤⎦ dθ
2 0 2π
= 9 1 ⎢⎡ sin ( 2ωt + ωτ + 2θ )
2π
⎤
2 2π 2 + θ cosωτ ⎥
⎣ ⎦0
9 ⎡ ⎡ sin ( 2ω t + ωτ + 2θ ) ⎤ 2π ⎤
= [0 + 2πcosωt ] ⎢ ⎢ ⎥ = 0⎥
4π 2
⎣⎢ ⎣ ⎦0 ⎥⎦
9
RXX (τ ) = cosωτ
2
9
RXX ( 0 ) =
2
RYY (τ ) = E ⎣⎡ Y ( t ) Y ( t + τ ) ⎦⎤
⎡ ⎛ π⎞ ⎛ π ⎞⎤
=
E ⎢ 2cos ⎜⎝ω t + θ − ⎟ 2cos ω
⎜ ⎝t + ωτ + θ − ⎟
⎣ 2 ⎠ 2 ⎥ ⎠⎦
4
= E ⎡⎣ cos ( 2ωt + ωτ + 2θ − π ) + cosωτ ⎤⎦
2
2π
1
= 2 ∫ ⎡⎣ cos ( 2ωτ + ωτ + 2θ − π ) + cosωτ ⎤⎦ dθ
0
2π
1 ⎡ sin ( 2ω t + ωτ + 2θ − π ) ⎤
2π
= ⎢ + θ cosωτ ⎥
π 2
⎣ ⎦0
= 2cosωτ
13
π⎞
Similarly, R (τ ) = 3cos ⎛ωτ −
XY ⎜ 2⎟
⎝ ⎠
RXY (τ ) max
=3
By property (2) i.e.,RXX ( 0 ) RYY ( 0 ) > RXY (τ )
⎛ π⎞
∴ 9 ≥ 3cos ωτ − , ∀τ
.2 ⎜ ⎟
2 2
⎝ ⎠
In this care RXY (τ ) takes on its maximum possible value of 3 at
π n2π
τ= + , ∀n
2ω ω
Since it is periodic
R (0) R (0) = R (τ )
2
XX YY XY max
b). State and prove Wiener – Khinchine Theorem
Solution:
Statement:
If X T (ω ) is the Fourier transform of the truncated random process defined as
⎧⎪ X (t ), − T ≤ t ≤ T
XT (t ) = ⎨
⎪⎩0 , otherwise
Where { X (t )} is a real WSS process with power spectral function SXX (ω ) , then
SXX (ω ) = lim ⎡ 1 E X (ω )
2 ⎤
{ }
T → ∞ ⎢ ⎣ 2T T ⎥⎦
Proof:
∞
Given XT (ω ) =
∫ X (t ) e dt
−iωt
T
−∞
T
= ∫ X (t ) e−iωt dt
−T
Now X T(ω ) = X (T ω ) X (T−ω )

2
T T
= ∫ X ( t 1) e −iωt1 dt 1 ∫ X (t )e 2
−iωt 2
dt 2
−T −T
= X (t )X (t e −iω (t1 −t 2 ) dt dt
T T
)
∫∫ 1 2 1 2
{X }=
T T
(ω ) E ⎡ X (t ) X (t
−T −T2
∴E
) ⎤e−iω (t −t ) dt dt
1 2
T ∫∫
−T −T
⎣ 1 2 ⎦ 1 2
= R (t − t
T T
)e −iω ( t1 −t 2 )
dt dt
∫∫ 1 2 1 2
−T −T
14
T T
= ∫ ∫ 0 (t1 − t2 )dt1dt2 ---------------- (1)

−T −T
{X (t)} is WSS its autocorrelation is a function of time difference ie., t1 t2 =
T
A( T ,T ) B
T T
Q
t1 t 2 = d
S
P
D R C(T , T )
T
The double integral (1) is evaluated over the area of the square ABCD bounded by the
t1 = −T ,T & t2 = −T ,T as shown in the figure
We divide the area of the square into a number of strips like PQRS, where PQ is given by
t1 − t2 = τ and RS is given by t1 − t2 = τ + dτ .
When PQRS is at A, t1 − t2 = −T −T = −2T and t1 − t2 = 2T and at C
∴τ varies from 2T to 2T such that the area ABCD is covered.
Now dt1dt2 = elemental area of the t1, t2 plane
=Area of the strip PQRS (2)
At P , t2 = T , t1 = τ + t2 = τ − T At
Q , t1 = T , t2 = t1 =T PC =
CQ = 2T −τ ,τ > 0
PC = CQ = 2T + τ ,τ < 0
RC = SC = 2T −τ − dτ , τ > 0
When τ > 0 ,
Area of PQRS = ΔPCQ − ΔRCS
1 1
= ( 2T − τ ) − ( 2T − τ − 2τ )
2 2
2 2
= ( 2T − τ ) dτ Omitting ( dτ )
2 -------------------------
(3)
From (2) & (3)
15
dt1dt2 = (2T − τ dτ ) − − − − − (4)

Using (4) in (1)
{X }
2T
(ω ) ∫ 0 (τ ) ( 2T − τ ) dτ
2
E T =
−2T
{ } τ ⎞
2T
1 E X (ω ) ⎛
T
2
= ∫ 0 (τ ) ⎜1− ⎟ dτ
2T ⎝ 2T ⎠
}
−2
( w) = lim 2T 0 (τ ) dτ −
T 2
∴
lim 1
E {X ∫
lim τ dτ
T → ∞ 2T T →∞ T → ∞ ∫ 2T
T
−2T
∞
= ∫ 0 (τ ) dτ
−∞
∞
= ∫ R (τ ) e −iωτ dτ
−∞
= S (ω )
This theorem provides an alternative method for finding S (ω ) for a WSS Process.
Problem 23. a). Define the spectral density S (ω ) of a real valued stochastic
processes{X (t ) : t ≥ 0}, obtain the spectral density of {Y (t ) : t ≥ 0},Y (t ) = α X (t ) when
1
α is increment of X (t ) such that P (α = 1) = P (α = −1) = .
2
Solution:
If X (t )} is a stationary process {either in the strict serve or wide sense with
autocorrelation function R τ , then the Fourier transform of R τ is called the Power
spectral density function of X (t )} denoted by S (w )
∞
Thus S ( ω ) = ∫ R (τ ) e −iωτ dτ
−∞
1
Given P{α = 1} = P {α = −1} =
2
1 1
∴ E (α ) = 1. + (−1). = 0
2 2
2 1 2 1
∴ E α( =) (1) . + ( −1) . = 1
2
2 2
E ⎡⎣ Y ( t ) ⎤⎦ = E ⎡⎣ α X ( t ) ⎤⎦
= E ( α ) E ⎡⎣ X ( t ) ⎤⎦ ⎡⎣ α X ( t ) areindependent ⎤⎦
E ⎡⎣ Y ( t ) ⎤⎦ = 0 ⎡⎣ E ( α ) = 0⎤⎦
RYY (τ ) = E ⎡⎣ Y ( t ) Y ( t + τ ) ⎤⎦
= E ⎡⎣ α X ( t ) α X ( t + τ ) ⎤⎦
16
= E ⎡⎣ α 2 X ( t ) X ( t + τ ) ⎤⎦
= E ⎡⎣α 2 ⎤ ⎦E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
= 1 E ⎡⎣ X ( t ) X ( t + τ ) ⎤⎦
= (1) P ⎡⎣ X ( t )( t + τ ) = 1⎤⎦ + (−1) P ⎡⎣ X ( t )( t + τ ) = −1⎤⎦
e −λT ( λT ) e −λT ( λT )
n
= (1) ∑ n! + (−1) ∑ n!
n=even n=odd
= e −λT coshλτ − e −λT sinhλτ

= e−λτ [coshλT − sinhλT ]
= e−λT e−λT = e−2λT
∞
S ( ω ) = ∫ R (τ )e −iωτ dτ
−∞
∞
= ∫ e−2 λ τ e−i ωτ dτ
−∞
0 ∞
∫ e e dτ + ∫ e 0 e dτ
2 λτ −iωτ −2 λτ −iωτ
= −∞
0 ∞
( 2 λ −iω )τ −( 2 λ +i ω )τ
= ∫e dτ + ∫ e dτ
−∞ 0
0 ∞
⎡ ( 2λ −iω )τ ⎤ ⎡ e−( 2 λ +iω )τ ⎤
=⎢e ⎥ +⎢ ⎥
2λ − iω
⎣ ⎦ −∞ ⎣ − ( 2λ + iω ) ⎦ 0
1 1
= +
2λ − iω 2λ + iω
2λ − iω + 2λ − iω
=
4λ 2 − ω 2
4λ
= 2
4λ + ω 2
b). Show that (i) S (ω ) ≥ 0 & (ii) S (ω ) = S (−ω ) where S (ω ) is the spectral density of a
real valued stochastic process.
Solution:
(i) S (ω ) ≥ 0
Proof:
If possible let S (ω ) < 0 at ω = ω0
ε ε
i.e., let S (ω ) < 0 in ω0 – < ω < ω 0 + , where ε is very small
2 2
Let us assume that the system function of the convolution type linear system is
17
⎧ ε ε
⎪1 , ω0 − < ω < ω0 +
H ( ω ) =⎨ 2 2
⎪⎩0 , elsewhere
Now SYY (ω ) = H (ω ) 2S (ω )
⎧ XX
⎪S ( ) 0 – ε < ω < ω + ε
ω , ω
= ⎨ XX 0
2 2
⎪⎩0 , elsewhere
E ⎡⎣ Y 2 ( t ) ⎤⎦ = R YY (0)
∞
1
=
2π ∫ SYY (ω )dω
−∞
ε
ω 0+
2
= 1 SXX (ω )dω
2π
ω−
∫ ε
0
2
ε
= S (ω )
2π XX 0
[ SXX (ω ) Considered a constant S XX ( ω0 ) band is
narrow] E ⎡⎣ Y 2 ( t ) ≥ 0, S ( ω ) ≥ 0 which is contrary to our initial assumption.
xx 0
SXX (ω ) ≥ 0 everywhere, since ω = ω0 is arbitrary

(ii) S (ω ) = S (−ω )
Proof:
∞
Consider S ( −ω ) = ∫ R (τ )eiωτ dτ
−∞
Let τ = −u then dτ = −du and u varies from and
−∞
S (−ω ) = ∫ R (−u )e−iωu (−du )

∞
∞
= ∫ R (−u )e−iωudu
−∞
∞
= ∫ R ( u ) e −iωu du ⎡⎣ R (τ ) is an even function of τ ⎤⎦

∞
= S (ω )
Hence S (ω ) is even function of ω
Problem 24. a). An autocorrelation function R ( τ ) of {X (t ); −∞ < t < ∞} in given by
ce −α τ , c > 0, α > 0 . Obtain the spectral density of R ( τ ).
Solution:
Given R (τ ) = ce−α τ
18
By definition
∞
S ( ω ) = ∫ R (τ ) e −iωτ dτ
−∞
∞
= ∫ ce−α τ e −iωτ dτ
−∞
∞ ∞
−α ( −τ )
= ∫ ce e −iωτ
dτ + ∫ ceα (τ ) e −iωτ dτ
−∞ 0
0 ∞ ⎤
⎡ (α −iω )τ −(α −iω )τ
= c⎢ ∫ e dτ + ∫ e dτ ⎥
⎣ −∞ 0 ⎦
= c ⎨⎢ ⎡ e( α −iω )τ ⎥⎤ + ⎢⎡ e−(α +iω )τ ⎥ ⎤ ⎬∞ ⎫⎪
0
α − iω −(α + iω )
⎩⎪ ⎣ ⎦ −∞ ⎣ ⎦ 0 ⎪⎭
⎡ 1 1 ⎤
=c +
⎢ ⎣ α − iω α + iω ⎥ ⎦
⎡ α + iω + α − iω ⎤
=c
⎣ ⎢ α −ω ⎥⎦
2 2
2αc
S (ω ) =
α 2 + ω2
b). Given that a process X (t )} has the autocorrelation function
RXX (τ ) = Ae−α τ cos(ω τ )0 where A > 0 , > 0 and ω0 are real constants, find the power
spectrum of X (t).
Solution:
∞
By definition S ( ω ) = ∫ R (τ ) e − iωτ dτ
−∞
∞
= ∫ Ae−α τ cos ( ω τ )0 e −iωτ dτ

= A ∞ e −α τ cos ( ω − τ ) [ cosωτ − isinωτ ] dτ
−∞
∫ −∞
0
= A e −α τ cosω τ cosωτ + A
∫ e cosω τ ( −sinωτ ) dτ
−α τ
∫ ∞
−∞
0
−∞
∞
0
∫ −ατ∞ 0
= 2 A e cosω τ cosωτ dτ
⎡ cos (ω + ω )τ + cos (ω − ω )τ ⎤
0
∞
= 2A ∫ e−ατ ⎢
⎥ dτ
0 0
−∞ ⎣ 2 ⎦
∞ – a
Using ∫e
0
ax
cosbxdx = a2 + b2
19
⎡ α α ⎤
S (ω ) = A ⎢ 2⎥
⎢ α + (ω + ω ) + α + (ω − ω ) ⎥
2 2 2
⎣ 0 0 ⎦
= Aα ⎡⎢ α 2 + (ω 1 + ω ) + α 2 + (ω 1− ω 2 ⎤⎥
2
⎣⎢ 0 0) ⎥⎦
Problem 25. a). Find the power spectral density of the random binary transmission
⎪⎧1− τ
process whose autocorrelation function is R (τ ) = for τ ≤ 1 .
⎨
⎪0
⎩ elsewhere
Solution:
∞
By definition S ( ω ) = ∫ R (τ )e −iωτ dτ
−∞
1
= ∫ ⎡⎣ 1− τ ⎤⎦ e − iωτ dτ
−1
0 1
= ∫ [1+ τ ] e −iωτ d τ + ∫ [1− τ ] e −iωτ dτ

−1 0
0 1
= ∫ [1+ τ ] e −iωτ dτ + ∫ [1− τ ] e −iωτ dτ

−1 0
−i ωτ 0 1
⎡ e −iωτ
e ⎤ ⎡ e −iωτ e−i ωτ ⎤
= ⎢ (1+ τ ) – + ⎢ (1− τ ) –(−1) 2 2 ⎥
−iω i2ω2 ⎥ −iω iω
⎣ ⎦ −1 ⎣ ⎦0
⎡ 2 – −iω −iω
e − e2 ⎤ ⎥
= ⎢ ω2 ω
⎣ ⎦
1
= [2 − 2cosω ]
ω2 2
= (1− cosω )
ω2 ⎡ ⎛2
2 ω ⎞⎤
=
1− ⎜1− sin 2 ⎟⎥
ω22 ⎢ ⎣⎡ ⎝ 2 ω ⎤ ⎠⎦
= 2sin
ω2 ⎢ ⎣ 2⎥⎦
4 ω
= sin
ω2 2 2
⎡ ω ⎤
⎢ sin 2 ⎥
S (ω ) = ⎢
⎢ ω ⎥⎥
⎣ 2 ⎦
20
−α τ
b). Show that the power spectrum of the autocorrelation function e ⎡⎣ 1+ α τ ⎤⎦ is
4α 3
.
(α + w )
2
2 2
Solution:
∞
By definition, S ( ω ) = ∫ R (τ )e −iωτ dτ
−∞
∞
= ∫ e −α τ ⎡⎣ 1+ α τ ⎤⎦ e −iωτ dτ
−∞
0 ∞
= ∫ e(α −iω )τ (1− ατ ) dτ + ∫ e −(α +iω )τ (1+ ατ ) dτ

−∞ 0
∞
⎡ e(α −iω )τ e(α −iω )τ ⎤ ⎡ e(α +iω )τ e−(α +iω )τ ⎤
0
= ⎢ (1− ατ ) − ( −α ) 2 ⎥
+ ⎢ (1+ ατ ) − (α ) 2 ⎥
α − iω ( α − iω ) − (α + iω ) ( α + iω )
⎢⎣ ⎥⎦ −∞ ⎢⎣ ⎥⎦ 0
⎡ 1 α ⎥ ⎤ ⎡ ⎢ 1 α ⎥ ⎤
= ⎢ α − iω + (α − iω ) 2 + 0 + α + iω + (α + iω ) 2
⎣⎢ ⎥⎦ ⎢⎣ ⎥⎦
⎡ 1 ⎡ ⎤
= + 1 ⎤ +α 1
+
1
α − iω α + iω ⎥ ⎦ ⎢⎣ (α − iω ) (α + iω )2 ⎥ ⎦
2
⎡ α + iω 2 + α − iω 2 ⎤
2α ( ) ( )
= 2 + α ⎢ ⎥⎥
(α + ω )
2
2 2
α + ω2
⎢
⎣ ⎦
2α
= 2 2 +
2α α 2 − ω2 ( )
α +ω ( )
2
α 2 + ω2
2α (α + ω ) + 2α (α − ω )
2 2 2 2
=
(α + ω ) 2 2 2
2α (α + ω + α − ω )
2 2 2 2
=
(α + ω ) 2 2 2
4α 3
= (α 2 − ω 2 )2
Problem 26. a). Find the spectral density function whose autocorrelation function is
⎧ τ
⎪1− ;τ ≤T
given by R (τ ) = ⎨ τ
⎪0 ; elsewhere
⎩
Solution:
21
S ( ω ) = ∫ R (τ )e −iωτ dτ
−∞
T
⎡ 1− τ ⎤ −iωτ
= ∫⎢ ⎥ e dτ
−T ⎣
τ ⎦
T
⎛ τ ⎞
= ∫ ⎜1− ⎟ [cosωτ − isinωτ dτ]
−T ⎝
T ⎠
T
⎛ τ ⎞
= ⎜1− ⎟ cosωτ dτ
∫ T⎠
−T ⎝
T
⎛ τ ⎞
= 2 ⎜1− ⎟ cosωτ dτ
∫0 ⎝ T ⎠
T
⎛ τ⎞
= 2 ⎜1− ⎟ cosωτ dτ
∫0 ⎝ T ⎠ T
⎡⎛ τ ⎞ sinωτ ⎛ −1 ⎞⎛ −cosωτ ⎞⎤
= 2 ⎢ ⎜1− ⎟ – ⎜ τ ⎟⎜
T ω ω 2 ⎟⎥
⎣⎝ ⎠ ⎝ ⎠⎝ ⎠⎦ 0
⎡ −cosωT 1 ⎤
=2 +
⎣⎢ ωT ω2T ⎥ ⎦
2
= [
2 1− cosωT
]
ω222T 2 ⎛ ωT ⎞
= ω T 2 sin ⎜ ⎟
2
⎝ ⎠
⎛ ωT ⎞
4sin2
⎜2 ⎟
S (ω ) = ⎝ ⎠
ω2T
b). Find the power spectral density function whose autocorrelation function is given by
A2
R (τ ) = cos (ω0τ )
2
Solution:
∞
S ( ω ) = ∫ R (τ ) e −iωτ dτ
−∞
∞ 2
A
=
∫ cos ( ω 0τ ) e −iωτ dτ
−∞
2
∞
A2
=
2 −∞
∫ cos (ω τ ) [cosωτ − isinωτ ] dτ
0
∞ ∞
A ⎡ 2
⎤
⎢ ⎡ cos ( ω − ω0 )τ + cos ( ω + ω0 ) τ ⎤⎦ dτ − i ∫ ⎡⎣ sin ( ω + ω0 ) τ + sin ( ω − ω0 ) ⎤⎦ dτ ⎥
4 ∫⎣
=
⎣ −∞ −∞ ⎦
22
A2 ⎡ ∞ ∞
= ⎢ ∫ ⎡⎣ cos ( ω + ω0 ) τ − isin ( ω + ω0 ) τ ⎤⎦ dτ + ∫ ⎡⎣ cos ( ω − ω0 ) τ − isin ( ω − ω0 ) τ ⎤⎦ dτ

4 ⎣ −∞ −∞
∞ ∞ ⎤
2
⎡
= A ⎢ e−i(ω +ω )τ dτ + ∫ e −i(ω −ω )τ dτ ⎥
∫
0 0
4 ⎣ −∞ −∞ ⎦
By definition of Dirac delta function
1 ∞ iωτ
S (ω ) = e dτ
2π −∞∫
∞
2πδ ( ω ) = ∫ e −iωτ dτ
−∞
(1) reduces to
A2
S (ω ) = ⎡⎣ 2πδ ( ω + ω0 ) + 2πδ ( ω − ω0 ) ⎤⎦
4
π A2
S (ω ) = ⎡ δ (ω + ω0 ) + δ (ω − ω0 )⎤ ⎦
4 ⎣
Problem⎧ b27. a). If the power spectral density of a WSS process is given by
(a − ω ); ω ≤ a
⎪
S (ω ) = ⎨ a find its autocorrelation function of the process.
⎪0 ;ω >a
⎩
Solution:
∞
1
R (τ ) = S ( ω ) eiωτ dτ
2π ∫ −∞
b ( a − ω ) eiωτ dτ
a
1
= ∫
2π a −a
a
1 b
= ∫ ( a − ω )(cosωτ + isinωτ d)ω
2π a−a a
2 b
( a − ω ) cosωτ d ω
2π ∫ 0a
=
sinωτ cosωτ ⎤
a
b ⎡
= ⎢ (a −ω) – ⎥
πa ⎣ τ τ
2
⎦0
b ⎡ cosaτ 1 ⎤
= − +
⎢
πa ⎣ τ 2
τ2 ⎥⎦
= [
b 1− cosaτ
]
π aτ 2 2 ⎛ aτ ⎞
b 2sin
= π aτ 2 ⎜2⎟
⎝ ⎠
23
b).The power spectral density function of a zero mean WSS process { X (t )} is given
⎧⎪1 ; ω < ω0 ⎞
by S ( ω ) = . Find R (τ ) and show also that X (t ) & X ⎛t + π are
⎨ ⎜ ⎟
⎪0
⎩ ; elsewhere ⎝ ω0⎠
uncorrelated
Solution:
1∞
R (τ ) = S ( ω ) e iωτ d ω
2π −∞ ∫
ω0
1
eiωτ d ω
2π −∫ω
=
0 ω
= 1 ⎢ ⎡ e ωτ ⎥ ⎤
0
i
2π iτ
⎣ 0
⎦ −ω
πτ1 e −2 e−iω0τ ⎤ ⎥
⎢ ⎡ iω0τ
=
⎣ ⎦
sinω0τ
=
πτ ⎞
To show that X (t ) X ⎛t + π are uncorrelated we have to show that the auto
& ⎜ ⎟
⎝ ω0 ⎠
covariance is zero.
⎞⎤
⎡ ⎛ π =0
i.e., C X (t ) X t +
⎢⎣ ⎜⎝
ω 0⎟⎥⎠⎦
Consider,
⎡ ⎛ π ⎞⎤ ⎛π ⎞ ⎡ ⎛ π ⎞⎤
C ⎢ X (t ) X ⎜ t + ⎟⎥ = R XX ⎜ − E ⎡ ⎣ X (t )⎦ ⎤ E ⎢ X ⎜ t + ⎟⎥
ω ω ⎟ ω
⎣ ⎝ 0 ⎠⎦ ⎝ 0 ⎠ ⎣ ⎝ 0 ⎠⎦
= RXX ⎜ π − 0
⎛ ⎞
ω ⎟
⎝0⎠
⎡ ⎛ π ⎞⎤ sinπ
C ⎢ X (t ) X ⎜ t + ⎟⎥ = =0
ω πτ
⎣ ⎝ 0 ⎠ ⎦
Hence X ( t ) X ⎛t + π ⎞ are uncorrelated.
& ⎜ ⎟
⎝ ω0⎠
Problem 28. a). The power spectrum of a WSS process { X (t )} is given
1
by S (ω ) = . Find the autocorrelation function and average power of the process.
(1+ ω ) 2 2
Solution:
1 ∞
R (τ ) = S ( ω ) eiωτ d ω
∫
2π −∞
24
∞ 1
1 eiωτ d ω ------- (1)
= ∫
2π −∞ (1+ ω 2 )
2
The integral in (1) is evaluated by contour integration technique as given below.

eiazdz
Consider ∫ , where C is the closed contour consisting of the real axis from R to
C( 1+ z )
2
R and the upper half of the Ole , Z = R

1+ Z 2 = 0
Z 2 = −1
Z = ±i
Z = i is a double pole.
1 Lt
d ⎡
( z − i ) + ( z )⎤
2
[Res] Z = i =
( z −1)! z → i dz ⎣ ⎦
⎡ iaz ⎤
1 Lt d ⎢ z − i 2 e 2 ⎥
= 1! z → i dz ⎢ ( ) ( z 2 +1) ⎥
⎣ ⎦
iaz
Lt d e
=
z → i dz ( z + i ) 2
Lt ⎡ ( z + i ) i a eiaz − eiaz 2 ( z + i ) .1 ⎤
2
= ⎢
z →i⎢ ⎣ ( z + i )4 ⎦⎥
Lt ⎡ ( z + i ) i a eiaz − 2eiaz ⎤
= ⎢
z →i⎢ ⎣ ( z + i )3 ⎦⎥
e−a (−2a − 2) 2e−a (1+ a ) e−a (1+ a )
= = =
8 ( −i ) 8i 4i
n
Res
By Cauchy residue theorem, f (z)dz = 2 i f (z) and taking limits R
c
z=a
i=1 i
and using Jordan’s lemma

eiax 2 ie a (1+ a) e a (1+ a)
dx = = ----- (2)
(1+ x )
2
2 4i 2
Using (2) in (1)
1 (1+ τ ) e−τ π (1 + τ ) e−τ
R (τ ) = =
2π 2 4
Average power of { X ( t )} = ⎡⎣ R (τ ) ⎤⎦ τ =0
⎡ (1+ τ ) e −τ ⎤ 1
=⎢ = 4 = 0.25
4⎥
⎣ ⎦ τ =0
25
b). Suppose we are given a cross – power spectrum defined by

a + jbω , −W < ω < W
⎧
SXY (ω ) = ⎪⎨ W where ω > 0 , a & b are real constants. Find the cross
⎪⎩0 , elsewhere
correlation
Solution:
1 ∞
RXY ( τ ) =
∫ SXY (ω ) e jωτ d ω
2π −∞
1 W
= ∫ ⎜⎛a + jbω ⎞⎟e jωτ dω
2π −W ⎝ W⎠
W
+ jb ⎢ ⎡ ω e ωτ – e jωτ ⎤
ωτ
= a ⎢⎡ e ⎥⎤
j j W
2π jτ 2π W jτ j 2τ 2 ⎥
⎣ ⎦ −W ⎣ ⎦ −W
jb ⎡ We τ – − jWτ
e jWτ We − jWτ e 2 ⎥⎤
jWτ − jWτ jW
a ⎡e −e ⎤⎥
= 2π ⎢ jτ + 2π W ⎢ jτ + τ 2 + jτ τ
⎣ ⎦ ⎣ ⎦
asinπτ(Wτ ) bW ⎢ ⎡ e − e2
jW τ − jW τ ⎤ ⎥ ⎢
jb 2 ⎡ e − e2 ⎤
jW τ − jW τ ⎥
= + π Wτ + π Wτ
⎣ ⎦ ⎣ ⎦
asin (Wτ ) b cos (Wτ ) bsin (W τ )
= + −
1 πτ⎡ ⎛ b ⎞ πτ π Wτ 2 ⎤
= 2 ⎢ ⎜ aτ − ⎟sin (Wτ ) + bτ cos (Wτ ) ⎥
πτ W
⎣⎝ ⎠ ⎦
Problem 29. a). The cross-power spectrum of real Random process { X (t )} and {Y (t )}
is given by S (ω ) = a + jbω; ω < 1 . Find the cross correlation function.
XY ⎨
⎪0
⎩ ; elsewhere
Solution:
1∞
RXY (τ ) = ∫ SXY ( ω ) e jωτ d ω
2π −∞
1
1
=
∫ ( a + jbω ) e jωτ d ω
2π −1
1
= a ⎢ ⎡ e ωτ ⎥ ⎤ + b ⎢ ⎡ ωe ωτ – e jωτ ⎤ 1
j j
2π jτ 2π jτ j 2τ 2 ⎥
⎣ ⎦ −1 ⎣ ⎦ −1
a ⎢ ⎡ e jτ − e− jτ ⎤ ⎥ jb ⎢ ⎡ e jτ jτ − jτ – − jτ
= 2π + 2π jτ + eτ 2 + ejτ e 2 ⎥⎤
jτ τ
⎣ ⎦ j ⎣ j ⎦
a jb ⎡ e τ − e− τ ⎤ + jb ⎡ e jτ − e − jτ ⎤
=
πτ
[ sinτ ] + 2π ⎢ jτ
⎥
2π
⎢
τ2 ⎥
⎣ ⎦ ⎣ ⎦
asinτ bcosτ bsinτ
= + −
πτ πτ πτ 2
26
1
= ⎡ ( aτ − b ) sinτ + bτ cosτ ⎤
πτ2⎣ ⎦
b). Find the power spectral density function of a WSS process with autocorrelation
function. R (τ ) = e −ατ .
2
Solution:
∞
SXX (ω ) = RXX (τ ) e−iωτ dτ

∫
−∞
∞
∫ e e dτ
− aτ −iωτ
= −∞
2
∞
(
− aτ 2 +iωτ )
= ∫e dτ
−∞
∞ ⎛ iωτ ⎞
−a τ 2 + ⎟
⎝⎜ ⎠
= ∫e a
dτ
−∞
∞ − a ⎛τ 2 + i ω ⎞
τ
⎜ 2. . ⎠⎟
∫e
⎝
= a
dτ
−∞
⎡⎛ 2 2iω i 2ω 2 ⎞ i2ω 2 ⎤
∞ −
a ⎢ ⎜τ + t+ − ⎥
⎢ ⎣ ⎝ 2a 2 2 a 2 ⎟ ⎠ 2 2 a 2 ⎥⎦
dτ
= ∫e
−∞
⎡ ⎛ iω ⎞2 ω2 ⎤
∞
−a ⎢ ⎜τ + ⎟ + 4a2 ⎥
⎢ ⎣ ⎝ 2a ⎠
dτ
∫e
⎦
=
−∞
⎡ ⎛ iω ⎞2 ⎤ ω2
∞
−a ⎢ ⎜τ + ⎟ ⎥ −
= ∫e
−∞
⎢ ⎣ ⎝ 2a ⎠ ⎥ ⎦ 4a 2 .e dτ
2
ω2 ∞ ⎛ iω ⎞
− −a ⎜τ + ⎟
∴ SXX (ω ) = e 2a
dτ
∫e
⎝ ⎠
4a
−∞
iω ⎞
2
⎛ ⎛ iω ⎞
Now let, a τ + ⎟ = u2 i.e. u = a τ + ⎟
⎜ ⎜
⎝ 2a ⎠ ⎝ 2a ⎠
∴ du = a dτ
and as, τ → −∞, u → −∞ and as τ → ∞, u → ∞
ω2 ∞
(ω ) = e
−
e−u du
2
∴ SXX
ω 2
4a
∫−∞ a
−
4a ∞
e 2
a .2∫ e du
= −u
2
.... e −u is even
0
−ω 2
∞
2e 4a π 2
dx = x
∫e
−X
= ... Standard Result :
a 2 0 2
27
π −ω
2
∴ SXX (ω ) =
ae
4a
Problem 30. a). The autocorrelation function of the random telegraph signal process is
given by R (τ ) = a 2e −2γ τ . Determine the power density spectrum of the random telegraph
signal.
Solution:
∞ ∞
SXX (ω ) = ∫ RXX (τ )e−iωτ dτ = ∫ae

2 −2 γ τ
e −iωτ dτ
−∞ −∞
0 ∞
= a 2 ∫ e −2 γ τ e−iωτ dτ + a ∫ e2 −2γ τ
e −iωτ dτ
−∞ 0
0 ∞
= a 2 ∫ e −2 γ ( −τ ) e−iωτ dτ + a 2 ∫ e−2 γ (τ ) .e −iωτ dτ

−∞ 0
0 ∞
= a 2 ∫ e( 2 γ −iω )τ dτ + a 2 ∫ e − ( 2 γ +iω )τ
dτ
−∞ 0
0 ∞
⎡ e( 2γ −iω )τ ⎤ ⎡ e−( 2 γ −iω )τ ⎤
=a ⎢2
+a ⎢
2
(⎣ 2γ − iω ) ⎦ −∞ ⎣ − ( 2γ − iω ) ⎦ 0
⎥ 2
⎥ a2
=
a
( 2γ − iω )
(
e0 – e−∞ − ) e −∞ – e0 ( )
( 2γ + iω )
a2 a2
=
2γ − iω
( )
1− 0 − (0 −1)
( 2γ + iω )
⎡ 1 1 ⎤ ⎡ ⎢2γ + iω + 2γ − iω ⎤ ⎥
= a 2 ⎢ 2γ − iω + 2γ + iω ⎥ = a 2 ( 2γ ) + ω 2
2
⎣ ⎦ ⎢⎣ ⎥⎦
2⎡ 4γ ⎤ 4a 2γ
i.e., SXX (ω ) = a ⎢ 2 2 ⎥
= 2
⎣ 4γ + ω ⎦ 4γ + ω
2
b). Find the power spectral density of the random process {x (t )} if E {x (t )} = 1

and R xx(τ ) = 1+ e−α τ .
Solution:
∞
S (ω ) = ∫ R (τ ) e
XX
−iωτ
dτ
−∞
∞ ∞ 0 ∞
= ∫ ⎡ 1+
⎣ e ⎤ e⎦ dτ
−α τ −iωτ
∫ e dτ + ∫ e−∞e dτ + ∫ e e0 dτ
−iωτ ατ −iωτ − ατ −iωτ
−∞
= −∞
1 1
= δ (ω ) + +
α − iω α + iω
2α
= δ (ω ) +
α 2 + ω2
*******
28
Unit.5. Linear System with Random Inputs
UNIT – V LINEAR SYSTEMS WITH RANDOM INPUTS
PART - A
Problem 1. If the system function of a convolution type of linear system is given by

⎧1
⎪ for t ≤ a
h (t ) = ⎨ 2a find the relation between power spectrum density function of
⎪0 for t > a
⎩
the input and output processes.
Solution:a
− sin a ω
H (ω ) = h (t ) e i ω t dt =
∫
−a aω
We know that S YY (ω ) = H (ω ) S (ω )
2
XX
sin2 aω
⇒ SYY (ω ) = SXX (ω ) .
a2 ω 2
Problem 2. Give an example of cross-spectral density.
Solution:
The cross-spectral density of two processes X (t ) and Y (t ) is given by
⎧⎪ p + iqω, if ω < 1
S
XY
(ω ) = ⎨
⎪⎩0, otherwise
⎧ A, 0 ≤ t ≤ 1
Problem 3. If a random process X (t ) is defined as X (t ) = ⎨ , where A is a
⎩0, otherwise
random variable uniformly distributed from −θ to θ . Prove that autocorrelation function
θ2
of X (t ) is .
3
Solution:
RXX ( t, t + τ ) = E ⎣⎡ X ( t ) .X ( t + τ ) ⎦⎤
⎡⎣ X ( t ) is cons tan t ⎤ ⎦
= E ⎡⎣ A2 ⎤⎦
But A is uniform in ( −θ ,θ )
1
∴ f (θ ) = , −θ < a < θ
2θ
θ
∴RXX ( t, t + τ ) = ∫ a2 f ( a ) da
−θ θ
θ 1 1 ⎡ a3 ⎤
=
∫
−θ
a2 .
2θ
dθ =
2θ ⎢⎣ 3 ⎥⎦ −θ
1
1 ⎡ 3 1 θ2
θ − ( −θ ) ⎤ = θ3 =
3
=
.2
6θ ⎣ ⎦ 6θ 3
1
Problem 4. Check whether is a valid autocorrelation function of a random
1+ 9τ 2
process.
1
Solution: Given R (τ ) =
1+ 9τ 2
1 1
∴ R (− τ ) = = = R (τ )
( )
1+ 9 −τ 2 1+ 9τ 2
∴ R (τ ) is an even function. So it can be the autocorrelation function of a random
process.
Problem 5. Find the mean square value of the process X(t) whose power density spectrum
4
is .
4 + ω2
Solution:
4
Given SXX (ω ) =
4 +ω2
1 ∞
Then RXX (τ ) = ∫ SXX (ω ) eiτω dω
2π −∞
Mean square value of the process is E ⎡ ⎣ X 2 (t )⎦ ⎤ = RXX (0)
∞
1
2π −∞∫ XX ( )
= S ω dω
1 ∞ 4
2π ∫−∞ 4 + ω 2
= dω
= 4∫∞ 12 dω ⎡ 1
⎢ 4 + ω 2 is even
⎤
π 4+ω ⎥
∞ ⎣ ⎦
ω⎤
0
4⎡ 1 2
= . ⎢ tan −1 = (
tan −1 ∞ − tan −1 0 )
π 2⎣ ⎥
2⎦0 π
2 π
= . =1
π2 ⎧1
; 0≤t ≤T
Problem 6. A Circuit has an impulse response given by h ( t ) = ⎪⎨ T find the
⎪⎩0 ; elsewhere
relation between the power spectral density functions of the input and output processes.
Solution:
T
H (ω ) = ∫ h (t ) e−iωtdt
0
2
T
1
= e−iω tdt
∫T
0
T
1 ⎡ e−iωt ⎤
= ⎢ ⎥
T ⎣ −iω ⎦ 0
1 ⎢⎡ −e −iωTiω+1⎤ ⎥
=T
⎣ ⎦
= (1− e )
−i ωT
Tiω
SYY (ω ) = H (ω ) 2 S (ω )
XX
(1− e ) −iωT 2
= SXX (ω )
ω 2T 2
Problem 7. Describe a linear system.
Solution:
Given two stochastic process { X 1 ( t )} and{X 2 (t )}, we say that L is a linear
transformation if
L ⎡⎣ a1 X 1 ( t ) + a2 X 2 ( t ) ⎤⎦ = a1 L ⎡⎣ X1 ( t ) ⎤⎦ + a2 L ⎡⎣ X 2 ( t
)⎤⎦
Problem 8. Given an example of a linear system.
Solution: x (t ) .
Consider the system f with output tx (t ) for an input signal
i.e . y (t ) = f ⎣ ⎡ X (t )⎤ ⎦ = t x (t )
Then the system is linear.
For any two inputs x1 (t ) , x2 (t ) the outputs are tx1 (t ) and tx2 (t ) Now
f ⎡⎣ a1 x1 ( t ) + a2 x2 ( t ) ⎤⎦ = t ⎡⎣ a1 x1 ( t ) + a2 x2 ( t ) ⎤⎦
= a1tx1 (t ) + a2tx2 (t )
= a1 f ( x1 ( t ) ) + a2 f ( x2 ( t ) )
∴the system is linear.
Problem 9. Define a system, when it is called a linear system?
Solution:
Mathematically, a system is a functional relation between input x (t ) and output y (t ) .
Symbolically, y (t ) = f ⎣ ⎡ x (t )⎤ ⎦ , −∞ < t < ∞..
The system is said to be linear if for any two inputs x1 (t ) and x2 (t ) and constants
a1, a2 , f ⎡⎣ a1 x1 ( t ) + a2 x2 ( t ) ⎤⎦ = a1 f ⎡⎣ x1 ( t ) ⎤⎦ + a2 f ⎡⎣ x2 ( t ) ⎤⎦ .
Problem 10. State the properties of a linear system.
Solution:
Let X1 (t ) and X 2 (t ) be any two processes and a and b be two constants.
3
If L is a linear filter then

L ⎡⎣ a1 x1 ( t ) + a2 x2 ( t ) ⎤⎦ = a1L ⎡⎣ x1 ( t ) ⎤⎦ + a2 L ⎡⎣ x2 ( t ) ⎤⎦ .
Problem 11. Describe a linear system with an random input.
Solution:
We assume that X (t ) represents a sample function of a random process{X (t )}, the
system produces an output or response Y (t ) and the ensemble of the output functions
forms a random process{Y (t )}. The process {Y (t )} can be considered as the output of
the system or transformation f with { X (t )} as the input the system is completely
specified by the operator f .
Problem 12. State the convolution form of the output of linear time invariant system.
Solution:
If X (t ) is the input and h (t ) be the system weighting function and Y (t ) is the output,
∞
then Y (t ) = h (t ) * X (t ) = ∫ h (u ) X (t − u ) du
−∞
Problem 13. Write a note on noise in communication system.
Solution:
The term noise is used to designate unwanted signals that tend to disturb the transmission
and processing of signal in communication systems and over which we have incomplete
control.
Noise
Uncorrelated Noise Correlated Noise
Internal Noise External Noise
White Shot Partition Atmospheric Man made

Noise, Noise Noise Noise Noise
Thermal
Noise
Problem 14. Define band-limited white noise.
Solution:
Noise with non-zero and constant density over a finite frequency band is called band-
limit white noise i.e.,
⎧ N0 ω ≤ ω
⎪ ,
SNN (ω ) = ⎨ 2 B
⎪⎩0 , otherwise
Problem 15. Define (a) Thermal Noise (b) White Noise.
Solution:
4
(a) Thermal Noise: This noise is due to the random motion of free electrons in a
conducting medium such as a resistor.
(or)
Thermal noise is the name given to the electrical noise arising from the random
motion of electrons in a conductor.
(b) White Noise(or) Gaussian Noise: The noise analysis of communication
systems is based on an idealized form of noise called White Noise.
PART-B
Problem 16. A random process X (t ) is the input to a linear system whose impulse
response is h (t ) = 2e−t , t ≥ 0 . If the autocorrelation function of the process is
RXX (τ ) = e −2 τ , find the power spectral density of the output process Y (t ) .
Solution:
Given X (t ) is the input process to the linear system with impulse response
h (t ) = 2e−t , t ≥ 0
So the transfer function of the linear system is its Fourier transform
∞
−∞
∞
⎡⎣ 2e − t , t ≥ 0⎤⎦
= ∫−∞2e−te−iωt dt
∞
= 2∫ e−(1+iω )t dt
0
∞
⎡ e−(1+iω )t ⎤
= 2⎢ ⎥
⎣ − (1+ iω ) ⎦ 0
−2 2
=
1+ iω
[0 −1] =
1+ iω
Given RXX (τ ) = e −2τ
∴ the spectral density of the input is

∞
SXX (ω ) = RXX (τ )e −iωτ dt

∫−∞
∞
= ∫ e−2 τ e −iωτ dτ
−∞
0 ∞
∫ e e dτ + ∫ e 0 e dτ
2τ −iωτ −2τ −iωτ
= −∞
0 ∞
( 2−iω )τ − ( 2+iω )τ
= ∫e dτ + ∫ e dτ
−∞ 0
5
0 ∞
⎡ e( 2−iω )τ ⎤ ⎡ e−( 2−i ω )τ ⎤
=⎢ ⎥ +⎢ ⎥
⎣ 2 − iω ⎦ −∞ ⎣ − ( 2 + iω ) ⎦ 0
1 1
= [1− 0]− [0 −1]
2 − iω 2 + iω
1 1
= +
2 − iω 2 + iω
2 + iω + 2 − iω 4
= =
(2 + iω )( 2 − iω ) 4 + ω 2
We know the power spectral density of the output process Y (t ) is given by
S YY (ω ) = H (ω ) S (ω )
2
XX
2
4
= 2
1+ iω 4 + ω
2
4 4
=
(1+ ω ) 4 + ω2
2
16
=
( )(
1+ ω 2 4 + ω 2 )
Problem 17. If Y ( t ) = A cos ( ω0 t + θ ) + N ( t ) , where A is a constant , θ is a random
variable with uniform distribution in (−π ,π ) and N (t ) is a band-limited Gaussian white
, for ω − ω < ω
⎧ N0
⎪
noise with a power spectral density SNN (ω ) = ⎨ 2 0 B
. Find the power
⎪⎩0, elsewhere
spectral density of Y (t ) . Assume that N (t ) and θ are independent.
Solution:
Given Y ( t ) = A cos ( ω0 t + θ ) + N ( t )
N (t ) is a band-limited Gaussian white noise process with power spectral density
N0
SNN (ω ) = , ω − ω 0 < ωB ie. ω0 − ωB < ω < ω0 + ωB
2
∞
Required SYY (ω ) = ∫ R (τ ) e
YY
−iωτ
dτ
−∞
Now RYY (τ ) = E ⎡⎣ Y ( t ) Y ( t + τ )⎤⎦

{
= E ⎡⎣ A cos (ω0t + θ ) + N ( t )⎤⎦ ⎡⎣ A cos ( ω0t + ω0τ + θ ) + N ( t + τ )⎤⎦ }
6
⎧⎪ A2 cos (ω t + θ ) cos (ω t + ω τ + θ ) + N ( t ) N ( t + τ ) + A cos (ω t + θ ) N ( t + τ )⎫⎪

=E⎨ 0 0 0 0
⎬
⎪+ A cos ( ω0t + ω0τ + θ ) N ( t )
2⎩
⎪⎭
= A E ⎡ cos (ω t + θ ) .cos (ω t + ω τ + θ )⎤ + E ⎡ N (t ) N (t + τ )⎤
⎣ 0 0 0 ⎦ ⎣ ⎦
+ AE ⎣⎡ cos ( ω0 t + θ ) ⎤⎦ E ⎡⎣ N ( t + τ ) ⎤⎦
+ AE ⎡⎣ cos ( ω0 t + ω0 + θ ) ⎤⎦ E ⎡⎣ N ( t ) ⎤⎦ [ θ and N (t ) are independent]
A2
=
2 {
E ⎡⎣ cos ( 2ω0t + ω0 t + 2θ ) ⎤⎦ + cos ω0τ + RNN (τ ) }
+ A E ⎡⎣ cos ( ω0 t + θ ) ⎤⎦ E ⎡⎣ N ( t + τ ) ⎤⎦
+ A E ⎡⎣ cos ( ω0 t + ω0τ + 2θ ) ⎤⎦ E ⎡⎣ N ( t ) ⎤⎦ }
1
Since θ is uniformly distributed in (−π ,π ) the pdf of θ is f (θ ) = , −π < θ < π
2π
π
∴E ⎡⎣ cos ( ω0 t + θ ) ⎤⎦ = ∫ cos ( ω0 t + θ ) f (θ ) dθ
−π
π
1
= ∫ [cos ω t.cos θ − sin ω t.sin θ ]
−π
0 0
2π
dθ
π π
1 ⎡ ⎤
= ⎢ cos ω 0 ∫
t cos θ d θ − sin ω0 ∫
t sin θ dθ⎥
2π
1 ⎣ −π −π ⎦
= ⎡ cos ω t [sin θ ]π – sin ω t.0 ⎤ = 0
2π ⎣
0
−π 0 ⎦
π
1
Similarly E ⎡⎣ cos ( 2ω0 t + ω0τ + 2θ ) ⎦⎤ = cos ( 2ω0t + ω0τ + 2θ )
∫ −π
2π
dθ
π
1
=
2π ∫ ⎡⎣ cos ( 2ω t + ω τ ) cos 2θ − sin ( 2ω t + ω τ ) sin 2θ ⎤⎦ dθ
−π
0 0 0 0
π π
1 ⎧ ⎫
= ⎨cos ( 2ω0 t + ω0τ ) ∫ cos 2θ dθ − sin ( 2ω0 t + ω0τ ) ∫ sin 2θ dθ ⎬
2π
⎩ −π π −π ⎭
1⎧⎪ ⎡ sin 2θ ⎤ ⎪⎫
= ⎨cos ( 2ω0 t + ω0τ ) . ⎢ − sin ( 2ω0 t + ω0τ ) .0 ⎬ = 0
2π ⎪⎩ ⎣ 2 ⎥ −π ⎪⎭
A2
∴ R (τ ) =
YY
cos ω0τ + RNN (τ )
2 ⎤ −iωτ
∞
⎡ A2
∴ SYY ( ω ) = ∫ ⎢ cos ω0τ + RNN (τ ) ⎥ e dτ
2
−∞ ⎣ ⎦
2 ∞ ∞
A
= 2 cos ω τ 0.e −iωτ d τ +
∫ ∫ RNN (τ ) e−i ωτ dτ
−∞ −∞
π A2
= { δ (ω − ω ) + δ (ω + ω ) } + S ( ω )
0 0 NN
2
7
π A2 {δ (ω − ω ) + δ (ω + ω )}+ N0 ,ω − ω < ω < ω + ω

= 0 0 0 B 0 B
2 2
Problem 18. Consider a Gaussian white noise of zero mean and power spectral density
N0 1
applied to a low pass RC filter whose transfer function is H ( f ) = . Find
2 1+ i2π fRC
the autocorrelation function.
Solution:
The transfer function of a RC circuit is given. We know if X (t ) is the input process and
Y (t ) is the output process of a linear system, then the relation between their spectral
densities is SYY (ω ) = H (ω ) 2S (ω )
XX
The given transfer function is in terms of frequency f

SYY ( f ) = H ( f ) (f)
2
S XX
SYY ( f ) = 1 N0
1+ 4π f R C 22
2 2 2
1 ∞
∴RYY (τ ) = ∫ SYY ( ω ) eiωτ d ω
2π −∞
∞
1 ei22πτ f2 2 2 N 0 d f
= ∫
2π −∞ 1+ 4π f R C 2
N ∞ ei ( 2πτ ) f
= 4π0 −∞ 2 2 2 ⎛
∫ 4π R C ⎜ 1 + f 2 ⎞⎟ df
4π 2 R 2 C 2
⎝ ⎠
N ∞
ei ( 2πτ ) f
=
16π R C
23 02
−∞ ⎛ 1 ⎞
∫ 2 df
⎜ + f2
⎝ 2 π ⎠
⎟ ∞ π −ma
RC eimx
We know from contour integration ∫ dx = e
−∞ a2 + x2 a
τ
N0 π −
∴R (τ ) = e 2π RC
16π 3 R2 C 2
YY
1
2π RC
N0 τ
–
= 2π 2 RC e 2π RC
16π 3 R2 C 2
τ
N0 –
= e 2π RC
8π RC
Problem 19. A wide-sense stationary noise process N(t) has an autocorrelation function
RNN (τ ) = Pe−3 τ , where P is a constant Find its power spectrum
Solution:
8
Given the autocorrelation function of the noise process N (t ) is RNN (τ ) = Pe−3 τ .

∞
∴SNN (ω ) = ∫ RXX (τ ) e−iωτ dτ

−∞
∞
= ∫ P e−3 τ e −iωτ dτ
−∞
∞
⎧0 ⎫
= P ⎨ ∫ e e dτ + ∫ e e dτ ⎬
⎩ −∞ 0 ⎭
∞
⎧ 0 ( 3−iω )τ ⎫
= P⎨ ∫ e dτ + ∫ e− ( 3+iω )τ dτ ⎬
⎩ −∞ 0 ⎭
0 ∞
⎧⎪ ⎡ e( ) ⎤
3−i ω τ ⎡ e( ) ⎤
− 3+i ω τ
= P ⎨⎢ ⎥ +⎢ ⎥
⎪⎩ ⎣ 31− iω ⎦ −∞ ⎣ − ( 31 + iω ) ⎦ 0
⎧ ⎫
= P ⎨ 3 − iω (1− 0 ) − 3 + iω ( 0 −1)⎬
⎩ ⎭
⎧ 1 1 ⎞
=P +
⎨⎩ ⎟
3 − iω 3 + iω ⎠⎬
⎧⎨⎪( 33 −
+ iiωω )(
+ 33−+iωiω ) ⎪
⎫
= 9+ω
6P 2
=P
⎩⎪ ⎪⎭
Problem 20. A wide sense stationary process X(t) is the input to a linear system with
impulse response h (t ) = 2e−7t ,t ≥ 0 . If the autocorrelation function of X(t) is
RXX (τ ) = e −4 τ , find the power spectral density of the output process Y(t).
Solution:
Given X(t) is a WSS process which is the input to a linear system and so the output
process Y(t) is also a WSS process (by property autocorrelation function)
Further the spectral relationship is S YY (ω ) = H (ω ) S (ω )
2
XX
Where SXX (ω ) = Fourier transform of RXX (τ )

∞
∫ R (τ )e
−iωτ
= XX
dτ
−∞
∞
−∞
0 ∞
∫ e e dτ + ∫ e 0 e dτ
= −∞
0 ∞
τ ( 4−iω ) −τ ( 4+iω )
= ∫e dτ + ∫ e dτ
−∞ 0
9
0 ∞
= ⎢⎡ eτ ( 4−iω ) ⎤⎥ + ⎢⎡ e −τ ( 4+iω ) ⎥⎤
4 − iω 4 + iω
⎣ ⎦ −∞ ⎣ ⎦0
1 1
= ⎡⎣ e − e ⎤⎦ +
0 −∞ ⎡⎣ e −∞ − e 0 ⎤⎦
4 − iω 4 + iω
1 1
= –
4 − iω 4 + iω
4 + iω − 4 + iω 2iω
SXX (ω ) = =
(4 − iω )( 4 + iω ) 16 + ω 2
H (ω ) = Fourier transform of h (t )
∞
= ∫ h (t )e−iωtdt
−∞
∞
⎡⎣ h ( t ) = 0 if t < 0⎤⎦
= ∫0 2e−7t .e−iωt dt
∞
= 2∫ e −( 7+iω ) t dt
0
∞
⎡ e−(7+iω )t ⎤
= 2⎢ ⎥
⎣ − ( 7 + iω ) ⎦ 0
−2 2
= ⎡⎣ e −∞ − e0 ⎤⎦ =
7 + iω 7 + iω
2 2
∴ H (ω ) = =
7 + iω 49 + ω 2
4
∴ H (ω ) =
2
49 + ω 2
Substituting in (1) we get the power spectral density of Y(t),
4 2iω 8iω
SYY (ω ) = . = .
49 + ω 16 + ω
2 2
(
49 + ω 2 16 + ω 2 )( )
Problem 21. A random process X(t) with RXX (τ ) = e −2 τ is the input to a linear system
whose impulse response is h (t ) = 2e−t , t ≥ 0 . Find cross correlation RYY (τ ) between the
input process X(t) and the output process Y(t).
Solution:
The cross correlation between input X(t) and output Y(t) to a linear system is
RXY (τ ) = RXX (τ ) h (τ )
Taking Fourier transforms, we get
SXY (ω ) = RXX (ω ) H (ω )
Given RXX (τ ) = e −2 τ
10
∴ SXX (ω ) = ∫ RXX (τ )e−iωτ dτ

−∞
∞
−∞
0 ∞
∫ e e dτ + ∫ e 0 e dτ
= −∞
0 ∞
= ∫ e( 2−iω )τ dτ + ∫ e −( 2+iω )τ dτ
−∞ 0
0 ∞
⎡ e( 2−iω )τ ⎤ ⎡ e−( 2+iω )τ ⎤
=⎢ ⎥ +⎢ ⎥
⎣ 2 − iω ⎦ −∞ ⎣ − ( 2 + iω ) ⎦ 0
1 1
= (1− 0) − (0 −1)
2 − iω 2 + iω
1 1 4
= + =
2 − iω 2 + iω 4 + ω 2
∞
∴ H (ω ) = ∫ h (t ) e−iωtdt
−∞
∞
∫ 2 e e dt
−t −iωt
= −∞
∞
∞
−( 1+iω ) t
⎡ e−(1+iω )t ⎤
= 2∫e dt = 2 ⎢ ⎥
0
⎣ − (1+ iω ) ⎦ 0
−2
= [0 −1] = 2
1+ iω 1+ iω
4 2
∴SXY (ω ) = .
4 + ω 2 1+ iω
= 8
(2 + iω )( 2 − iω )(1+ iω )
8 A B C
Let = + +
(2 + iω )( 2 − iω )(1+ iω ) 2 + iω 2 − iω 1+ iω
∴8 = A ( 2 + iω )(1+ iω ) + B (2 + iω )(1+ iω ) + C (2 + iω )( 2 − iω )
Put ω = i 2, then 8 = A ( 4 )( −1 ) ⇒ A = −2
8
ω = i then 8 = C (1)(3) ⇒ C =
3
2
ω = −i 2, then 8 = B ( 4 )( 3 ) ⇒ B =
3
2
−2 8/3
∴ SXY (ω ) = + 3 +
2 + iω 2 − iω 1+ iω
11
Taking inverse Fourier transform

⎛ −2 ⎞ ⎜ ⎛ 3 2 ⎟⎞ 1 ⎛ 8 / 3 ⎞
∴ R (τ ) = F − 1 + +F −1 + F−
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ 2 + iω ⎠ ⎜ 2 − iω ⎟ ⎝ 1+ iω ⎠
XY
⎝ ⎠
2 8
= −2.e − τ μ (τ ) + e τ μ ( −τ ) + e −τ μ (τ )
2 2
3 3
Problem 22. If X(t) is a band limited process such that SXX (ω ) = 0 ,where ω > σ prove
that 2 ⎡⎣ R XX ( 0 ) − RXX (τ ) ⎤⎦ ≤ σ 2τ 2 R XX ( 0)
Solution:
Given SXX (ω ) = 0 , ω > σ
⇒ SXX (ω ) = 0 if ω < −σ or ω > σ
1 ∞
RXX ( τ ) =
∫ SXX (ω ) eiτ ω dω
2π −∞
1 σ S ( ω ) eiτω dω
=
2π ∫
−σ
XX
σ
1
=
2π ∫ S XX ( ω )( cosτω + i sinτω ) dw
−σ
1 ⎧σ σ
⎫
= ⎨ ∫ S XX ( ω ) cosτω dw + i ∫ S XX ( ω ) sin τω dw⎬
2π
⎩ −σ −σ ⎭
σ σ σ
∴ ∫ S XX ( ω ) cosτ ω d ω = 2∫ S XX (ω ) cos τ ω d ω and ∫ S XX (ω ) sin τ ω d ω = 0

−σ 0 −σ
1 σ
∴ RXX (τ ) = 2 ∫ S XX ( ω ) cosτω d ω
2π 0
σ
1
S ( ω )cosτω d ω
π ∫0 XX
=
1σ
∴ RXX (0 )= ∫ SXX ( ω) dω ...(1)
π 0
1σ 1 σ
∴ RXX ( 0 ) − RXX (τ ) = ( )
π −∫σ
SXX ω dω −
π ∫ S XX (ω ) cos ωτ dω
−σ
σ
1
= ∫ S XX ( ω )(1− cos ωτ ) d ω
π1 −σ ⎛ ωτ ⎞
= σ S (ω ) 2 sin2 dω
π ∫ XX ⎜ 2 ⎟
−σ ⎝ ⎠
We know that sin θ ≤ θ
2 2
12
⎛ ωτ ⎞ ⎛ ωτ ⎞
2
ω 2τ 2 σ 2τ 2 ⎡ 0 ≤ ω ≤ 2σ , ω ≤σ ⎤ ⎦
⎟≤⎜
2 2
∴2 sin ⎜ ⎟ = < ⎣
⎝2 ⎠ ⎝2 ⎠ σ 2 222
1 σ τ
∴ RXX ( 0 ) − RXX (τ ) ≤
π ∫0 XX ( ) 2
.S ω dω
σ
1 σ 2τ 2
≤
π 2 ∫ SXX (ω )dω
0
σ
τσ 2 2
≤
2π ∫ SXX (ω ) dω
0
σ
τσ 2 2
≤
2π ∫ SXX (ω ) dω
0
≤σ τ
2 2
RXX (0) [Using (1)]
2π
∴2 ⎡⎣ R XX (0) − RXX (τ ) ⎤⎦ ≤ σ 2τ 2 R XX (0)
Problem 23. The autocorrelation function of the Poisson increment process is given by
⎧λ 2 for τ >∈
R (τ ) = ⎨⎪ λ⎛ τ ⎞ . Prove that its spectral density is given by
⎪ λ 2
+ ⎜1− ⎟ for τ ≤∈
⎩ ∈ ⎝ ∈ ⎠
ωt
4λ sin2
S ( ω ) = 2πλ 2δ ( ω ) + 2 .
Solution:
⎪⎧λ 2 for τ > −∈ or τ >∈
Given the autocorrelation function R (τ ) = ⎛ ⎞
⎨2 λ τ
⎪λ + ∈⎜1− ⎟
∈ ⎠
for −∈≤ τ ≤∈
⎩ ⎝
∞
∴The spectral density is given by S ( ω ) = ∫ R (τ ) e −iτω dτ , by definition

−∞
−∈ ∈ ∞
= ∫ R (τ ) e −iτω dτ + ∫ R (τ ) e −iτω dτ + ∫ R (τ ) e −iτω dτ

−∞ −∈ ω
∞
−∈
⎡ λ⎛ ⎞⎤
τ
∈
= ∫ λ e dτ + ∫ ⎢ λ +
2 − iτω 2
⎜1− ⎟⎥ e−iτω dτ + λ 2 e −iτω dτ
−∞ ⎢
−∈ ⎣
∈ ⎝ ∈ ⎠⎥ ⎦ ∫ω
−∈ ∞
∈
= ∫ λ 2 e −iτω dτ + ∫ λ 2 e−iτω dτ + ∫ λ 2 e − iτω dτ + ∫
∈
λ ⎛ 1− τ ⎞ −iτω
⎜ ⎟ e dτ
−∞ −∈ ∈
∈
−∈ ⎝
∈⎠
−∞
λ ∈⎛ τ ⎞ −iτω
= ∫ λ e dτ + ∈ ∫ ⎜1− ∈ ⎟ e dτ
2 − iτω
−∞ −∈ ⎝ ⎠
13
λ ∈⎛ τ ⎞
( )
= F λ 2 + ∈ ∫ ⎜1− ⎟( cosτω − i sin τω dτ)
−∈ ⎝ ∈ ⎠
Where F (λ 2 ) is the Fourier transform of λ 2
λ ⎧⎪ ∈ ⎛ ⎞ ∈⎛ ⎞ ⎫⎪
( )
S (ω ) = F λ +2
⎨∫ ⎜ 1− τ ⎟ cos τω dτ − i ∫⎜ 1− τ ⎟ sin τω d τ ⎬ .....(1)
∈ ⎩⎪−∈ ⎝ ∈⎠ −∈ ⎝
∈ ⎠ ⎪⎭
⎛ τ ⎞ ⎛ τ ⎞
But ⎜1− ⎟ cosτω is an even function of τ and ⎜1− ⎟ sin τω is an odd function of τ
⎝ ∈ ⎠ ⎝ ∈ ⎠
∈⎛ ⎞ ∈⎛ ⎞ ∈⎛ τ⎞
τ τ
∴ ∫ ⎜1− ⎟ sin τω dτ = 0 and ∫ ⎜1− ⎟ cosτω dτ = 2 ∫ ⎜1− ⎟ cos τω dτ ( τ > 0; τ =τ )
∈ ∈ ∈ ⎝ ⎠
−∈ ⎝ ⎠ λ ∈⎛ τ ⎞ −∈ ⎝ ⎠ 0
( )
S (ω ) = F λ 2 +
∈ 0⎝
2∫ ⎜1− ⎟ cos τω dτ
∈⎠
∈
2λ ⎡ ⎛ τ ⎞⎛ sin τω ⎞ ⎛ −1 ⎞⎛ − cos τω ⎞⎤
= F λ + ⎢ ⎜ 1−⎟⎜
2
∈ ( ) ∈
−
ω ⎟ ⎜ ∈ ⎟⎜ ω2 ⎟⎥ [by Bernoulli’s formula]
⎣⎝ ⎠⎝ ⎠ ⎝ ⎠⎝ ⎠⎦ 0
2λ ⎡ 1 ⎛ τ ⎞
( ) 1 ⎤ ∈
=F λ + 2
⎜1− sin ⎟ τω − cosτω
∈ ⎢ω ∈ ∈ω 2 ⎥
⎣ ⎝ ⎠ ⎦0
2λ ⎡
= F (λ ) +
1 ⎤
2
0− ( cos ∈ ω − cos 0 )
∈ ⎢ ⎣ ∈ω 2 ⎥⎦
2λ 1
= F λ2 + ( ) [1− cos ∈ ω ]
∈ ∈ω2
2λ ∈ω
= F (λ 2 ) + .2 sin2
∈ ω2 2
2
4λ 2 ∈ω
S (ω ) = F (λ ) +
2
sin .......(1)
∈ ω2
2
2
To find the value of F (λ 2 ) , we shall find the inverse Fourier transform of S (ω ) ,
R (τ ) = F − 1 ( S (ω ))
1∞
= S ( ω ) eiωτ d ω
2π −∞ ∫
Consider S ( ω ) = 2πλ 2δ ( ω ) , where δ ( ω ) is the unit impulse function.
∞
1
R (τ ) = 2πλ 2δ ( ω ) ei ωτ d ω
2π −∞ ∫
∞
=λ ∫ δ (ω ) eiωτ d ω
2
−∞
∞
= λ .1
2
[ ∫ φ (t )δ (t ) dt = φ (0)
−∞
14
=λ 2
⇒ ∫ ei ωτ δ ( ω ) dω = e0 = 1 as φ ( ω ) = eiτω ⎤⎦
−∞
Thus λ 2 = R (τ ) Taking Fourier transform

F ( λ 2 ) = F ( R (τ ) ) = S ( ω ) = 2πλ 2δ ( ω )
4λ ⎛ ω ∈⎞
Substituting in (1) we get S ( ω ) = 2πλ δ ( ω ) + sin 2
2
.
∈2 ω 2 ⎜ 2 ⎟
⎝ ⎠
Problem 24. Suppose X(t) be the input process to a linear system with autocorrelation
1
RXX (τ ) = 3δ (τ ) and the impulse response H (ω ) = , then find(i) the
6 + iω
autocorrelation of the output process Y(t). (ii) the power spectral density of Y(t).
Solution:
1
Given RXX (τ ) = 3δ (τ ) and H (ω ) =
6 + iω
∞
∴ SXX (ω ) = ∫ R (τ ) e
XX
−iωτ
dτ
−∞
∞
= ∫ 3δ (τ ) e −iωτ dτ
−∞
∞
= 3 ∫ δ (τ ) e −iωτ dτ
−∞
∞
We know
∫ δ (τ )φ (τ ) = φ (0 )
−∞
Here φ (τ ) = e −iωτ ∴φ ( 0 ) = 1
∴ SXX (ω ) = 3.1 = 3
We know the spectral relation between input and output process is
S YY (ω ) = H (ω ) S (ω )
2
XX
1
But H (ω ) =
2
36 + ω 2
3
∴ SYY (ω ) = which is the power spectral density of Y (t )
36 + ω 2
Now the autocorrelation of Y (t ) is R (τ ) = F −1 ( S (ω ))
YY YY
⎛ 3 ⎞
R (τ ) = F − 1
YY ⎜ 36 + ω 2 ⎟
⎝ ⎠
We know F −1 ⎛ 2α ⎞ = e−α τ
⎝⎜ α 2 + ω 2 ⎟ ⎠ ⎞
∴ R (τ ) = 3
F −1 ⎛ 2.6 [Here α = 6]
YY ⎜ 62 + ω 2 ⎟
2.6 ⎝ ⎠
15
1 −6 τ
= e
4
∞
(ii) SYY (ω ) = ∫ R (τ ) e
YY
−iωτ
dτ
−∞
∞ 1 −6 τ −iτω
= ∫ 4 e e dτ
−∞
∞
1⎧ 0 ⎫
= ⎨ ∫ e( 6τ −iω )τ dτ + ∫ e−( 6+iω )τ dτ ⎬
4 ⎩−∞ 0 ⎭
+ ⎢⎡ e−( 6+iωτ ) ⎥ ⎤ ∞ ⎫⎬⎪
0
= 1 ⎧⎨⎢
⎪ ⎡ e( ) ⎤⎥
6−iω τ
4 6 − iω − ( 6 + iω )
⎩⎪ ⎣ ⎦ −∞ ⎣
1
⎦ 0 ⎪⎭
⎫
1 ⎧ 1 (1− 0) −
= ⎨ ( 0 −1)
⎬
4 ⎩ 6 − iω 6 + iω ⎭
1 ⎧⎨ 1 1 ⎫ ⎬
= +
4 ⎩ 6 − iω 6 + iω ⎭
1 ⎧⎨⎪ 6 + iω + 6 − iω ⎫⎪⎬
= 4 ( 6 − iω )( 6 + iω )
⎩⎪ ⎪⎭
1 ⎧ 12 ⎫ 3
= ⎨ 2⎬ = 2 .
4 ⎩ 36 + ω ⎭ 36 + ω
Problem 25. Show that the power spectrum SYY (ω ) of the output of a linear system with
system function H (ω ) is given by S YY (ω ) = S (ω ) H (ω ) where SXX (ω ) is the

2
XX
power spectrum of the input.

Solution:
∞
If {X (t )} is a WSS and if y (t ) = ∫ h ( α )X (t − α ) dα
−∞
We shall prove that S YY (ω ) = S (ω ) H (ω )

2
XY .
∞
Consider Y (t ) = ∫ X (t − α ) h (α ) dα
−∞
∞
X ( t + τ ) Y ( t ) = ∫ X ( t + τ ) X ( t − α ) h (α ) d α
−∞
∞
E ⎡⎣ X ( t + τ ) Y ( t ) ⎤⎦ = ∫ E ⎣⎡ X ( t + τ ) X ( t − α ) ⎤⎦ h ( α ) dα
−∞
∞
RYX ( −τ ) = ∫ RXX (τ + α ) h (α ) dα
−∞
∞
RXY (τ ) = ∫ RXX (τ − β ) h ( − β ) d β
−∞
16
R XY (τ ) = R XX (τ ) *h ( −τ )
∞
Y ( t ) Y ( t − τ ) = ∫ X ( t − α ) Y ( t − τ ) h ( α ) dα
−∞
∞
∴E ⎡⎣ Y ( t ) Y ( t − τ ) ⎤⎦ = ∫ RXY (τ − α )h ( α ) d α
−∞
Assuming that { X (t )} & {Y (t )} are jointly WSS

RYY (τ ) = RXY (τ ) * h (τ ) ------------ (2)
Taking Fourier transform of (1) we get
SXY (ω ) = SXX (ω ) H * (ω ) -------- (3)
Where H *( ω ) is the conjugate of H (ω )
Taking Fourier transform of (2) we get
SYY (ω ) = SXY (ω ) H (ω ) ------------ (4)
Inserting (3) in (4)
SYY (ω ) = SXX (ω ) H * (ω ) H (ω )
SYY (ω ) = SXX (ω ) H (ω ) 2
Problem 26. A system has an impulse response h ( t ) = e−β tU ( t ) , find the power spectral
density of the output Y (t ) corresponding to the input X (t ) .
Solution:
Given h ( t ) = e −β t , t ≥ 0
∞
−∞
∞
H (ω ) = ∫ e −β t e − iωt dt
0
∞
= ∫ e−t ( β +i ω ) dt
0
∞
⎡ e −t ( β +iω ) ⎤
=⎢ ⎥
⎣ − ( β + iω ) ⎦ 0
1
H (ω ) =
β + iω
1
H * (ω ) =
β − iω
H (ω ) = H (ω ) H * (ω )
2
⎛ 1 ⎞⎛ 1 ⎞
=
⎜⎝ ⎟⎜ ⎟
β + iω ⎠⎝β − iω ⎠
17
1
=
β +α2
2
(ω ) = H (ω ) (ω )
2
∴S YY S XX
SXX (ω )
SYY (ω ) = 2 2
β +α
Problem 27. If X (t ) is the input voltage to a circuit and Y (t ) is the output voltage,
{ X (t )} is a stationary random process with μx = 0 and RXX (τ ) = e−2 τ . Find μ y , SXX (ω )
1
and SYY (ω ) , if the system function is given by H (ω ) = 2 2 .
ω +2
Solution:
Given Mean ⎡ ⎣ X (t )⎤ ⎦ = μX = 0
∞
Y (t ) = ∫ h (α ) X (t − α ) dα
−∞
∞
E ⎡⎣ Y ( t ) ⎤⎦ = ∫ h (α ) E ⎡⎣ X ( t − α ) ⎤⎦ dα = 0
−∞
∞
SXX (ω ) = RXX (τ ) e −i ωτ dτ
∫
−∞
∞ ∞ ∞
= ∫ e−2 τ e −iωτ dτ = ∫ cosωτ dτ

e −2 τ = 2∫ e −2τ cosωτ dτ
−∞ −∞ 0
⎡ 2 ⎤ 4
S
XX
(ω ) = 2 =
⎢⎣ ω 2 + 4 ⎥⎦ ω2 + 4
H (ω ) = 1
ω +4 2
2
S (ω ) = H ( ω ) S
(ω ) = ⎛⎜ 1 ⎞ 4
⎟ 2
⎝ ω + 4 ⎠ω + 4
YY XX 2
4
SYY (ω ) =
(ω )
2
2
+4
Problem 28. X (t ) is the input voltage to a circuit (system) and Y (t ) is the output
voltage. {X (t )} is a stationary random process with μx = 0 and RXX (τ ) = e−α τ . Find
R
μ y , Syy (ω ) & Ryy (τ ) if the power transfer function is H (ω ) =
R + iLω
Solution:
∞
Y (t ) = ∫ h (α ) X (t − α ) dα
−∞
18
E ⎡⎣ Y ( t ) ⎤⎦ = ∫ h (α ) E ⎡⎣ X ( t − α ) d α ⎤⎦
−∞
E ⎡⎣ Y ( t ) ⎤⎦ = 0
∞
SXX (ω ) = ∫R XX (τ ) e−iωτ dτ
−∞
0 ∞
∫ e e dτ + ∫ e 0 e dτ
ατ −iωτ −ατ −iωτ
= −∞
= ⎢⎡ e(α −iω )τ ⎥⎤ 0 + ⎢⎡ e−(α +iω )τ ⎤ ⎥∞ = + =
α − iω − ( α + iω ) α − i1 +1iω α 2 +2ωα 2
⎣ ⎦ −∞ ⎣ ⎦0 ω α
SYY (ω ) = SXX (ω ) H (ω )
2
2α R
=
α + ω R + L 2ω 2
22 2
Consider,
2α R2 A B
= 2 2 + 2 2 2
(α + ω )( R + L ω ) α + ω R + L ω
2 2 2 2 2
By partial fractions
⎛ ⎞
⎜ R2 ⎟
⎛ R2 2α ⎜ 2 ⎟
2α ⎜ 2 2 2 ⎟ ⎞ ⎜ α − R2 ⎟
R −Lα 2
SYY (ω ) = ⎝ ⎠+ ⎝ L ⎠
α2 + ω
2 2
R + L 2ω
2 2 2
⎛R⎞ ⎛ R2 ⎞
2α ⎜ ⎟ 2α ⎜ 2 ⎟
⎝ L ⎠ 1
. α 2 + ω2 + ⎝L ⎠2 . 1
= ⎛ R ⎞2 ⎛ R ⎞ ⎛ R ⎞
2
α 2 − ⎜ ⎟ ⎜ ⎟ + ω2
⎜ ⎟ −α
2
⎝L⎠ ⎝L ⎠ ⎝L ⎠
1 1
= λ. 2 2 + μ.
α +ω ⎛R⎞2
⎜ ⎟ +ω
2
⎝L⎠
λ ∞ eiωτ dω + μ ∞∫ eiωτ dω
RYY (τ ) =
2π
∫ α 2
+ ω 2
2 π ⎛ R ⎞
2
−∞ −∞
⎜ ⎟ +ω
2
⎝ ⎠L
By contour integration technique we know that
eiaz dz = e ab , a > 0
2 2
z +b b
19
⎛ R ⎞2 ⎛ R ⎞2
⎜L⎟α −⎛ ⎞ τ
R
⎜L⎟ L
∴ RYY (τ ) = ⎛ R⎝⎞2 ⎠ −α τ
e +
⎝ ⎠R 2 ⎜ ⎟
⎛ ⎞
e⎝ ⎠
⎜ ⎟ − α 2
α 2
− ⎜ ⎟
⎝L⎠ ⎝L⎠ .
Problem 29. X (t )is the i/P voltage to a circuit and Y (t ) is the O/P voltage. X (t ) is a
stationary random process with zero mean and autocorrelation RXX (τ ) = e−2 τ . Find the
1
mean of Y (t ) and its PSD if the system function H (ω ) = .
Jω + 2
Solution:
1
H (ω ) =
Jω + 2
1
⇒ H ( 0) =
2
E ⎡⎣ Y ( t ) ⎤⎦ = E ⎡⎣ X ( t ) ⎤⎦ .H ( 0 ) = 0
1
H (ω ) =
2
ω +4 2
4
S ( ω ) = F ⎡ R (τ ) ⎤ = F ⎡ e−2 τ ⎤ =
XX ⎣ XX ⎦
⎣ ⎦ ω2 + 4
4
SYY ( ω ) = H (ω ) S XX (ω ) =
2
( )
2
ω2 + 4
0 2
( ) ∫ ( 2 + τ ) ( 9 + 2eτ ) d τ + ∫ ( 2 − τ ) (9 + 2e−τ ) d τ
∴E Y 2 =
−2 0 0 2
⎡ 9τ 2 ⎤ ⎡ 9τ 2 ⎤
= ⎢ 18τ + 4e +
τ + 2 e (τ −1) ⎥
τ
+ ⎢ 18τ − 4eτ − 2 e −τ (τ +1) ⎥
⎣ 2 ⎦ −2 ⎣ 2 ⎦0
∴E (Y2 ) = 40.542
V ar ( Y ) = E (Y 2 ) − ⎡⎣ E ( Y ) ⎤⎦
2
= 40.542 − 36 = 4.542
1
Problem 30. Consider a system with transfer function . An input signal with
1+ iω
autocorrelation function mδ (τ ) + m 2 is fed as input to the system. Find the mean and
mean-square value of the output.
Solution:
1
Given, H (ω ) = and RXX (τ ) = m δ (τ ) + m 2
1+ iω
SXX ( ω ) = m + 2π m 2δ ( ω )
We know that, SYY (ω ) = H (ω ) S (ω )
2
XX
20
2
1
= ⎡ m + 2π m 2δ ( ω ) ⎤⎦
1+1iω ⎣
= ⎡ m + 2π m 2δ ( ω ) ⎤
1+ ω 2 ⎣ ⎦
RYY (τ ) is the Fourier inverse transform of SYY (ω ).
m −τ
So, RYY (τ ) = e + m2
2
lim
RXX (τ ) = X
2
We know that
τ →∞
2
So X = M 2
X =m
Also H ( 0 ) = 1
We know that Y = 1, m = m
m
Mean-square value of the output = Y = RYY ( 0 ) = + m2
2
2
Problem 31. If the input to a time-invariant, stable, linear system is a WSS process,
prove that the output will also be a WSS process.
Solution:
Let X (t ) be a WSS process for a linear time variant stable system with Y (t ) as the
output process.
∞
Then Y (t ) =
−∞
∫ h (u ) X (t − u ) du where h(t) is weighting function or unit impulse
response.
∞
∴ E ⎡⎣ Y ( t ) ⎤⎦ = ∫ E ⎡⎣ h ( u ) X ( t − u ) ⎤⎦ du
−∞
∞
= ∫ h (u ) E ⎣ ⎡ X (t − u )⎤ ⎦ du
−∞
Since X (t ) is a WSS process, E ⎡ ⎣ X (t )⎦ ⎤ is a constant for any t.

μX
∴ E ⎡ ⎣ X (t − u )⎤ ⎦ = μX
∞
∴ E ⎡⎣ Y ( t ) ⎤⎦ = ∫ h ( u ) μ X du
−∞
∞
= μX
∫ h (u ) du
−∞
∞
Since the system is stable ,

∫ h (u ) du is finite
−∞
∴ E ⎡⎣ Y ( t ) ⎤⎦ is a constant.
21
Now RYY ( t, t + τ ) = E ⎡⎣ Y ( t ) Y ( t + τ ) ⎤⎦
⎡ ∞ ∞
⎤
= E ⎢ ∫ h ( u1 ) X ( t − u1 ) du1 ∫ h ( u2 ) X ( t + τ − u2 ) du2 ⎥
⎣ −∞ −∞ ⎦
⎡ ∞ ∞ ⎤
= E ⎢ ∫ ∫ h ( u1 ) h ( u2 ) X ( t − u1 ) X ( t + τ − u2 ) du1u2 ⎥
⎣ −∞ −∞ ⎦
∞∞
= ∫ ∫ h ( u1 ) h ( u2 ) E ⎡⎣ X ( t − u1 ) X ( t + τ − u2 )⎤⎦ du1u2
−∞ −∞
Since X (t ) is a WSS process, auto correlation function is only a function time difference.
∞∞
∴ RYY ( t, t + τ ) = ∫ ∫ h ( u1 ) h ( u2 )RXX (τ + u1 − u2 ) du1 du2

−∞ −∞
When this double integral is evaluated by integrating w.r. to u1 , u2 , the R.H.S is only a
function of τ .
∴ RYY ( t, t + τ ) is only a function of time difference τ .
Hence Y (t ) is a WSS process.
∞
Problem 32. Let X (t ) be a WSS and if Y (t ) =

∫ h (u ) X (t − u ) du
−∞
then show that
a) RXY (τ ) = h (τ ) * RXX (τ )
b) RYX (τ ) = h ( −τ ) * RXX (τ )
c) Ryy (τ ) = h (τ ) * Rxy (τ )
Where * denotes the convolution and H * (ω ) is the complex conjugate of H (ω ) .
Solution:
Given X (t ) is WSS∴E ⎣ ⎡ X (t )⎤ ⎦ is constant and
RXX ( t, t + τ ) = RXX (τ )
∞
Y (t ) = ∫ h (u ) X (t − u ) du
−∞
Now RXY ( t, t + τ ) = E ⎡⎣ X ( t ) Y ( t + τ ) ⎤⎦
⎡ ∞ ⎤
= E ⎢ X ( t ) ∫ h ( u ) .X ( t + τ − u ) du ⎥
⎣∞ −∞ ⎦
⎡ ⎤ ∞
= E ⎢ ∫ h ( u ) X ( t ) X ( t + τ − u ) du ⎥ = ∫ h ( u ) .E ⎡⎣ X ( t ) X ( t + τ − u ) ⎤⎦ du
⎣ −∞ ⎦ −∞
Since X (t ) is a WSS Process,
E ⎡⎣ X ( t ) X ( t + τ − u ) ⎤⎦ = RXX (τ − u )
∞
∴ RXY ( t, t + τ ) = ∫ h ( u ) RXX (τ − u ) du
−∞
22
⇒ RXY (τ ) = RXX (τ )* h (τ )
(b). Now RYX (τ ) = RXY ( −τ )
= RXY ( −τ ) * h ( −τ ) [ from (i) ]
= RXX (τ ) * h ( −τ ) [Since RXX (τ ) is an even function of τ ]
(c). RYY ( t, t − τ ) = E ⎣⎡ Y ( t ) Y ( t − τ ) ⎤⎦
⎡∞ ⎤
= E ⎢ ∫ h ( u ) X ( t − u ) du Y ( t − τ ) ⎥
⎣ −∞∞ ⎦
⎡ ⎤
= E ⎢ ∫ X ( t − u ) Y ( t − τ ) h ( u ) du ⎥
⎣ −∞ ⎦
∞ ∞
= ∫ E ⎡⎣ X ( t − u ) Y ( t − τ ) ⎤⎦ h ( u ) du = ∫ RXY (τ − u ) h ( u ) du
−∞ −∞
It is a function of τ only and it is true for any τ .

∴ RYY (τ ) = RXY (τ ) * h (τ )
Problem 33. Prove that the mean of the output of a linear system is given by
μY = H (0) μX , where X (t ) is WSS.
Solution:
We know that the input X (t ) , output Y (t ) relationship of a linear system can expressed
as a convolution Y (t ) = h (t ) * X (t )
∞
= ∫ h (u )( t − u ) du
−∞
Where h (t ) is the unit impulse response of the system.

∴the mean of the output is
⎡ ∞ ⎤ ∞
E ⎡⎣ Y ( t ) ⎤⎦ = E ⎢ ∫ h ( u ) X ( t − u ) du ⎥ = ∫ h ( u ) E ⎡⎣ X ( t − u ) ⎤⎦ du
⎣ −∞ ⎦ −∞
Since X(t) is WSS, E ⎡ ⎣ X (t )⎦ ⎤ = μX is a constant for any t.
E ⎡ ⎣ X (t − u )⎤ ⎦ = μX
∞ ∞
∴ E ⎣⎡ Y ( t ) ⎤⎦ = ∫ h ( u ) μ X du = μ X ∫ h ( u ) du
−∞ −∞
We know H (ω ) is the Fourier transform of h (t ) .

∞
i.e. H (ω ) =
∫ h (t ) dt
−∞
∞ ∞
put ω = 0 ∴H (0) = ∫ h (t ) dt = ∫ h (u ) du
−∞ −∞
∴E ⎡⎣ Y ( t ) ⎤⎦ = μ X H ( 0 ) .
23
UNIT –I RANDOM VARIABLES
PART-A
Problem1. X and Y are independent random variables with variance 2 and 3. Find the variance
of 3X + 4Y .
Solution:
V (3X + 4Y ) = 9Var( X ) +16Var(Y ) + 24Cov( XY )
= 9× 2 +16 × 3 + 0 (∴ X & Y are independent cov( XY ) = 0 )
= 18 + 48 = 66.
Problem 2. A Continuous random variable X has a probability density function F ( x) = 3x2;
0 ≤ x ≤ 1. Find ‘a’ such that P ( x ≤ a ) = P ( x > a )
Solution:
We know that the total probability =1
Given P ( X ≤ a ) = P ( X > a ) = K (say)
Then K + K = 1
1
K=
2
1 1
ie P ( X ≤ a ) = & P ( X > a ) =
2 2
1
Consider P ( X ≤ a) =
2
a
1
i.e. f ( x) dx =
∫0 2
a 1
∫0 3x dx = 2
2
a
⎛ x3 ⎞ 1
3⎜ ⎟ =
⎝ 3 ⎠0 2
1
a3 =
2
1/ 3
⎛1⎞
a=⎜ ⎟ .
⎝2⎠
⎧Cxe− x ; if x > 0
Problem 3. A random variable X has the p.d.f f ( x) given by f ( x) = ⎨
⎩0 ; if x ≤ 0
Find the value of C and cumulative density function of X .
1
Solution:
∞
Since
∫ f ( x) dx = 1
−∞
∞
∫ Cxe−xdx = 1
0
( )( )
∞
C ⎡⎣ x −e − x − e− x ⎤ ⎦= 1
0
C =1
⎧xe − x ; x > 0
∴ f x( =) ⎨
⎩ 0 ;x ≤ 0
−t x
x x
⎡ −t −t ⎤ −x −x
C.D.F F ( x ) = ∫ f ( x )dt = ∫ te dt = ⎣ −te – e ⎦ 0 = −xe − e +1

0 0
= 1− (1+ x) e −x .
⎧1
⎪ ( x +1); −1 < x < 1
Problem 4. If a random variable X has the p.d.f f ( x ) = ⎨2 .
⎪
⎩ 0 ; otherwise
Find the mean and variance of X .
Solution:
1
11 11 2
Mean= ∫ xf ( x) dx = 2 ∫ (x x +1 ) dx =
2 ∫ x + x dx ( )
−1 −1 −1
1
2
1 ⎜⎛ x3 x ⎞ 1
= 2 3 + ⎟ =
⎝ 2 ⎠−1 3
2 1
(
μ ′ = 1 x f ( x)dx = 1 x + x dx = 1 ⎡ x
3 2 4
+ x 3
⎤⎥
1
)
⎢
2 ∫−1 ∫
2 −1 2⎣4 3 ⎦ −1
1 ⎡ 1 1 1 1⎤
= ⎢ + − +
2 ⎣ 4 3 4 3⎦
1 2 1
= .⎥ =
2 3 3
()
2
Variance = μ2′ − μ ′ 1
1 1 3 −1 2
= − = = .
3 9 9 9
⎧2e−2 x ; x ≥ 0
Problem 5. A random variable X has density function given by f ( x) = ⎨ .
⎩0 ; x < 0
Find m.g.f
Solution:
2
∞ ∞
( )
M X (t ) = E etx = etx f ( x ) dx = e 2e dx
∫ ∫
tx −2 x
0 0
∞
= 2∫ e( t −2 ) x dx
0
∞
⎡ (t −2)x ⎤ 2
= 2⎢ e ⎥ = ,t < 2 .
t −2 2−t
⎣ ⎦0
Problem 6. Criticise the following statement: “The mean of a Poisson distribution is 5 while the
standard deviation is 4”.
Solution: For a Poisson distribution mean and variance are same. Hence this statement is
not true.
Problem 7. Comment the following: “The mean of a binomial distribution is 3 and variance is 4
Solution:
In binomial distribution, mean >variance but Variance < Mean
Since Variance = 4 &Mean = 3, the given statement is wrong. ⎛ 1⎞ ⎛ 1⎞
Problem8. If X and Y are independent binomial variates B 5, and B 7,
⎜ 2⎟ ⎜ 2⎟
⎝ ⎠ ⎝ ⎠
find P[ X + Y = 3]
Solution:
1
X + Y is also a binomial variate with parameters n1 + n2 = 12 & p =
2
3 9
⎛1⎞ ⎛1⎞
∴ P [ X + Y = 3] = 12C3 ⎜ ⎟ ⎜ ⎟
⎝2 ⎠ ⎝ 2 ⎠
55
= 10
2
4
Problem 9. If X is uniformly distributed with Mean1and Variance , find P[ X > 0]
3
Solution:
If X is uniformly distributed over (a, b ) , then
E ( X ) = b + a and V ( X ) = ( b − a )
2
2 12
b+a
∴ =1⇒a+b=2
2
(b − a) 4
2
= ⇒ ( b − a ) = 16
2
⇒
12 3
⇒ a + b = 2 & b − a = 4 We get b = 3, a = −1
∴a = −1& b = 3 and probability density function of x is
3
⎧1
⎪ ; −1 < x < 3
f ( x) = ⎨ 4
⎪⎩0 ;Otherwise
0
1 1 0 1
P [ x < 0] = ∫ dx = [ x ]−1 = .
−
4 4 4
1
Problem 10. State the memoryless property of geometric distribution.
Solution:
If X has a geometric distribution, then for any two positive integer ' m ' and ' n '
P ⎡ X > m + n X > m⎤ = P [ X > n] .
⎣ ⎦
Problem 11. X is a normal variate with mean = 30 and S.D = 5
Find the following P [26 ≤ X ≤ 40]
Solution:
X N (30, 52 )
∴ μ = 30 & σ = 5
X −μ
Let Z = be the standard normal variate
σ ⎡ 26 − 30 40 − 30 ⎤
P [26 ≤ X ≤ 40] = P ≤Z≤
⎣⎢ 5 5 ⎥⎦
= P [−0.8 ≤ Z ≤ 2] = P [−0.8 ≤ Z ≤ 0] + P [0 ≤ Z ≤ 2]
= P [0 ≤ Z 0.8] + [ 0 ≤ z ≤ 2]
= 0.2881 + 0.4772 = 0.7653 .
⎡ 3⎤
Problem 12. If X is a N ( 2, 3) Find P Y ≥ where Y +1 = X .
⎢⎣
2⎥ ⎦
Solution:
⎡ 3⎤ ⎡ 3⎤
P Y ≥ = P X −1 ≥
⎣ ⎢ 2⎥ ⎦ ⎢⎣ 2⎥ ⎦
= P[ X ≥ 2.5] = P [Z ≥ 0.17]
= 0.5 − P [0 ≤ Z ≤ 0.17]
= 0.5 − 0.0675 = 0.4325
1
Problem 13. If the probability is that a man will hit a target what is the chance that he
4
will hit the target for the first time in the 7th trial?
Solution:
The required probability is
P[FFFFFFS ] = P ( F ) P ( F ) P ( F ) P ( F ) P ( F ) P ( F ) P ( S )
6
⎛ 3 ⎞ . ⎛ 1 ⎞ = 0.0445 .
=q p=⎜ ⎟ ⎜ ⎟
6
⎝4⎠ ⎝4⎠
4
Hence p = Probability of hitting target and q = 1− p .
Problem 14. A random variable X has an exponential distribution defined by p.d.f.
f ( x ) = e−x , 0 < x < ∞ . Find the density function of Y = 3X + 5 .
Solution:
dy dx 1
y = 3x + 5 ⇒ =3⇒ =
dx dy 3
P.d.f of y h ( y ) = f ( x) dx
Y X
dy
1
h
Y
( y) = e−x.
3
y −5 ⎞
−⎛
y −5 1 ⎜ ⎟
Using x = we get hY ( y ) = e ⎝ 3 ⎠ , y > 5( x > 0 ⇒ y > 5)
3 3
Problem 15. If X is a normal variable with zero mean and variance σ , Find the p.d.f of y = e−x
2
Solution:
Given Y = e−x
1
f ( x) =
2
1 x
−
e 2σ
2
σ 2π
1 1
h ( y ) = f ( x) dx (log y )
2
–
1 × .
= 2σ 2
e
dy σ 2π
Y X
y
PART-B
Problem 16. A random variable X has the following probability function:
Values of X ,
X : 0 1 2 3 4 5 6 7
P( X ) : 0 K 2K 2K 3K K2 2K 2 7K 2 + K
Find (i) K , (ii) Evaluate P ( X < 6 ) , P ( X ≥ 6) and P (0 < X < 5)
(iii). Determine the distribution function of X .
(iv). P (1.5 < X < 4.5 X > 2)
(v). E (3x − 4) ,Var(3x − 4)
Solution(i):
7
Since ∑ P ( X ) = 1,
x=0
K + 2K + 2K + 3K + K 2 + 2K 2 + 7K 2 + K = 1
10K2+9K−1=0
1
K= or K = −1
10
1
As P ( X ) cannot be negative K =
10
Solution(ii):
5
P ( X < 6) = P ( X = 0) + P ( X = 1) +... + P ( X = 5)
1 81
= + 2 + 2 + 3 + 1 + ... =
10 10 10 10 100 100
Now P ( X ≥ 6) = 1− P ( X < 6)
81 19
= 1− =
100 100
Now P (0 < X < 5) = P ( X = 1) + P ( X = 2) + P ( X = 3) = P ( X = 4)
= K + 2K + 2K + 3K
8 4
= 8K = = .
10 5
Solution(iii):
The distribution of X is given by FX ( x) defined by
F X ( x) = P ( X ≤ x)
X : 0
1 2 3 4 5 6 7
1 3 5 4 81 83
FX (x) : 0 1
10 10 10 5 100 100
⎧x
; x = 1, 2, 3, 4, 5
Problem 17. (a) If P ( x) = ⎪
⎨15
⎪⎩0 ; elsewhere
Find (i) P { X = 1or2} and (ii) P{1/ 2 < X < 5 / 2 x > 1}
(b) X is a continuous random variable with pdf given by
⎧ Kx in 0 ≤ x ≤ 2
⎪2K in 2 ≤ x ≤ 4
⎪
F(X)= ⎨
⎪6K − Kx in 4 ≤ x ≤ 6
⎪⎩0 elsewhere
Find the value of K and also the cdf FX ( x) .
Solution:
(a) i) P ( X = 1 or 2) = P ( X = 1) + P ( X = 2)
1 2 3 1
= + = =
15 15 ⎧⎛151 5 5⎞ ⎫
P < X < ∩ ( X > 1)
⎨⎜2 ⎬
⎛1 5
⎞ ⎩⎝ 2 ⎟⎠ ⎭
ii) P ⎜ < X < / x > 1⎟ =
⎝2 2 ⎠ P ( X > 1)
P{( X = 1or2) ∩ ( X > 1)}
=
P ( X > 1)
6
P ( X = 2)
=
1− P ( X = 1)
1
2 /15 2 /15 2 = .
= = =
1− (1/15) 14 /15 14 7
∞
Since
∫ F ( x) dx = 1
∞
2 4 6
∫ Kxdx + ∫ 2Kdx + ∫ ( 6k − kx )dx = 1

0 2 4
⎡ ⎛ x 2 ⎞2 6
⎛ x2 ⎞ ⎤
6
⎟ + ( 2x ) + ∫ ⎜ 6x −
4
K ⎢⎜ ⎟ ⎥ =1
2
2
⎢ ⎣ ⎝ ⎠0 4⎝ 2 ⎠4 ⎥⎦
K ⎡⎣ 2 + 8 − 4 + 36 −18 − 24 + 8⎤⎦ = 1
1
8K = 1 K=
8
x
We know that FX ( x ) =
∫ f ( x) dx
−∞
x
If x < 0 , then FX ( x ) = ∫ f ( x) dx = 0
−∞
x
If x ∈(0, 2) , then FX ( x ) =
∫ f ( x) dx
−∞
0 x
F X ( x) = ∫ f ( x) dx + ∫ f ( x) dx
−∞ 0
0 x 0
1x
= ∫ 0dx + ∫ Kxdx = ∫ 0dx + ∫ xdx
−∞ 0 −∞
8 0
x
2 2
⎛x ⎞ x
=⎜ ⎟ = ,0≤x≤2
⎝ 16 ⎠0 16
x
If x ∈ ( 2, 4) , then FX ( x ) = ∫ f ( x) dx
−∞
0 2 x
F X ( x) = ∫ f ( x) dx + ∫ f ( x) dx + ∫ f ( x) dx
−∞ 0 2
0 2 x
= ∫ 0dx + ∫ Kxdx + ∫ 2Kdx

−∞ 0 2 2 x
2 x ⎛ x2 ⎞ ⎛x⎞
= ∫ dx + ∫ dx = ⎜ ⎟ + ⎜ 4 ⎟
x 1
8 4 16
0 2 ⎝ ⎠0 ⎝ ⎠2
7
1 x 1
= + −
4 4 2
x 4 x −1 , 2 ≤ x < 4
= − =
4 16 4
0 2 4 x
If x ∈ ( 4, 6) , then FX ( x) = ∫ 0dx + ∫ Kxdx + ∫ 2Kdx + ∫ k (6 − x ) dx

−∞ 0 2 4
2 4 x
x 1 1
8(
=∫ dx + ∫ 6 − x ) dx
dx + ∫
0
8 2
4 4
2 4 x
⎛ x 2 ⎞ ⎛ x ⎞ + ⎛ 6x − x 2 ⎟⎞
=⎜ ⎟ +⎜ ⎟
16 4 ⎜ 8 16
⎝ ⎠0 ⎝ ⎠2 ⎝ ⎠4
1 2
= + 1− 1 + 6x − x − 3 +1
4 2 8 16
4 +16 − 8 +12x − x2 − 48 +16
=
16
−x 2 +12x − 20
= ,4 ≤ x ≤ 6
16
0 2 4 6 ∞
If x > 6 , then FX ( x ) = ∫ 0dx + ∫ Kxdx + ∫ 2Kdx + ∫ k ( 6 − x ) dx + ∫ 0dx

−∞ 0 2 4 6
=1, x ≥ 6
⎧0 ;x≤0
⎪ 2
⎪x ;0 ≤ x ≤ 2
⎪16
( x) = ⎪⎨ ( x −1)
1
∴ FX ;2≤ x≤ 4
⎪4
⎪−1
( )
⎪16 20 −12x + x ; 4 ≤ x ≤ 6
2
⎪
⎩⎪ 1 ;x ≥ 6
Problem18. (a). A random variable X has density function
⎧ K
⎪ ,−∞ < x < ∞
f ( x) = ⎨1 + x2 . Determine K and the distribution functions. Evaluate the
⎩⎪ 0 , Otherwise
probability P ( x ≥ 0) .
⎧2x , 0 < x < 1
(b). A random variable X has the P.d.f f ( x ) = ⎨
⎛ 1⎞ ⎛1 1⎞ ⎩ 0 ,⎛Otherwise
3 1⎞
Find (i) P X < (ii) P <x< (iii) P X > / X >
⎜ 2⎟ ⎜4 2⎟ ⎜ 4 2⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠
Solution (a):
8
∞
Since
∫ F ( x)dx = 1
−∞
∞
K
∫ 1+ x dx = 1 2
−∞
∞ dx
K∫ =1
1+ x 2
∞
∞
(
K tan−1 x )−∞
=1
⎛ π ⎛ π ⎞⎞
K − − =1
⎜ 2 ⎜ 2 ⎟⎟
⎝ ⎝ ⎠⎠
Kπ = 1
1
K=
π
x x
K
FX ( x ) = ∫ f ( x)dx = ∫ 1+ x dx 2
−∞ −∞
1
( tan −1 x )
x
=
1π ⎡ ⎛ π ⎞⎤
−∞
= tan −1 x − −
π1⎢⎣ ⎡ π ⎜ 2 ⎟⎥
⎝ ⎤ ⎠⎦
= + tan −1 x , −∞ < x < ∞
π ⎢⎣ 2 ∞ ⎥⎦
1 ∞ dx 1 −
P ( X ≥ 0) = ∫ =
π 01+ x2 π
(tan 1 x) 0
1 ⎛π ⎞ 1
= − tan−1 0 = .
π ⎜2 ⎟
2
Solution (b): ⎝ ⎠
⎛ 1/ 2 1/ 2 2 1/ 2 2 ×1 1
(i) P ⎜ x < ⎟⎞ = ∫ f ( x ) dx = ∫ 2xdx = 2 ⎜ ⎛ ⎟x ⎞ =
1 =
2 2 8 4
⎝ ⎠ 0 0 ⎝ ⎠0
⎛1 1 ⎞ 1/ 2 1/ 2 2 1/ 2
(ii) P ⎜ < x < ⎟ = ∫ f ( x ) dx = ∫ 2xdx = 2 ⎜ ⎛ x⎟ ⎞
4 2 2
⎝ ⎠ 1/ 4 1/ 4 ⎝ ⎠1/ 4
⎛1 1
= 2 − ⎞ ⎛1 1 ⎞ 3
= − = .
⎜ ⎟ ⎜ ⎟
⎝ 8 32 ⎠ 3⎝ 4 161 ⎠ 16
⎛ ⎞ ⎛ 3⎞
P⎜X > ∩X > P X>
⎛ 3 1 ⎞ 4 2⎟ ⎜ 4⎟
(iii) P ⎜ X > / X > ⎝ ⎠ ⎝ 1 ⎞⎠
2 ⎠⎟ = = P⎛X >
⎛ 1 ⎞
⎝ 4 P X>
⎜ 2 ⎟ ⎜ 2⎟
⎝ ⎠ ⎝ ⎠
9
1
⎛ 3⎞ 1 ⎛ x2 ⎞ 1 9 7
4 ∫ f ( x ) dx = ∫ 2xdx = 2 ⎜ 2 ⎟
P⎜ X > ⎟ = = 1− =
⎝ ⎠3/ 4 3/ 4 ⎝ 2 ⎠13/ 4 16 16
⎛ 1 ⎞ 1 1 ⎛x ⎞ 1 3
P⎜ X > ⎟ =
2 ∫ () ∫f x dx = 2xdx = 2 ⎜ ⎟
2
= 1 −
4
=
4
⎝ ⎠ 1/ 2 1/ 2 ⎝ ⎠1/ 2
⎛ 3 1⎞ 7 7 4 7
P ⎜ X > / X > ⎟ = 163 = 16× 3= 12 .
4 2
⎝ ⎠
4
⎧Ke −3x , x > 0
Problem 19.(a).If X has the probability density function f ( x) = ⎨
⎩0 , otherwise
find K , P[0.5 ≤ X ≤ 1]and the mean of X .
(b).Find the moment generating function for the distribution whose p.d.f is
f ( x ) = λ e−λ x , x > 0 and hence find its mean and variance.
Solution:
∞
Since
∫ f ( x) dx = 1
−∞
∞
∫ Ke−3xdx = 1
0
∞
⎡ e−3x ⎤
K⎢ ⎥ =1
⎣ − ⎦0
K 3
=1
3
K =3
e−3 − e−1.5 ⎡ −1.5 −3 ⎤
1 1
−3x
P (0.5 ≤ X ≤ 1) = ∫ f ( x ) dx = 3 ∫ e dx = 3 −3 =⎣e –e ⎦
0.5 0.5
∞ ∞
Mean of X = E ( x ) = xf ( x ) dx = 3 xe −3x dx
∫ 0
∫ 0
∞
⎡ ⎛ −e−3x ⎞ ⎛ e−3x ⎞⎤ 3×1 1
= 3⎢ x⎜ −1
⎟ ⎜ ⎟⎥ = =
⎣ ⎝ 3 ⎠ ⎝ 9 ⎠⎦ 0 9 3
1
Hence the mean of X = E ( X ) =
∞
3 ∞
M X (t ) = E etx =( ) ∫ etx f ( x) dx = ∫ λ e − λ x tx
e dx
0 0
∞
= λ ∫ e− x(λ −t )dx
0
10
∞
⎡ e− x(λ −t ) ⎤ λ
=λ⎢ ⎥ =
⎣ −(λ − t) ⎦ 0 λ − t
⎡ d ⎤
Mean = μ ′ = M (t ) ⎡⎢ λ 2⎥⎤ 1
1 ⎢ dt X = (λ − t) =λ
⎥⎣
⎦
t =0 ⎢⎣ ⎥⎦ t =0
⎡ d2 ⎤
μ′= ( ) = ⎡⎢ λ ( 2) 3 ⎤⎥ = 22
2 ⎢ dt 2 X t ⎥
M (λ − t) λ
⎣ ⎦ t =0 ⎢⎣ ⎦⎥ t =0
()
2
2 1 1
Variance = μ2′ − μ ′ = − = .
λ λ λ2
1 2 2
Problem 20. (a). If the continuous random variable X has ray Leigh density
F ( x ) = ⎜⎛ x2 e − 2α ⎟×U ( x ) find E ( x )nand deduce the values of E X and Var( X ) .
2
x ⎞
⎜α
2
⎟
( )
⎝ ⎠
⎧1 − x
⎪ e 2 ,x > 0
(b). Let the random variable X have the p.d.f f ( x) = ⎨2 .
⎪⎩0 , otherwise.
Find the moment generating function, mean & variance of X .
Solution:
⎧1 if x > 0
(a) Here U ( x ) = ⎨
⎩0 if x ≤ 0
∞
( )
E x n = ∫ xn f (x)dx
0
∞ − x2
= xn x2 e2α dx
∫0 α
2
x2
Put =t, x = 0, t = 0
2α 2
x
dx = dt x = α,t = ∞
α2
( )
∞
= ∫ 2α 2 t
n/2
e −t dt ⎡ x= 2α . t ⎤
⎣ ⎦
0
∞
= 2n / 2 α n ∫0t n / 2 e−t dt
E xn ( ) ⎛n ⎞
= 2n/ 2 α n Γ +1 − (1)
⎜2 ⎟
⎝ ⎠
Putting n =
1/ 21 in⎛(1)
E ( x ) = 2 α Γ ⎜ ⎟=
3 ⎞ we get ⎛ 1 ⎞
2α Γ ⎜ +1 ⎟
2 2
⎝ ⎠ ⎝ ⎠
11
1 ⎛1⎞
= 2α Γ ⎜ ⎟
2
2 ⎝ ⎠
α ⎛1⎞
π [ Γ = π
= 2 ⎜2 ⎟
⎝ ⎠
π
∴ E ( x) = α
2
Putting n = 2 in (1), we get
( )
E x 2 = 2α 2 Γ ( 2 ) = 2α 2 [ Γ ( 2 ) = 1]
( )
∴Var ( X ) = E X 2 − ⎡⎣ E ( X ) ⎤⎦
2
π
= 2α 2 − α 2
2
⎛ π ⎞ 2 ⎛ 4 −π ⎞ 2 .
= 2− α = α
⎜ 2 ⎟ ⎜ 2 ⎟
⎝ ∞ ⎠ ⎝ ∞ ⎠
( )
(b) M X (t ) = E etx = ∫ etx f ( x) dx = ∫ etx 12 e− x / 2dx
−∞ 0
∞
∞ −⎛ t 1 ⎞ x
⎡⎢ ⎝ ⎜⎛ t − 21 ⎟⎞⎠x ⎥ ⎤
1 ⎜ ⎟ 1 e 1 1
= ∫e ⎝
2
dx = ⎢
⎠ ⎥ = , if t < .
20 2 ⎢ ⎛ t − 1 ⎞⎥ 1− 2t 2
⎢ ⎜ 2 ⎟⎥
⎣⎝ ⎠⎦ 0
⎡ d ⎤ ⎡ 2 ⎤
E ( X ) = dt M X (t ) = =2
⎣ ⎦ ⎢ 2⎥
(1− 2t ) ⎥⎦ t =0
2 ⎡ d2 t =0 ⎣⎢ 8 ⎤
⎤ ⎡
( )
E X = ⎢ 2 M X (t )⎥ =⎢ 3⎥ =8
⎣ dt ⎦ t =0 ⎢ ⎣ (1− 2t ⎥⎦ t =0
)
( )
Var ( X ) = E X 2 − E ( X ) ⎤⎦ = 8 − 4 = 4 .
2
Problem 21. (a).The elementary probability law of a continues random variable is

f ( x ) = y e−b(x−a) , a ≤ x ≤ ∞, b > 0 where a, b and y are constants. Find y the rth moment
0 0 0
about point x = a and also find the mean and variance.

(b).The first four moments of a distribution about x = 4 are 1,4,10 and 45 respectively.
Show that the mean is 5, variance is 3, μ 3 = 0 and μ 4 = 26 .
Solution:
Since the total probability is unity,
∞
−∞
∫ f ( x) dx = 1
∞
−b( x−a)
y0 ∫ e dx = 1
0
12
∞
⎡ e−b( x−a) ⎤
y0 ⎢ ⎥ =1
⎣ −b 1 ⎦ 0
⎛ ⎞
y ⎜ ⎟= 1
0 ⎝
b⎠
y0 = b.
∞
μ ′ r( rth moment about the point x = a ) = ∫ ( x − a ) f ( x ) dx

r
−∞
∞
= b ∫ ( x − a ) e − b ( x − a ) dx
r
Put x − a = t , dx = dt , when x = a, t = 0 , x = ∞, t = ∞
∞
= b ∫ t r e − bt dt
0
Γ ( r +1) r!
=b (r +1)
=
b br
In particular r = 1
1
μ′ =
1
b
2
μ 2′ =
b2
1
Mean = a + μ ′ = a +
1
b
()
2
Variance = μ2′ − μ ′ 1
2 1 = 1 .
= −
b2 b2 b2
b) Given μ ′ =1 1, μ ′ =2 4, μ ′ = 10,
3
μ ′ = 45
4
μ ′r = rth moment about to value x=4

Here A = 4
Here Mean = A + μ ′1= 4 +1 = 5
()
2
Variance = μ2 = μ2′ − μ ′ 1
= 4 −1 = 3 .
()
3
μ 3 = μ ′ 3− 3μ ′μ ′2+ 21 μ ′ 1
= 10 − 3 ( 4 )(1) + 2 (1) = 0
3
( ) − 3( μ ′ )
2 4
μ 4 = μ ′ 4− 4μ ′ μ ′3+ 61 μ ′ μ ′ 2 1 1
13
= 45 − 4 (10 )(1) + 6 ( 4 )(1) − 3 (1)
2 4
μ4 = 26 .
Problem 22. (a). A continuous random variable X has the p.d.f
f ( x ) = kx2e−x , x ≥ 0. Find the rth moment of X about the origin. Hence find mean and
variance of X.
(b). Find the moment generating function of the random variable X, with probability
density function f ( x ) = 2 − x for 10≤≤xx<<21.Also find μ ′ , μ ′ .
⎪⎧ x for
⎨ 1 2
⎪0 otherwise
⎩
Solution:
∞
Since
∫ Kx2e−xdx = 1
0
∞
⎡ ⎛ e− x ⎞ ⎛ e− x ⎞ ⎛ e−x ⎞⎤
⎟ − 2x ⎜ ⎟+ 2⎜ ⎟⎥ = 1
2
K⎢x ⎜
⎣ ⎝ ⎠−1 ⎝ 1 ⎠ ⎝ −1 ⎠⎦ 0
2K = 1
1
K= .
2
∞
μr′ = ∫ x f ( x)dx
r
1∞
= x e dx
2 ∫0 r +2 − x
∞
1 −x (r +3)−1 ( r + 2)!
=
2 ∫0
e x dx =
2
3!
Putting n = 1 , μ ′ = = 3
1
2
′ 41
n = 2,μ = = 12
2
2
∴ Mean = μ ′ = 1
3
()
2
Variable= μ 2′ − μ ′ 1
i.e. μ 2 = 12 − ( 3) = 12 − 9
2
∴ μ2 = 3.
∞
(b) M X (t ) =
∫ etx f ( x)dx
−∞
1 2
= ∫ e xdx + ∫ etx ( 2 − x ) dx
tx
0 1
14
1
⎛ xetx etx ⎞ ⎡ etx ⎤
2
etx
= − 2 ⎟ + ⎢ ( 2 − x ) − (−1) 2 ⎥
⎜ t t t t
⎝ ⎠0 ⎣ ⎦1
t t 2t t t
e e 1 e e e
= − 2 + 2 + 2 − − 2
t t 2 t t t t
⎛ et −1 ⎞
=⎜ ⎟
⎝ t ⎠ 2
⎡ t t2 t3 ⎤
= ⎢ 1+ + + + ... −1⎥
⎣ 1! 2! 3! 2
⎦
⎡ t t2 t3 ⎤
= ⎢ 1+ + + + ...⎥
⎣ 2! 3! 4! ⎦
t
′
1μ = coeff . of
=1
1!
t2 7
μ2 ′ = coeff . of 2! = 6
.
x−θ
1 −
Problem 23. (a).The p.d.f of the r.v. X follows the probability law: f ( x) = e θ ,
2θ
−∞ < x < ∞ . Find the m.g.f of X and also find E ( X ) and V ( X ) .
(b).Find the moment generating function and rth moments for the distribution. Whose
p.d.f is f ( x) = Ke−x , 0 ≤ x ≤ ∞ . Find also standard deviation.
Solution:
M (t ) = E (etx ) = ∞ etx f ( x ) dx = ∞ 1
e
– θx−θ etxdx
X ∫
−∞
∫ 2θ
−∞
θ ( x−θ ) ∞ ( − x −θ )
= ∫ 1e θ etxdx + ∫ 1 e θ etxdx
−∞ 2θ ⎛ 1⎞ θ 2∞θ ⎛1 ⎞
−1 θ x t+ −x −t
M X (t ) = e ⎝⎜ θ ⎟⎠ dx + e ⎜θ ⎟
e
2θ
∫ 2θ ∫ θ
e ⎝ ⎠
dx
−∞
⎛ 1⎞
θ t+ −θ
⎛1
−1
⎞
⎜
−1 θ⎟ ⎜θ ⎟
⎝ ⎠ ⎠ ⎝
e e e e
= +
2θ ⎛ t + 1 ⎞ 2θ ⎛ 1 − t ⎞
⎜ θ⎟ ⎜θ ⎟
⎝θ t ⎠ ⎝ ⎠ θt
e eθ t e = θ t ⎡ − θ 2 ⎤ −1
= + =
e ⎣1 ( t )
2 (θ t +1) 2 (1− θ t ) 1− θ t
2 2 ⎦
⎡⎢ 1 θ 2t 2 ⎤
+ θ t + 2! + ...⎥⎡ + θ t + θ t + ...⎤⎦
2 2 4 4
=
⎣ 1
⎣ ⎦
3θ t
2 2
= 1+ θ t + + ...
15
2!
16
E ( X ) = μ ′1= coeff . of t in M X (t ) = θ
t2
μ 2′ = coeff . of in M X (t ) = 3θ
2
2!
( )
2
Var ( X ) = μ2′ − μ ′ 1 = 3θ 2 − θ 2 = 2θ 2 .
b)
Total Probability=1
∞
∴ ∫ ke − x dx = 1
0
∞
⎡ e−x ⎤
k⎢ ⎥ =1
⎣ − ⎦0
1 k =1
∞ ∞
e(t −1)xdx
M
X
( t ) = E ⎡⎣ etx ⎤
⎦
= etxe − x dx
∫ =
∫
0 0
∞
⎡ (t −1)x ⎤ 1
=⎢ e ⎥ = ,t <1
t −1 1− t
⎣ ⎦0
= (1− t ) = 1+ t + t 2 + ... + t r + ...∞
−1
μ′= t2
=
1 coeff . of r!
r!
When r = 1, μ 1′ = 1! = 1
r = 2 , μ ′2= 2! = 2
Variance = μ ′ 2− μ ′ 1= 2 −1 = 1
∴ Standard deviation=1.
Problem 24. (a). Define Binomial distribution Obtain its m.g.f., mean and variance.
(b). (i).Six dice are thrown 729 times. How many times do you expect atleast 3 dice
show 5 or 6 ?
(ii).Six coins are tossed 6400 times. Using the Poisson distribution, what is the
approximate probability of getting six heads x times?
Solution:
a) A random variable X said to follow binomial distribution if it assumes only non
negative values and its probability mass function is given by
P ( X = x ) = nCx p q , x = 0,1, 2,..., n and q = 1− p.
x n −x
M.G.F.of Binomial distribution:-

M.G.F of Binomial Distribution about origin is
n
M X ( t ) = E ⎡⎣ etx ⎤ ⎦ = ∑ etx P ( X = x )
x=0
n
= ∑ nC xx P xq n − xe t x
x=0
17
n x
= ∑ nCx ( )q
pet n− x
x=0
( )
n
M X (t) = q + pet
Mean of Binomial distribution
Mean = E ( X ) = M X ′ (0)
(
= ⎡ n q + pet ) pet ⎤
n−1
= np Since q + p = 1
⎢⎣ ⎥⎦ t =0
( )
E X 2 = M X′ (0)
(
= ⎡ n ( n −1) q + pet ) ( pet ) ( ) ⎤
n−2 2 n−1
+ npet q + pet
⎢⎣ ⎥⎦ t =0
( )
E X 2 = n (n −1) p2 + np
= n2 p2 + np (1− p ) = n2 p2 + npq
( )
Variance = E X 2 − ⎡⎣ E [ X ]⎤⎦ = npq
2
Mean = np ;Variance = npq

b) Let X :the number of times the dice shown 5 or 6
1 1 1
P [5 or 6] = + =
6 6 3
1 2
∴ P = and q =
3 3
Here n = 6
To evaluate the frequency of X ≥ 3
By Binomial theorem,
r 6−r
⎛1⎞ ⎛ 2 ⎞
P [ X = r ] = 6Cr ⎜ ⎟ ⎜ ⎟ where r = 0,1, 2...6 .
⎝ 3⎠ ⎝ 3 ⎠
P [ X ≥ 3] = P (3) + P ( 4 ) + P (5) + P (6)
3 3 4 2 5 6
= 6C3 ⎜ ⎟⎛ 1⎜⎞ ⎟⎛ 2 ⎞ + 6C ⎛ 1 ⎞ ⎛ 2 ⎞ + 6C ⎛ 1 ⎞ ⎛ 2 ⎞+ 6C
4⎜ ⎟ ⎜ ⎟ 5⎜ ⎟ ⎜ ⎟ 6 ⎜ ⎟
⎛1 ⎞
3 3 3 3 3 3 3
⎝ ⎠⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝⎠
= 0.3196
∴Expected number of times atleast 3 dies to show 5 or 6 = N × P [ X ≥ 3]
= 729 × 0.3196 = 233 .
25. (a). A die is cast until 6 appears what is the probability that it must cast more then
five times?
(b). Suppose that a trainee soldier shoots a target an independent fashion. If the
probability that the target is shot on any one shot is 0.8.
(i) What is the probability that the target would be hit on 6th attempt?
(ii) What is the probability that it takes him less than 5 shots?
Solution:
18
1
Probability of getting six =
6
1 1
∴p= & q = 1−
6 6
Let x : No of throws for getting the number 6.By geometric distribution
P [ X = x] = qx−1 p, x = 1, 2, 3....
Since 6 can be got either in first, second……throws.
To find P [ X > 5] = 1− P [ X ≤ 5]
5 5 x −1 1
⎛ ⎞
= 1− ∑ ⎜ ⎟ .
x =1 ⎝ 6 ⎠ 6
⎡ ⎛ 1 ⎞ ⎛ 5 ⎞⎛ 1 ⎞ ⎛ 5 ⎞2 ⎛ 1 ⎞ ⎛ 5 ⎞3 ⎛ 1 ⎞ ⎛ 5 ⎞4 ⎛ 1 ⎞⎤
= 1− ⎢ ⎜ ⎟ + ⎜ ⎟⎜ ⎟ + ⎜ ⎟ ⎜ ⎟ + ⎜ ⎟ ⎜ ⎟ + ⎜ ⎟ ⎜ ⎟⎥
⎣ ⎢ ⎝ 6 ⎠ ⎝ 65 ⎠⎝ 6 ⎠ ⎝ 6 ⎠ ⎝ 6 ⎠ ⎝ 6 ⎠ ⎝ 6 ⎠ ⎝ 6 ⎠ ⎝ 6 ⎠⎥ ⎦
1 ⎡ ⎛5⎞ ⎤
⎢ 1− ⎜ ⎟ ⎥
6⎢ ⎣ ⎝6 ⎠ ⎥ ⎦ ⎛ 5 ⎞ 5 = 0.4019
= 1− = ⎜ ⎟
5 ⎝6 ⎠
1−
6
b) Here p = 0.8, q = 1 − p = 0.2
P[ X = r] = qr−1 p, r = 0,1, 2...
(i) The probability that the target would be hit on the 6th attempt = P [ X = 6]
= ( 0.2 ) ( 0.8) = 0.00026
5
(ii) The probability that it takes him less than 5 shots = P[ X < 5]
4
4 r −1
r −1
=
∑q
r =1
p = 0.8∑(0.2)
r =1
= 0.8[1+ 0.2 + 0.04 + 0.008] = 0.9984

Problem 26. (a). State and prove the memoryless property of exponential distribution.
(b). A component has an exponential time to failure distribution with mean of 10,000
hours.
(i). The component has already been in operation for its mean life. What is the
probability that it will fail by 15,000 hours?
(ii). At 15,000 hours the component is still in operation. What is the probability
that it will operate for another 5000 hours.
Solution:
a) Statement:
If X is exponentially distributed with parameters λ , then for any two positive
integers‘s’ and‘t’, P [ x > s + t / x > s] = P [ x > t ]
Proof:
⎧λ e −λ x , x ≥ 0
The p.d.f of X is f (x )= ⎨
⎩0 , Otherwise
19
∞
∴P X > t ∞
⎡ −λx ⎤ − λt
= λ −λx
[ ] ∫t e dx = ⎣ −e ⎦ t = e
P [ x > s + t ∩ x > s]
∴ P [ X > s + t / x > s] =
P[ x > s]
P[ X > s + t] e − λ ( s +t ) − λt
= = =
P [ X > s] e− λs
e
= P[x > t ]
b) Let X denote the time to failure of the component then X has exponential distribution
with Mean = 1000 hours.
1 1
∴ = 10, 000 ⇒ λ =
λ 10, 000
⎪ 1 e − 10,000 , x ≥ 0
⎧ x
The p.d.f. of X is f ( x ) = ⎨10, 000

⎪ 0 , otherwise
⎩
(i) Probability that the component will fail by 15,000 hours given it has already been in
operation for its mean life = P [ x < 15, 000 / x > 10, 000]
P [10, 000 < X < 15, 000]
=
P [ X > 10, 000]
15,000
∫ f ( x) dx
e−1 − e−1.5
= 10,000
∞
= −1
e
∫ f ( x) dx
10,000
0.3679 − 0.2231
= = 0.3936 .
0.3679
(ii) Probability that the component will operate for another 5000 hours given that
it is in operational 15,000 hours = P[ X > 20, 000 / X > 15, 000]
= P [ x > 5000] [By memoryless prop]
∞
∫ f ( x) dx = e = 0.6065
−0.5
=
5000
27. (a). The Daily consumption of milk in a city in excess of 20,000 gallons is
1
approximately distributed as a Gamma variate with parameters α = 2 and λ = .
10, 000
The city has a daily stock of 30,000 gallons. What is the probability that the stock is in
sufficient on a particular day?
20
(b). The lifetime (in hours) of a certain kxpiece of equipment is a continuous r.v. having
⎧xe − , 0 < x < ∞
range 0 < x < ∞ and p.d.f.is f ( x) = ⎨ . Determine the constant K and
⎩0 , otherwise
evaluate the probability that the life time exceeds 2 hours.
Solution:
a) Let X be the r.v denoting the daily consumption of milk (is gallons) in a city
Then Y = X − 20, 000 has Gamma distribution with p.d.f.
y
1 –
f ( y) = y2−1e 10,000
,y≥0
(10, 000 ) Γ ( 2 )
2
y
–
ye 10,000
f (y ) = , y ≥ 0.
(10, 000)2
∴the daily stock of the city is 30,000 gallons, the required probability that the stock is
insufficient on a particular day is given by
P [ X > 30, 000] = P [Y > 10, 000]
y
∞ ∞ –
10,000
= ye
∫ g ( y ) dy = ∫ (10, 000) 2
dy
10,000 10,000
y dy
put Z = then dz =
10, 000 10, 000
∞
∴ P [ X > 30, 000] = ∫ ze−zdz

1
∞ 2
= ⎡⎣ −ze − z − e − z ⎤⎦ =
1 e
b) Let X the life time of a certain piece of equipment.
⎧xe − kx , 0 < x < ∞
Then the p.d.f. f x( )= ⎨
⎩0 , Otherwise
∞
To find K , ∫ f ( x) dx = 1
0
∞
∫e
−kx
0
x2−1dx = 1
Γ( 2)
=1⇒ K 2 =1 ∴K =1
2
K
⎧xe − x , 0 < x < ∞
∴ f x( =) ⎨
⎩0 , Otherwise
P[Life time exceeds 2 hours] = P[ X > 2]
∞
= ∫ f ( x ) dx
2
21
∞
= ∫ xe−xdx
2
( ) ( )
∞
= ⎣⎡ x −e− x − e − x ⎦⎤
2
= 2e + e = 3e = 0.4060
−2 −2 − 2
Problem 28. (a). State and prove the additive property of normal distribution.
t2
(b). Prove that “For standard normal distribution N (0,1), M X (t ) = e 2 .

a) Statement:
If X , X ,..., X are n independent normal random variates with mean ( μ ,σ 2
),
1 2 n 1 1
(μ 2
,σ 2
2
) ,… (μ n
,σ n
2
) then X + X +... + X
1 2 n
also a normal random variable with mean
⎞
⎛n μ , ∑σ .n 2
⎜∑ i ⎟ i
⎝ i=1 i =1 ⎠
Proof:
We know that. M X + X +...+ X 1 2 n
(t ) = M X (t ) M X (t )...M X (t )
1 2 n
t2σ 2
μi t + i
But M X ( t ) = e 2
, i = 1, 2 ... n
i
t 2σ 2
t2 σ 2
t2σ 2
μ1 t + 1 μ2 t + 2
μn t + n
M X +X 1 2 +...+ X n
(t ) = e 2
e 2 ...e 2
(σ 2
+σ 22 +...+ σ 2
)t 2
(μ )t +
1 n
+ μ +...+μ
=e
1 2 n
2
n
n ∑σ i
2 2
t
∑μ t+ i=1
=e i=1
i 2
By uniqueness MGF, X1 + X 2 + ... + Xn follows normal random variable with

⎛n n
2⎞
parameter ⎜ ∑ μi , ∑ σ i ⎟ .
⎝ i=1 i =1 ⎠
This proves the property.
b) Moment generating function of Normal distribution
= M X ( t ) = E ⎡⎣ e−tx ⎤ ⎦
1 ⎜ ⎟
= ∫∞etx e 2 1 ⎛ σx− μ ⎞dx
2
⎝ ⎠
σ 2π −∞
x−μ
Put z = then dz = σdx, − ∞ < Z < ∞
σ
∞ 2
t(σ z + μ )− z
1
∴ MX (t ) = ∫ e 2
dz
2π −∞ ⎛ 2 ⎞
∞ − z −tσ z
e μt ⎜2 ⎟
= ∫e ⎝ ⎠ dz
2π −∞
22
1 σ ⎛t
2 2 ⎞ σ 2t 2 ∞ 1
μt ∞ − ( z −tσ )
2
+⎜ μt − ( z −tσ )
2
2
= e ∫e dz = e e ∫e
2 ⎝ 2⎟ ⎠ 2
dz
2π −∞ 2π −∞
∞ 1
− ( z −tσ )
2
1
the total area under normal curve is unity, we have ∫e 2
dz = 1
2π −∞
μt + σ 2t2
Hence M X (t ) = e 2
∴For standard normal variable N (0,1)
t2
M X (t ) = e 2
Problem 29. (a). The average percentage of marks of candidates in an examination is 45
will a standard deviation of 10 the minimum for a pass is 50%.If 1000 candidates appear
for the examination, how many can be expected marks. If it is required, that double that
number should pass, what should be the average percentage of marks?
(b). Given that X is normally distribution with mean 10 and
probability P [ X > 12] = 0.1587 . What is the probability that X will fall in the
interval (9,11) .
Solution:
a) Let X be marks of the candidates
Then X N (42,102 )
X − 42
Let z =
10
P [ X > 50] = P [Z > 0.8]
= 0.5 − P [0 < z < 0.8] = 0.5 − 0.2881 = 0.2119
Since 1000 students write the test, nearly 212 students would pass the
examination.
If double that number should pass, then the no of passes should be 424.
We have to find z1 , such that P [Z > z1 ] = 0.424
∴ P [0 < z < z1] = 0.5 − 0.424 = 0.076
From tables, z = 0.19
50 − x1
∴z = ⇒ x = 50 −10z = 50 −1.9 = 48.1
1 1 1
10
The average mark should be 48 nearly.
b) Given X is normally distributed with mean μ = 10.
x−μ
Let z = be the standard normal variate.
σ
12 −10 2
For X = 12, z = ⇒z=
σ σ
2
Put z1 =
σ
Then P [ X > 12] = 0.1587
23
P [Z > Z1 ] = 0.1587
∴0.5 − p [0 < z < z1] = 0.1587
⇒ P [ 0 < z < z1 ] = 0.3413
From area table P[ 0 < z < 1] = 0.3413
2
∴ Z1 = 1 ⇒ = 1
σ
To find P [9 < x < 11]
1 1
For X = 9, z = − and X = 11, z =
2 2
∴ P [9 < X < 11] = P [−0.5 < z < 0.5] = 2P [0 < z < 0.5] = 2 × 0.1915 = 0.3830
Problem. (a). In a normal distribution,31 % of the items are under 45 and 8% are over
64.Find the mean and standard deviation of the distribution.
(b). For a certain distribution the first moment about 10 is 40 and that the 4th moment
about 50 is 48, what are the parameters of the distribution.
Solution:
a) Let μ be the mean and σ be the standard deviation.
Then P [ X ≤ 45] = 0.31 and P [ X ≥ 64] = 0.08
45 − μ
When X = 45 , Z = = −z
σ 1
z1
∴ z1 is the value of z corresponding to the area

∫φ ( z )dz = 0.19
0
∴ z1 = 0.495
45 − μ = −0.495σ ---(1)
64 − μ
When X = 64 , Z = =z
σ
2
z2
∴ z2 is the value of z corresponding to the area

∫ φ ( z )dz = 0.42
0
∴ z2 = 1.405
64 − μ = 1.405σ ---(2)
Solving (1) & (2) We get μ = 10 (approx) & σ = 50 (approx)
b) Let μ be mean and σ 2 the variance then μ ′1 = 40 about A=10
∴Mean A + μ ′ =1 10 + 40 10+40
⇒ μ = 50
Also μ 4= 48 ⇒ 3σ 4 = 48 ⇒ σ 2 = 4
∴ The parameters are Mean = μ = 50 and S.D = σ = 2 .
24

Random Process Characteristic Equation NOTES

Uploaded by

Copyright:

Available Formats

Random Process Characteristic Equation NOTES

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Random Process Characteristic Equation NOTES

Uploaded by

Copyright:

Available Formats

Unit.3.

Classification of Random Processes

UNIT – III RANDOM PROCESSES

Problem 1. Define I & II order stationary Process

Problem 2. Define wide-sense stationary process

Problem 3. Define a strict sense stationary process with an example

A nth order stationary process becomes a SSS process when n → ∞ .

Problem 5. When are two random process said to be orthogonal?

7. Write the postulates of a poisson process?

8. When is a poisson process said to be homogenous?

Probability during 1-min interval, no customer arrives = P { X (t ) = 0} = e−2 .

10. Define ergodic process.

11. Define a Gaussian process.

Where μi = E { X ( ti )} and Λ is the n

12. Define a Markov process with an example.

13. Define a Markov chain and give an example.

Problem 14. What is a stochastic matrix? When is it said to be regular?

(ii). If T is discrete and S is continuous, the random process is called a continuous

b). Consider the random process X (t ) = cos(t + φ ) , where φ is uniformly distributed in

⎡ ∞ ⎛ at n−1 −∑∞n ⎜ ⎛ at⎟ n−1 ⎤ ⎥

= cosλt E [cosλt] − sinλt E [siny] = 0

Problem 19. a). If a random process { X (t )} is defined by {X (t )} = sin (ω t+ Y ) where

E [ X (t)] = ∫ X (t) f ( y)dy

= 2 cos (t1 − t2 ) = is a function of time difference.

= E ( Acoswt1 + Bsinwt2 )( Acoswt 2 + Bsinwt2 )

= E A2coswt coswt + ABcoswt sinwt + BAsinwt c2oswt + B sinwt sinwt

= E ( B 2 )cosωt 1cosωt 2 – E (BA)cosωt sin 1 ωt + 2E ( A )sinωt sin1ωt

RYY (t1, t2 ) =is a function of time difference.

RXY (t1, t2 ) = is a function of time difference.

b). Write a note on Binomial process.

(ii). Sn is a binomial random variable so, P(Sn = m) = nC mpmqn m

i.e., The Poisson process possesses Markov property.

b). State and establish the properties of Poisson process.

M X (t) (u ) = e and M X (t) (u ) = e

M X (t)+ X (t) (u) = M X (t) (u)M X

By uniqueness of moment generating function, the process X 1 (t ) + X 2 (t )} is a

Recall that E { X 2 (t )} for a poisson process {X (t )} with parameter λ is given by

Problem 23. a). If the process {N (t ) : t ≥ 0} is a Poisson process with parameter λ ,

This is a linear differential equation.

Now taking n =1 we get

eλt P1( t ) = λ ∫ P0 ( t )e λ dt ---------------- (3)

P2 (t)e t = P1 (t)e tdt

= e −λt ∑n=1 ( n −1)!

Since X (t )} is a process of independent increments.

= lim 1 ⎢⎡ sin ( 2ωt + ωt + 2θ )

b). If the WSS process { X (t )} is given by X (t ) = 10 cos(100t + θ ) where θ is uniformly

∴{X (t )} is correlation ergodic.

= E ⎡⎣ cos (ω t + 0 ) cos (ωt + ωτ + 0 )⎤⎦

1 ⎡⎢ sin (2ωt + ωτ + 20 ) + tcos (ωτ )⎥ ⎤

∴{X (t )} is correlation ergodic.

If the are uncomelated then,

Now the joint density of X t & X t is

b). If { X ( t )} is a Gaussian process with μ ( t ) = 10 & C ( t1 , t2 ) = 16 e− t −t1 2

(from normal tables)

X (10) X (6) is also a normal R V With mean μ (10) − μ (6) = 10 −10 = 0

= 0.26 × 0.7 + 0.34 × 0.2 + 0.29 × 0.1 = 0.279

P2 > 0, P1 > 0, P2 > 0

The Markov chain is irreducible

The TPM of the chain is

P(1) = P(0)P = (0, 1, 0)