1|Page

UNIT V

LINEAR SYSTEMS WITH RANDOM INPUTS

Linear time invariant system - System transfer function – Linear systems with random inputs – Auto
correlation and cross correlation functions of input and output – white noise.

PART – A

1. Define a system. When is it called a linear system? (April/May 2003)(June, 2012)

A system is a functional relationship between the input 𝑋(𝑡) and the output 𝑌(𝑡). The functional relationship
is written as 𝑦(𝑡) = 𝑓[𝑥(𝑡)] .

If 𝑓[𝑎1 𝑋1 (𝑡) ± 𝑎2 𝑋2 (𝑡)] = 𝑎1 𝑓[𝑋1 (𝑡)] ± 𝑎2 𝑓[𝑋2 (𝑡)] then f is called a linear system.

2. Write a note on linear system(Nov./Dec. 2004) (May/June 2007) (April/May 2008)(May/June 2013)
If 𝑓[𝑎1 𝑋1 (𝑡) ± 𝑎2 𝑋2 (𝑡)] = 𝑎1 𝑓[𝑋1 (𝑡)] ± 𝑎2 𝑓[𝑋2 (𝑡)] then f is called a linear system.

If 𝑌(𝑡 + ℎ) = 𝑓[𝑋(𝑡 + ℎ)] where 𝑌(𝑡) = 𝑓[𝑋(𝑡)] then 𝑓 is called a time – invariant system or 𝑋(𝑡) and Y(t)
are said to form a time invariant system.

If the output 𝑌(𝑡1 ) at a given time 𝑡 = 𝑡1 depends only on 𝑋(𝑡1 ) and not on any other past or future values
of 𝑋(𝑡), then the system 𝑓 is called a memoryless system. If the value of the output 𝑌(𝑡) at 𝑡 = 𝑡1 depends only
on the past values of the input 𝑋(𝑡), 𝑡 ≤ 𝑡. (ie) 𝑌(𝑡1 ) = 𝑓[𝑋(𝑡); 𝑡 ≤ 𝑡1 ], then the system is called a casual
system.

3. State the properties of linear system.(april, 2010) (Nov./Dec. 2003)

The properties of linear system are

(i) If a system is such that its input 𝑋(𝑡) and its output 𝑌(𝑡) are related by a convolution integral, then the
system is a linear time invariant system.
(ii) If the input to a time-invariant, stable linear system is a WSS process, the output will also be a WSS
process.
(iii) The power spectral densities of the input and output processes in the system are connected by the
relation sYY (ω) = |H(ω)|2 sXX (ω), where H(ω) is the Fourier transform of unit impulse response
function ℎ(𝑡).
4. Define system weighting function.(APRIL, 2004)
If the output 𝑌(𝑡) of a system is expressed as the convolution of the input 𝑋(𝑡) and a function ℎ(𝑡) (ie)
∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢 then ℎ(𝑡) is called the system weighting function.

5. What is unit impulse response of a system? Why is it called so?(APRIL, 2004)

∞
If a system is of the form 𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢 then the system weighting function ℎ(𝑡) is also
called unit impulse response of the system. It is called so because the response (output) 𝑌(𝑡) will be ℎ(𝑡),
when the input 𝑋(𝑡) = the unit impulse function 𝛿(𝑡).

6. If the input of a linear system is a Gaussian random process, comment about the output random process.(April,
2006)
2|Page

If the input of a linear system is a Gaussian random process, then the output will also be a Gaussian random
process.

∞
7. If the input 𝑋(𝑡) of the system 𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢 is the unit impulse function, (june, 2006)
Prove that 𝑌(𝑡) = ℎ(𝑡)
∞
Given 𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢

Put 𝑋(𝑡) = 𝛿(𝑡)

Therefore 𝑋(𝑡 − 𝑢) = 𝛿(𝑡 − 𝑢)

∞
𝑌(𝑡) = ∫−∞ ℎ(𝑡 − 𝑢) 𝛿(𝑢)𝑑𝑢 [By the property of convolution]

=ℎ(𝑡 − 0) = ℎ(𝑡)
𝑢
1 ∞
8. If a system is defined as 𝑌(𝑡) = 𝑇 ∫0 𝑋(𝑡 − 𝑢)𝑒 −𝑇 𝑑𝑢, find its unit impulse function.(june, 2006)
𝑢
1 ∞ −
Given 𝑌(𝑡) = ∫0 𝑋(𝑡 − 𝑢)𝑒 𝑇 𝑑𝑢
𝑇

∞
1 −𝑢
𝑌(𝑡) = ∫ 𝑒 𝑇 𝑋(𝑡 − 𝑢) 𝑑𝑢
0 𝑇
𝑡
1 −
The unit impulse function ℎ(𝑡) = {𝑇 𝑒 𝑡>0 .
𝑇

0 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
∞
9. If {𝑋(𝑡)} and {𝑌(𝑡)} in the system 𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢 are WSS processes, how are their
autocorrelation functions related?(June, 2007)
𝑅𝑌𝑌 (𝜏) = 𝑅𝑋𝑌 (𝜏) ∗ ℎ(−𝜏)

𝑅𝑋𝑌 (𝜏) = 𝑅𝑋𝑋 (𝜏) ∗ ℎ(−𝜏) , where * denotes convolution.

∞
10. If the input and output of the system 𝑌(𝑡) = ∫−∞ ℎ(𝑢) 𝑋(𝑡 − 𝑢)𝑑𝑢 are WSS processes, how are their power
spectral densities related?(June, 2007)
sYY (ω) = |H(ω)|2 sXX (ω)

Where H(ω) is the Fourier transform of ℎ(𝑡)

11. Define the power transfer function or system function of the system.(April, 2008)
The power transfer function or system function of the system is the Fourier transform of the unit impulse
function of the system.
1
𝑓𝑜𝑟 |𝑡| ≤ 𝑐
12. If the system function of a convolution type of linear system is given by ℎ(𝑡) = { 2𝑐
Find the
0 𝑓𝑜𝑟 |𝑡| > 𝑐
relation between power spectral density functions of the input and output processes. (April, 2011)(April, 2009)

∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡
𝑐 1
𝐻(𝜔) = ∫−𝑐 2𝑐 𝑒 −𝑖𝜔𝑡 𝑑𝑡
1 𝑐
= 2𝑐 ∫−𝑐 𝑒 −𝑖𝜔𝑡 𝑑𝑡
3|Page

1 𝑐
= 2𝑐 ∫−𝑐 (cos 𝜔 𝑡 − 𝑖𝑠𝑖𝑛 𝜔𝑡) 𝑑𝑡

1 𝑐 𝑖 𝑐
= 2𝑐 ∫−𝑐 cos 𝜔𝑡 𝑑𝑡 + 2𝑐 ∫−𝑐 𝑠𝑖𝑛 𝜔𝑡 𝑑𝑡

1 𝑐 𝑖
= ∫ cos 𝜔𝑡 𝑑𝑡 + (0)
2𝑐 −𝑐 2𝑐

since the first and second integrand are even and odd functions

2 𝑐 1 𝑐 1 𝑠𝑖𝑛𝜔𝑡 𝑐 𝑠𝑖𝑛𝜔𝑐
= ∫ cos 𝜔𝑡 𝑑𝑡 = ∫ cos 𝜔𝑡 𝑑𝑡 = [ ] =
2𝑐 −𝑐 𝑐 −𝑐 𝑐 𝜔 0 𝜔𝑐

sYY (ω) = |H(ω)|2 sXX (ω)

𝑠𝑖𝑛𝜔𝑐 2
sYY (ω) = | | sXX (ω)
𝜔𝑐

sin2 𝜔𝑐
sYY (ω) = s (ω)
𝜔 2 𝑐 2 XX
13. What is thermal noise? By what type of random processes is it represented?(April, 2009)
Thermal noise is the noise because of the random motion of free electrons in conducting media such as a
resistor. It is represented by Gaussian random processes.

14. Define white noise.(April, 2011)(april, 2009)

Let 𝑋(𝑡) be a sample function of a WSS noise process, then {𝑋(𝑡), 𝑡 ∈ 𝑇} is called the white noise if the power
𝑁0
density spectrum of {𝑋(𝑡), 𝑡 ∈ 𝑇} is constant at all frequencies. (ie) 𝑠𝑁𝑁 (𝜔) = 2
where 𝑁0 is a real positive
constant.
𝑁
15. If the power spectral density of white noise is 20 , find its autocorrelation function.(April, 2010)
∞
𝑁0 𝑁0
𝐹 [ 𝛿(𝜏)] = ∫ 𝛿(𝜏) 𝑒 −𝑖𝜔𝜏 𝑑𝜏
2 −∞ 2

𝑁0 ∞ 𝑁0 𝑁0
= ∫ 𝛿(𝜏) 𝑒 −𝑖𝜔𝜏 𝑑𝜏 = (1) =
2 −∞ 2 2

𝑁0 𝑁0
Therefore 𝑅𝑁𝑁 (𝜏) = 𝐹 −1 [ ]= 𝛿(𝜏)
2 2

∞
∫−∞ 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 = 1

16. If the input to a linear time invariant system is white noise {𝑁(𝑡)} what is power spectral density function of the
output?(pril,2009)
If the input to a linear time invariant system is white noise {𝑁(𝑡)}, then the power spectral density of the output
sYY (ω) is given by

sYY (ω) = |H(ω)|2 sXX (ω)

𝑁0
sYY (ω) = |H(ω)|2
2

Where {𝑌(𝑡)} is the output process and 𝐻(𝜔) is the power transfer function.
4|Page

17. Prove that 𝑌(𝑡) = 2𝑋(𝑡) is linear

18. Check whether the system 𝑌(𝑡) = 𝑥 3 (𝑡)
19. Check whether the system 𝑌(𝑡) = 𝛼 𝑋(𝑡)
Solution: Let 𝑦1 (𝑡) and 𝑦2 (𝑡) be the output signal corresponding to the input signal and 𝑥1 (𝑡) and 𝑥2 (𝑡)
respectively

𝑌1 (𝑡) = 𝛼 𝑋1 (𝑡) and 𝑌2 (𝑡) = 𝛼 𝑋2 (𝑡)

For any scalar 𝑐1 and 𝑐2 the output signal for the input signal 𝑋(𝑡) = 𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)
𝑌(𝑡) = 𝛼 𝑋(𝑡)

𝑌(𝑡) = 𝛼 [𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)]

𝑌(𝑡) = 𝑐1 𝛼𝑋1 (𝑡) + 𝑐2 𝛼𝑋2 (𝑡)

𝑌(𝑡) = 𝑐1 𝑌1 (𝑡) + 𝑐2 𝑌2 (𝑡)

For this we get the given system is linear.

20. Check whether the system 𝑌(𝑡) = 𝛼 𝑋(𝑡)

Solution: Let 𝑦1 (𝑡) and 𝑦2 (𝑡) be the output signal corresponding to the input signal and 𝑥1 (𝑡) and 𝑥2 (𝑡)
respectively

𝑌1 (𝑡) = 𝛼 𝑋1 (𝑡) and 𝑌2 (𝑡) = 𝛼 𝑋2 (𝑡)

For any scalar 𝑐1 and 𝑐2 the output signal for the input signal 𝑋(𝑡) = 𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)
𝑌(𝑡) = 𝛼 𝑋(𝑡)

𝑌(𝑡) = 𝛼 [𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)]

𝑌(𝑡) = 𝑐1 𝛼𝑋1 (𝑡) + 𝑐2 𝛼𝑋2 (𝑡)

𝑌(𝑡) = 𝑐1 𝑌1 (𝑡) + 𝑐2 𝑌2 (𝑡)

For this we get the given system is linear.

21. check whether the system 𝑌(𝑡) = 𝑋 2 (𝑡)

Solution: Let 𝑦1 (𝑡) and 𝑦2 (𝑡) be the output signal corresponding to the input signal and 𝑥1 (𝑡) and 𝑥2 (𝑡)
respectively

𝑌1 (𝑡) = 𝑋12 (𝑡) and 𝑌2 (𝑡) = 𝛼 𝑋22 (𝑡)

For any scalar 𝑐1 and 𝑐2 the output signal for the input signal 𝑋(𝑡) = 𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)
𝑌(𝑡) = 𝛼 𝑋(𝑡)

𝑌(𝑡) = 𝛼 [𝑐1 𝑋1 (𝑡) + 𝑐2 𝑋2 (𝑡)]2

𝑌(𝑡) = 𝑐12 𝑋12 (𝑡) + 𝑐22 𝑋22 (𝑡) + 2𝑐1 𝑐2 𝑋1 (𝑡)𝑋2 (𝑡)

𝑌(𝑡) ≠ 𝑐1 𝑌1 (𝑡) + 𝑐2 𝑌2 (𝑡)

For this we get the given system is not linear.

1. Show that 𝒔𝒀𝒀 (𝝎) = |𝑯(𝝎)|𝟐 𝒔𝑿𝑿 (𝝎) where 𝒔𝑿𝑿 (𝝎)and 𝒔𝒀𝒀 (𝝎) are the power spectral density
functions of the input 𝑿(𝒕) and the output 𝒀(𝒕) and 𝑯(𝝎) is the system transfer function.
Solution:
We know that linear time invariant system 𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡)
5|Page

∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢 − − − − − (1)
Power spectral density of the out put 𝑌(𝑡) is
∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ 𝑅𝑌𝑌 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 − − − − − (2)
Since 𝑋(𝑡) is a WSS process and 𝑌(𝑡) is also WSS Process.
𝑅𝑌𝑌 (𝜏) = 𝐸[𝑌(𝑡)𝑌(𝑡 + 𝜏)]
∞ ∞
𝑅𝑌𝑌 (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 ) 𝑑𝑢1 𝑑𝑢2 )

∞ ∞
= ∫−∞ ∫−∞ ℎ(𝑢1 )ℎ(𝑢2 )𝐸[𝑋(𝑡 − 𝑢1 ) 𝑋(𝑡 + 𝜏 − 𝑢2 )] 𝑑𝑢1 𝑑𝑢2

Since 𝑋(𝑡) is a WSS process ⇒ 𝑅𝑋𝑋 (𝜏) = 𝐸[𝑋(𝑡)𝑋(𝑡 + 𝜏)]

∞ ∞
𝑅𝑌𝑌 (𝜏) = ∫−∞ ∫−∞ ℎ(𝑢1 )ℎ(𝑢2 ) 𝑅𝑋𝑋 (𝜏 + 𝑢1 − 𝑢2 ) 𝑑𝑢1 𝑑𝑢2 − − − − − (3)

put equation (3) in (2)

∞ ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ ∫−∞ ∫−∞ ℎ(𝑢1 )ℎ(𝑢2 ) 𝑅𝑋𝑋 (𝜏 + 𝑢1 − 𝑢2 ) 𝑒 −𝑖𝜔𝜏 𝑑𝑢1 𝑑𝑢2 𝑑𝜏

∞ ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ ℎ(𝑢1 )𝑑𝑢1 ∫−∞ ℎ(𝑢2 ) 𝑑𝑢2 ∫−∞ 𝑅𝑋𝑋 (𝜏 + 𝑢1 − 𝑢2 ) 𝑒 −𝑖𝜔𝜏 𝑑𝜏

by change of variable put 𝑢 = 𝜏 + 𝑢1 − 𝑢2

𝑑𝑢 = 𝑑𝜏 𝑢 = 𝜏 + 𝑢1 − 𝑢2

⇒ 𝜏 = 𝑢 − 𝑢1 + 𝑢2

𝜏 = −∞ ⇒ 𝑢 = −∞ 𝜏 = ∞ ⇒ 𝑢 = ∞
∞ ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ ℎ(𝑢1 )𝑑𝑢1 ∫−∞ ℎ(𝑢2 ) 𝑑𝑢2 ∫−∞ 𝑅𝑋𝑋 (𝑢) 𝑒 −𝑖𝜔(𝑢−𝑢1 +𝑢2 ) 𝑑𝑢

∞ ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ ℎ(𝑢1 )𝑑𝑢1 ∫−∞ ℎ(𝑢2 ) 𝑑𝑢2 ∫−∞ 𝑅𝑋𝑋 (𝑢) 𝑒 −𝑖𝜔𝑢 𝑒 𝑖𝜔𝑢1 𝑒 −𝑖𝜔𝑢2 𝑑𝑢

∞ ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ ℎ(𝑢1 )𝑒 𝑖𝜔𝑢1 𝑑𝑢1 ∫−∞ ℎ(𝑢2 ) 𝑒 −𝑖𝜔𝑢2 𝑑𝑢2 ∫−∞ 𝑅𝑋𝑋 (𝑢) 𝑒 −𝑖𝜔𝑢 𝑑𝑢

We know that Fourier transform of ℎ(𝑡) is 𝐻(𝜔)

∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡

∞
𝐻 ∗ (𝜔) = ∫−∞ ℎ(𝑡)𝑒 𝑖𝜔𝑡 𝑑𝑡 (Complex conjugate)

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏
6|Page

𝑠𝑌𝑌 (𝜔) = 𝐻(𝜔)𝐻 ∗ (𝜔) 𝑠𝑋𝑋 (𝜔) Using 𝑧𝑧 = |𝑧|2

𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

2. If the input to a time invariant stable linear system is a wide sense stationary process, prove that
the output will also be a wide sense stationary process.
OR

Show that input 𝑿(𝒕) is a WSS process then output 𝒀(𝒕) also WSS process.

Proof: We know that input and output are related by

𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡) (Convolution)

∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢 ------ (1)

∞
𝐸[𝑌(𝑡)] = 𝐸[∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢]

∞
= ∫−∞ ℎ(𝑢)𝐸[𝑋(𝑡 − 𝑢)]𝑑𝑢

Since 𝑋(𝑡) ia a WSS process Mean is constant for any time t

𝐸[𝑋(𝑡 − 𝑢)] = 𝜇𝑥 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

∞
𝐸[𝑌(𝑡)] = ∫−∞ ℎ(𝑢) 𝜇𝑥 𝑑𝑢

∞
𝐸[𝑌(𝑡)] = 𝜇𝑥 ∫−∞ ℎ(𝑢) 𝑑𝑢 = 𝑓𝑖𝑛𝑖𝑡𝑒

∞
Since the system is stable ∫−∞ ℎ(𝑢) 𝑑𝑢 = 𝑓𝑖𝑛𝑖𝑡𝑒

Since 𝐸[𝑌(𝑡)] Constant

Next we show that the auto correlation

𝑅𝑌𝑌 (𝑡, 𝑡 + 𝜏) Depends on 𝜏

𝑅𝑌𝑌 (𝑡, 𝑡 + 𝜏) = 𝐸[𝑌(𝑡)𝑌(𝑡 + 𝜏)

∞ ∞
𝑅𝑌𝑌 (𝑡, 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 ) 𝑑𝑢1 𝑑𝑢2 )
7|Page

∞ ∞
= ∫−∞ ∫−∞ ℎ(𝑢1 )ℎ(𝑢2 )𝐸[𝑋(𝑡 − 𝑢1 ) 𝑋(𝑡 + 𝜏 − 𝑢2 ) ]𝑑𝑢1 𝑑𝑢2

Since 𝑋(𝑡) is a WSS process, auto correlation function is only a function of time difference
𝐸[𝑋(𝑡 − 𝑢1 )𝑋(𝑡 + 𝜏 − 𝑢2 ) = 𝑅𝑋𝑋 (𝜏 + 𝑢1 − 𝑢2 )

∞ ∞
𝑅𝑌𝑌 (𝑡, 𝑡 + 𝜏) = ∫−∞ ∫−∞ ℎ(𝑢1 )ℎ(𝑢2 ) 𝑅𝑋𝑋 (𝜏 + 𝑢1 − 𝑢2 ) 𝑑𝑢1 𝑑𝑢2

Double Integral is evaluated by integrating with respect to 𝑢1 𝑎𝑛𝑑 𝑢2 , the RHS is only a function of 𝜏.

∞
3. If 𝑿(𝒕) is a WSS process and if 𝒀(𝒕) = ∫−∞ 𝒉(𝒖)𝑿(𝒕 − 𝒖)𝒅𝒖 then 𝑹𝑿𝒀 (𝝉) = 𝑹𝑿𝑿 (𝝉) ∗ 𝒉(𝝉) Where *
denote the convolution
Solution: Given
∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢
𝑅𝑋𝑌 (𝜏) = 𝐸[𝑋(𝑡). 𝑌(𝑡 + 𝜏)]
∞
𝑅𝑋𝑌 (𝜏) = 𝐸[𝑋(𝑡). ∫−∞ ℎ(𝑢)𝑋(𝑡 + 𝜏 − 𝑢)𝑑𝑢]
∞
= ∫−∞ ℎ(𝑢)𝐸[𝑋(𝑡) 𝑋(𝑡 + 𝜏 − 𝑢)]𝑑𝑢 𝑡+𝜏−𝑢−𝑡 = 𝜏−𝑢
∞
= ∫−∞ ℎ(𝑢) 𝑅𝑋𝑋 (𝜏 − 𝑢) 𝑑𝑢 𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡)

𝑅𝑋𝑌 (𝜏) = 𝑅𝑋𝑋 (𝜏) ∗ ℎ(𝜏) by convolution.

∞
4. If 𝑿(𝒕) is a WSS process and if 𝒀(𝒕) = ∫−∞ 𝒉(𝒖)𝑿(𝒕 − 𝒖)𝒅𝒖 then 𝑹𝒀𝒀 (𝝉) = 𝑹𝑿𝒀 (𝝉) ∗ 𝒉(−𝝉) Where *
denote the convolution
Solution: Given
∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢
𝑅𝑌𝑌 (𝜏) = 𝐸[𝑌(𝑡). 𝑌(𝑡 + 𝜏)]
∞
𝑅𝑌𝑌 (𝜏) = 𝐸(∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢) 𝑌(𝑡 + 𝜏) 𝑑𝑢)
∞
= ∫−∞ 𝐸[𝑋(𝑡 − 𝑢) 𝑌(𝑡 + 𝜏)] ℎ(𝑢)𝑑𝑢 𝑡+𝜏+𝑢−𝑡 =𝜏+𝑢
∞
= ∫−∞ 𝑅𝑋𝑌 (𝜏 + 𝑢) ℎ(𝑢) 𝑑𝑢

Put 𝑢 = −𝛼 ⇒ 𝑑𝑢 = −𝑑𝛼

−∞
= ∫∞ 𝑅𝑋𝑌 (𝜏 − 𝑢) ℎ(−𝑢) (−𝑑𝑢)

∞
= ∫−∞ 𝑅𝑋𝑌 (𝜏 − 𝑢) ℎ(−𝑢) 𝑑𝑢 𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡)

𝑹𝒀𝒀 (𝝉) = 𝑹𝑿𝒀 (𝝉) ∗ 𝒉(−𝝉) by convolution.

∞
5. If the Input 𝑿(𝒕) and output 𝒀(𝒕) are related by 𝒀(𝒕) = ∫−∞ 𝒉(𝒖)𝑿(𝒕 − 𝒖)𝒅𝒖 Then the system is linear
time invariant system.
Solution: Now first we prove the linearity
Consider 𝑋(𝑡) = 𝑎1 𝑋1 (𝑡) + 𝑎2 𝑋2 (𝑡)
8|Page

∞
Then 𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢
∞
= ∫−∞ ℎ(𝑢)[𝑎1 𝑋1 (𝑡 − 𝑢) + 𝑎2 𝑋2 (𝑡 − 𝑢)]𝑑𝑢
∞ ∞
= 𝑎1 ∫−∞ ℎ(𝑢) 𝑋1 (𝑡 − 𝑢) 𝑑𝑢 + 𝑎2 ∫−∞ ℎ(𝑢)𝑋2 (𝑡 − 𝑢)𝑑𝑢
= 𝑎1 𝑌1 (𝑡) + 𝑎2 𝑌2 (𝑡)
Hence the system is linear
Now we prove that the system is a time invariant system
∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢
Replacing 𝑡 by 𝑡 + 𝑘
∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 + 𝑘 − 𝑢)𝑑𝑢
= 𝑌(𝑡 + 𝑘)

6. A system has an impulse response function 𝒉(𝒕) = 𝒆−𝜷𝒕 𝒖(𝒕) find the power spectral density of the
output 𝒀(𝒕) corresponding to the input 𝑿(𝒕).
Solution.

Let 𝑋(𝑡) be the input and 𝑌(𝑡) be the output, the power spectral density of out put 𝑌(𝑡)

𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

0 𝑡<0
We know that unit step function 𝑢(𝑡) = {
1 𝑡≥0

Given ℎ(𝑡) = 𝑒 −𝛽𝑡 𝑢(𝑡)

0 𝑡<0
ℎ(𝑡) = { −𝛽𝑡
𝑒 𝑡≥0

𝐻(𝜔) is the Fourier transform of the function ℎ(𝑡)

∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡

∞
𝐻(𝜔) = ∫0 𝑒 −𝛽𝑡 𝑒 −𝑖𝜔𝑡 𝑑𝑡 since ℎ(𝑡) = 𝑒 −𝛽𝑡 𝑡≥0

∞
𝐻(𝜔) = ∫0 𝑒 −(𝛽+𝑖𝜔)𝑡 𝑑𝑡

∞
𝑒 −(𝛽+𝑖𝜔)𝑡
=[ ]
−(𝛽+𝑖𝜔)
0

1 1
= 0 − −(𝛽+𝑖𝜔) = (𝛽+𝑖𝜔)

1 𝛽−𝑖𝜔 𝛽−𝑖𝜔
𝐻(𝜔) = (𝛽+𝑖𝜔) = (𝛽+𝑖𝜔)(𝛽−𝑖𝜔) = (𝛽2 +𝜔2 )

𝛽 2 +𝜔2 1
|𝐻(𝜔|2 = (𝛽2 2 )2 =
+𝜔 𝛽 2 +𝜔2
9|Page

Power spectral density of the output 𝑌(𝑡) is 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

1
𝑠𝑌𝑌 (𝜔) = 𝑠 (𝜔) Hence proved
𝛽 2 +𝜔2 𝑋𝑋

7. A wide sense stationary random process 𝑿(𝒕). With autocorrelation function 𝑹𝑿𝑿 (𝝉) = 𝑨𝒆−𝜶|𝝉| . Where 𝑨
and 𝜶 are real positive constants is applied to the input of a linearly time invariant system with impulse
response 𝒉(𝒕) = 𝒆−𝒃𝒕 𝒖(𝒕) where 𝒃 is a real positive constant. Find the power spectral density of the output
𝒀(𝒕) of the system.
Solution

Let 𝑋(𝑡) be the input and 𝑌(𝑡) be the output, the power spectral density of out put 𝑌(𝑡)

𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

0 𝑡<0
We know that unit step function 𝑢(𝑡) = {
1 𝑡≥0

Given ℎ(𝑡) = 𝑒 −𝑏𝑡 𝑢(𝑡)

0 𝑡<0
ℎ(𝑡) = { −𝑏𝑡
𝑒 𝑡≥0

𝐻(𝜔) is the Fourier transform of the function ℎ(𝑡)

∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡

∞
𝐻(𝜔) = ∫0 𝑒 −𝑏𝑡 𝑒 −𝑖𝜔𝑡 𝑑𝑡 since ℎ(𝑡) = 𝑒 −𝑏𝑡 𝑡≥0

∞
= ∫0 𝑒 −(𝑏+𝑖𝜔)𝑡 𝑑𝑡

∞
𝑒 −(𝑏+𝑖𝜔)𝑡
= [ −(𝑏+𝑖𝜔) ]
0

1 1
=0− =
−(𝑏+𝑖𝜔) (𝑏+𝑖𝜔)

1 𝑏−𝑖𝜔 𝑏−𝑖𝜔
𝐻(𝜔) = (𝑏+𝑖𝜔) = (𝑏+𝑖𝜔)(𝑏−𝑖𝜔) = (𝑏2 +𝜔2 )

𝑏2 +𝜔2 1
|𝐻(𝜔|2 = (𝑏2 2 )2 =
+𝜔 𝑏2 +𝜔2

Power spectral density of the in put 𝑋(𝑡) is

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

Given 𝑅𝑋𝑋 (𝜏) = 𝐴𝑒 −𝛼|𝜏| .

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝐴𝑒 −𝛼|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏
10 | P a g e

∞
= ∫−∞ 𝐴𝑒 −𝛼|𝜏| (𝑐𝑜𝑠𝜔𝜏 − 𝑖𝑠𝑖𝑛𝜔𝜏) 𝑑𝜏

∞ ∞
= 𝐴 ∫−∞ 𝑒 −𝛼|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 − 𝑖𝐴 ∫−∞ 𝑒 −𝛼|𝜏| 𝑠𝑖𝑛𝜔𝜏 𝑑𝜏

Even Odd

∞
= 𝐴 ∫−∞ 𝑒 −𝛼|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 − 0

∞ ∞ 𝑎
= 2𝐴 ∫0 𝑒 −𝛼𝜏 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 ∴ ∫0 𝑒 −𝑎𝑥 𝑐𝑜𝑠𝑏𝑥 𝑑𝑥 = 𝑎2 +𝑏2

𝛼 2𝐴𝛼
= 2𝐴 𝛼 2 +𝜔2
= 𝛼 2 +𝜔2

Power spectral density of the output 𝑌(𝑡) is 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

1 2𝐴𝛼 2𝐴𝛼
𝑆𝑌𝑌 (𝜔) = 𝑏2 +𝜔2 𝛼2 +𝜔2 = (𝑏2 +𝜔2 )(𝛼2 +𝜔2 )

8. 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 mean zero 𝝁𝒙 = 𝟎. And autocorrelation function 𝑹𝑿𝑿 (𝝉) = 𝒆−𝟐|𝝉| . Find the mean 𝝁𝒀 and
𝟏
power spectrum 𝑺𝒀𝒀 (𝝎) of the output if the system transfer function is given by 𝑯(𝝎) = 𝝎+𝟐𝒊

Solution:

Given 𝜇𝑥 = 0

𝑅𝑋𝑋 (𝜏) = 𝑒 −2|𝜏|

1
𝐻(𝜔) =
𝜔+2𝑖

We know that for a linear time invariant system

𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡)

∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢

∞
Mean 𝐸[𝑌(𝑡)] = 𝐸[∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢]

∞
𝐸[𝑌(𝑡)] = ∫−∞ ℎ(𝑢)𝐸[𝑋(𝑡 − 𝑢)]𝑑𝑢

Given 𝐸[𝑋(𝑡)} = 𝜇𝑥 = 0 𝑋(𝑡) is stationary for al t

𝐸[𝑋(𝑡 − 𝑢)] = 𝜇𝑥 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

∞ ∞
𝐸[𝑌(𝑡)] = ∫−∞ ℎ(𝑢) 𝜇𝑥 𝑑𝑢 = ∫−∞ ℎ(𝑢) 0 𝑑𝑢 = 0

Mean of 𝑌(𝑡) is zero

11 | P a g e

Power spectral density of the in put 𝑋(𝑡) is

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

Given 𝑅𝑋𝑋 (𝜏) = 𝑒 −2|𝜏|

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑒 −2|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏 -

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑒 −2|𝜏| (𝑐𝑜𝑠𝜔𝜏 − 𝑖𝑠𝑖𝑛𝜔𝜏) 𝑑𝜏

∞ ∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑒 −2|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 −𝑖 ∫−∞ 𝑒 −2|𝜏| 𝑠𝑖𝑛𝜔𝜏 𝑑𝜏

Even Odd

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑒 −2|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 - 0

∞ ∞ 𝑎
𝑆𝑋𝑋 (𝜔) = 2 ∫0 𝑒 −2𝜏 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 ∴ ∫0 𝑒 −𝑎𝑥 𝑐𝑜𝑠𝑏𝑥 𝑑𝑥 = 𝑎2 +𝑏2

2 4
=2 =
22 +𝜔2 4+𝜔2

1 𝜔−2𝑖 𝜔−2𝑖
Given 𝐻(𝜔) = 𝜔+2𝑖 = (𝜔+2𝑖)(𝜔−2𝑖) = (𝜔2 +22 )

𝜔2 +22 1
|𝐻(𝜔)|2 = 2 =
(𝜔 +22 )2 (𝜔2 +22 )

Power spectral density of the output 𝑌(𝑡) is 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

1 4 4
𝑆𝑌𝑌 (𝜔) = (𝜔2 =
+4) 4+𝜔2 (4+𝜔2 )(4+𝜔2 )

4
𝑆𝑌𝑌 (𝜔) = (4+𝜔2 )2

𝟏
𝟎<𝒕<𝑻
9. A circuit has unit impulse response given by 𝒉(𝒕) = { 𝑻 Evaluate 𝒔𝒀𝒀 (𝝎) in terms of
𝟎 𝒆𝒍𝒔𝒆𝒘𝒉𝒆𝒓𝒆
𝒔𝑿𝑿 (𝝎)
Solution:
Let 𝑋(𝑡) be the input and 𝑌(𝑡) be the output, the power spectral density of out put 𝑌(𝑡)
𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

𝐻(𝜔) is the Fourier transform of the functions ℎ(𝑡)

∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡

𝑇1
= ∫0 𝑇
𝑒 −𝑖𝜔𝑡 𝑑𝑡
12 | P a g e

1 𝑇
= 𝑇 ∫0 𝑒 −𝑖𝜔𝑡 𝑑𝑡

1 𝑇
= 𝑇 ∫0 (𝑐𝑜𝑠𝜔𝑡 − 𝑖𝑠𝑖𝑛𝜔𝑡) 𝑑𝑡

1 𝑠𝑖𝑛𝜔𝑡 𝑐𝑜𝑠𝜔𝑡 𝑇
= 𝑇[ 𝜔
+𝑖 𝜔
]
0

1 𝑠𝑖𝑛𝜔𝑇 𝑐𝑜𝑠𝜔𝑇 1
= ( +𝑖 −0−𝑖 )
𝑇 𝜔 𝜔 𝜔

1
𝐻(𝜔) = 𝜔𝑇 (𝑠𝑖𝑛𝜔𝑇 + 𝑖(𝑐𝑜𝑠𝜔𝑇 − 1))

1
|𝐻(𝜔)|2 = 2 2 (sin2 𝜔𝑇 + (𝑐𝑜𝑠𝜔𝑇 − 1)2 )
𝜔 𝑇

1
|𝐻(𝜔)|2 = 2 2 (sin2 𝜔𝑇 + 𝑐𝑜𝑠 2 𝜔𝑇 + 1 − 2 cos 𝜔𝑇)
𝜔 𝑇

1
|𝐻(𝜔)|2 = 2 2 (1 + 1 − 2 cos 𝜔𝑇)
𝜔 𝑇

1
|𝐻(𝜔)|2 = 2 2 (2 − 2 cos 𝜔𝑇)
𝜔 𝑇

𝜔𝑇 𝜔𝑇 𝜔𝑇 𝜔𝑇 2
2 2 2(sin2 ) 4sin2 sin2 𝑠𝑖𝑛
|𝐻(𝜔)|2 = (1 − cos 𝜔𝑇) = 2
= 2
= (𝜔𝑇)2
2
=[ 𝜔𝑇
2
]
𝜔2 𝑇 2 𝜔2 𝑇 2 𝜔2 𝑇 2
4 2

Power spectral density of the output 𝑌(𝑡) is 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

𝜔𝑇 2
𝑠𝑖𝑛
𝑠𝑌𝑌 (𝜔) = [ 𝜔𝑇
2
] 𝑠𝑋𝑋 (𝜔)
2

10. Let 𝑿(𝒕)be the input voltage to a circuit system and 𝒀(𝒕) be the output voltage .If 𝑿(𝒕) is stationary
Random process with mean zero 𝝁𝒙 = 𝟎. And Auto correlation function 𝑹𝑿𝑿 (𝝉) = 𝒆−𝜶|𝝉| .Find (1)
𝑹
𝑬[𝒀(𝒕)] (2) 𝑺𝑿𝑿 (𝝎) (3) The spectral density of 𝒀(𝒕) If the power Transfer function 𝑯(𝝎) = (4)
𝑹+𝒊𝑳𝝎

auto correlation of 𝒀(𝒕)

Solution:

We know that input and output are related by

𝑌(𝑡) = ℎ(𝑡) ∗ 𝑋(𝑡) (Convolution)

∞
𝑌(𝑡) = ∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢 ------ (1)

∞
𝐸[𝑌(𝑡)] = 𝐸[∫−∞ ℎ(𝑢)𝑋(𝑡 − 𝑢)𝑑𝑢]

∞
= ∫−∞ ℎ(𝑢)𝐸[𝑋(𝑡 − 𝑢)]𝑑𝑢

Since 𝑋(𝑡) is a Stationary with zero mean for any time t

13 | P a g e

𝜇𝑥 = 𝐸[𝑋(𝑡)] = 0

𝐸[𝑋(𝑡 − 𝑢)] = 0

∞
𝐸[𝑌(𝑡)] = ∫−∞ ℎ(𝑢) 0 𝑑𝑢 = 0

𝐸[𝑌(𝑡)] = 0 Mean of 𝑌(𝑡) is zero 𝜇𝑦 = 𝐸[𝑌(𝑡)] = 0

(2) ⟹ Given 𝑅𝑋𝑋 (𝜏) = 𝑒 −𝛼|𝜏| .

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞
= ∫−∞ 𝑒 −𝛼|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏 -

∞
= ∫−∞ 𝑒 −𝛼|𝜏| (𝑐𝑜𝑠𝜔𝜏 − 𝑖𝑠𝑖𝑛𝜔𝜏) 𝑑𝜏

∞ ∞
= ∫−∞ 𝑒 −𝛼|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 −𝑖 ∫−∞ 𝑒 −𝛼|𝜏| 𝑠𝑖𝑛𝜔𝜏 𝑑𝜏

Even Odd

∞
= ∫−∞ 𝑒 −𝛼|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 - 0

∞ ∞ 𝑎
= 2 ∫0 𝑒 −𝛼|𝜏| 𝑐𝑜𝑠𝜔𝜏 𝑑𝜏 ∴ ∫0 𝑒 −𝑎𝑥 𝑐𝑜𝑠𝑏𝑥 𝑑𝑥 =
𝑎 2 +𝑏2

𝛼
=2 𝛼 2 +𝜔2

2𝛼
𝑆𝑋𝑋 (𝜔) =
𝛼 2 +𝜔2

𝑅 𝑅(𝑅−𝑖𝐿𝜔) 𝑅(𝑅−𝑖𝐿𝜔)
3) ⟹ Given 𝐻(𝜔) = 𝑅+𝑖𝐿𝜔 = (𝑅+𝑖𝐿𝜔)(𝑅−𝑖𝐿𝜔) = 𝑅2 +𝐿2 𝜔2

𝑅 2 (𝑅2 +(𝐿𝜔)2 ) 𝑅2
|𝐻(𝜔)|2 = (𝑅2 +𝐿2 𝜔2 )2
= (𝑅2 +𝐿2 𝜔2 )

Power spectral density of the output 𝑌(𝑡) is 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)

𝑅2
𝑆𝑌𝑌 (𝜔) = (𝑅2 +𝐿2 𝜔2 ) 𝑠𝑋𝑋 (𝜔)
𝑅2 2𝛼 2𝑅2 𝛼
𝑆𝑌𝑌 (𝜔) = (𝑅2 +𝐿2 𝜔2 ) 𝛼 2 +𝜔2
= (𝑅2 +𝐿2 𝜔2 )(𝛼2 +𝜔2 )

4) ⟹ Auto correlation of 𝑌(𝑡)

1 ∞
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑌𝑌 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 2𝑅2 𝛼
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ (𝑅2 +𝐿2 𝜔2 )(𝛼2 +𝜔2 ) 𝑒 𝑖𝜔𝜏 𝑑𝜔

2𝑅2 𝛼 ∞ 𝑒 𝑖𝜔𝜏
𝑅𝑌𝑌 (𝜏) = ∫ 𝑑𝜔
2𝜋 −∞ (𝑅 +𝐿 𝜔2 )(𝛼2 +𝜔2 )
2 2
14 | P a g e

Method of partial fraction

1 𝐴 𝐵
(𝑅2 +𝐿2 𝜔2 )(𝛼 2 +𝜔2 )
= (𝑅2 +𝐿2 𝜔2 )
+ (𝛼2 +𝜔2 )

1 = 𝐴(𝛼 2 + 𝜔2 ) + 𝐵(𝑅 2 + 𝐿2 𝜔2 )
Put 𝜔2 = − 𝛼 2 ⇒ 1 = 𝐵(𝑅 2 − 𝐿2 𝛼 2 )
1
⇒ 𝐵 = (𝑅2 −𝐿2 𝛼2 )

𝑅2 𝑅2
Put 𝜔2 = − ⇒ 1 = 𝐴 (𝛼 2 − )
𝐿2 𝐿2
1
⇒ 𝐴= 𝑅 2
(𝛼 2 − 2 )
𝐿

2𝑅2 𝛼 ∞ 𝑒 𝑖𝜔𝜏 2𝑅2 𝛼 ∞ 𝑒 𝑖𝜔𝜏

𝑅𝑌𝑌 (𝜏) = 𝑅2
∫−∞ (𝑅2 +𝐿2 𝜔2 ) 𝑑𝜔 + 2𝜋(𝑅2 −𝐿2 𝛼2 ) ∫−∞ (𝛼2 +𝜔2 ) 𝑑𝜔
2𝜋(𝛼 2 − 2 )
𝐿

2𝑅2 𝛼 ∞ 𝑒 𝑖𝜔𝜏 2𝑅2 𝛼 ∞ 𝑒 𝑖𝜔𝜏

= 𝑅2
∫−∞ 𝑅 2
𝑑𝜔 + 2𝜋(𝑅2 −𝐿2 𝛼2 ) ∫−∞ (𝛼2 +𝜔2 ) 𝑑𝜔
2𝜋𝐿2 (𝛼 2 − 2 ) (( ) +𝜔2 )
𝐿 𝐿

𝑅
𝑅2 𝛼 1 𝑅2 𝛼 1 ∞ 𝑒 𝑖𝑚𝑧 𝜋
= 𝑅2 𝑅
𝑒 −|𝜏| 𝐿 + (𝑅2 𝑒 −|𝜏|𝛼 ∫−∞ 𝑧2 +𝑎2 𝑑𝑧 = 𝑎 𝑒 −|𝑚|𝑎
𝐿2 (𝛼 2 − 2 ) 𝐿 −𝐿2 𝛼2 ) 𝛼
𝐿

𝑅
𝑅𝛼 𝑅2
= 𝑅2
𝑒 −|𝜏| 𝐿 + (𝑅2 −𝐿2 𝛼2 ) 𝑒 −|𝜏|𝛼
𝐿(𝛼 2 − 2 )
𝐿

𝑵𝟎
11. Consider the white noise of zero mean and power spectral density 𝟐
applied to a low pass RC filter where
𝟏
transfer function 𝑯(𝒇) = 𝟏+𝟐𝝅𝒊𝒇𝑹𝑪 .Find the output spectral density and auto correlation function of the

output.
Solution:
1
Given 𝐻(𝑓) = 1+2𝜋𝑖𝑓𝑅𝐶 𝜔 = 2𝜋𝑓
1
Given 𝐻(𝜔) = 1+𝑖𝜔𝑅𝐶 𝜔 = 2𝜋𝑓 Angular function
𝑁0
Spectral density function of the input of filter is 2
= 𝑆𝑋𝑋 (𝜔)

Relation between the input and output spectral density is given by

𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)
1 1−𝑖𝜔𝑅𝐶 1−𝑖𝜔𝑅𝐶
𝐻(𝜔) = 1+𝑖𝜔𝑅𝐶 = (1+𝑖𝜔𝑅𝐶)(1− 𝑖𝜔𝑅𝐶) = (1+𝜔2 𝑅2 𝐶 2 )

𝜔2 𝑅 2 𝐶 2 1
|𝐻(𝜔)| = (1+𝜔2 =
𝑅2 𝐶 2 )2 1+𝜔2 𝑅2 𝐶 2

Output spectral density is given by 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

1 𝑁0
𝑆𝑌𝑌 (𝜔) = 1+𝜔2 𝑅2 𝐶 2 2
1 ∞
Auto correlation of the output 𝑅𝑌𝑌 (𝜏) = ∫ 𝑆 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
2𝜋 −∞ 𝑌𝑌
1 ∞ 1 𝑁0
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 1+𝜔2 𝑅2 𝐶 2 2
𝑒 𝑖𝜔𝜏 𝑑𝜔
1 𝑁0 ∞ 1
= 2𝜋 2
∫−∞ 1+𝜔2 𝑅2 𝐶 2 𝑒 𝑖𝜔𝜏 𝑑𝜔
15 | P a g e

𝑁0 ∞ 𝑒 𝑖𝜔𝜏 𝑁 ∞ 𝑒 𝑖𝜔𝜏
= 4𝜋
∫−∞ 1+𝜔2 𝑅2 𝐶 2 𝑑𝜔 = 4𝜋𝑅20𝐶 2 ∫−∞ 1 𝑑𝜔
( 2 2+𝜔2 )
𝑅 𝐶

𝑁0 𝜋 1 ∞ 𝑒 𝑖𝑚𝑧 𝜋
= 4𝜋𝑅2 𝐶 2 1
𝑒 −|𝜏| 𝑅𝐶
∫−∞ 𝑧2 +𝑎2 𝑑𝑧 = 𝑎 𝑒 −|𝑚|𝑎
𝑅𝐶

𝑁0 |𝜏|
= 4𝑅𝐶
𝑒− 𝑅𝐶

This is the auto correlation function of the output function.

𝑵𝟎
12. A White Gaussian noise 𝑋(𝑡) with zero mean and spectral density 𝟐
is applied to low-pass RC filter shown in

the figure

Determine the auto correlation of the output 𝑌(𝑡)

Solution
1
The transfer function of the Rc filter is given by 𝐻(𝜔) =
1+𝑖𝜔𝑅𝐶
𝑁0
Spectral density function of the input of filter is 2
= 𝑆𝑋𝑋 (𝜔)

Relation between the input and output spectral density is given by

𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)
1 1−𝑖𝜔𝑅𝐶 1−𝑖𝜔𝑅𝐶
𝐻(𝜔) = 1+𝑖𝜔𝑅𝐶 = (1+𝑖𝜔𝑅𝐶)(1− 𝑖𝜔𝑅𝐶) = (1+𝜔2 𝑅2 𝐶 2 )

𝜔2 𝑅 2 𝐶 2 1
|𝐻(𝜔)| = (1+𝜔2 =
𝑅2 𝐶 2 )2 1+𝜔2 𝑅2 𝐶 2

Output spectral density is given by 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

1 𝑁0
𝑆𝑌𝑌 (𝜔) = 1+𝜔2 𝑅2 𝐶 2 2
1 ∞
Auto correlation of the output 𝑅𝑌𝑌 (𝜏) = ∫ 𝑆 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
2𝜋 −∞ 𝑌𝑌
1 ∞ 1 𝑁0
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 1+𝜔2 𝑅2 𝐶 2 2
𝑒 𝑖𝜔𝜏 𝑑𝜔
1 𝑁0 ∞ 1
= 2𝜋 2
∫−∞ 1+𝜔2 𝑅2 𝐶 2 𝑒 𝑖𝜔𝜏 𝑑𝜔
𝑁0 ∞ 𝑒 𝑖𝜔𝜏 𝑁 ∞ 𝑒 𝑖𝜔𝜏
= 4𝜋
∫−∞ 1+𝜔2 𝑅2 𝐶 2 𝑑𝜔 = 4𝜋𝑅20𝐶 2 ∫−∞ 1 𝑑𝜔
( 2 2+𝜔2 )
𝑅 𝐶

𝑁0 𝜋 1 ∞ 𝑒 𝑖𝑚𝑧 𝜋
= 4𝜋𝑅2 𝐶 2 1
𝑒 −|𝜏| 𝑅𝐶
∫−∞ 𝑧2 +𝑎2 𝑑𝑧 = 𝑎 𝑒 −|𝑚|𝑎
𝑅𝐶

𝑁0 |𝜏|
= 4𝑅𝐶
𝑒− 𝑅𝐶

This is the auto correlation function of the output function.

16 | P a g e

−𝑡
1
13. A linear system is described by the impulse response ℎ(𝑡) = 𝑒 𝑅𝑐 𝑢(𝑡) .Assume an input signal whose
𝑅𝑐
autocorrelation function is 𝐵𝛿(𝜏).Find the autocorrelation mean and power of the output.
Solution:
Let 𝑋(𝑡) be the input and 𝑌(𝑡) be the output, the power spectral density of out put 𝑌(𝑡)
𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑆𝑋𝑋 (𝜔)
0 𝑡<0
We know that unit step function 𝑢(𝑡) = {
1 𝑡≥0
−𝑡
1
Given ℎ(𝑡) = 𝑒 𝑅𝑐 𝑢(𝑡)
𝑅𝑐
0 𝑡<0 1
ℎ(𝑡) = { 1 −𝑡
𝛽 = 𝑅𝐶
𝑒 𝑅𝑐 𝑡≥0
𝑅𝑐
0 𝑡<0
ℎ(𝑡) = { −𝑡
𝛽𝑒 𝛽 𝑡≥0
𝐻(𝜔) is the Fourier transform of the function ℎ(𝑡)
∞
𝐻(𝜔) = ∫−∞ ℎ(𝑡)𝑒 −𝑖𝜔𝑡 𝑑𝑡
−𝑡
∞
𝐻(𝜔) = ∫0 𝑒 −𝛽𝑡 𝑒 −𝑖𝜔𝑡 𝑑𝑡 since ℎ(𝑡) = 𝛽𝑒 𝛽 𝑡≥0

∞
𝐻(𝜔) = 𝛽 ∫0 𝑒 −(𝛽+𝑖𝜔)𝑡 𝑑𝑡

∞
𝑒 −(𝛽+𝑖𝜔)𝑡
= 𝛽[ ]
−(𝛽+𝑖𝜔)
0

𝛽 𝛽
= 0 − −(𝛽+𝑖𝜔) = (𝛽+𝑖𝜔)

𝛽 𝛽(𝛽−𝑖𝜔) 𝛽(𝛽−𝑖𝜔)
𝐻(𝜔) = = =
(𝛽+𝑖𝜔) (𝛽+𝑖𝜔)(𝛽−𝑖𝜔) (𝛽 2 +𝜔2 )

𝛽 2 (𝛽 2 +𝜔2 ) 𝛽2
|𝐻(𝜔|2 = (𝛽 2 +𝜔2 )2
= 𝛽 2 +𝜔2

Power spectral density of the input 𝑋(𝑡) is

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞ 𝛼
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝐵 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 ∫−𝛼 𝛿(𝜏)𝜙(𝜏)𝑑𝜏 = 𝜙(0)

∞
= 𝐵 ∫−∞ 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 here 𝜙(𝜏) = 𝑒 −𝑖𝜔𝜏 ⇒ 𝜙(0) = 1

= 𝐵(1)

Spectral relation between input and output process 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

𝛽2 𝐵𝛽 2
𝑠𝑌𝑌 (𝜔) = 𝛽2 +𝜔2 (𝐵) = 𝛽2 +𝜔2
17 | P a g e

Auto correlation of the output 𝑌(𝑡) is

1 ∞
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑌𝑌 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 𝐵𝛽 2
= 2𝜋 ∫−∞ 𝛽2 +𝜔2 𝑒 𝑖𝜔𝜏 𝑑𝜔

𝐵𝛽 2 ∞ 𝑒 𝑖𝜔𝜏 ∞ 𝑒 𝑖𝑚𝑧 𝜋
= ∫
2𝜋 −∞ 𝛽 2 +𝜔2
𝑑𝜔 ∫−∞ 𝑧 2+𝑎2 𝑑𝑧 = 𝑎 𝑒 −|𝑚|𝑎

𝐵𝛽 2 𝜋
= 2𝜋 𝛽
𝑒 −|𝜏|𝛽
𝐵𝛽 −|𝜏|𝛽
𝑅𝑌𝑌 (𝜏) = 2
𝑒

14. If 𝒀(𝒕) = 𝑨 𝐜𝐨𝐬( 𝝎𝟎 𝒕 + 𝜽) + 𝑵(𝒕) , where 𝑨 is a constant 𝜽 a random variable with a uniform distribution
in (−𝝅, 𝝅) and 𝑵(𝒕) is a band limited Gaussian white noise with a power spectral density𝑺𝑵𝑵 (𝝎) =
𝑵𝟎
𝒇𝒐𝒓|𝝎 − 𝝎𝟎 | < 𝝎𝜷
{ 𝟐 . Find the power spectral density of 𝒀(𝒕). Assume that 𝑵(𝒕) and 𝜽 are
𝟎 𝒆𝒍𝒔𝒆 𝒘𝒉𝒆𝒓𝒆
independent.
Solution:

Given 𝑌(𝑡) = 𝐴 cos( 𝜔0 + 𝜃) + 𝑁(𝑡)

𝑁(𝑡) is a band limited Gaussian white noise with a power spectral density
𝑁0
for |𝜔 − 𝜔0 | < 𝜔𝛽
𝑆𝑁𝑁 (𝜔) = { 2 .
0 else where

𝑁0
for − 𝜔𝛽 < (𝜔 − 𝜔0 ) < 𝜔𝛽
𝑆𝑁𝑁 (𝜔) = { 2
0 else where

𝑁0
for 𝜔0 − 𝜔𝛽 < 𝜔 < 𝜔0 + 𝜔𝛽
𝑆𝑁𝑁 (𝜔) = { 2
0 else where

𝑆𝑌𝑌 (𝜔) = Fourier transform of 𝑅𝑌𝑌 (𝜏)

∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ 𝑅𝑌𝑌 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

𝑅𝑌𝑌 (𝜏) = 𝐸[𝑌(𝑡)𝑌(𝑇 + 𝜏)]

𝑌(𝑡) = 𝐴 cos( 𝜔0 𝑡 + 𝜃) + 𝑁(𝑡)

𝑌(𝑡 + 𝜏) = 𝐴 cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) + 𝑁(𝑡 + 𝜏)

𝐸[𝑌(𝑡)𝑌(𝑇 + 𝜏)] = 𝐸[𝐴 cos( 𝜔0 𝑡 + 𝜃) + 𝑁(𝑡)] [𝐴 cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) + 𝑁(𝑡 + 𝜏)]

= 𝐸[𝐴2 cos( 𝜔0 𝑡 + 𝜃) cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) + 𝐴 𝑁(𝑡) cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) +

𝐴 cos( 𝜔0 𝑡 + 𝜃)𝑁(𝑡 + 𝜏) + 𝑁(𝑡) 𝑁(𝑡 + 𝜏)]
18 | P a g e

𝐴2
= 2
𝐸[ 2 cos( 𝜔0 𝑡 + 𝜃) cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃)] + 𝐴 𝐸[𝑁(𝑡) cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃)] +

𝐴 E[cos( 𝜔0 𝑡 + 𝜃)𝑁(𝑡 + 𝜏)] + 𝐸[𝑁(𝑡) 𝑁(𝑡 + 𝜏)]

2𝑐𝑜𝑠𝐴𝑐𝑜𝑠𝐵 = cos(𝐴 + 𝐵) + cos(𝐴 − 𝐵)

=
𝐴2 𝐴2
2
𝐸[cos(2 𝜔0 𝑡 + 2𝜃 + 𝜔0 𝜏)] + 2
𝐸[cos 𝜔0 𝜏] + 𝐴𝐸[𝑁(𝑡) ]AE [cos( 𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃)] + 𝐴 E[cos( 𝜔0 𝑡 +

𝜃)] 𝐴𝐸[𝑁(𝑡 + 𝜏)] + 𝐸[𝑁(𝑡) 𝑁(𝑡 + 𝜏] -------- (A)

Now 𝜃 a Random variable with a uniform distribution in (−𝜋, 𝜋) the probability distribution of 𝜃 if given by
1
𝑓(𝜃) = −𝜋 <𝜃 <𝜋
2𝜋

𝜋
E[cos( 𝜔0 𝑡 + 𝜃)] = ∫−𝜋 cos(𝜔0 𝑡 + 𝜃) 𝑓(𝜃) 𝑑𝜃

1 𝜋
= 2𝜋 ∫−𝜋 cos(𝜔0 𝑡 + 𝜃) 𝑑𝜃

1 𝜋
= [sin(𝜔0 𝑡 + 𝜃) ] −𝜋 =0 − − − − −(1)
2𝜋

𝜋
E[cos( 𝜔0 𝑡 + 𝜔𝜏 + 𝜃)] ∫−𝜋 cos(𝜔0 𝑡 + 𝜔𝜏 + 𝜃) 𝑓(𝜃) 𝑑𝜃

1 𝜋
= 2𝜋 ∫−𝜋 cos(𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) 𝑑𝜃

1
= 2𝜋 [ sin(𝜔0 𝑡 + 𝜔0 𝜏 + 𝜃) ]𝜋−𝜋 = 0 − − − − −(2)

𝜋
E[cos(2 𝜔0 𝑡 + 𝜔0 𝜏 + 2𝜃)] = ∫−𝜋 cos(2𝜔0 𝑡 + 𝜔0 𝜏 + 2𝜃) 𝑓(𝜃) 𝑑𝜃

1 𝜋
= ∫−𝜋 cos(2𝜔0 𝑡 + 𝜔0 𝜏 + 2𝜃) 𝑑𝜃
2𝜋

1 sin(2𝜔0 𝑡+𝜔0 𝜏+2𝜃) 𝜋

= 2𝜋 [ 2
] = 0 − − − − −(3)
−𝜋

W K T R YY (τ) = E[Y(t)Y(T + τ)]

Similarly

R NN (τ) = E[N(t)N(T + τ)] − − − − − (4)

Put (1) , (2), (3) and (4) by (A)

𝐴2
𝑅𝑌𝑌 (𝜏) = 2
cos 𝜔0 𝜏 + 𝑅𝑁𝑁 (𝜏)

∞
𝑆𝑌𝑌 (𝜔) = ∫−∞ 𝑅𝑌𝑌 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏
19 | P a g e

∞ 𝐴2
𝑆𝑌𝑌 (𝜔) = ∫−∞ ( cos 𝜔0 𝜏 + 𝑅𝑁𝑁 (𝜏)) 𝑒 −𝑖𝜔𝜏 𝑑𝜏
2

𝐴2 ∞ ∞
𝑆𝑌𝑌 (𝜔) = ∫
2 −∞
cos 𝜔0 𝜏 𝑒 −𝑖𝜔𝜏 𝑑𝜏 +∫−∞ 𝑅𝑁𝑁 (𝜏) 𝑒 −𝑖𝜔𝜏 𝑑𝜏

𝐴2 ∞ 𝑒 𝑖𝜏𝜔𝑜 +𝑒 −𝑖𝜏𝜔𝑜 ∞
𝑆𝑌𝑌 (𝜔) = ∫
2 −∞ 2
𝑒 −𝑖𝜔𝜏 𝑑𝜏 +∫−∞ 𝑅𝑁𝑁 (𝜏) 𝑒 −𝑖𝜔𝜏 𝑑𝜏

𝐴2 ∞ −𝑖𝜔𝜏 𝑖𝜏𝜔
= ∫ 𝑒
4 −∞
𝑒 𝑜 + 𝑒 −𝑖𝜔𝜏 𝑒 −𝑖𝜏𝜔𝑜 𝑑𝜏 + 𝑆𝑁𝑁 (𝜔)

𝐴2 ∞ −(𝜔−𝜔 )𝑖𝜏
= ∫ 𝑒
4 −∞
𝑜 + 𝑒 −𝑖(𝜔+𝜔𝑜 )𝜏 𝑑𝜏 + 𝑆𝑁𝑁 (𝜔)

𝜋𝐴2 1 ∞ −(𝜔−𝜔 )𝑖𝜏 𝜋𝐴2 1 ∞ −𝑖(𝜔+𝜔 )𝜏

= ∫ 𝑒 𝑜 𝑑𝜏 + ∫ 𝑒 𝑜 𝑑𝜏 + 𝑆𝑁𝑁 (𝜔)
2 2𝜋 −∞ 2 2𝜋 −∞

𝜋𝐴2 𝜋𝐴2 𝑁
= 2
𝛿(𝜔 − 𝜔0 ) + 2
𝛿(𝜔 + 𝜔0 ) + 20
𝑁0 1 ∞ −𝑖𝜔𝜏
𝑠𝑁𝑁 (𝜔) = ∫ 𝑒 𝑑𝜏 = 𝛿(𝜔)
2 2𝜋 −∞

15. A wide sense stationary noise process 𝑵(𝒕) has an auto correlation function 𝑹𝑵𝑵 (𝝉) = 𝝆𝒆−𝟑|𝝉| where 𝝆 is
constant. Find the power spectrum.
Solution: Given auto correlation of noise process 𝑁(𝑡) is 𝑅𝑁𝑁 (𝜏) = 𝜌𝑒 −3|𝜏|

∞
𝑆𝑁𝑁 (𝜔) = ∫−∞ 𝑅𝑁𝑁 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞
𝑆𝑁𝑁 (𝜔) = ∫−∞ 𝜌𝑒 −3|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞
= ∫−∞ 𝜌𝑒 −3|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞
= ∫−∞ 𝜌𝑒 −3|𝜏| (cos 𝜔𝜏 − sin 𝜔𝜏)𝑑𝜏

∞ ∞
= ∫−∞ 𝜌𝑒 −3|𝜏| cos 𝜔𝜏 𝑑𝜏 − ∫−∞ 𝜌𝑒 −3|𝜏| sin 𝜔𝜏 𝑑𝜏

∞ ∞ 𝑎
= 2𝜌 ∫0 𝑒 −3|𝜏| cos 𝜔𝜏 𝑑𝜏 − 0 ∴ ∫0 𝑒 −𝑎𝑥 𝑐𝑜𝑠𝑏𝑥 𝑑𝑥 = 𝑎2 +𝑏2

3
= 2𝜌 32 +𝜔2

6𝜌
𝑆𝑋𝑋 (𝜔) = 32 +𝜔2

16. Find the power spectral density of the random telegraph signal.
Solution:
The Auto correlation function of the random telegraph signal 𝑅𝑋𝑋 (𝜏) = 𝑒 −2𝜆|𝜏|

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏
20 | P a g e

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑒 −2𝜆|𝜏| 𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞
= ∫−∞ 𝑒 −2𝜆|𝜏| (cos 𝜔𝜏 − sin 𝜔𝜏)𝑑𝜏

∞ ∞
= ∫−∞ 𝑒 −2𝜆|𝜏| cos 𝜔𝜏 𝑑𝜏 − ∫−∞ 𝑒 −2𝜆|𝜏| sin 𝜔𝜏 𝑑𝜏

∞ ∞ 𝑎
= 2 ∫0 𝑒 −2𝜆𝜏 cos 𝜔𝜏 𝑑𝜏 − 0 ∴ ∫0 𝑒 −𝑎𝑥 𝑐𝑜𝑠𝑏𝑥 𝑑𝑥 =
𝑎2 +𝑏 2

4𝜆
= 4𝜆2 +𝜔2

17. Assume a random process 𝑿(𝒕) is given as input to a system with transfer function 𝑯(𝝎) = 𝟏 for
𝑵𝟎
−𝝎𝟎 < 𝒙 < 𝝎𝟎 . If the auto correlation function of the input process is 𝜹(𝝉).Find the auto correlation
𝟐

function of the output process

Solution:

𝑵𝟎
Given 𝑅𝑋𝑋 = 𝜹(𝝉)
𝟐

𝐻(𝜔) = 1 for −𝝎𝟎 < 𝒙 < 𝝎𝟎

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞ 𝑅𝑋𝑋 (𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞ 𝑁0
= ∫−∞ 2
𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏

𝑁0 ∞ 𝛼
= 2
∫−∞ 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 ∫−𝛼 𝛿(𝜏)𝜙(𝜏)𝑑𝜏 = 𝜙(0)

𝑁0
𝑆𝑋𝑋 (𝜔) = 2
(1) here 𝜙(𝜏) = 𝑒 −𝑖𝜔𝜏 ⇒ 𝜙(0) = 1

Power spectral density of 𝑌(𝑡) 𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

𝑁0
𝑠𝑌𝑌 (𝜔) =
2
1 ∞
Output of auto correlation function 𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑌𝑌 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 𝑁0 𝑖𝜔𝜏
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 2
𝑒 𝑑𝜔
𝑁 𝜔
= 4𝜋0 ∫−𝜔0 𝑒 𝑖𝜔𝜏 𝑑𝜔
0

𝑁 𝜔
= 4𝜋0 ∫−𝜔0 cos 𝜔𝜏 + sin 𝜔𝜏 𝑑𝜔
0

𝑁0 𝜔0 𝑁 𝜔
= ∫ cos 𝜔𝜏
4𝜋 −𝜔0
𝑑𝜔 +4𝜋0 ∫−𝜔0 sin 𝜔𝜏 𝑑𝜔
0
2𝑁0 𝜔0
= ∫ cos 𝜔𝜏
4𝜋 0
𝑑𝜔 +0
𝑁 sin 𝜔𝜏 𝜔0
= 2𝜋0 [ 𝜏
]
0
𝑁0 sin 𝜔0 𝜏
𝑅𝑌𝑌 (𝜏) = 2𝜋𝜏
21 | P a g e

18. Suppose 𝑋(𝑡) be the input process to a linear system with auto correlation 𝑅𝑋𝑋 (𝜏) = 3 𝛿(𝜏) , and the impulse
1
response 𝐻(𝜔) = 6+𝑖𝜔 , then find the (1) auto correlation of the output process 𝑌(𝑡) (2).Power spectral density

𝑌(𝑡) .
Solution:
Given auto correlation𝑅𝑋𝑋 (𝜏) = 3 𝛿(𝜏)
1 6−𝑖𝜔 6−𝑖𝜔
𝐻(𝜔) = 6+𝑖𝜔 = (6+𝑖𝜔)(6−𝑖𝜔) = 62 +𝜔2
1
|𝐻(𝜔)|2 = 2 2
6 +𝜔

∞ 𝛼
𝑆𝑋𝑋 (𝜔) = ∫−∞ 3 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 ∫−𝛼 𝛿(𝜏)𝜙(𝜏)𝑑𝜏 = 𝜙(0)

∞
= 3 ∫−∞ 𝛿(𝜏)𝑒 −𝑖𝜔𝜏 𝑑𝜏 here 𝜙(𝜏) = 𝑒 −𝑖𝜔𝜏 ⇒ 𝜙(0) = 1

= 3(1)

Spectral relation between input and output process 𝑆𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

1 3
𝑠𝑌𝑌 (𝜔) = 62 +𝜔2 (3) = 62 +𝜔2

Auto correlation of the output of 𝑌(𝑡) is

1 ∞
𝑅𝑌𝑌 (𝜏) = ∫ 𝑆 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
2𝜋 −∞ 𝑌𝑌
1 ∞ 3
= ∫ 𝑒 𝑖𝜔𝜏 𝑑𝜔
2𝜋 −∞ 62 +𝜔2

3 ∞ 𝑒 𝑖𝜔𝜏 ∞ 𝑒 𝑖𝑚𝑧 𝜋
= 2𝜋 ∫−∞ 62 +𝜔2 𝑑𝜔 ∫−∞ 𝑧 2+𝑎2 𝑑𝑧 = 𝑎 𝑒 −|𝑚|𝑎

3 𝜋
= 2𝜋 6 𝑒 −|𝜏|6
1
𝑅𝑌𝑌 (𝜏) = 4 𝑒 −|𝜏|6
1
19. Consider a system with transfer function . An input signal with auto correlation function 𝑚δ(τ) + m2 is
1+𝑖𝜔

fed as input to the system. Find the mean and mean square value of the output.
Solution:
Given auto correlation 𝑅𝑋𝑋 (𝜏) = 𝑚δ(τ) + m2
1 1−𝑖𝜔 1−𝑖𝜔
𝐻(𝜔) = 1+𝑖𝜔 = (1+𝑖𝜔)(6−𝑖𝜔) = 1+𝜔2
1
|𝐻(𝜔)|2 =
1+𝜔2

Spectral density of the output

∞
𝑆𝑋𝑋 (𝜔) = ∫−∞(𝑚δ(τ) + m2 )𝑒 −𝑖𝜔𝜏 𝑑𝜏

∞ ∞
= 𝑚 ∫−∞ δ(τ)𝑒 −𝑖𝜔𝜏 𝑑𝜏 + 𝑚2 ∫−∞ 𝑒 −𝑖𝜔𝜏 𝑑𝜏
22 | P a g e

∞ 1 ∞
= 𝑚 ∫−∞ δ(τ)𝑒 −𝑖𝜔𝜏 𝑑𝜏 + 2𝜋 𝑚2 ∫ 𝑒 −𝑖𝜔𝜏 𝑑𝜏
2 𝜋 −∞
𝛼
∫−𝛼 𝛿(𝜏)𝜙(𝜏)𝑑𝜏 = 𝜙(0)

𝑆𝑋𝑋 (𝜔) = 𝑚(1) + 2𝜋𝑚2 𝛿(𝜔) here 𝜙(𝜏) = 𝑒 −𝑖𝜔𝜏 ⇒ 𝜙(0) = 1

1 ∞
∫ 𝑒 −𝑖𝜔𝜏
2 𝜋 −∞
𝑑𝜏 = δ(ω)

Spectral relation between input and output process 𝑠𝑌𝑌 (𝜔) = |𝐻(𝜔)|2 𝑠𝑋𝑋 (𝜔)

1
𝑠𝑌𝑌 (𝜔) = 1+𝜔2 (𝑚 + 2𝜋𝑚2 𝛿(𝜔))

𝑅𝑌𝑌 (𝜏) is the Fourier inverse transform of 𝑆𝑌𝑌 (𝜔)

1 ∞
𝑅𝑌𝑌 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑌𝑌 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 1
= 2𝜋 ∫−∞ 1+𝜔2 (𝑚 + 2𝜋𝑚2 𝛿(𝜔)) 𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 𝑚𝑒 𝑖𝜔𝜏 1 ∞
= 2𝜋 ∫−∞ 1+𝜔2 𝑑𝜔 +2𝜋 ∫−∞ 2𝜋𝑚2 𝛿(𝜔)) 𝑒 𝑖𝜔𝜏 𝑑𝜔
𝑚 ∞ 𝑒 𝑖𝜔𝜏 ∞
= 2𝜋 ∫−∞ 1+𝜔2 𝑑𝜔 +𝑚2 ∫−∞ 𝛿(𝜔)) 𝑒 𝑖𝜔𝜏 𝑑𝜔

𝑚 𝜋 ∞ 𝑒𝑖𝑚𝑧 𝜋 ∞
𝑅𝑌𝑌 (𝜏) =
2𝜋 1
𝑒 −|𝜏| + 𝑚2 (1) ∫−∞ 𝑧2 +𝑎2 𝑑𝑧 = 𝑎 𝑒−|𝑚|𝑎 ∫−∞ 𝛿(𝜔)) 𝑒𝑖𝜔𝜏 𝑑𝜔 = 1
𝑚 −|𝜏|
𝑅𝑌𝑌 (𝜏) = 2
𝑒 + 𝑚2

20. Find the auto correlation function of the Gaussian white noise.
𝑁0
Solution: The power spectral density of the Gaussian white noise is 𝑆𝑋𝑋 (𝜔) = 2
where 𝑁0 is the positive real
valued constant.
1 ∞
𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞ 𝑁0 𝑖𝜔𝜏
= 2𝜋 ∫−∞ 2
𝑒 𝑑𝜔
𝑁0 1 ∞ 𝑖𝜔𝜏
= ∫ 𝑒 𝑑𝜔
2 2 𝜋 −∞
𝑁0 1 ∞ 𝑖𝜔𝜏
= 2
𝛿(𝜏) ∫ 𝑒
2 𝜋 −∞
𝑑𝜔 = 𝛿(𝜏)

The system is time invariant. Hence the system is linear time invariant system

21. If 𝑿(𝒕) is a band limited process such that 𝑺𝑿𝑿 (𝝎) = 𝟎 when |𝝎| > 𝜎 prove that 𝟐[𝑹𝑿𝑿 (𝟎) −
𝑹𝑿𝑿 (𝝉)] ≤ 𝝈𝟐 𝝉𝟐 𝑹𝑿𝑿 (𝟎)
Solution: by definition of auto correlation
1 ∞
𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔)𝑒 𝑖𝜔𝜏 𝑑𝜔
1 ∞
𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔)(cos 𝜔𝜏 + 𝑖 sin 𝜔𝜏) 𝑑𝜔
1 ∞ 1 ∞
𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) cos 𝜔𝜏 𝑑𝜔 + 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) sin 𝜔𝜏 𝑑𝜔
1 ∞ 1 ∞
𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) cos 𝜔𝜏 𝑑𝜔 + 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) sin 𝜔𝜏 𝑑𝜔
23 | P a g e

1 ∞
= 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) cos 𝜔𝜏 𝑑𝜔 + 0 (odd function)
1 ∞
𝑅𝑋𝑋 (0) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) 𝑑𝜔 − − − (∗) cos 0 = 1

1 ∞
𝑅𝑋𝑋 (0) − 𝑅𝑋𝑋 (𝜏) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔)[1 − cos 𝜔𝜏] 𝑑𝜔 Since 𝑆𝑋𝑋 (𝜔) = 0 & |𝜔| > 𝜎

1 ∞ 𝜔𝜏
= 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) (2 sin2 2
) 𝑑𝜔 − − − − − (1)
𝑠𝑖𝑛𝜃
since | | ≤1 |𝑠𝑖𝑛𝜃| ≤ 𝜃
𝜃
∴ sin 𝜃 ≤ 𝜃 2
2

𝜔𝜏 𝜔2 𝜏 2 𝜔𝜏 𝜔2 𝜏 2
∴ sin2 2
≤ 4
⇒ 2 sin2 2
≤ 2
− − − − − (2) Substituting (2) in (1) we
get
1 ∞ 𝜔2 𝜏 2
𝑅𝑋𝑋 (0) − 𝑅𝑋𝑋 (𝜏) ≤ 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) 2
𝑑𝜔
𝜔2 𝜏 2 1 ∞
≤ 2
∫ 𝑆 (𝜔)
2𝜋 −∞ 𝑋𝑋
𝑑𝜔

𝜔2 𝜏 2 1 ∞
𝑅𝑋𝑋 (0) − 𝑅𝑋𝑋 (𝜏) ≤ 2
𝑅𝑋𝑋 (0) by 𝑅𝑋𝑋 (0) = 2𝜋 ∫−∞ 𝑆𝑋𝑋 (𝜔) 𝑑𝜔 − (∗)

2[𝑅𝑋𝑋 (0) − 𝑅𝑋𝑋 (𝜏)] ≤ 𝜎 2 𝜏 2 𝑅𝑋𝑋 (0)

24 | P a g e

Bernoulli’s Rule ∫ 𝑢 𝑑𝑣 = 𝑢𝑣 − 𝑢1 𝑣 / + 𝑢2 𝑣 // − 𝑢3 𝑣 // + ⋯.

𝑒 𝑎𝑥
1. ∫ 𝑒 𝑎𝑥 cos 𝑏𝑥 𝑑𝑥 = 2 2 [𝑎 𝑐𝑜𝑠𝑏𝑥 + 𝑏𝑠𝑖𝑛𝑏𝑥]
𝑎 +𝑏
𝑒 𝑎𝑥
2. ∫ 𝑒 𝑎𝑥 sin 𝑏𝑥 𝑑𝑥 = 2 2 [𝑎 𝑠𝑖𝑛𝑏𝑥 − 𝑏𝑐𝑜𝑠𝑏𝑥]
𝑎 +𝑏

3. 2 sin 𝐴 cos 𝐵 = sin(𝐴 + 𝐵) + sin(𝐴 − 𝐵)

4. 2 cos 𝐴 sin 𝐵 = sin(𝐴 + 𝐵) − sin(𝐴 − 𝐵)
5. 2 cos 𝐴 𝑐𝑜𝑠 B = cos(𝐴 − 𝐵) + cos(𝐴 + 𝐵)
6. 2 𝑠𝑖𝑛𝐴 𝑠𝑖𝑛 B = cos(𝐴 − 𝐵) − cos(𝐴 + 𝐵)
e x  e x ei   e  i 
7. sinh x  sin  
2 2i

e x  e x ei   e  i 
8. cosh x  cos  
2 2

𝑒 𝑎𝑥
9. ∫ 𝑒 𝑎𝑥 cos 𝑏𝑥 𝑑𝑥 = 2 2 [𝑎 𝑐𝑜𝑠𝑏𝑥 + 𝑏𝑠𝑖𝑛𝑏𝑥]
𝑎 +𝑏
𝑒 𝑎𝑥
10. ∫ 𝑒 𝑎𝑥 sin 𝑏𝑥 𝑑𝑥 = [𝑎 𝑠𝑖𝑛𝑏𝑥 − 𝑏𝑐𝑜𝑠𝑏𝑥]
𝑎2 +𝑏2
25 | P a g e

∞
1. ∫𝑜 𝑒− 𝑎𝑥 cos 𝑏𝑥 𝑑𝑥 =
𝑎
a)
𝑎2 +𝑏2
𝑏
b)
𝑎2 +𝑏2
1
c)
𝑎2 +𝑏2
d) None
∞
2. ∫𝑜 𝑒− 𝑎𝑥 sin 𝑏𝑥 𝑑𝑥 =
𝑏
a)
𝑎2 +𝑏2
𝑎
b)
𝑎2 +𝑏2
1
c)
𝑎2 +𝑏2
d) None
∞
3. ∫−∞ e−iωτ dτ =
a) 2πδ(ω)
b) δ(ω)
c) 1
d) None
1 ∞ −iωτ
4. ∫ e
2𝜋 −∞
dτ =
a) δ(ω)
b) 2πδ(ω)
c) 1
d) None
5. Find Power spectral density if auto correlation is given
∞
a) SXX (ω) = ∫−∞ R XX (τ)e−iωτ dτ
1 ∞
b) SXX (ω) = ∫ R (τ)e−iωτ dτ
2π −∞ XX
∞
c) SXX (ω) = ∫−∞ R XX (τ)eiωτ dτ
d) None
6. Find auto correlation if is Power spectral density given
1 ∞
a) R XX (τ) = ∫ S (ω)eiωτ dω
2π −∞ XX
∞
b) R XX (τ) = ∫−∞ SXX (ω)eiωτ dω
∞
c) R XX (τ) = ∫−∞ SXX (ω)e−iωτ dω
d) None
7. The auto correlation of stationary random processes is given by𝑅𝑋𝑋 (𝜏) = 𝑎𝑒 −𝑏|𝜏| , 𝑏 > 0
Find the spectral density function.
2ab
a) SXX (ω) =
b2 +ω2
2b
b) SXX (ω) =
b2 +ω2
2a
c) SXX (ω) =
b2 +ω2
d) None
26 | P a g e

8. The auto correlation of stationary random processes is given by𝑅𝑋𝑋 (𝜏) = 𝑒 −𝛼|𝜏| , Find the
spectral density function.
2α
a) SXX (ω) =
α2 +ω2
2b
b) SXX (ω) =
a2 +ω2
2a
c) SXX (ω) =
b2 +ω2
d) None
9. The auto correlation of stationary random processes is given by𝑅𝑋𝑋 (𝜏) = 𝑎2 𝑒 −2𝛼|𝜏| , Find
the spectral density function.
4a2 α
a) SXX (ω) =
4α2 +ω2
4α
b) SXX (ω) =
4α2 +ω2
4a2
c) SXX (ω) =
4α2 +ω2
d) None
10. The auto correlation of stationary random processes is given by𝑅𝑋𝑋 (𝜏) = 𝜌𝑒 −𝜌|𝜏| , Find the
spectral density function.
2ρ2
a) SXX (ω) =
𝜌2 +ω2
ρ2
b) SXX (ω) =
𝜌2 +ω2
2ρ2
c) SXX (ω) =
𝜌2 −ω2
d) None
11. Find the power spectral density of a random signal with auto correlation function e−λ|τ|
2λ2
a) SXX (ω) =
λ2 +ω2
λ2
b) SXX (ω) =
λ2 +ω2
2λ2
c) SXX (ω) =
λ2 −ω2
d) None
1 − |τ| |τ| ≤ 1
12. Autocorrelation function of an ergodic process X(t)isby R (τ) = { .
0 |τ| > 1
Obtain the spectral density of X .
2(1−cos ω)
a) SXX (ω) =
ω2
2(1+cos ω)
b) SXX (ω) =
ω2
2(1−sin ω)
c) SXX (ω) =
ω2
d) None
13. The autocorrelation function of the random processX(t) is given by
|τ|
1− |τ| ≤ T
R(τ) = { T . Find the power spectrum of the process X(t).
0 |τ| > T
27 | P a g e

2(1−cos ωT)
a) SXX (ω) =
Tω2
2(1+cos ωT)
b) SXX (ω) =
𝑇ω2
2(1−sin ωT)
c) SXX (ω) =
Tω2
d) None
14.