𝑯𝑾#𝟏

Q1. What are the basic components of Communication systems?

A communication system typically consists of the following basic components:

1. Information Source: This is the origin of the message or information to be transmitted.

It can be a person, a device, or any entity that generates information.
2. Transmitter: The transmitter processes the information from the source and converts it
into a suitable format for transmission over the communication channel. This often
involves modulation, which is the process of superimposing the information signal onto a
carrier wave.
3. Channel: The channel is the medium through which the transmitted signal travels from
the transmitter to the receiver. It can be wired (like a cable or fiber optic cable) or
wireless (like radio waves, microwaves, or infrared).
4. Receiver: The receiver receives the transmitted signal and demodulates it to extract the
original information. This involves processes like amplification, filtering, and
demodulation.
5. Destination: This is the final recipient of the information, such as a person, a device, or
another system.

Q2. How can we measure the performance of an analog communication system and a
digital communication system?

Analog Communication Systems

Signal-to-Noise Ratio (SNR):

o Measures the ratio of signal power to noise power.

o Higher SNR indicates better signal quality and less interference.

Digital Communication Systems

Bit Error Rate (BER):

o Measures the number of bits received in error compared to the total number of
bits transmitted.
o Lower BER indicates better system performance.
Q3. A random variable 𝑿 is defined by the following distribution function

𝟎, 𝒙<𝟎
𝟑
𝑭𝒙 (𝒙) = {𝑨𝒙 , 𝟎 ⩽ 𝒙 ⩽ 𝟏𝟎
𝑩, 𝟏𝟎 < 𝒙

(a) Find the proper values for 𝑨&𝑩

The distribution function 𝐹𝑋 (𝑥) satisfies the following properties

1 𝐹𝑋 (𝑥) is non-decreasing and lies between 0 and 1(0 ≤ 𝐹𝑥 (𝑥) ≤ 1𝜆.

2 𝐹𝑋 (𝑥) is a cumulative distribution function (CDF), so:

lim 𝐹𝑋 (𝑥) = 1 and lim 𝐹𝑋 (𝑥) = 0

𝑥→∞ 𝑥→−∞

For the piecewise function

0 for 𝑥 < 0
𝐹𝑋 (𝑥) = {𝐴𝑥 3 for 0 ≤ 𝑥 ≤ 10
𝐵 for 𝑥 > 10

• At 𝑥 = 10, 𝐹𝑋 (𝑥) must equal 1 (since 𝐹𝑋 (𝑥) → 1 ≤ 𝑥 → ∞ ):

1
𝐴(10)3 = 1 ⟹ 𝐴 =
1000

• At 𝑥 > 10, 𝐹𝑥 (𝑥) remains constant and equals 𝐵, 50𝐵 = 1.

Thus, the values are

1
𝐴= , 𝐵=1
1000

(b) Obtain and plot the PDF 𝒇𝒙 (𝒙)

The PDF 𝑓𝑋 (𝑥) is the derivative of the CDF 𝐹𝑋 (𝑥) :

𝑑
𝑓𝑋 (𝑥) = 𝐹 (𝑥).
𝑑𝑥 𝑋

Using the piecewise definition of 𝐹𝑋 (𝑥) :

0 for 𝑥 < 0
𝑓𝑥(𝑥) = {3𝐴𝑥 2 for 0 ≤ 𝑥 ≤ 10
0 for 𝑥 > 10
 1
Substituting 𝐴 = 1500 :

0 for 𝑥 < 0
3
𝑓𝑥(𝑥) = { for 0 ≤ 𝑥 ≤ 10
cos 𝑥 2
0 for 𝑥 > 10

Plot:
The function is parabolic for 0 ≤ 𝑥 ≤ 10 and zero elsewhere. I can create the plot if needed.

(c) Find the mean and variance of 𝒙

The mean 𝜇 is given by:

∞
𝜇 = 𝔼[𝑋] = ∫ 𝑥𝑓𝑥 (𝑥)𝑑𝑥
−∞

Substituting 𝑓𝑋(𝑥) :

10 10
3 3
𝜇=∫ 𝑥⋅ 𝑥 2 𝑑𝑥 = ∫ 𝑥 3 𝑑𝑥
0 1000 1000 0

Compute the integral:

10 10
𝑥4 104
∫ 𝑥 3 𝑑𝑥 = [ ] = = 2500
0 4 0 4

Thus:

3
𝜇= ⋅ 2500 = 7.5
1000

The variance 𝑑2 is:

𝜎 2 = 𝔼[𝑋 2 ] − (𝔼[𝑋])2

First, compute 𝔼[𝑋 2 ] :

∞ 10 10
3 3
𝔼[𝑋 2 ] = ∫ 𝑥 2 𝑓𝑥 (𝑥)𝑑𝑥 = ∫ 𝑥2 ⋅ 𝑥 2 𝑑𝑥 = ∫ 𝑥 4 𝑑𝑥
−∞ 0 1000 1000 0

Compute the integral

 10 10
4
𝑥5 105
∫ 𝑥 𝑑𝑥 = [ ] = = 20000
0 5 0 5

Thus:

3
𝔼[𝑋 2 ] = ⋅ 20000 = 60
1000

Now compute 𝜎 2 :

∇2 = 𝐸(𝑋 2 ] − (𝐸[𝑋])2 = 60 − (7.5)2 = 60 − 566.25 = 3.75

(d) What is the probability that 𝟑 ⩽ 𝒙 ⩽ 𝟕 ?

The probability is:

𝑃(3 ≤ 𝑋 ≤ 7) = 𝐹𝑋 (7) − 𝐹𝑋 (3) −

From the CDF:

1
𝐹𝑥 (𝑥) = 𝑥 3 for 0 ≤ 𝑥 ≤ 10
1000

Substitute:

1 343 1 27
𝐹𝑋 (7) = ⋅ 73 = = 0.343, 𝐹𝑋 (3) = ⋅ 33 = = 0.027
1000 1000 1000 1000

Thus:

𝑃(3 ≤ 𝑋 ≤ 7) = 0.343 − 0.027 = 0.316.

Q4. Two random variables 𝒙 and 𝒚 are distributed according to

𝑘𝑒 −𝑥−𝑦 𝑥⩾𝑦⩾0
𝑓𝑥,𝑦 (𝑥, 𝑦) = {
0 othar wise

(a) Find the value of the constant 𝑘 :

The total probability for a joint PDF must equal 1:

∞ ∞
∫ ∫ 𝑓𝑥, 𝑦(𝑥, 𝑦)𝑑𝑥𝑑𝑦 = 1
−∞ −∞

For the region 𝑥 ≥ 𝑦 ≥ 0, the limits of integration are

• 𝑦 ranges from 0 to ∞.

• 𝑥 ranges from 𝑦 to ∞ (since 𝑥 ≥ 𝑦 ).

Thus:

∞ ∞
∫ ∫ 𝑘𝑒 −𝑥−𝑦 𝑑𝑥𝑑𝑦 = 1
0 𝑦

Compute the inner integral with respect to 𝑥 :

∞ ∞
∫ 𝑒 −𝑥−𝑦 𝑑𝑥 = 𝑒 −𝑦 ∫ 𝑒 −𝑥 𝑑𝑥 = 𝑒 −𝑦 [−𝑒 −𝑥 ]∞
𝑦 =𝑒
−𝑦
⋅ 𝑒 −𝑦 = 𝑒 −2𝑦 .
𝑝 𝑦

Now campute the outer integral:

∞ ∞
∫ 𝑘𝑒 −2𝑦 𝑑𝑦 = 𝑘 ∫ 𝑒 −2𝑦 𝑑𝑦
0 0

The integral of 𝑒 −2𝑦 is

∞ ∞
−𝑒 −2𝑦 1
∫ 𝑒 −2𝑦 𝑑𝑦 = [ ] =
0 2 0 2

Thus:

1
𝑘⋅ =1 ⟹ 𝑘=2
2

(b) Find the marginal density functions 𝑓𝑋 (𝑥) and 𝑓𝑌 (𝑦) :

Marginal PDF of 𝑋𝑟 , 𝑓(𝑥) :

∞
𝑓𝑋(𝑥) = ∫ 𝑓𝑋,𝑌 (𝑥, 𝑦)𝑑𝑦.
−𝑥

From the limits, 𝑥 ≥ 𝑦 ≥ 0,50𝑦 ranges from 0 to 𝑥 :

𝑥
𝑓𝑥(𝑥) = ∫ 2𝑒 −𝑥−𝑦 𝑑𝑦
0

Factor out 𝑒 −𝑥 :

𝑥
𝑓𝑋 (𝑥) = 2𝑒 −𝑥 ∫ 𝑒 −𝑦 𝑑𝑦
0

The integral of 𝑒 −8 is:

𝑧
∫ 𝑒 −𝑦 𝑑𝑦 = [−𝑒 −𝑦 ]0𝑧 = 1 − 𝑒 −𝑥
0

Thus:

𝑓𝑥(𝑥) = 2𝑒 −𝑥 (1 − 𝑒 −𝑥 ), 𝑥 ≥ 0

Marginal PDF of 𝑌, 𝑓𝑌 (𝑦) :

∞
𝑓𝑌 (𝑦) = ∫ 𝑓𝑋,𝑌 (𝑥, 𝑦)𝑑𝑥
−∞

From the lirrits, 𝑥 ≥ 𝑦 ≥ 0,50𝑥 ranges from 𝑦 to ∞ :

∞
𝑓𝑌 (𝑦) = ∫ 2𝑒 −𝑥−𝑦 𝑑𝑥
𝑦

Factor out 𝑒 −𝑦 :

∞
𝑓𝑌 (𝑦) = 2𝑒 −𝑦 ∫ 𝑒 −𝑥 𝑑𝑥
𝑦

The integral of 𝑒 −𝑥 is:

∞
∫ 𝑒 −𝑥 𝑑𝑥 = [−𝑒 −𝑥 ]∞
𝑦 =𝑒
−𝑦
𝑦

Thus:
 𝑓𝑌 (𝑦) = 2𝑒 −𝑦 − 𝑒 −𝑦 = 2𝑒 −2𝑦 , 𝑦 ≥ 0

(c) Are 𝑋 and 𝑌 independent?

Two random variables 𝑋 and 𝑌 are independent if:

𝑓𝑋,𝑌 (𝑥, 𝑦) = 𝑓𝑋 (𝑥)𝑓𝑌 (𝑦), ∀𝑥, 𝑦

Here, the jaint PDF 𝑓𝑋, 𝑌(𝑥, 𝑦) = 2𝑒 −𝑥−𝑦 (in the region 𝑥 ≥ 𝑦 ≥ 0) does not factorize into the
product of 𝑓𝑋 (𝑥) and 𝑓𝑌 (𝑦) over the entire region of 𝑥 and 𝑦.

For example, 𝑓𝑋 (𝑥) and 𝑓𝑌 (𝑦) include terms dependent on limits (1 − 𝑒 −𝑧 and 2𝑒 −2𝑦 .
Thus, 𝑋 and 𝑌 are not independent.
(d) Find 𝑓𝑋∣𝑌 (𝑥 ∣ 𝑦) :

The conditional PDF 𝑓𝑋∣𝑌 (𝑥 ∣ 𝑦) is given by:

𝑓𝑋𝑌 (𝑥, 𝑦)
𝑓𝑋𝑌 (𝑥 ∣ 𝑦) = .
𝑓𝑌 (𝑦)

Substitute the expressions for 𝑓𝑋,𝑌 (𝑥, 𝑦) and 𝑓𝑌 (𝑦) :

• 𝑓𝑋, 𝑌(𝑥, 𝑦) = 2𝑒 −𝑥−𝑝 for 𝑥 ≥ 𝑦 ≥ 0

• 𝑓𝑌 (𝑦) = 2𝑒 −2 for 𝑦 ≥ 0.

Thus:

2𝑒 −𝑥−𝑦
𝑓𝑋∣𝑌 (𝑥 ∣ 𝑦) = = 𝑒 −(𝑥−𝑚) , 𝑥 ≥ 𝑦 ≥ 0
2𝑒 −2𝑦

The conditional PDF is:

𝑒 −(𝑥−𝑦) , 𝑥 ≥ 𝑦 ≥ 0
𝑓𝑋∣𝑌 (𝑥 ∣ 𝑦) = {
0, otherwise.
Q5. (a) sketch the ensemble of random processes.

𝒙(𝒕) = 𝒂 𝐜𝐨𝐬(𝝎𝒕 + 𝚽)

where 𝒘 and 𝚽 are constants and 𝒂 is R.V. uniformly distributed in the range (−𝑨, 𝑨).
(b) By observing the ensemble, determine whether this is a stationary or non-stationary
process. Give your reasons

(a) Sketch the ensemble of random processes

An ensemble of random processes is a collection of different realizations of the process x(t)

generated by varying the random variable a.

For this case:

a takes different values within the range (-A,A).

Each realization of x(t) is a sinusoidal wave, with the same frequency 4 , phase 4 , but
different amplitudes a.

To sketch the ensemble:

Horizontal axis represents time t.

Each realization corresponds to a sinusoidal curve with amplitude chosen randomly from
(-A,A).

Key characteristics for the sketch:

All realizations have the same frequency ω.

All realizations are shifted by the same phase Φ.

The amplitude varies linearly between -A and A.

(Since this is textual, you can visualize sinusoidal waves with varying amplitudes centered at 0. )

(b) Stationarity of the process

Definition of Stationarity:
A random process is stationary if its statistical properties (mean, variance, autocorrelation) do
nat depend an time
Analyze 𝑥(𝑡) :

1 Mean (𝔼[𝑦(𝑡)]:

𝔼[𝑥(𝑡)] = 𝔼[𝑎]cos (𝜔𝑡 + Ψ).

Since 𝑎 is uniformly distributed in (−𝐴, 𝐴), its mean is:

−𝐴 + 𝐴
𝐸[𝑎] = =0
2

Thus:

𝔼[𝑥(𝑡)] = 0 − cos (𝜔𝑡 + J) = 0

The mean is time-invariant.

2. Variance (Var [𝑥(𝑡)] : Variance is

Var [𝑥(𝑡)] = 𝔼[𝑥 2 (𝑡)] − (𝔼[𝑥(𝑡)])2

Since 𝐸[𝑥(𝑡)] = 0 :

Var [𝑥[𝑡)] = 𝔼[𝑥 2 (𝑡)]

Compute E[𝑥 2 (𝑡)]:

𝑥 2 (𝑡) = 𝑎2 cos 2 (𝜔𝑡 + 4)

and:

𝔼[𝑥 2 (𝑡)] = 𝔼[𝑎2 ] ⋅ cos2 (𝜔𝑡 + Φ)

For a uniformly distributed in ( −𝐴, 𝐴 ), the second moment is:

1 𝐴 2 𝐴2
𝔼[𝑎2 ] = ∫ 𝑎 𝑑𝑢 =
2𝐴 −𝐴 3

Thus:

𝐴2
𝔼[𝑥 2 (𝑡)] = − cos2 (𝜔𝑡 + I)
3

Since cos 2 (𝜔𝑡 + Φ) varies with 𝑡, the variance depends on time.

3. Autocorrelation (𝑅2 (𝑡1 , 𝑡2 )) : The autocorrelation is

𝑅𝑥 (𝑡1 , 𝑡2 ) = 𝐸[𝑥(𝑡1 )𝑥(𝑡2 )]

Substitute 𝑥(𝑡) = 𝑎cos (𝜔𝑡 + 1):

𝑅2 (𝑡1 , 𝑡2 ) = 𝔼[𝑎2 ]cos (𝜔𝑡1 + Φ)cos (𝜔𝑡2 + Φ)

𝐴2
Since 𝐸[𝑎2 ] = 3
, the autocorrelation depends on 𝑡1 and 𝑡2 , indicating a non-stationary process.
Q6. Show that if the PSD of a random process 𝒙(𝒕) is band-limited to 𝑩 𝐇𝐳, And if:

𝒏 𝟏 𝒏=𝟎
𝑹𝒙 ( ) = {
𝟐𝑩 𝟎 𝒏 = ∓𝟏, ∓𝟐, ∓𝟑, …

Then 𝒙(𝒕) is a white band limited process:

𝝎
That is 𝑺𝒙 (𝝎) = 𝒌 𝐫𝐞𝐜𝐭 (𝟒𝝅𝜷) ?
𝒏
"try to think as 𝝉 = 𝒓 = ∓ 𝟐𝑩 "

Step 1: Given Information

1 Autocorrelation function:
𝑛 1, 𝑛=0
𝑅𝑥 ( ) = {
2𝐵 0, 𝑛 = ±1, ±2, ±3, …

This implies that the autocorrelation 𝑅𝑎 (𝜏) is a periodic train of impulses (non-zero only at
2𝐵
specific multiples of 𝜏 = ± 2𝐵.
2. PSD and Autocorrelation relationship:

The PSD 𝑆𝑧 (𝜔) is the Fourier Transform of the autocorrelation 𝑅𝑥 (𝜏) :

∞
𝑆𝑥 (𝜔) = ∫ 𝑅2 (𝜏)𝑒 −𝑗𝜔𝜏 𝑑𝜏
−∞

3 Goal: Prove that:

𝜔
𝑆𝑥 (𝜔) = 𝑘rect ( )
4𝜋𝐵

where rect is a rectangular function centered at 𝜔 = 0 with a bandwidth 4𝜋𝐵.

1
Using the given conditions for 𝑅𝑥 (2𝐵), we interpret 𝑅𝑥 (𝜏) as:

𝑅2 (𝜏) = 𝛿(𝜏),

where 𝛿(𝜏) is the Dirac delta function. This arises because 𝑅𝑥 (𝜏) is nocizero only when 𝜏 = 0,
and zero otherwise.

𝑛
However, because 𝑇 = 25 is mentioned, consiber the more general interperetation:

∞
𝑛
𝑅𝑥 (𝜏) = ∑ 𝛿 (𝜏 − ).
2𝐵
𝑛=𝑥
 9
This form matches the periodicity implied by the given 𝑅= (2𝐵) -

Step 3: Compute the PSD 𝑆2 (𝜔)

The PSD is computed 𝑎𝑠 the Fourier Transform of 𝑅𝑥 (𝜏) :

∞
𝑆𝑥 (𝜔) = ∫ 𝑅𝑥 (𝜏)𝑒 −𝑗𝜔𝑡 𝑑𝜏
−∞

Substitute 𝑅− (𝜏) :

∞ ∞
𝑛
𝑆2 (𝜔) = ∫ [ ∑ 𝛿 (𝜏 − )] 𝑒 −𝑗𝜔𝜏 𝑑𝜏
−∞ 𝑛−−∞ 2𝐵

Exchange the summation and the integral:

∞ ∞
𝑛
𝑆𝑥 (𝜔) = ∑ ∫ 𝛿 (𝜏 − ) 𝑒 −𝑗𝜔𝜏 𝑑𝜏
−∞ 2𝐵
𝑛−−∞

Using the sifting property of the delta function

∞
𝑛
∫ 𝛿 (𝑥 − ) 𝑒 −𝑗𝜏 𝑑𝜏 = 𝑒 −𝑗𝜔𝑡
−∞ 2𝐵

we get:

𝑆𝑧 (𝜔) = ∑ 𝑒 −𝑗≠
↓+∞

Step 4: Simplify 𝑆𝑥 (𝜔)

The summation:

∑ 𝑒 −𝑗𝜔
𝑛=−∞

is a Fourier series representation of a periodic rectangular function in frequency domain.

𝜔
Specifically, it corresponds to a rectangular function rect ( ) with width 4𝜋𝐵 and height
4𝜋𝐵

proportional to a constant 𝑘.

Thus:

𝜔
𝑆𝑥 (𝜔) = 𝑘rect ( ),
4𝜋𝐵
where 𝑘 is a scaling constant determined by normalization.

Step 5: Interpret the Result

1 Band-limited PSD: The PSD 𝑆2 (𝜔) is nonaero only in the range |𝜔| ≤ 2𝜋𝐵, confirming
the band-limited nature of the process.

2 White spectrum within the band: Within the range |𝜔| ≤ 2𝜋𝐵, 𝑆2 (𝜔) is constant,
consistent with the definition of a white process.
Q7. The PSD of narrow-band noise 𝒏(𝒕)Is is given by:

The carrier frequency is 5 Hz.

1 Find the PSB of the in-phase and quadrature-phase components of 𝒏(𝒕).

2 Find their cross-spectral densities:

𝑩
𝑮𝒏𝒄 (𝒇) = 𝑮𝒏𝒔 (𝒇) = 𝑮𝒏 (𝒇 − 𝒇𝒄 ) + 𝑮𝒏 (𝒇 + 𝒇𝒄 ), 𝒇<
𝟐
Given Information:

1 Carrier frequency 𝑓𝑐 = 5 Hz.

2 The PSD of the in-phase and quadrature-phase components is related to the PSD of 𝑛(𝑡).
𝐺𝑛 (𝑓).

3 Narrowband noise is described in terms of positive and negative frequency components:

𝐵
𝐺𝑛 (𝑓) = 𝐺𝑛𝑛 (𝑓) = 𝐺𝑛 (𝑓 − 𝑓𝑐 ) + 𝐺𝑛 (𝑓 + 𝑓𝑐 ), |𝑓| <
2

4 We are tasked with finding:

• PSDs of the in-phase and quadrature-phase components {𝐺𝑛 (𝑓) and 𝐺𝑛, (𝑓) )

• Their crass-spectral density.

Step 1: Analyze the PSD of In-Phase and Quadrature-Phase Components

The in-phase component 𝑛𝑐 (𝑡) and quadrature-phase component 𝑛, (𝑡) are formed by
modulating 𝑛(𝑡) using cosine and sine functions, respectively. Their PSDs are given by:

𝐺𝑛 (𝑓) = 𝐺𝑛2 (𝑓) = 𝐺𝑛 (𝑓 − 𝑓𝑐 ) + 𝐺𝑛 (𝑓 + 𝑓𝑐 )

This result comes from the properties of spectral shifting caused by modulation:

• 𝐺𝑤 (𝑓 − 𝑓𝑐 ) carresponds to the lower sideband (shifted down by 𝑓 l).

• 𝐺𝑛 (𝑓 + 𝑓𝑐 ) carresponds to the upper sideband (shifted up by 𝑓𝑐 )

Thus, the PSD of both 𝑛− (𝑡) and 𝑛𝑠 (𝑡) is identical and symmetric around 𝑓 = 0.
Interpretation:
𝐸
For a narrowband noise process, the bandwith is limited to |𝑓| < 2 . The PSDs 𝐺n (𝑓) and 𝐺𝑛, (𝑓)

are nonzero in this range.

Step 2: Cross-Spectral Density
The cross-spectral density between the in-phase and quadrature-phase components, 𝐺1,𝑛, (𝑓).
measures the correlation between 𝑛− (𝑡) and 𝑛3 (𝑡) in the frequency domain.

For narrowband noise:

• The in-phase and quadrature-phase components are uncorrelated.

• Hence, the cross-spectral density is zero:

𝐺𝑛,𝑤 , (𝑓) = 0.

This is due to the orthogonality of cosine and sine functions in modulation.

Step 3: Summary

1 PSD of In-Phase and Quadrature-Phase Components

𝐵
𝐺𝑛 (𝑓) = 𝐺𝑛2 (𝑓) = 𝐺𝑛 (𝑓 − 𝑓𝑐 ) + 𝐺𝑛 (𝑓 + 𝑓𝑐 ), |𝑓| <
2

2 Cross-Spectral Density:

𝐺𝑛+𝑥1 (𝑓) = 0

These results are consistent with the properties of narrowband noise in communication systems.