Birla Institute of Technology & Science, Pilani

1st Semester 2019-2020

Comprehensive Exam-Part B (Open Book)
Course No: EEE F311 Course Title: Communication Systems
Maximum Marks: 50 Duration: 120 Minutes
Date: 09-12-2019

Write down assumptions and definitions wherever there is an ambiguity.

Please write the final answer and important steps in rectangular boxes.

Q1. An analog passband transmission is shown in the figure below. The bandpass filter has a bandwidth 2B
centered at fc . The message signal m(t) is real-valued signal with amplitude varying between −Am to Am ,
and has an average power Pm . The bandwidth of m(t) is B. The square-law device obeys r(t) = 2x(t)+x2 (t).
[8 Marks].

x(t) r(t)
Square-Law
m(t) + Device BPF y(t)
Modulated Signal
Cos 2 π fct

(a) Find the average transmitted power of the modulated signal y(t) in terms of Pm .
(b) Find the modulation index of the modulated signal y(t).
(c) Draw a system block diagram for coherent demodulation.
(d) What minimum value Am will ensure non-coherent demodulation?

Q2. An M-ary PAM communication system transmits data at a rate 24000 bps. The communication channel
passes signals with frequencies from 600 Hz to 6600 Hz. [6 Marks].

(a) If M = 16, find the pulse (in time-domain) satisfying the Nyquist criteria for zero-ISI.
(b) If M = 4, find the pulse (in time-domain) satisfying the Nyquist criteria for zero-ISI.
(c) If M = 8, find BER (in numerical value) if Eb /N0 = 3.5.

Q3. A communication system uses 4-ary signaling where the signal set is as shown in the figure below. For parts
(a) to (e), assume that there is no effect of channel. [12 Marks].

s1(t) s2(t)
s3(t) s4(t)
A/sqrt(2)
A/sqrt(2)
A/sqrt(2)
0 1 2 t 0 1 2 t 0 2 t 0 1 2 t
1
-A/sqrt(2) -A/sqrt(2) -A/sqrt(2)

(a) Find average energy of the transmitted signals.

(b) Sketch matched filter f3 (t) corresponding to transmitted signal s3 (t).

1
 (c) Find the signal at the output of the matched filter f3 (t) if s1 (t) is transmitted.
(d) Find minimum number of basis signals and express the signals s1 (t), s2 (t), s3 (t), and s4 (t) using the
basis signals.
(e) Find the BER (in terms of signal amplitude A and Q-function) under AWGN of PSD N0 /2.
(f) Sketch the matched filters (using basis signals) if there is a distortionless channel h(t) = δ(t − 5).
Q4. A message signal x(t) = A cos(2πfm t) is received under additive white Gaussian noise having two-sided
PSD of N0 /2 W/Hz. The receiver is realized by two differentiator followed by a unity gain LPF filter of
bandwidth B i.e., F (f ) = 1, −B ≤ f ≤ B. Note that fm < B and that y(t) contains both the signal and
noise components. [6 Marks].

x(t)=Acos 2πfmt
+ d/dt d/dt F(f) y(t)
Differentiator Differentiator
n(t)

(a) Find the signal power at the output of LPF.

(b) Find the average noise power at the output of LPF.
Q5. X(t) is a random process with auto-correlation function Rx (τ ) = e−|τ | . The process is applied to a linear
system where X(t) is input and Y (t) is output as shown in the figure below. [4 Marks].
(a) Find the transfer function H(f ) of the system.
(b) Find Sy (1) i.e., the PSD of the random process Y (t) for f = 1 .

+
X(t) Y(t)
+ d/dt
+
X(t-1)
Delay = 1

Q6. Consider a linear systematic block encoder where three information bits d = (d1 , d2 , d3 ) is complemented
with three parity check bits as c1 = d2 ⊕ d3 , c2 = d1 ⊕ d3 , and c3 = d1 ⊕ d2 . [6 Marks].
(a) Find the code word for the data 101.
(b) Construct the parity check matrix Q such that GQT = 0.
(c) Decode the received word 100001.

Q7. Determine the capacity of the channel described by the channel matrix as shown in the figure below. Note
that the input symbols are not equiprobable. [4 Marks].

0 0

1/2

1 1
1/2

Q8. Consider a power-constrained additive noise channel Y = X + Z, where X and Y are the input and the
output signals and Z is the noise. Let S be the power of signal X and N be the power of noise Z. Take
channel bandwidth B = 1 Hz. Find the channel capacity if the additive noise Z is exponentially distributed
i.e. fZ (z) = λe−λz , z ≥ 0. [4 Marks].

2
Solution
Q1. After calculations and passing signal at centered frequency fc : y(t) = 2 cos 2πfc t + 2m(t) cos 2πfc t

(a) Average power: 4

2 + 4 P2m = 2 + 2Pm
(b) Modulation index = Am
(c) Multiplication with cos 2πfc t then LPF yield 1 + m(t), then pass through DC blocker to get m(t). Thus
the block diagram should consists of multiplier with cos 2πfc t, followed by LPF, then a DC blocker.
(d) |2 + 2m(t)| ≥ 0. Using (2 + 2m(t)) cos 2πfc t or the modulation index, the minimum value of Am = 0.

Q2. Rb = 24000 bps, B = 6000.

2B
(a) If M = 16, Rs = 6000 sps. Using Rs = 1+α , we get α = 1. This gives raised cosine pulse p(t) =
cos πRs t
Rs 1−4R 2 t2 sinc(πRs t), where Rs = 6000.
s
2B
(b) If M = 4, Rs = 12000 sps. Using Rs = 1+α , we get α = 0. This gives sinc pulse p(t) = sinc(πRs t),
where Rs = 12000.
q
(c) If M = 8, SER= 2 M Q[ 6M
M −1 log2 M Eb 7 7
2 −1 ( N )]. This gives 4 Q(1) or 4 × 0.1587 or 0.2777 [1 Mark]. BER
0
0.2777/(log2 M ) = 0.2777/3 = 0.0926 [1 Mark].

Q3. (a) (A2 + A2 + A2 + A2 )/4 = A2 .

(b) Sketch of f3 (t) is exact same as s3 (t).
(c) Output at the matched filter. A sketch will show a triangular pulse of amplitude −A2 at t = 2.



 0, t < 0
− A2 t, 0 ≤ t ≤ 2

y(t) = A2 2
 2 (t − 4), 2 ≤ t ≤ 4



0, t > 4

(d) Number of basis signals 2: φ1 (t) : rectangular pulse of height 1 and width 1: time-axis: 0 : 1, φ1 (t) :
rectangular pulse of height 1 and width 1 time-axis: 1 : 2. [1 Mark] s1 (t) = √A2 φ1 (t) + √A2 φ2 (t),
s2 (t) = √A2 φ1 (t) − √A2 φ2 (t), s3 (t) = − √A2 φ1 (t) − √A2 φ2 (t), s4 (t) = − √A2 φ1 (t) + √A2 φ2 (t).[1 Mark] Note
that there are other basis signals possible. However, there must be 2 basis signals which
must be of unit energy and orthogonal to each other.
(e) The above basis signals can be used 
toqrepresent
 signals
q as  a 4-QPSK/4-QAM constellation point. Error
A2 A2
probability of 4-QAM/4-QPSK: 2Q N0 − Q2 N0 .
(f) Matched filter 1: Shift φ1 (t) right 5 units. Matched filter 1: Shift φ2 (t) right 5 units.

Q4. (a) Double derivative on signal: −(2πfm )2 cos 2πfm t. Signal power = 8A2 π 4 fm
4.

(b) Noise PSD at the output due to two transfer functions of differentiation: N0 /2(2πf )4 . Noise power
RB
(N0 /2)16π 4 −B f 4 df = 16 4 5
5 N0 π B .

Q5. (a) H(f ) = j2πf + j2πf e−j2πf

2 32π 2
(b) Sx (f ) = 1+4π 2 f 2
, Sy (f ) = Sx (f )|H(f )|2 . Sy (1) = 1+4π 2
= 7.8024.

Q6. (a) The code word: 101101.

(b) Given:  
0 1 1 1 0 0
QT = 1 0 1 0 1 0
1 1 0 0 0 1
(c) Decoded word 100.

3
Q7. Capacity 0.322 bits/ symbol.
Let P (X = 1) = p, P (X = 0) = 1 − p. H(Y ) = −(1 − p + p/2) log2 (1 − p + p/2) − (p/2) log2 (p/2) =
−(1 − p/2) log2 (1 − p/2) − (p/2) log2 p/2.
H(Y /X) = (−p/2) log2 1/2 − (p/2) log2 (1/2).
I(X; Y ) = H(Y ) − H(Y /X).
To find p, we differentiate with I(X; Y ) and equate to zero. This gives p = 2/5. Using this, capacity is 0.322
bits/ symbol.

Q8. Capacity C = log2 (2πeS

R∞ + N ) − 2H(n) bps. Here H(n) is the differential entropy of exponential random
variable. H(n) = − 0 λe −λz log2 (λe−λz )dz = log2 λe . Hence, C = log2 (2πeS + N ) − 2 log2 λe bps. Capacity
in bits per symbol will be C = 12 log2 (2πeS + N ) − log2 λe .