Documente Academic
Documente Profesional
Documente Cultură
6
Number of Questions
5
Marks 1
4
Marks 2
3 Total number of questions
0
2015 2014 2013 2012 2011 2010 2009
Year Topic
2015 Amplitude modulation
Basics of TDMA, FDMA and CDMA and GSM
Delta modulation
Binary symmetric channel
Autocorrelation
Probability of error calculations
Amplitude modulation and demodulation systems
Random signals and noise: probability, random variables
2014 Amplitude modulation and demodulation systems
CDMA
Fundamentals of information theory and channel capacity theorem
Binary symmetric channel
Frequency modulation (FM) for low noise conditions
Power spectral density
Autocorrelation
Probability of error calculations
Pulse code modulation (PCM)
PSK
Random variables and binary symmetric channel
2013 Digital modulation schemes
Bandwidth consideration and probability of error calculations for these schemes
Fundamentals of information theory and channel capacity t heorem
2012 Digital modulation schemes
Bandwidth consideration and probability of error calculations for these schemes
Signal-to-Noise Ratio (SNR): Calculations for amplitude modulation (AM) and frequency modulation
(FM) for low noise conditions
Modulation technique
Power spectral density
Entropy
2011 Digital modulation schemes
Signal-to-Noise Ratio (SNR): Calculations for amplitude modulation (AM) and frequency modulation
(FM) for low noise conditions
Fundamentals of information theory and channel capacity theorem
Power spectral density
2010 robability density function
P
Realizations of analog c ommunication systems
Power spectral density
Analysis of modulated signal
2009 Amplitude modulation and demodulation systems
Realizations of analog c ommunication systems
Signal-to-Noise R
atio (SNR): Calculations for amplitude modulation (AM) and frequency modulation
(FM) for low noise conditions
Fundamentals of information theory and channel capacity theorem
In this chapter different topics related to random signals and noise including random variables, probability density
function, autocorrelation, power spectral density and different types of noise are covered.
A random variable is a function that associates a unique In other words, a discrete random variable can assume
numerical value with every outcome of an experience. only certain discrete values.
As an example, let x(.) be a function that maps sample For a general case, where random variable x can take
points z1, z2, z3 … zm into real numbers x1, x2, x3 … xn. values x1, x2, x3 … xn and the random variable y can take
In this case, x can be a random variable that takes on values y1, y2, y3 … ym, we have
values x1, x2, x3 … xn. The probability of random vari-
able x taking value xi is Px(xi). å å Pxy (xi , y j ) = 1 (42.2)
i j
Random variables can be classified as discrete and
continuous random variables.
This follows from the fact that the summation on the
left side is the probability of the union of all possible
42.1.1 Discrete Random Variable outcomes and must be unity.
If x and y are two random variables, then the condi-
A random variable is discrete if there exists a denumer- tional probability that x = xi given that y = yj is denoted
able sequence of discrete numbers xi such that by P (xi y j ) and the conditional probability that y = yj
xy
given that x = xi is denoted by Py x (y j xi ). The relation Other properties of the PDF px(x) are as follows:
between these conditional probabilities is given by the +¥
following expression: 1. ò px (x)dx = 1
−¥
å Px y (xi yj ) = å Py x (yj xi ) = 1 (42.3)
2. px (x) ≥ 0
i j
∫ e−x
1 2
/2
Fx (x) = dx (42.10)
A random variable that can assume any value over a 2p
−∞
continuous set is referred to as a continuous random
The given Gaussian function of Eq. (42.9) has zero mean
variable. In other words, a continuous random variable
and unit variance. For a generalized Gaussian random
can assume any value in a certain interval. The prop-
variable with mean m and variance s,
erties of the CDF Fx(x) described for discrete random
e−(x−m) / 2s 2
2
variables are applicable for continuous random variables 1
px (x) = (42.11)
as well.
s 2p
The probability density function (PDF) of a random x
−(x−m ) / 2s 2
∫e
2
1
variable x is defined as the derivative of the CDF Fx(x). Fx (x) = dx (42.12)
s 2p −∞
Therefore,
dFx (x) 42.1.4 Joint Distribution
= px (x) (42.8)
dx
For two random variables x and y, the CDF Fxy(x, y) is
The probability that the random variable x is in the defined as follows:
Fxy (x, y)= P (x ≤ x and y ≤ y )
interval (x, x + Δx) is px(x)Δx. ∆
(42.13)
If we have two functions g1(x) and g2(y) and the func- The autocorrelation function of an energy signal and
tion g(x, y) = g1(x)g2(y), then the mean of their product its energy spectral density (ESD) form a Fourier trans-
is given by form pair, that is,
y g (t ) ¬¾¾®y g (f )
FT
+¥ +¥
(42.40)
g1(x)g2 (y) = ò ò g1(x)g2 (y)pxy (x, y)dxdy (42.33) The time autocorrelation function of a real power signal
−¥ −¥ g(t) is defined as follows:
If x and y are independent, then + T/ 2
ò
1
Rg (t ) = lim g(t)g(t - t )dt
T ®¥ T
g1(x)g2 (y) = g1(x)g2 (y) (42.34)
−T/ 2
+T/ 2
ò
1
42.1.8 Moments = lim g(t)g(t + t )dt (42.41)
T ®¥ T
−T/ 2
The nth moment of a random variable x is defined as
The time autocorrelation function of a complex power
the mean value of xn. Therefore, the nth moment of x
signal g(t) is defined as follows:
(with zero mean, i.e. x = 0) is given by the following
expression: +T /2
Rg (t ) = lim ò g(t)g *(t −t )dt
1
+¥
n ∆ T ®¥ T
x = ò x px (x)dx n
(42.35) −T / 2
-¥ +T /2
− ò g * (t)g(t + t )dt (42.42)
1
= lim
The nth central moment of a random variable x with T ®¥ T
−T / 2
mean x is defined as follows:
+¥ For real power signal g(t), the autocorrelation function
∆
(x - x)n = ò (x - x)n px (x)dx (42.36) is defined as
-¥ +T /2
ò
1
The second central moment of a random variable x is Rg (t ) = lim g(t)g (t− t )dt
T ®¥ T
referred to as variance of x and is denoted by s x2 . −T / 2
s x is known as the standard deviation of random vari- For real g(t), the autocorrelation function is an even
able x and is given by the following expression: function of t.
For a power signal, the autocorrelation function and
s x2 = (x − x)2 = x2 − x
2
(42.37) the power spectral density (PSD) form a Fourier trans-
form pair, that is,
Therefore, variance of x is equal to the mean square
Rg (t ) ¬¾¾® Sg (f )
FT
value minus the square of the mean. When the mean is
(42.43)
zero, the variance is the mean square.
The PSD function is the Fourier transform of the A white noise process, W(t), is defined by
autocorrelation function Rx(t) of the process x(t), that is, N
SW (w ) = o −¥ < w < ¥
Rx (t ) ¬¾¾® Sx (f ) 2
FT
(42.45)
where SW (w) is the power spectral density and No is a real
Equation (42.45) is referred to as the Wiener-Khintchine constant (referred to as the intensity of the white noise).
theorem. The corresponding autocorrelation function is given by
No
Therefore, the autocorrelation function is given by the RW (t ) = d (t )
following expression: 2
+¥ where d (t ) is the Dirac delta.
Rx (t ) = ò Sx (f )e j2pft df (42.46) The average power of white noise
−¥
¥
ò
1 No
42.3.1 Power of a Random Process Pavg = EW 2 (t) = dw ® ¥
2p −¥
2
The average power Px of a wide-sense random process
The autocorrelation function and the PSD of a white noise
x(t) is its mean square value x2 : process is depicted in Fig. 42.2 (a) and (b), respectively.
+¥
ò
SW (ω)
Px = x2 = Rx (0) = Sx (f )df (42.47)
−¥ N
2
The power Px is the area under the PSD curve.
42.4 NOISE
0 ω
With reference to an electrical system, noise may be (a)
defined as any unwanted form of energy which tends
to interfere with proper reception and reproduction of RW (τ)
wanted signal. In other words, noise is random, undesir-
able electrical energy that enters the communications
system via the communicating medium or is generated
δ(τ)
N
internally and interferes with the transmitted message. 2
Noise may be classified into following two categories:
1. External noises: It is the noise whose sources
are external. External noise includes atmospheric
noises, extraterrestrial noises, man-made noises or 0 τ
industrial noises. (b)
2. Internal noise in communication: These are
the noises which get generated within the receiver or Figure 42.2 | White noise process (a) autocorrelation
communication system. These include thermal noise and (b) power spectral density.
or white noise or Johnson noise, shot noise and so on. Some important properties of white noise are:
42.4.1.1 Band-limited White Noise A noise process is called Gaussian noise if its amplitude
has a Gaussian probability density function (PDF).
A noise process which has non-zero constant PSD over a Gaussian white noise is a white noise (constant power
finite frequency band and zero elsewhere is called band- spectral density) with Gaussian distributed amplitude.
limited white noise. Thus the WSS process X(t) is band-
limited white noise if 42.4.3 Thermal noise
No
SX (w ) =
−B <w < B The noise generated in any resistance due to random
2 motion of electrons is called thermal noise. It is also
For example, thermal noise which has constant PSD up referred to as white noise or Johnson noise. Thermal
to very high frequency is better modelled by a band- noise is given by
Pn µ TB or Pn = KTB
limited white noise process.
The corresponding autocorrelation function RX (t ) is where
given by
N B sin Bt Pn = Maximum noise power output of a resistor (W).
RX (t ) = o K = Boltzmann’s constant = 1.38 × 10-23 J/K.
2p Bt
T = Absolute temperature (K).
42.4.1.2 Coloured Noise B = Bandwidth over which noise is measured (Hz).
A noise process which is not white is called coloured 42.4.4 Flicker Noise
noise. Thus the noise process X(t) with autocorrelation
function Flicker noise, also represented as a 1/f, is a type of elec-
RX (t ) = a2e−b t b>0 tronic noise with a pink power density spectrum. It occurs
in almost all electronic devices and can show up with a
and PSD
variety of other effects, such as impurities in a conductive
2a2b
SX (w ) =
channel, generation and recombination noise in a transis-
b2 + w 2 tor due to base current, and so on. 1/f noise in current or
voltage is always related to a direct current because it is
is an example of a coloured noise. a resistance fluctuation, which is transformed to voltage
or current fluctuations via Ohm’s law.
42.4.1.3 White Noise Sequence
42.4.5 Shot Noise
A random sequence W [n ] is called a white noise sequence if
Shot noise or Poisson noise is a type of electronic
No
SW (w ) = −p ≤ w ≤ p
noise which originates from the discrete nature of electric
2 charge.
IMPORTANT FORMULAS
å Pxy (xi , y j ) = Py (yj ) 14. Bayes’ rule for continuous random variables
i py x (y x)px (x)
å Pxy (xi , y j ) = Px (xi )
px y (x y) =
py (y)
j
+¥
ò px y (x y)dx = 1
5. Cumulative distribution function
Fx (x) = P (x ≤ x) −¥
6. Probability density function (PDF) 15. Two continuous random variables x and y are said
to be independent when
dFx (x)
= px (x) px y (x y) = px (x)
dx
7. Properties of PDF and py x (y x) = py (y)
+¥
ò
16. For two independent random variables x and y,
px (x)dx = 1
−¥
pxy (x, y) = px (x)py (y)
px (x) ≥ 0
17. For two independent random variables x and y,
8. The CDF Fx(x) of the Gaussian random variable Fxy (x, y) = Fx (x)Fy (y)
with zero mean and unit variance is
x 18. The mean value of a random variable x, also
∫ e dx
1 −x2 / 2 referred to as the average value or the expected
value, is defined as x = E[x] = å xiPx (xi )
Fx (x) =
2p −∞
i
9. For a Gaussian random variable with mean m and
variance s, 19. If the random variable x is continuous, then
+¥
e−(x−m) / 2s 2
2
1
px (x) =
s 2p
x = E[x] = ò xpx (x)dx
x −¥
∫ e−(x−m) / 2s 2
2
1
and Fx (x) = dx 20. The nth moment of a random variable x is defined
s 2p −∞ as the mean value of xn, that is,
10. For two random variables x and y, the CDF Fxy(x, y) +¥
∆
is defined as xn = ò xn px (x)dx
-¥
(
∆ P x ≤ x and y ≤ y
Fxy (x, y) = )
21. The nth central moment of a random variable x is
11. Joint PDF pxy(x, y) is defined as
defined as
¶2 +¥
pxy (x, y) = F (x, y) ∆
¶x¶y xy (x − x)n = ò (x − x)n px (x)dx
12. For two random variables x and y, given that the −¥
joint density is pxy(x, y), the individual probability 22. s x is known as the standard deviation of a random
densities or the marginal densities px(x) and py(y)
variable x and is given by s x2 = (x − x)2 = x2 − x
2
can be obtained from pxy(x, y) as follows:
+¥ 23. The autocorrelation function yg(t) of a real energy
px (x) = ò pxy (x, y)dy signal g(t) is
−¥
+¥
+¥
and py (y) = ò pxy (x, y)dx y g (t ) = ò g(t)g(t + t )dt
−¥
−¥
24. For complex energy signals, the autocorrelation 31. A white noise process W(t) is defined by
function is
No
SW (w ) = −¥ < w < ¥
+¥ +¥ 2
y g (t ) = ò g(t)g*(t −t )dt = ò g*(t)g(t + t )dt
where SW (w) is the power spectral density, No is a
−¥ −¥
real constant and called the intensity of the white
noise. The corresponding autocorrelation function
25. The autocorrelation function of an energy signal is given by
and its ESD form a Fourier transform pair,
No
that is, RW (t ) = d (t )
2
y g (t ) ¬¾¾®y g (f )
FT
where d (t ) is the Dirac delta.
26. The autocorrelation function of a real power signal 33. The average power of white noise
g(t) is
¥
ò
No
dw ® ¥
1
Pavg = EW 2 (t) =
2p
+T /2
ò
1
Rg (t ) = lim g(t)g(t − t )dt −¥
2
T ®¥ T −T / 2
+T / 2 34. The PSD and corresponding autocorrelation func-
ò
1
= lim g(t)g(t + t )dt tion of band-limited white noise is given by
T ®¥ T
−T / 2
No
SX (w ) = −B < w < B
2
27. The autocorrelation function of a complex power
signal g(t) is and
+T /2 No B sin Bt
RX (t ) =
ò
1
Rg (t ) = lim g(t)g* (t −t )dt 2p Bt
T ®¥ T
−T / 2
+T /2 35. The PSD and corresponding auto correlation function
ò
1
= lim g*(t)g(t + t )dt for coloured noise sequence are respectively given by
T ®¥ T
−T / 2
2a2b
28. For a power signal, the autocorrelation function SX (w ) = and RX (t ) = a2e−b t b>0
b +w
2 2
and the PSD form a Fourier transform pair, that is,
Rg (t ) ¬¾¾® Sg (f )
FT 36. The PSD and corresponding auto correlation func-
tion for white noise are respectively given by
29. The PSD Sx(f) of a random process x(t) is defined No
as the ensemble average of the PSDs of all sample SW (w ) = −p ≤ w ≤ p
2
functions, that is,
And
é X (f ) 2ù No
Sx (f ) = lim ê T ú W/Hz RW (m) = d [m ]
T ®¥ ê ú
T 2
ë û
37. Thermal noise is given by
30. The average power Px of a wide-sense random pro-
Pn µ TB or Pn = KTB
cess x(t) is its mean square value x2 , that is,
where Pn is the maximum noise power output of
+¥
ò
2 a resistor (W), K is Boltzmann’s constant (J/K),
Px = x = Rx (0) = Sx (f )df
T is absolute temperature (K) and B is the band-
−¥
width (Hz) over which noise is measured.
SOLVED EXAMPLES
1. The variance of a random variable x is s x2. Then the 3. For the distribution function given in Question 2,
variance of -kx (where k is a positive constant) is the value of the variance is
3 1
(a) s x2 (b) −ks x2 (a) (b)
2 3
(c) ks x2 (d) k2s x2 3
(c) (d)
1
4 4
Solution. We have Solution. Variance is given by
Var(−kx) = s 2 = E[(−kx)2 ] Var(x) = E[x2 ] − (E[x])2
Therefore, ∞ ∞
2
= ∫ x f (x)dx − ∫ xf (x)dx
2
s 2 = E[k2 x2 ] = k2 E[x2 ] = k2s x2 −∞
−∞
Ans. (d)
2. Let x be a continuous random variable with distri- E[x2] is given by
∞ ∞ ∞
2 3 1
bution function F(x) as given below. The value of
E[x] is E[x2 ] = ∫ x2f (x)dx = ∫ x x4 dx = 3 ∫ x2 dx
x ≤1 −∞
0 1 1
F (x) = 1 Therefore,
1− 3 x>1
x
x −1
x
1 1
E[x] = 3 lim = − (3) lim − = 3
x →∞ −1 x →∞ x 1
3 1 1
(a) (b)
3
2 3 2
3
Var(x) = 3 − =
3 1 2 4
(c) (d)
4 4 Ans. (c)
Solution. The density function f(x) is the derivative 4. If X is a random variable with mean m and vari-
of the distribution function, that is, ance s 2, and if Z is defined as Z = X − [(X − m ) s ].
E[Z] is given by
d d 1 3
f (x) = F (x) = 1− 3 = 4 (a) m
dx x x (b) 0
(c) m - s (d) m + s
dx
15
x(2 − x − y) 0 < x < 1, 0 < y < 1
f (x, y) = 2
0 otherwise
0 0.5 1 1.5 2 x
The probability distribution function is
The condition density of x, given that y = y, where
0 < y < 1 is x2 æ x2 ö
(a) x − , 0 < x ≤ 2 (b) ç x − ÷, x > 2
4 ç 4 ÷ø
(a) 0 (b) 1 è
6x(2 − x − y) 6x(2 − x − y) x æ xö
(c) 1 − , 0 < x ≤ 2 (d) ç 1 − ÷ , x > 2
2
(c) (d)
4 − 3y 4 + 3y è 2ø
Solution. For 0 < x < 1 and 0 < y < 1, we have Solution. The probability density function given in
the figure can be expressed as
fx ( x y ) =
f (x, y) f (x, y)
= +¥ x
1− , 0 ≤ x ≤ 2
fx (x) = 2
fy (y)
ò f (x, y)dx
0,
−¥ otherwise
The probability distribution function is given by 12. For the random variable x given in Question 11,
+¥ the probability in the range 0.5 < x ≤ 1.5 is
æ aö
x
ò fx (a )da = ò ç 1 − ÷ da , 0 < x ≤ 2
è 2ø (a) 2 (b) 1.5
−¥ 0 (c) 1 (d) 0.5
æ x2 ö
= çx − ÷ , 0 < x ≤ 2 Solution. Probability in the range 0.5 < x ≤ 1.5 is
è 4ø
P = fx(1.5) - fx(0.5) = 0.5
Ans. (a) Ans. (d)
ò f (x)dx = 1
Solution. Mean value −¥
Therefore,
æ xö
2
x = ò x ç 1 − ÷ dx = = 0.67
2
ò c (1 − x ) dx = 1
+1
è 2 ø 3 2
0
Ans. (0.67) −1
x3
1
2. For the data given in Question 1, the mean-squared
value of x is or, cx − c = 1
3
−1
Solution. Mean-squared value
Solving the above equation, we get
æ xö
2
E[x ] = ò x ç 1 − ÷ dx = = 0.67
2
c ö ææ c − ccö ö− æ −c + c ö = 1
2 2
è 2ø æ
çè c − ÷ø − çèçè−c +3 ÷ø ÷ø =çè 1 ÷
3
0
3 3 3ø
Ans. (0.67)
3 3
3. The PDF of a Gaussian random variable x is given Therefore, cTherefore,
= = 0.75 c = = 0.75
4
by Px (x) = (1 3 2p )e−(x−4)
2
18 4
. Find the probabil-
ity of the event {x = 4}.
Ans. (0.75)
PRACTICE EXERCISE
(a)
1
(b)
1 (a) {1, 1, 2, 3} (b) {3, 5, 9, 15, 9, 5, 3}
a a 2 (c) {3, 2, 1, 1} (d) {15, 9, 3, 9, 15}
(2 Marks)
1 1
15. Given two processes x (t) = A cos (2pfc t + q ) and
(c) (d)
a +1 (a + 1)2
x2 (t) = B sin (2pfc t + q ) , where q is an indepen-
1
(2 Marks)
dent random variable uniformly distributed over
7. The autocorrelation of a sinusoid is s the interval (0, 2p). Cross-correlation of the given
(a) sine pulse (b) sinusoid processes is
(c) p1 p5 + p2 p3 + p1 p3 p6 + p2 p4 p5 − p1 p2 p5 p6 e−1
for x = 0, 1,…
-p p p p − p p p p − p p p p − p p p p
(c) (d) None of the above
x!
1 3 5 6 1 2 4 5 1 2 3 6 2 4 5 6 (2 Marks)
+p p p p p + p p p p p
1 2 3 5 6 1 2 4 5 6 31. For the data given in Question 30, the marginal
(d) Cannot be determined from the given data. distribution of Y is
e−2 2y e−2y
(2 Marks)
27. If pi = p for all i, then the probability that there is (a) for y = 0, 1,… (b) for y = 0, 1,…
y! y!
some functioning path from node 1 to node 4, is
e−3 3y e−2y
(a) 2p2. (c) for y = 0, 1,… (d) for y = 0, 1,…
y! 2y !
(b) 2p2(1 + p + p3) - 5p4. (2 Marks)
(c) p2(1 + p + p3) - 3p4.
32. For the data given in Question 30, the joint prob-
(d) Cannot be determined from given data. ability mass function of X and Y - X is
(1 Mark)
e−3
28. The density function of a continuous random vari- (a) for x, z = 0, 1,…
able X is given by x!z !
c(x + x) for 0 < x < 1 e−1
f (x) = (b) for x, z = 0, 1,…
0 otherwise x!z !
e−2
The value of the constant c is (c) for x, z = 0, 1,…
x!z !
3 6 5 4
(a) (b) (c) (d) (d) None of the above
7 7 7 7 (2 Marks)
(2 Marks)
33. The relationship between X and Y - X is:
29. For the data given in Question 28, the probability
(a) They are independent.
density of 1/X is
(b) They are linearly dependent.
6 1 1 (c) They are exponentially dependent.
for y > 1
(a) 7 y 3 y 2 y
+ (d) Cannot be determined from the given data.
(2 Marks)
0 otherwise
34. For the data given in Question 30, the correlation
3 1 1
between X and Y is
for y > 1
(b) 7 y 3 y 2 y
+ 1 1 1 1
(a) (b) (c) (d)
0 otherwise 3 5 7 2
(2 Marks)
5 1 1
3 + 2 for y > 1 35. In a buffer there are a geometrically distributed
(c) 7 y y y number of messages waiting to be transmitted over
0 otherwise a communication channel, where the parameter p
of the geometric distribution is known. Your mes-
4 1 1 sage is one of the waiting messages. The messages
3 + 2 for y > 1
(d) 7 y y y
are transmitted one by one in a random order. Let
the random variable X be the number of messages
0 otherwise that are transmitted before your message. What is
(2 Marks) the expected value of X?
11 1
− 1 (b) − 1
2 p
(a)
p
30. The joint probability mass function of the lifetimes
X and Y of two connected components in a machine
can be modelled by p(x, y) = e−2 x !(y − x)!, for 11 1
2 p
+ 1 (d) + 1
x = 0, 1, … and y = x, x + 1, …. The marginal
(c)
p
distribution of x is
(2 Marks)
e−2 2e−2
(a) for x = 0, 1,… (b) for x = 0, 1,… 36. A stationary random variable x(t) has the following
autocorrelation function Rx (t ) = s 2e−m t , where
x! x!
1. Find the mean for a random variable x having PDF 5. Find the variance of the random variable in
as shown in the following figure. Question 3.
(1 Mark)
px(x)
6. If the variance s d2 of d[n] = x[n] - x[n - 1] is one-
1/4 tenth the variance s x2 of a stationary zero-mean
discrete-time signal x[n], then find the normalized
autocorrelation function Rxx (k) / s x2 at k = 1.
−1 1 2 3 (2 Marks)
(2 Marks) 7. Two resistors of 50 Ω and 100 Ω at temperatures
100 K and 50 K, respectively, are connected in series.
2. For the data given in Question 1, find the variance.
Their equivalent noise temperature is _____K.
(1 Mark)
(1 Mark)
3. A continuous random variable has a PDF
8. In a binary transmission channel, a 1 is transmit
f (x) = Kx2e-x , x ≥ 0. Find the value of K. ted with probability 0.8 and a 0 with probability
(2 Marks) 0.2. The conditional probability of receiving a 1
given that a 1 was sent is 0.95 and of receiving a 0
4. Find the mean value of the random variable in
when a 0 was sent is 0.99. What is the probability
Question 3.
that a 1 was sent when receiving a 1?
(1 Mark)
(2 Marks)
æ3 ö
1
1. (b) For 0 < x < 1, fx (x) = ò ç + xy ÷ dy = +
3 x
By differentiating Fx / y (a) with respect to a, the
è4 ø 4 2
0 density function of x/y is given by
æ ö
1
For 0 < x < 1, fx (x) = ò ç + xy ÷ dy = +
3 3 x
è4 ø 1
4 2 fx / y (a) = ,0 < a < ∞
(a + 1)2
0
fxy (x, y)
and fy (y x) = 7. (b) The autocorrelation function of a signal x(t) is
fx (x)
given by
3 + 4xy
0 < y < 1, 0 < x < 1 +¥
= 3 + 2x
0 otherwise
f xx (t) = ò x(t )x(t − t)dt
−¥
+¥
1 Given that the signal x(t) = Asinwt, therefore f xx (t) = ò A sin wtA
1 1
2. (d) P y > < x < + dx
2 2 2 −¥
+¥
∞ +¥
1 f xx (t) = ò A sin wtA sin w (t − t )dt
1
A2
= ∫ fy y dy = ∫ ò
(3 + 2y) 9
dy = = (cos w t
2 4 16 −¥ 2 −¥
+¥
1/ 2 1/ 2
A2 +p
ò (cos w t − cos 2wt cos w t − sin 2wt sin wt)dt
3. (d) For a narrow-band noise with Gaussian quadra- 2
K ò (cos
= A
2 =
ture components, the PDF of its envelope will be −¥ 2 −p
Rayleigh. 2 +p
+p
K ò (cos wt − cos 2wt cos wt − sin 2wt sin wt)dt
A
A2
ò
=
4. (b) Autocorrelation function of energy signal is 2 =K cos w
−p
conjugate symmetric, that is, Rx (t ) = Rx*(− t ) . 2 −p
2 +p
2 −òp
cos wtdt = K ¢ cos wt
If the function is real, then the autocorrelation func- A
tion has even symmetry, that is, Rx (t ) = Rx (−t ).
=K
As the energy function is real, therefore its autocor-
relation function has even symmetry. 8. (a) For mutually exclusive events A and B,
5. (d) Autocorrelation function and PSD make a P (A È B) = P (A) + P (B)
Fourier transfer pair, that is,
9. (b) Half-wave rectification can be represented as
Rx (t ) ¬¾¾® Gx (w )
FT
y = x for x ≥ 0
Therefore, for the given function 0 for x < 0
éæ sin f ö2 ù
−1 é 2 æ f öù
So,
−1
Rx (t ) = F êç ÷ ú = F êsin c ç ÷ ú
êëè f ø úû ë è p øû
e−y
1 1
f (y) = d (y) +
2
2N
Inverse Fourier transform of square of sine function 2 2pN
is always a triangular signal in time domain. ∞
E[y 2 ] = ∫y
2
6. (d) The distribution function of x/y for a > 0 is f (y)dy
0
x ∞
Fx / y (a) = P ≤ a 1
∫y e−y
1
2 d (y) +
2
2 2N
y = dy
2pN
∞ ay 0
Hence, Similarly,
¥
ò Nt e
1 2 −t2 2 1
E[y 2 ] = N dt Var(Y) =
2pN 0
18
¥ Therefore, correlation between x and y is
ò t e dt =
N 2 −t 22 N
= − 1 36
2p rxy =
1
2 =−
( 1 18 × 1 18 )
0
2
10. (b) The joint density of x and y is
13. (d) As,
1
fx, y (x, y) = =2 Cov(ax, by)
Area ∆ r axby =
Var(ax) Var(by)
11. (d) We have
Cov(x, y)
+¥ 1−x = ab = r xy
ò fx, y (x, y ¢)dy ¢ = ò 2dy = 2(1 − x)
fx (x) = a Var(x)b Var(by)
−¥ 0 Therefore, the correlation would not change if the
+¥ 1−y units are measured in inches.
fy (y) = ò fx, y (x ¢, y)dx ¢ = ò 2dx = 2(1 − y)
14. (b) Given that x[n] = {1, 1, 2, 3}
−¥ 0
Expected value of x is The autocorrelation sequence of x[n] is
¥
x2 x3
1
å
1
E[x] = 2∫ x(1 − x)dx = 2 − =
1 rxx [k ] = x [m] x [m − k ] = rxx [−k ]
0 2 3
0
3 m =−¥
2p −∫p
0 0
sin(2pfct ) + sin[2pfc (2t + t + 2q )]dq
AB 1
=−
é y2 2 y4 ù
1 1 2
= ò y(1 − 2y + y 2 )dy = ê − y 3 + ú
0 êë 2 3 4 ûú
0 =−
AB
sin(2pfc t )
2
1
=
12 16. (a) The ACF at the output of the RC integrator is
1 æ1ö æ1ö 1 −t
Cov(x, y) = E[xy] - E[x]E[y] = − ç ÷ ×ç ÷ = − N0
exp
12 è 3 ø è 3 ø 36 4RC RC
x3 x 4
1 1
E[x2 ] = 2∫ x2 (1 − x)dx = 2 − =
1 17. (a) Gaussian probability density of random vari-
0 3 4 0 6 able x is given by
1 é −(x − m)2 ù
1 1 exp ê ú
2
1 px (x) =
Var(x) = E[x ] − (E[x]) = − =
2 2
2ps êë 2s
2
úû
6 3 18
1 x2 Py x (0 1) = Py x (1 0) = Pe
px (x) = exp − (1)
2p 2
and Py x (0 0) = Py x (1 1) = 1 − Pe
Given that
x2 Also, Px (1) = Q and Px (0 ) = (1 − Q)
px (x) = K exp − (2)
2
We know that
Py (y j ) = å Px (xi )Py x (y j xi )
By comparing Eqs. (1) and (2), we have
1 i
K=
2p Therefore,
18. (c) The PDF of the envelope of narrow-band
Py (1) = Px (0)Py x (1 0) + Px (1)Py x (1 1)
Gaussian noise is Rayleigh.
= (1 − Q)Pe + Q(1 − Pe )
19. (a) The PSD and power of a signal, both are
directly proportional to the square of the ampli-
tude of the signal. Therefore, option (a) is correct 23. (c) Py (0) = Px (0)Py x (0 0) + Px (1)Py x (0 1)
= (1 - Q)(1 − Pe ) + QPe
20. (a) Given that g(t) = e−at u(t)
Autocorrelation of a function g(t) is given by the 24. (d) P (e ) = å P (xi )P (e xi )
expression i
+¥
= Px (0)P (e 0) + Px (1)P (e 1)
Rg (t ) = ò g(t)g(t − t )dt = 0.4(10 -6 ) + 0.6(10 -4 ) = 60.4 × 10−6
−¥
Therefore, autocorrelation for the given function is 25. (b) Given that P(e 0) = 10−6. Therefore, one out
+∞ ∞
e−at
of 1 million received 0s will be in error.
∫ e−at e−a(t−t )u(t)dt = eat ∫ e−2at dt = Given that P(e 1) = 10−4 . Therefore, one out of
−∞ 0
2a
10000 received 1s will be in error.
21. (b) Given that The error probability P(e) of the channel is
60.4 × 10-6. Therefore, one out of 16556 digits will
t
g(t) = AΠ be in error.
T
Therefore, statements S1 and S3 are correct.
Autocorrelation of a function g(t) is given by
26. (a) There are four paths from node n1 to n4. These
+¥ paths are path (l1, l5), path (l2, l6), path (l1, l3, l6)
Rg (t ) = ò g(t)g(t − t )dt = Rg (−t ) and path (l2, l4, l5). Let Aj be the event that the
−¥ jth path is functioning. The desired probability is
T given by P (A1 È A2 È A3 È A4 ).
For t < T , Rg (t ) = ò A dt = A (T − t )
2 2
Therefore,
t
∞ 4
For t ≥ T , Rg (t ) = ∫ g(t)g(t − t )dt = 0 P (A1 È A2 È A3 È A4 ) = å P (Aj )
j =1
T
3 4 2 3 4
Therefore, −å å P (Aj Ak ) + å å å P (Aj Ak Al )
j =1 k = j + 1 j =1 k = j +1 l =k + 1
A2 (T − t ) t < T −P (A1A2 A3 A4 )
Rg (t ) =
0 t ≥T
22. (a) Let x be the transmitted digit and y be the Hence, the probability that there is some func-
received digit. tioning path from node n1 to node n4 is equal to
We know that P (X = x, Y − X = z) = P (X = x)
+¥ P (Y − X = z) for all x, z
ò f (x)dx = 1
Therefore, X and Y - X are independent.
−¥
Therefore, 34. (d) E[XY] = E[X]E[Y - X] + E[X2], as X and Y
1 are independent.
ò c(x + x )dx = 1
As X and (Y - X) are both Poisson distributed
0
with expected value 1, therefore E[XY] = 1 + 2 = 3.
Therefore, 7/6 c = 1 or c = 6/7 The expected value and variance of the Poisson
distributed variable X are given by E[X] = 1 and
29. (a) P (Y ≤ y) s2(X) = 1.
ì æ 1ö æ 1ö æ1ö The expected value and variance of the random
ïP X ≤ ÷ = 1− P çX ≤ ÷ = 1 - F ç ÷ for y >1
= í çè yø è yø èyø
variable Y are E[Y] = E[Y - X + X] = E[Y - X] +
ï 0 E[X] = 1 + 1 = 2 and s 2(Y ) = s 2(Y - X)+ s 2(X)
î for y ≤ 1
= 1 + 1 = 2.
where F(x) is the probability distribution function Therefore,
of X.
3 − 1× 2 1
p(X , Y ) = =
By differentiation, the density function g(y) of Y 1× 2 2
is given by
6 1 1 6 1 1
35. (a) Let the random variable Y denote the number
f × = 3 + 2 for y > 1
g(y) = 7 y y 2
of messages waiting in the buffer.
7y y y
Then, P (Y = y) = p(1 − p)y −1 for y ≥ 1
0 otherwise
By law of conditional probability
30. (c) The marginal distribution of X is
¥
¥
e -2 å P (X = x Y = y)p(1 − p)y −1
P (X = x) = å
P (X = x) =
y = x x !(y − x)!
y = x +1
¥
¥
e−2 e−1 å p(1 − p)y −1 for x = 0, 1, ¼
1
å
1
for x = 0, 1,…
=
= = y
x! y = x +1
k =0 k ! x!
Using the formula P(Z = x/y |Y = y) = P(Z = x/y) = 0.5. This gives
¥ n P(X = x, Y = y) = 0.5 P(Y = y).
å
u
= − ln(1 − u) for u <1 Also, by P(X = 1) = P(Y = 1, Z = 1) + P(Y = -1,
n =1 n !
Z = -1) and the independence of Y and Z, it fol-
the expression for P(X = x) can be written as lows that P(X = 1) = 0.52 + 0.52. This shows that
p x P(X = x) = 0.5 for x = -1, 1 and so by P(X = x,
∑ y p(1 − p)y −1
1
P (X = x) = − −
1 − p
ln p Y = y) = 0.5 P(Y = y), P(X = x, Y = y) = P(X = x)
y =1 P(Y = y) for x, y ∈ {-1, 1}, proving that X and Y
Using the fact that the discrete uniform distribu- are independent.
tion on 0, 1, … y - 1 has an expected value (1/2) 39. (a) By the same logic given in Question 38, X and
(y - 1), the expected value of X is calculated from Z are independent.
∞
E[X ] = ∑ E[X Y = y ]p(1 − p)y −1 40. (b) X is not independent of Y + Z as P(X = 1,
Y + Z = 0) = 0 and P(X = 1)P(Y + Z = 0) > 0.
y =1
∞
11
∑ 2 (y − 1)p(1 − p)y −1 = 2 p − 1
1 41. (d) The marginal density of x is
=
x
fx (x) = ∫
6 (x − y)dy = 3x2 for 0 < x < 1
y =1
ò ò xKx e
3 3 2 −x
1 (x − 1)
+3
∫
1 4. Mean value E[x] = xf[x]dx = dx
= (x − 1)2 dx =
4 4 3 −¥ 0
−1 −1 ¥
4 = 0.5 ò x3e−x dx = 3
= = 1.33
3 0
Ans. (1.33) Ans. (3)
¥ ¥
ò ò x Kx e
2 −x
5. E[x2 ] = x2f (x)dx = 2
dx
−¥ 0
R1 R2
¥ Vn1
2 2
Vn2
= 0. 5 ò x e 4 −x
dx = 12
0
Total noise voltage Vn2 = Vn12 + V n22
Variance of x is Var(x) = E[x2] - (E[x])2 = 12 - 9 = 3 As Vn12 = 4KT1BR1 and Vn22 = 4KT2 BR2 and Vn2 = 4KTe BR = 4
Ans. (3) Vn2 = 4KTe BR = 4KTe B(R1 + R2 )
Therefore,
6. s d2 = E[[x[n ] − x[n − 1]]2 ] = E[x[n ]]2 + E[x[n − 1]]2
4KTe B(R1 + R2 ) = 4KT1BR1 + 4KT2 BR2 (1)
- 2E[x[n ]x[n − 1]]
s x2 R1T1 + R2 T2
s d2 = Te = (2)
10 R1 + R2
s x2 50 ´ 100 + 100 × 50
= s x2 + s x2 − 2Rxx (1) (Given that k = 1) Te =
50 + 100
= 66.67 K
10 Ans. (66.67)
Therefore,
8. Let H be the event that a 1 is sent and E be the
19 2 R (1) 19 event that a 1 is received. The desired posterior
2Rxx (1) = s or xx2 = = 0.95 probability P (H E ) satisfies
10 x sx 20
1. The noise at the input to an ideal frequency detec- 2. Let X and Y be two statistically independent
tor is white. The detector is operating above random variables uniformly distributed in the
threshold. The power spectral density of the noise ranges (-1, 1) and (-2, 1), respectively. Let
at the output is Z = X + Y. Then the probability that (Z ≤ -2) is
(a) raised cosine (b) flat
(c) parabolic (d) Gaussian 1
(a) Zero (b)
(GATE 2003: 1 Mark) 6
Ans. (d)
4. Let Y and Z be the random variables obtained by
As we know, the area under fy(y) = 1. Therefore, sampling X(t) at t = 2 and t = 4, respectively. Let
from the above figure K ¢ = 1 3 W = Y - Z. The variance of W is
The distribution function of Z = X + Y is (a) 13.36 (b) 9.36
(c) 2.64 (d) 8.00
fx(x) * fy(y) (GATE 2003: 2 Marks)
Solution. Given that W = Y - Z
1 therefore, s W
2
= E[Y 2 ] + E[Z 2 ] − 2E[Y × Z ]
3
Y is sampled at t = 2 and Z at t = 4,
−3 −2 −1
therefore, s W = s y2 + s z2 − 2Rxx (2) = 8 + 8 − 2[4e−0.2 2 + 1]
2 z 2
sW
2
= s y2 + s z2 − 2Rxx (2) = 8 + 8 − 2[4e−0.2 2 + 1]
Point -2 lies in the middle of -1 and -3. Therefore,
the value of the fx(x) * fy(y) function at Z = -2 is
Hence,
1/6. Therefore, sW
2
= 2.64
Ans. (c)
Prob [Z ≤ −2] = Area of graph [z ≤ −2] 5. The distribution function Fx (x) of a random vari-
1 1 1 able x is shown in the following figure. The prob-
= × ×1 = ability that x = 1 is
2 6 12
Fx(x)
Ans. (d)
Common Data for Questions 3 and 4: Let X
be the Gaussian random variable obtained by sam-
1.0
pling the process x(t) at t = ti and let
0.55
∞
∫
1 −y 2/2
Q(a ) = e dy
2p
0.25
a
x
−2 0 1 3
0 0
Solution. Probability that x = 1 is given by
1
P(x = 1) = Fx(x = 1+) - Fx(x = 1-) + ∫ (x − 0.7)2dx = 0.039
= 0.55 - 0.25 = 0.30 0. 3
Ans. (d) Therefore, the root mean square value of the quan-
tization noise = 0.039 = 0.198
6. A 1mW video signal having a bandwidth of Ans. (b)
100 MHz is transmitted to a receiver through a
cable that has 40 dB loss. If the effective one-sided 8. Noise with uniform power spectral density of
noise spectral density at the receiver is 10-20 W/Hz, No W/Hz is passed through a filter H(w) = 2exp
then the signal-to-noise ratio (SNR) at the receiver is (-jwtd) followed by an ideal low-pass filter of band-
width B Hz. The output noise power in watts is
(a) 50 dB (b) 30 dB
(c) 40 dB (d) 60 dB (a) 2 NoB (b) 4 NoB
(GATE 2004: 2 Marks) (c) 8 NoB (d) 16 NoB
(GATE 2005: 2 Marks)
Solution. Given that bandwidth B = 100 MHz,
power of signal Ps = 1MW and noise density at the Solution. We have
receiver, No = 10-20 W/Hz. Signal-to-noise ratio Signal-to-noise ratio at the output of the filter =
(SNR) of the transmitter is given by (SNR)out
10 -3
Ps Signal-to-noise ratio at the input of the filter =
SNR = = −20 = 109 (SNR)in
No B 10 × 100 × 106 Transfer function of the filter = H(w)
(SNR)dB = 10log 109 = 90 dB (SNR )out = H(w )
2
(SNR )in = 4No
Cable loss = 40 dB
Therefore, SNR at the receiver = (90 - 40) dB = 50 dB
Output noise power in watts
Ans. (a) PN = Bandwidth × (SNR )out = 4NoB
Ans. (b)
7. A random variable X with uniform density in the
interval 0 to 1 is quantized as follows: 9. Output of a communication channel is a random
If 0 ≤ X ≤ 0.3, xq = 0 variable v with the probability density function as
If 0.3 <X ≤ 1, xq = 0.7 shown in the following figure. The mean square
value of v is
where, xq is the quantized value of X. The root
mean square value of the quantization noise is p(v)
Quantization noise Qe = signal value − quantized value E[v2 ] = ò v2fv (v) dv (1)
= x − xq 0
From the figure,
Mean square value of quantization noise Kv
1
fv (v) =
4
= E[(x − xq )2 ] = ò (x − xq )2 fx (x) dx Now, 4K/2 = Area under the curve = 1
0
ò òx
1
4 E[x2 ] = x2fx (x)dx = 2
dx
æ vö
4
v4
E[v2 ] = ò v2 ç ÷ dv = −a −a
4
è 8ø 4×8
0 0 2 /3 2 /3
1 x3
ò 4 x dx = 4 3
1 2
or, 2
E[v ] = 8 E[x2 ] =
Ans. (c) −2 / 3 −2 / 3
4 −∫a
1 1 (d) series RLC filter
dx =
3 (GATE 2006: 2 Marks)
Hence,
2a 1 2 Solution. Given that
= or a =
4 3 3 16
Ans. (b) S(w ) =
16 + w 2
11. The quantization noise power for the quantization Therefore, transfer function
region between -a and +a in the figure is
4
4 1 H(s) =
(a) (b) 4+s
81 9
H(s) is the transfer function of a first order low-
5 2
(c) (d) pass filter. From the given options, option (a) is the
81 81 correct one.
(GATE 2005: 2 Marks) Ans. (a)
14. The parameters of the system obtained in Question Solution. Sample space of random variable
13 would be X = (−¥, ¥)
After transformation, the sample space of random
(a) First-order RL low-pass filter would have
R = 4 Ω, L = 1 H
variable Y = (0, 1)
Hence,
(b) First-order RC high-pass filter would have fY (y) = Ad (y) + B[d (y − 1)]
R = 4 Ω, C = 0.25 F Therefore, option (c) satisfies the given condition.
(c) Tuned LC filter would have L = 4 H, C = 4 F Ans. (b)
(d) Series RLC low-pass filter would have R = 1 Ω,
16. If E denotes expectation, the variance of a random
L = 4 H, C = 4 F
variable X is given by
(GATE 2006: 2 Marks)
(a) E[X2] - E2[X] (b) E[X2] + E2[X]
Solution. From the solution of Question 13, we (c) E[X2] (d) E2[X]
know that the system is a first-order low pass filter. (GATE 2007: 1 Mark)
The figure below shows the circuit of a generalized
low-pass filter. Solution. Variance of the random variable X is
L s x2 = E [X 2 ] − E 2 [X ]
Ans. (a)
R 17. If R(t) is the autocorrelation function of a real,
wide-sense stationary random process, then which
of the following is NOT true?
(a) R(t ) = R(−t )
Transfer function of the filter (b) R(t ) ≤ R(0)
R (c) R(t ) = −R(−t )
H( jw ) =
R + jwL (d) The mean square value of the process is R(0)
(GATE 2007: 1 Mark)
Comparing with the transfer function in Question 13,
we get R = 4 Ω and L = 1 H. Solution. Autocorrelation function of a real wide-
Ans. (a) sense stationary random process is an even func-
tion, therefore R(t ) = −R(−t ) is not true.
15. A uniformly distributed random variable X with Ans. (c)
probability density function fx (x) = (1 / 10)[u(x + 5) − u(x − 5)]
(x) = (1 / 10)[u(x + 5) − u(x − 5)], where u(.) is the unit step function 18. If S(f) is the power spectral density of a real, wide-
passed through a transformation given in the fol- sense stationary random process, then which of the
lowing figure. The probability density function of following is ALWAYS true?
the transformed random variable Y would be (a) S(0) ≥ S(f ) (b) S(f ) ≥ 0
¥
Y (c) S(-f) = -S(f ) (d) ò S(f ) df = 0
−¥
1
(GATE 2007: 1 Mark)
Solution. Power spectral density (PSD) of a real
X
−2.5 2.5 wide-sense stationary random process is always a
positive quantity, therefore S(f) ≥ 0.
1 Ans. (b)
(a) fy (y) = [u(y + 2.5) − u(y − 2.5)]
5 19. The probability density function (PDF) of a
(b) fy (y) = 0.5d (y) + 0.5d (y − 1) random variable X is as shown below.
PDF
(c) fy(y) = 0.25d (y + 2.5) + 0.25d (y − 2.5) + 0.5d (y) 1
2 fc2
(GATE 2008: 2 Marks) H(f ) =
f 2 + fc2
Solution. CDF function is given by
fc2
Output PSD = H(f ) ⋅ (Input PSD) = ⋅K
x 2
Fx (x) = ò fx (x) dx f 2 + fc2
−¥
Output noise power
where fx(x) is the probability density function.
Integral of increasing ramp signal is an increasing ¥ ¥
fc2
parabola and integral of decreasing ramp signal is = ò (output PSD) df = K ò f 2 + fc2
df
a decreasing parabola. −¥ −¥
Therefore, option (a) is the correct answer. = Kpfc ¥ ¥
fc2
Ans. (a) By substituting f = fc=
ò and solving,
tanq (output PSD)we
dfget
=K ò f 2 + fc2
df
20. Px (x) = M exp(−2 x ) + N exp(−3 x ) is the proba- −¥ −¥
bility density function for the real random variable Output noise power = Kpfc
X, over the entire x-axis. M and N are both positive Ans. (c)
real numbers. The equation relating M and N is 22. A white noise process x(t) with two-sided
2 1 PSD 1 × 10-10 W/Hz is input to a filter whose
(a) M + N = 1 (b) 2M + N = 1
3 3 magnitude squared response is shown in the fol-
(c) M + N = 1 (d) M + N = 3 lowing figure.
(GATE 2008: 2 Marks)
Solution. Given that the PDF is
H(f)2
Px (x) = M exp(−2 x ) + N exp(−3 x ) X(t) 1 Y(t)
We know that
¥ f
ò Px (x) dx = 1 −10 kHz 10 kHz
−¥
Therefore,
∞ The power of the output process y(t) is given by
∫ (Me−2 x + Ne−3 x )dx = 1
−∞
(a) 5 × 10-7 W (b) 1 × 10-6 W
¥ (c) 2 × 10-6 W (d) 1 × 10-5 W
ò (Me
−2 x
+ Ne−3x )dx =
1
or,
2 (GATE 2009: 1 Mark)
0
PSD of output P (X + Y = 2, X − Y = 0)
P (X + Y = 2 X − Y = 0) =
P (X − Y = 0)
Go (f ) = H(f ) × Gi (f ) = 1 × 10−10 H(f )
2 2
P (X − Y = 0) = P (X = 0, Y = 0) + P (X = 1, Y = 1)
Output noise power No + P (X = 2, Y = 2)
= P (X = 0) × P (Y = 0) + P (X = 1) × P (Y = 1)
+fo
τ τ
P(X = k) 0.1 0.2 0.4 0.2 0.1
Therefore,
Sy (f ) = 0
+ Ans. (d)
x(t) + d y(t)
dt Statement for Linked Answer Questions 27
+ and 28. Consider a baseband binary PAM receiver
shown in the following figure. The additive channel
D
Delay = 0.5 ms noise n(t) is white with power spectral densitySN(f)
= No/2 = 10-20 W/Hz. The low-pass filter is ideal
Let SY(f ) be the power spectral density of Y(t). with unity gain and cut-off frequency 1 MHz. Let
Which one of the following statements is correct? Yk represent the random variable y(tk). Yk = Nk if
transmitted bit bk = 0, and Yk =a + Nk if transmit-
(a) SY(f ) > 0 for all f ted bit bk = 1, where Nk represents the noise sample
(b) SY(f ) = 0 for f > 1 kHz value. The noise sample has a probability density
(c) SY(f ) = 0 for f = nfo, fo = 2 kHz, where n is function PN (n) = 0.5ae−a n . (This has mean zero
k
any integer and variance 2/a2.) Assume transmitted bits to be
(d) SY(f ) = 0 for f = (2n + 1)fo, fo = 1 kHz, where equiprobable and threshold z is set to a/2 =10-6 V.
n is any integer n(t)
(GATE 2010: 2 Marks)
r(t) y(t) y(tk)
Solution. The given block diagram is redrawn as +
x(t) LPF S/H bk
shown below:
Sampling Threshold
+ time z
x(t) y1(t) d
+ y(t)
dt Receiver
+
Delay where
0.5 ms 1 if y(tk) ≥ z
bk =
0 if y(tk) ≤ z
From the figure
27. The value of the parameter a (in V-1) is
y (t) = x(t) + x(t − 0.5 × 10−3 )
1
(a) 1010 (b) 107
Therefore, (c) 1.414 × 10-10 (d) 2 × 10-20
Y1(f ) = X(f )[1 + e−j2pf (0.5 ×10
−3
) (GATE 2010: 2 Marks)
]
Transfer function Solution. We have
Output noise PSD = Input noise PSD × H(f )
2
= 1 + e−jpf ×10
Y (f ) −3
H1(f ) = 1
SNO (f ) = SNI (f ) H(f ) = 10−20 H(f )
2 2
X(f )
Now, The following figure shows the curves for H(f) and
Y (f ) SNO(f)
H2 (f ) = = j2pf H(f)
Y1 (f )
Sx (ω)
1 1
W = N 0, 9 × + 4 ×
400d(w -104) 4 9
PDF of W
0 9 10 11 w (103 rad/s)
6000 6400
(a) , 0 (b) ,0
p p P (W ≥ 0)
6400 20 6000 20
(c) , (d) ,
p p 2 p p 2
W
(GATE 2012: 1 Mark)
Solution. We know that E[X2(t)] represents the
total power in the random signal.
Therefore, P (W ≥ 0) = Area under the curve from 0 to ¥ =
1
Therefore,
2
11×103
¥
1
≥
2p 9×∫103 x
1
P = E[X 2 (t)] = 2 × S (w ) dw
P (W 0) = Area under the curve from 0 to =
2 Ans. (b)
1 1 6400
= 400 + × 6 × 2 × 103 = 33. Consider two identically zero-mean random vari-
π 2 p ables U and V. Let the cumulative distribution
At w = 0, there is no frequency component pres- functions of U and 2V be F(x) and G(x), respec-
ent, hence dc value of the process is zero. Hence, tively. Then, for all values of x
|E[X(t)] = 0 (a) F(x) - G(x) ≤ 0 (b) F(x) - G(x) ≥ 0
(c) [F(x) - G(x)] . x ≤ 0 (d) [F(x) - G(x)] . x ≥ 0
Ans. (a)
32. Let U and V be two independent zero mean
(GATE 2013: 2 Marks)
Gaussian random variables of variances 1/4 and
1/9, respectively. The probability P(3V ≥ 2U) is
Solution. We have
4
(a) (b)
1 F(x) = P{x ≤ x} and G(x) = P{2x ≤ x} = P{x ≤ x/2}
9 2
For positive values of x, F(x) - G(x) ≥ 0
2
(c) (d)
5 For negative values of x, F(x) - G(x) < 0
3 9 But, [F(x) - G(x)] . x ≥ 0, for all values of x.
(GATE 2013: 2 Marks) Ans. (d)
In this chapter, we discuss the analog communication systems. The topics covered include amplitude and angle
modulation and demodulation systems, spectral analysis of these operations, superheterodyne receivers, elements
of hardware, realizations of analog communication systems, signal-to-noise ratio (SNR) calculations for amplitude
modulation (AM) and frequency modulation (FM) for low noise conditions.
mVc
m(t) v(t) = Vc cos w c t + cos(w c − w m )t
2
mVc
t + cos(w c + w m )t (43.4)
2
(a)
Carrier The frequency spectrum of an AM signal in case of a
single frequency modulating signal thus contains three
frequency components, namely, the carrier frequency
t component (wc), the sum frequency component (wc +
wm) and the difference frequency component (wc − wm).
(b) The sum component represents the upper sideband and
the difference component the lower sideband. Figure 43.2
shows the frequency spectrum.
AM
(c)
Amplitude
Figure 43.1 | Amplitude modulation. Vc
(a) Modulating signal, (b) Carrier signal, (c) Modulated (m/2)Vc (m/2)Vc
signal.
(DSBSC) or simply the DSB signal. The DSBSC signal, signal falls below this peak value, it falls quickly below
in turn, can be generated by multiplying the modulating the capacitor voltage (which is nearly equal to the peak
signal m(t) and the carrier coswct. Figure 43.4 shows the voltage), thus causing the diode to open. The capacitor
arrangement of generating the DSBSC signal. now discharges through the resistor R at a slow rate with
time constant RC. The same operation repeats in the
m(t) DSBSC signal other cycles also. The output voltage vc(t) therefore fol-
lows the envelope of the AM signal. The slow discharge
of the capacitor through the resistor allows the capacitor
voltage to follow a declining envelope.
Capacitor discharge between the positive peaks causes
coswct a ripple signal of frequency wc in the output. This ripple
can be reduced by choosing a larger time constant RC
so that the capacitor discharges very little between the
Figure 43.4 | Generation of DSBSC signal. peaks (RC >> 1/wc). However, very large value of RC
will make it impossible for the capacitor voltage to follow
Demodulation of the standard AM signal is very simple
a fast declining envelope. If B is the bandwidth of the
and is implemented by using what is known as envelope
message signal, then for proper detection of the signal by
detection technique. In a standard AM signal, when
envelope detector,
the amplitude of the unmodulated carrier signal is very
large, the amplitude of modulated carrier signal is pro-
2pB < << w c (43.9)
1 1 1
<< RC < , or
portional to the modulating signal. Demodulation in this
wc 2pB RC
case simply reduces to detection of envelope of modu-
lated carrier signal regardless of the exact frequency The output of the envelope detector is A + m(t) with a
or phase of the carrier. Figure 43.5 shows the envelope ripple frequency of wc. The DC term A can be blocked
detector circuit used for demodulating the standard AM out by a capacitor or a simple high-pass RC filter. The
signal. The capacitor C filters out the high-frequency ripple may be reduced by a low-pass RC filter.
carrier variations.
In general, for the envelope detection to properly
D detect the modulating signal m(t), the following two con-
ditions must be met:
Envelope
1. wc >> bandwidth of m(t)
Envelope LPF
DSBSC m(t)
signal
Demodulated
Demodulated signal
C
signal
coswct
signal and then suppress one of the sidebands by the The difference signal represents the upper sideband SSB
process of band pass filtering. This method known as fre- signal while the sum represents the lower sideband SSB
quency discrimination method is illustrated in Fig. 43.7. signal. For instance, if m(t) is taken as coswmt, then
In practice, this approach poses some difficulty because m′(t) would be sinwmt. The SSB signal in case of minus
the filter needs to have sharp cut-off characteristics. sign would then be
SSBFC signal cos w m t × cos w c t − sin w m t × sin w c t = cos(w m + w c )t
BPF
m(t)
and in case of plus sign, it would be
cos w m t ⋅ cos w c t + sin w m t ⋅ sin w c t = cos(w m − w c )t
coswct
43.2.4.3 R3E System (Single Sideband
Figure 43.7 | Frequency discrimination method for
Reduced Carrier System)
generating SSBFC signal.
R3E System is the single sideband reduced carrier system
Another method for generating an SSB signal is the also called pilot carrier system. Reinsertion of carrier
phase shift method. Figure 43.8 shows the basic block with much reduced amplitude before transmission is
schematic arrangement. The blocks labelled `−p/2’ are aimed at facilitating receiver tuning and demodulation.
phase shifters that add a lagging phase shift of p/2 to This reduced carrier amplitude is 16 or 26 dB below the
every frequency component of the signal applied at the value it would have had it not been suppressed in the
input to the block. Hence, it is a Hilbert transformer. first place. This attenuated carrier signal while retaining
Hilbert transform x̂ (t ) of a signal x(t) is defined by the the advantage of saving in power provides a reference
equation signal to help demodulation in the receiver.
∞
ò (t − s) ds
1 X (s)
x̂ (t ) =
p −∞ 43.2.4.4 J3E System (Single Sideband
Suppressed Carrier System)
Hilbert transformer is a system whose transfer function is
H(v)=-j × sgn(v), where sgn is the signum function given by
It is the single sideband suppressed carrier (SSBSC)
ì−1 v < 0 ü
sgn(v) = í ý
system. This is the system usually referred to as SSB,
î 1 v > 0þ in which carrier is suppressed by at least 45 dB in the
The output block can be either an adder or a subtrac- transmitter. It was not popular initially due to the
tor. If m(t) is the modulating signal and m′(t) is the requirement of high receiver stability. However, with
modulating signal delayed in phase by p/2, then the SSB the advent of synthesizer-driven receivers, it has now
signal produced at the output can be represented by the become the standard form of radio communication.
following expression: Generation of SSBFC signals was briefly described
xSSB (t) = m(t) cos w c t ± m¢(t) sin w c t (43.10) above under H3E systems. Suppression of carrier in an
AM signal is achieved in the building block known as
The output with `+’ sign is produced when the output the balanced modulator. Figure 43.9 shows the typical
block is an adder and with `−’ when the output block is circuit implemented using FETs. The modulating signal
a subtractor. is applied in push pull to a pair of identical FETs as
shown, and as a result, the modulating signals appearing
at the gates of the two FETs are 180° out of phase. The
carrier signal, as is evident from the circuit, is applied to
the two gates in phase. The modulated output currents
cos wct of the two FETs produced as a result of their respective
−
+ gate signals are combined in the centre tapped primary
−π/2
m(t) 2 SSB signal of the output transformer. If the two halves of the circuit
are perfectly symmetrical, it can be proved with the help
of simple mathematics that the carrier signal frequency
−π/2
will be completely cancelled in the modulated output and
the output would contain only the modulating frequency,
sum frequency and difference frequency components. The
Figure 43.8 | Phase shift method for generating SSBFC modulating frequency component can be removed from
signal. the output by tuning the output transformer.
43.2.4.5 B8E System
i1
RF in This system uses two independent sidebands with carrier
either attenuated or suppressed. This form of AM is
SSBSC also known as independent sideband (ISB) transmission
m(t) and is usually employed for point-to-point radio
signal
telephony.
i2
43.2.4.6 C3F System
Figure 43.9 | Balanced modulator. Vestigial sideband (VSB) transmission is the other
name for this system. It is used for transmission of
Demodulation of SSBSC signals can be implemented
video signal in commercial television broadcasting to
by using a coherent detector scheme as outlined in case
conserve bandwidth. It is a compromise between SSB
of demodulation of DSBSC signal in earlier paragraphs.
and DSB modulation systems in which a vestige or part
Figure 43.10 shows the arrangement.
of the unwanted sideband is also transmitted usually
with a full carrier along with the other sideband. The
typical bandwidth required to transmit a VSB signal
LPF is about 1.25 times that of an SSB signal. Figure 43.11
SSB signal Demodulated
signal shows the spectrum of transmitted signals in case of
NTSC TV standards followed in the United States,
Canada and Japan (Fig. 43.11a) and PAL TV stan-
cosωct dards followed in Europe, Australia and elsewhere
(Fig. 43.11b). As can be seen from the two figures, if
Figure 43.10 | Coherent detector for demodulation of
the channel width is from say A to B MHz, the picture
carrier is at (A + 1.25) MHz and the sound carrier is
DSBSC signal.
at (B − 0.25) MHz.
P C S
Relative 4.5
amplitude 0.75 4 0.25
3.58
A B
0 1 2 3 4 5 6
1.25 f(MHz)
(a)
P C S
Relative 5.5
amplitude 0.75 5 0.25
4.43
A B
0 1 2 3 4 5 6 7
1.25
f(MHz)
(b)
Figure 43.11 | (a) NTSC TV standard signal. (b) PAL TV standard signal.
VSB signal can be generated by passing a DSB signal following essential elements are common to all super-
through an appropriate sideband shaping vestigial side- heterodyne circuits: a receiving antenna; a tuned stage,
band filter as shown in Fig. 43.12. The demodulation which may optionally contain amplification (RF ampli-
scheme for VSB signal is shown in Fig. 43.13. fier); a variable frequency local oscillator; a frequency
mixer; a band pass filter and intermediate frequency (IF)
amplifier; and a demodulator plus additional circuitry
VSB
to amplify or process the original audio signal (or other
m(t) filter VSB signal transmitted information).
Receiving antenna
RF IF Audio
coswct amplifier Mixer Filter amplifier amplifier
f + 2fIF , if fLO > f (high side injection) instantaneous amplitude of the modulating or baseband
fimg = signal. The rate at which these frequency variations take
f − 2fIF , if fLO < f (low side injection) place is of course proportional to the modulating fre-
quency. If the modulating signal is expressed by vm =
The mixer will process not only the desired input signal Vmcoswmt, then the instantaneous frequency ( f ) of an
at f RF, but also all signals present at its inputs. There FM signal is mathematically expressed as follows:
will be many mixer products (heterodynes). Most other
signals produced by the mixer (such as due to stations at f = fc (1 + KV m cos w m t) (43.11)
nearby frequencies) can be filtered out in the IF ampli-
where fc is the unmodulated carrier frequency, Vm the
peak amplitude of modulating signal, wm the modulating
fier; that gives the superheterodyne receiver its superior
performance. However, if fLO is set to fRF + fIF, then an
frequency and K the constant of proportionality.
incoming radio signal at fLO + fIF will also produce a
heterodyne at fIF; this is called the image frequency and The instantaneous frequency is maximum when
must be rejected by the tuned circuits in the RF stage. coswmt = 1 and minimum when coswmt = −1. This gives
The image frequency is 2fIF higher (or lower) than fRF,
fmax = fc (1 + KVm ) (43.12)
so employing a higher IF frequency fIF increases the
receiver’s image rejection without requiring additional
selectivity in the RF stage. and fmin = fc (1 − KVm ) (43.13)
The intermediate frequency amplifier is tuned to a Frequency deviation (d ) is one of the important param-
fixed frequency, known as the intermediate frequency eters of an FM signal and is given by (fmax − fc) or (fc −
(fIF) that does not change as the receiving frequency fmin). This gives frequency deviation, as
changes. The IF amplifier is selective around its center
frequency fIF. The received signal is now processed d = KVm fc (43.14)
by the demodulator stage where the audio signal (or
other baseband signal) is recovered and then further Figures 43.15(a)−(c), respectively, show the modulating
amplified. signal (taken as a single tone signal in this case), the
unmodulated carrier and the modulated signal.
Superheterodyne receivers have essentially replaced
all previous receiver designs. The superheterodyne
receiver offers superior sensitivity, frequency stability m(t)
and selectivity. Compared with the tuned radio fre-
quency receiver (TRF) design, superheterodyne receivers
offer better stability because a tuneable oscillator is more
easily realized than a tuneable amplifier. Operating at a (a)
lower frequency, IF filters can give narrower passbands Carrier
at the same Q factor than an equivalent RF filter. A
fixed IF also allows the use of a crystal filter or similar
technologies that cannot be tuned.
One major disadvantage of the superheterodyne
receiver is the problem of image frequency. In het-
erodyne receivers, an image frequency is an undesired
(b)
input frequency equal to the station frequency plus
twice the intermediate frequency. The image frequency FM
results in two stations being received at the same time,
thus producing interference. Image frequencies can be
eliminated by sufficient attenuation on the incoming
signal by the RF amplifier filter of the superheterodyne
receiver.
(c)
43.4 FREQUENCY MODULATION
Figure 43.15 | Frequency modulation.
In frequency modulation (FM), the instantaneous fre- An FM signal can be mathematically represented by
quency of the modulation signal varies directly as the the following expression:
v(t) = A sin[w c t + (d fm )sin w m t ] function of mf. Also, J0(mf), J1(mf), J2(mf), J3(mf), …,
= A sin(w c t + mf sin w m t) (43.15) respectively, represent amplitude of carrier, first side-
band, second sideband, third sideband and so on. This
where mf is the modulation index = d/fm, and A is the is evident from the FM signal expression given above.
amplitude of the modulated signal, which in turn is equal
to the amplitude of the carrier signal.
1.0
Depth of modulation in case of an FM signal is defined
as the ratio of frequency deviation (d ) to maximum
Jn(mf) J0(mf)
0.8
allowable frequency deviation. Maximum allowable
J1(mf)
frequency deviation is different for different services and
0.6 J2(mf)
is also different for different standards even for a given J3(mf)
type of service using this form of modulation. J4(mf)
J5(mf)
0.4 J6(mf)
For instance, maximum allowable frequency deviation
for commercial FM radio broadcast is 75 kHz. It is
0.2
50 kHz for FM signal of TV sound in CCIR standards
and 25 kHz for FM signal of TV sound in FCC
standards. Therefore, 0 2 4 6 8 10
Depth of modulation for commercial FM radio broadcast
−0.2
d (in kHz)
=
75 −0.4
0 2 4 6 8 10
Depth of modulation for TV FM sound in CCIR stan-
d (in kHz)
mf
0.5
mf = 0.2
wc-wm wc wc+wm w
0.5
mf = 1
mf = 5
signal, as the modulation signal increases, the side- Deviation ratio D has the same significance for arbitrary
band power increases here too. But it does so only modulation as the modulation index mf for sinusoidal
at the cost of carrier power so that the total trans- modulation. The bandwidth in this case is given by the
mitted power remains constant. following expression:
5. In FM, the carrier component can disappear com- Bandwidth = 2(D + 1)w M (43.19)
pletely for certain specific values of mf for which
J0(mf) becomes zero. These values are 2.4, 5.5, 8.6, This expression for bandwidth is generally referred to as
11.8 and so on. Carson’s rule.
In case of D << 1, the FM signal is considered as a
43.4.2 Narrow-Band and Wide-Band FM narrow-band signal and the bandwidth is given by the
following expression:
An FM signal whether it is a narrow-band FM signal
or a wide-band FM signal is decided by its bandwidth Bandwidth = 2(D + 1)w ≈ 2w (43.20)
and in turn by its modulation index. For a modulation M M
voltage with peak amplitude Vn is present along with a ratio (SNR) of 12 dB, a unity modulation index for
carrier voltage of peak amplitude Vc, the noise voltage AM and FM systems, the noise performance of the two
amplitude modulates the carrier with a modulation index systems is more or less the same, slightly better in case
equal to Vn/Vc. It also phase modulates the carrier with of AM system.
a phase deviation equal to sin−1(Vn/Vc). This expression We shall now examine the effect of change in mod-
for phase deviation results when a single frequency noise ulating noise frequencies. We should remember that
voltage is considered vectorially and the noise voltage the noise frequency will interfere with the desired sig-
vector is superimposed on the carrier voltage vector as nals only when the noise difference frequency, that
shown in Fig. 43.18. is, the frequency produced by mixing action of car-
rier and noise frequencies lies within the pass band of
(wn − wc) the receiver. Now, a change in modulating and noise
difference frequencies does not have any effect on the
noise modulation index and signal modulation index in
case of AM. As a result, noise-to-signal ratio in case
Vn of AM remains unaltered. In case of FM, when the
noise difference frequency is lowered, there again is no
wc Vc effect of this change on the noise modulation index
as a constant noise to carrier voltage means a con-
stant phase modulation due to noise and hence a con-
stant noise modulation index. However, a reduction in
modulating frequency implies an increase in the signal
modulation index in the same proportion. This leads to
Figure 43.18 | Noise in FM signal. reduction in noise-to-signal ratio in the same propor-
tion. For instance, in the example considered earlier,
In case of voice communication, an FM receiver is if the modulating frequency were decreased from 15
not affected by the amplitude change as it can be kHz to 30 Hz, the noise-to-signal ratio in case of FM
removed in the receiver in the limiter circuit. Also, an will reduce to (0.253 × 30/15000) = 0.0005 or 0.05%
AM receiver will not be affected by the phase change. while noise-to-signal ratio in case of AM remains same
It is therefore the effect of phase change on the FM at 0.25 or 25%.
receiver and the effect of amplitude change on the AM
Figure 43.19 shows how noise at the receiver output
receiver that can be used as the yardstick for deter-
varies with noise difference frequency or noise sideband
mining the noise performance of the two modulation
frequency assuming that noise frequencies are evenly
techniques. Two very important aspects that need to
spread over the entire pass band of the receiver. Figure
be addressed in the comparison of the two communica-
43.19(a) is for the case where mf = 1 at the highest
tion techniques vis-à-vis their noise performance are
modulating frequency and Fig. 43.19(b) depicts the case
the effects of modulation index and the SNR at the
where mf = 5 at the highest modu-lating frequency.
receiver input. These are discussed as follows and illus-
A rectangular distribution in case of AM is obvious.
trated with examples.
A much better performance in case of FM for higher
modulation index has also been explained.
43.4.3.1 Effect of Modulation Index
An FM receiver uses a limiter circuit that precedes Noise has a greater effect on the higher modulating
the FM demodulator. The idea behind the use of lim- frequencies than it has on lower ones. This is because
iter circuit is the fact that any amplitude variations in of the fact that FM results in smaller values of phase
an FM signal are spurious and contain no intelligence deviation at the higher modulating frequencies whereas
information. As FM demodulator circuits to some extent the phase deviation due to white noise is constant for all
respond to amplitude variations, removing these ampli- frequencies. Because of this, SNR deteriorates at higher
tude variations result in a better noise performance in an modulating frequencies. If the higher modulating frequen-
FM receiver. This amplitude limiter acts on the stron- cies above a certain cut-off frequency were boosted at the
ger signals and tends to reject the weaker signals. Thus, transmitter prior to modulation according to a certain
when the SNR at the limiter input is very low, that is, known curve and then reduced at the receiver in the same
when the signal is weak, an FM system offers a poorer fashion after the demodulator, a definite improvement in
performance as compared to an AM system. An FM noise immunity would result. The process of boosting
system offers better performance with respect to an AM the higher modulating frequencies at the transmitter
system only when the SNR is above a certain threshold and then reducing them in the receiver are, respectively,
value, which is 8 (or 9 dB). This is clear from the curves known as pre-emphasis and de-emphasis. Figure 43.21
shown Fig. 43.20, which depict that: shows the pre-emphasis and de-emphasis curves.
1. The FM system offers full improvement over the
AM system when the SNR is about 3 dB greater + Pre-emphasis
than this threshold of 9 dB.
Response
2. The AM system has a definite advantage as (dB)
compared to the FM system for input SNR of less 0
than 9 dB.
3. Improvement of the FM system over the AM De-emphasis
system is visible for input SNR of greater than
9 dB. The quantum of improvement increases with −
increase in SNR till it reaches its maximum value f1 log f
f B C
2
Sb
= d ⋅ ⋅
f1 2fm N
9 12 (43.23)
S dB Nb
N in
where f1 is the cut-off frequency for pre-emphasis/
Figure 43.20 | SNR in FM and AM systems. de-emphasis curve.
43.4.4 Generation of FM Signals that carrier frequency tends to drift and therefore addi-
tional circuitry is required for frequency stabilization.
In case of an FM signal, the instantaneous frequency of The problem of frequency drift is overcome in crystal
the modulation signal varies directly as the instanta- controlled oscillator schemes.
neous amplitude of the modulating or baseband signal. While, as we all know, crystal control provides a very
The rate at which these frequency variations take place stable operating frequency, the exact frequency of oscil-
is of course proportional to the modulating frequency. lation in this case mainly depends upon the crystal char-
Though there are many possible schemes that can be acteristics and to a very small extent on the external
used to generate the signal characterized above, all of circuit. For example, a capacitor connected across the
them depend simply on varying the frequency of an crystal can be used to change its frequency typically
oscillator circuit in accordance with the modulating from 0.001% to 0.005%. The frequency change may be
signal input. linear only up to a change of 0.001%. Thus, a crystal
One of the possible methods is based on the use oscillator can be frequency modulated over a very small
of a varactor (a voltage variable capacitor) as a part range by a parallel varactor. The frequency deviation
of the tuned circuit of an LC oscillator. The resonant possible with such a scheme is usually too small to be
frequency of this oscillator will not vary directly with used directly. The frequency deviation in this case is
the amplitude of the modulating frequency as it is then increased by using frequency multipliers as shown
inversely proportional to the square root of the capaci- in Fig. 43.23.
tance. However, if the frequency deviation is kept
small, the resulting FM signal is quite linear. Figure +V
43.22 shows the typical arrangement when the modu-
lating signal is an audio signal. This is also known as
the direct method of generating an FM signal, as in this
case, the modulating signal directly controls the carrier Modulating RFC
frequency. signal FM
Crystal signal
C2 RFC RF Crystal Frequency
bypass oscillator multipliers
To oscillator
Varactor
tank circuit C1
Figure 43.23 | Crystal-control oscillator based scheme
Modulating for FM signal generation.
signal
Another approach that eliminates the requirement
of extensive chains of frequency multipliers in direct
crystal-controlled systems is an indirect method where
−V frequency deviation is not introduced at the source of
Figure 43.22 | LC-oscillator-based direct method of FM
RF carrier signal, that is, the oscillator. The oscillator in
this case is crystal controlled to get the desired stability
signal generation.
of the unmodulated carrier frequency and the frequency
Another direct method scheme that can be used for deviation is introduced at a later stage. The modulating
generation of an FM signal is the reactance modulator. signal phase modulates the RF carrier signal produced
In this, the reactance offered by a three-terminal active by the crystal controlled oscillator. As frequency is noth-
device such as a FET or a bipolar transistor forms a part ing but the rate of change of phase, phase modulation
of the tuned circuit of the oscillator. The reactance in of the carrier has the associated frequency modulation.
this case is made to vary in accordance with the modu- Introduction of a leading phase shift would lead to an
lating signal applied to the relevant terminal of the active increase in the RF carrier frequency and a lagging phase
device. For example, in case of FET, the drain-source shift results in a reduced RF carrier frequency. Thus, if
reactance can be shown to be proportional to the trans- the phase of the RF carrier is shifted by the modulat-
conductance of the device, which in turn can be made ing signal in a proper way, the result is a frequency-
to depend upon the bias voltage at its gate terminal. modulated signal. As phase modulation also produces
The main advantage of using a reactance modulator is little frequency deviation, a frequency multiplier chain is
that large frequency deviations are possible and thus less required in this case too. Figure 43.24 shows the typical
frequency multiplication is required. One of the major schematic arrangement of generating FM signal via the
disadvantages of both these direct method schemes is phase modulation route.
IF output
Modulating Phase Frequency FM transformer
signal modulator multiplier signal Cx Ix Cx Rx
Ix
Audio
Lx RL output
Output
Crystal from limiter
Ly
oscillator Cy Iy Cy Iy Ry
Resonance curve of
Ly tuned circuit Resonance curve of Positive output
Lx tuned circuit across RL
Voltage across
tuned circuits
Negative output
across RL
Large amplitude
input signal
that is, two signals 90° out of phase, to get the frequency Ea Ea Ea
discrimination property. One of the two signals is the
FM signal to be detected and its quadrature counterpart
is generated by using either a capacitor or an inductor Resultant Resultant Resultant
as shown in Fig. 43.27. The two signals here have been π/2 π/2 f π/2 + f
Eb Eb Eb
labelled as Ea and Eb. (a) (b) (c)
D1
Audio out
Eo
I1
E1 C1 R1
Ea L
Ep AFC
Eb
E2 C2 R2
I2
D2
(a)
E1 = Ep + Ea
E2 = Ep + Eb
E1 E1 E1
Ea Ea Ep Ea
φ
Ep Ep
φ φ
Eb Eb Eb
E2 E2 E2
D2
(a)
= Ep + Ea
E1 Systems
968 Chapter 43: Analog Communication
E2 = Ep + Eb
E1 E1 E1
Ea Ea Ep Ea
φ
Ep Ep
φ φ
Eb Eb Eb
E2 E2 E2
(b)
+
Output
Resonance Eo
voltage
Eo
Frequency
Useful (Linear)
Range
−
(c)
Figure 43.29 | Foster−Seeley frequency discriminator. (a) Circuit. (b) Phasor diagram. (c) Response curve.
Another commonly used FM detector circuit is the the frequency deviation, it also depends upon the aver-
ratio detector. This circuit has the advantage that it age carrier level.
is insensitive to short-term amplitude fluctuations in
the carrier and therefore does not require an additional E1=Ea−Eb D1
limiter circuit. The circuit configuration, as can be seen
from Fig. 43.30, is similar to the one given in case of the
Foster−Seeley frequency discriminator circuit except for Ea C1 R1
a couple of changes. These are reversal of diode connec- IF
input C3 E3
tions and addition of a large capacitor (C3). The time Eb
constant (R1 + R2)⋅C3 is much larger than the time Ec C2 R2
period of even the lowest modulating frequency of inter-
est. The detected signal in this case appears across the
E2=Ec−Eb D2
C1−C2 junction. The sum output across R1−R2 and
Audio
output
hence across C1−C2 remains constant for a given car- RL CL
rier level and also is insensitive to rapid fluctuations in
carrier level. However, if the carrier level changes very
slowly, C3 charges or discharges to the new carrier level.
The detected signal therefore is not only proportional to Figure 43.30 | Ratio detector.
Yet another form of FM detector is the one implemented 43.5 PHASE MODULATION
using a phase-locked loop (PLL). A PLL, as we know,
has a phase detector (usually a double balanced mixer), a
low-pass filter and an error amplifier in the forward path In phase modulation (PM), the instantaneous phase of
and a voltage-controlled oscillator (VCO) in the feed- the modulation signal varies directly as the instanta-
back path. The detected output appears at the output of neous amplitude of the modulating or baseband signal.
error amplifier as shown in Fig. 43.31. A PLL-based FM If the modulating signal is denoted by m(t), then the
detector functions as follows. instantaneous phase q(t) of the PM signal is mathemati-
cally expressed by the following expression:
Modulating q (t) = w c t + q 0 + kp m(t) (43.24)
FM IF signal O/P
where wc is the unmodulated carrier frequency, kp is the
input Low Error
Doubled pass amp
balanced mixer filter constant of proportionality and q0 is the initial phase of
the carrier signal.
If q0 = 0, then
q (t) = w c t + kp m(t)
f
(43.25)
Voltage
controlled The resulting PM signal is
oscilator
vPM (t) = A cos[w c t + kp m(t)] (43.26)
The instantaneous frequency wPM(t) in this case is given
by the following expression:
Figure 43.31 | PLL-based FM detector.
dq (t)
w PM (t) = = w c + kp m
(t) (43.27)
dt
The FM signal is applied to the input of the phase
detector. The VCO is tuned to a nominal frequency Therefore, in the phase modulation scheme, the instan-
equal to the unmodulated carrier frequency. The taneous angular frequency varies linearly with the deriv-
phase detector produces an error voltage depend- ative of the modulating signal.
ing upon frequency and phase difference between the
VCO output and instantaneous frequency of input FM
43.5.1 Relation between FM and PM
signal. As the input frequency deviates from the centre
frequency, the error voltage produced as a result of If a signal is an FM wave corresponding to m(t), it is also
a PM wave corresponding to ò m(a )da . Similarly a PM
frequency difference after passing through the low-pass
filter and error amplifier drives the control input of the
wave corresponding to m(t) is an FM wave correspond-
VCO to keep its output frequency always in lock with .
ing to m (t).
the instantaneous frequency of the input FM signal.
As a result, the error amplifier always represents the Figure 43.32 shows how the FM and PM are equiva-
detected output. The double-balanced mixer nature of lent and interchangeable.
phase detector suppresses any carrier-level changes and
therefore the PLL-based FM detector requires no addi-
tional limiter circuit. Modulating Phase
Integrator FM wave
wave modulator
A comparison of the three types of FM detectors would
reveal that Foster−Seeley type frequency discriminator
offers excellent linearity of response, is easy to balance
Ac cos(2pfct)
and the detected output depends only on frequency
deviation. But it needs high-gain RF and IF stages
to ensure limiting action. The ratio detector circuit, Modulating Ferquency
wave Differentiator PM wave
however, requires no additional limiter circuit; detected modulator
output depends both on frequency deviation and on
average carrier level. However, it is difficult to balance.
PLL-based FM detector offers excellent reproduction of Ac cos(2pfct)
modulating signal, is easy to balance and has low cost
and high reliability. Figure 43.32 | Interrelation between PM and FM signals.
IMPORTANT FORMULAS
6. For proper detection of the signal by envelope 11. In case of D << 1, the FM signal is considered as
detector narrow-band signal and the bandwidth is
Bandwidth = 2(D + 1)w M ≈ 2w M
1 1 1
<< RC < , or 2pB < << w c
wc 2pB RC
12. In case mf >> 1 (for sinusoidal modulation with
7. Instantaneous frequency ( f ) of an FM signal is modulating frequency wm) or D >> 1 (for arbi-
mathematically expressed as follows: trary modulation signal band limited to wM), the
f = fc (1 + KV m cos w m t) FM signal is termed as the wide-band FM and the
bandwidth in this case is given by the following
8. Frequency deviation for FM is d = KVm fc expression:
9. FM signal is expressed as follows: Bandwidth = 2mf w m or 2Dw M
v(t) = A{J (m ) sin w t 13. Instantaneous phase q(t) of the PM signal is math-
0 f c
+ J (mf )[sin(w + w )t − sin(w − w )t]
ematically expressed as follows:
1 c m c m
q(t) = wct + q0 + kpm(t)
+ J (m )[sin(w c + 2w )t − sin(w − 2w )t]
2 f m c m
+ J (m )[sin(w + 3w )t − sin(w − 3w )t] + }
14. The resulting PM signal is vPM(t) = Acos[wct +
3 f c m c m kpm(t)]
10. In case of a FM signal with an arbitrary modu- 15. The instantaneous frequency wPM(t) in this case is
lating signal m(t) band limited to (wM), deviation given by the following expression:
ratio (D) is given as:
dq (t)
w = w c + kp m(t)
.
Maximum frequency deviation (t) =
D= PM dt
Bandwidth of m(t)
SOLVED EXAMPLES
æ 0. 6 × 0. 6 ö
and PSB = 0.0625 × Pc Pt = 500 × ç 1 + ÷ = 500 × 1.18 = 590 W
è 2 ø
Therefore, saving in power Ans. (a)
P = lim ò u2 (t)dt
+ d f + 4 × 103 + 200 T ®¥
−T / 2
p
200
u2 (t) = A2 cos2 2p 4 × 103 + t
−jp / 6
d f + 4 × 10 −
250
p
3
+ 2e p
200
+ A2 cos2 2p 4 × 103 − t
+ 2e jp /6d f + 4 × 103 + 250 p
p 250 p
+ 4A2 sin2 2p 4 × 103 + t +
(d) None of the above p 3
250 p
Solution. From Quetion 8, we know that the + 4A2 sin2 2p 4 × 103 − t−
modulated signal is p 3
p
+ terms of cosine and sine fu
unctions
u(t) = A 2 cos 400t + 4 sin 500t + × (A cos 800pt )
3 in the first power
Taking Fourier transform of the above signal, we get Therefore, power content of the modulated signal is
200 200 A 2 A 2 4A 2 4A 2
d f − p + d f + p
= 5A 2
P = + + +
2 2 2 2
U (f ) = A
2 jp /3 250 2 −jp /3 250
d f − − e d
Ans. (d)
+ e f +
j p j p 11. In a superheterodyne AM receiver, the image chan-
1
2
( ) (
× d f − 4 × 103 + d f + 4 × 103
) nel selectivity is determined by
(a) the pre-selector and RF stages
(b) the pre-selector, RF and IF stages
Therefore, (c) the IF stages
200
d f − 4 × 10 − p
3 (d) All the stages
Solution. The image rejection should be achieved
200
+ d f − 4 × 10 +
3 before IF stage because once it enters into IF
p amplifier it becomes impossible to remove it from
−jp / 6 250 wanted signal.
+ 2e d f − 4 × 10 −
3
p So, image channel selectivity depends upon pre-
250
selector and RF amplifiers only.
jp / 6
+ 2e d f − 4 × 103 + The IF amplifiers help in rejection of adjacent
p
U (f ) =
A channel frequency and not image frequency.
2 200 Ans. (a)
+ d f + 4 × 103 −
p 12. In commercial TV transmission in India, picture
+ d f + 4 × 103 +
200 and speech signals are modulated, respectively, as
p (a) VSB and VSB (b) VSB and SSB
+ 2e−jp /6d f + 4 × 103 − 250
(c) VSB and FM (d) FM and VSB
p
Solution. In commercial TV transmission in India,
+ 2e jp /6d f + 4 × 103 + 250 picture signal is modulated using VSB modulation
p and speech or audio signal is modulated using FM
modulation.
Ans. (b) Ans. (c)
10. For the data given in Question 8, the power con- 13. In a DSB AM system, the carrier is c(t) = A cos(2pfc t)
tent of the modulated signal is c(t) = A cos(2pfc t) and the message signal is given by
1. The standard AM signal (A3E form of AM) broad- 2. A 4-GHz carrier is DSBSC modulated by a low-pass
cast from a station has an average percent of modu- message signal with maximum frequency of 2 MHz.
lation of 60%. If it is decided to shift to J3E form of The resultant signal is to be ideally sampled. Find
transmission, what would be average power saving the minimum frequency of the sampling impulse
in percentage if the signal strength in the reception train in Megahertz.
area is to remain unaltered?
Solution. Carrier frequency fc = 4 GHz = 4000 MHz
Solution. A3E is the double sideband AM with Maximum signal frequency fm = 2 MHz
full carrier. Total power in the modulated signal The maximum frequency of the upper sideband
would then be fH = fc + fm = 4000 ×106 + 2 ×106 = 4002 MHz
(0.6)2 The minimum frequency of the lower sideband
Pt = Pc 1 +
2 fL = fc − fm = 4000 ×106 − 2 ×106 = 3998 MHz
Substituting the value of m, we get Bandwidth of the modulated signal (BW) = fH − fL
= 4002 ×106 − 3998 ×106 = 4 MHz
(0.6)2 Therefore,
Pt = Pc 1 +
2
= 1.18Pc
4002 × 106
f
H
=
J3E is SSBSC AM.
BW 4 × 106
In case of J3E transmission, There, fH >> BW
In this case the minimum frequency of the sam-
m2
Pt = Pc ×
pling pulse train
4
fs(min) ≅ 2 BW = 2 × 4×106 = 8 MHz
Substituting the value of m, we get
Ans. (8)
(0.6)2
Pt = Pc × 3. The output signal from an AM modulator is
= 0.09Pc
4 u(t) = 5 cos(1800pt) + 20 cos(2000pt) + 5 cos(2200pt)
æ 1.18Pc − 0.09Pc ö
÷ × 100%
Find the modulation index.
Power saving = ç
è 1.18Pc ø Solution. Given that the AM modulated signal is
1.09
= × 100% = 92.3% u(t) = 5 cos(1800pt) + 20 cos(2000pt) + 5 cos(2200pt)
1.18
= 10 cos(2000pt) cos(200pt) + 20 cos(2000pt)
Ans. (92.3)
PRACTICE EXERCISE
1. A sinusoidal wave of amplitude 10 V and frequency (a) 3.41 V (b) 5.23 V (c) 4.53 V (d) 3.67 V
1 kHz is applied to an FM generator having a (1 Mark)
frequency sensitivity constant of 40 Hz/V. The 5. The maximum phase deviation for an angle-
frequency deviation is modulated signal given by
s(t) = cos[2p (2 × 106 t + 30 sin 150t + 40 cos 150t)] is
(a) 100 Hz (b) 200 Hz (c) 400 Hz (d) 500 Hz
(2 Marks)
(a) 100 (b) 50 (c) 50p (d) 100p
2. For the data given in Question 1, the FM modula- (2 Marks)
tion type is
6. Maximum frequency deviation for the angle-
(a) narrow-band FM modulated signal in Question 5 is
(b) wide-band FM
(c) dependent on the frequency of the carrier (a) 1000 Hz (b) 2500 Hz (c) 7500 Hz (d) 5000 Hz
waveform (1 Mark)
(d) None of the above 7. A 95 MHz carrier is frequency modulated by a
(1 Mark) sinusoidal signal and the modulating signal is
3. A carrier wave is frequency modulated using a sine such that maximum frequency deviation achieved
signal v(t) = Amsin(2pfm)t. In a certain experi- is 50 kHz. The modulation index and bandwidth
ment conducted with fm = 1 kHz and increasing of the modulated signal if the modulating signal
Am from zero, it is found that the carrier compo- frequency is 1 kHz are, respectively,
nent of FM wave is reduced to zero for the first (a) 50, 102 kHz (b) 51, 100 kHz
time when Am = 2 V. The frequency sensitivity of (c) 51, 102 kHz (d) 50, 100 kHz
the modulator is (2 Marks)
(a) 1000 Hz/V (b) 1220 Hz/V 8. For the data in Question 7, the modulation index
(c) 1530 Hz/V (d) 2450 Hz/V and bandwidth of the modulated signal if the modu-
(2 Marks) lating signal frequency is 100 kHz are, respectively,
4. For the carrier wave in Question 3, what is the (a) 0.5, 100 kHz (b) 1, 100 kHz
value of Am for which the carrier component is (c) 0.5, 300 kHz (d) 1, 300 kHz
reduced to zero for the second time? (1 Mark)
9. A carrier when frequency modulated by a certain (a) sinw 1t − cosw 2t (b) sinw 1t + cosw 2t
sinusoidal signal of 1 kHz produces a modulated (c) cosw 1t − sinw 2t (d) sinw 1t + cosw 2t
signal with a bandwidth of 20 kHz. If the same (1 Mark)
carrier signal is frequency modulated by another 16. An FM signal is represented by v(t) = 15cos[(108)pt
modulating signal whose peak amplitude is three + 6sin2p(103)t]. The maximum phase deviation in
times that of previous signal and frequency is one- radians is
half of the previous signal, the bandwidth of the
new modulated signal is (a) 5 (b) 6 (c) 8 (d) 9
(1 Mark)
(a) 40 kHz (b) 45 kHz (c) 50 kHz (d) 55 kHz
17. For the FM signal given in Question 16, the value
(2 Marks)
of maximum frequency deviation in Hertz is
10. In a 100% amplitude modulated signal, power in (a) 5000 (b) 1000 (c) 7500 (d) 6000
the upper sideband when the carrier power is 100 W (2 Marks)
and the modulation system is SSBSC is
18. An FM generator is fed with a tone modulating
(a) 25 W (b) 100 W (c) 50 W (d) 15 W signal Am cos(4p × 103 )t . Starting from 0.01 V, Am
(1 Mark) is gradually increased and when Am = 2 V, it has
11. Percentage saving in power of 100% modulated been found that the carrier component goes to zero
SSBSC AM signal as compared to DSB signal is for the first time. The frequency sensitivity of the
source is
(a) 100 (b) 66.67 (c) 150 (d) 83.33
(1 Mark) (a) 1.2 kHz/V (b) 1.8 kHz/V
(c) 2.4 kHz/V (d) 4.2 kHz/V
12. In a VSB system, modulating frequency of 3 MHz (2 Marks)
results in a sideband power of 25 W. If the carrier
19. In the Question 18, keeping Am constant at 2V,
power is 100 W, depth of modulation is
frequency fm is decreased till the carrier component
(a) 25% (b) 50% (c) 75% (d) 100% goes zero for the second time. The value of fm for
(2 Marks) this to happen is
13. The bandwidth of an FM signal in kilohertz, (a) 872 Hz (b) 231 Hz (c) 912 Hz (d) 2312 Hz
produced in a commercial FM broadcast with (1 Mark)
modulating signal frequency being in the range 20. Signal v(t) = 5[cos(106pt) − sin(103pt) × sin(106pt)]
of 50 Hz to 15 kHz and maximum allowable represents
frequency deviation being 75 kHz, is
(a) DSB suppressed carrier signal
(a) 200 (b) 210 (c) 450 (d) 460 (b) AM signal
(2 Marks) (c) SSB upper sideband signal
14. A message m(t) band limited to the frequency fm (d) Narrow-band FM signal
has a power of Pm. The power of the output signal (1 Mark)
in the following figure is 21. A 1 kHz tone is used to generate both an AM and
an FM signal. Unmodulated carrier amplitude is
the same for both the AM and FM. The modula-
Ideal low tion index mfFM of FM is 8. If the frequency compo-
pass filter nents at (fc ± 1000) Hz have the same magnitude
m(t)cos w 0t cut off f=fm Output in AM and FM, then the modulation index (mfAM)
pass band singal
of AM is (given that, J1(8) = 0.235)
gain = 1 (ω 0 >2pfm)
cos(w 0t + q) (a) 0.235 (b) 0.47 (c) 0.1175 (d) 0.74
(2 Marks)
value of wc so that the bandwidth of the transmit- 32. The image channel rejection in a superheterodyne
ted signal is 1% of the carrier frequency wc is receiver comes from
(a) 1 MHz (b) 500 kHz (c) 2 MHz (d) 4 MHz (a) IF stages only
(2 Marks) (b) RF stages only
24. The maximum power efficiency of an AM modula- (c) Detector and RF stages only
tor is (d) Detector, RF and IF stages
(1 Mark)
(a) 25% (b) 50% (c) 33% (d) 100%
(1 Mark) 33. A DSBSC signal is generated using the carrier
cos(wct + q) and modulating signal x(t). The enve-
25. Which of the following demodulator(s) can be used lope of the DSBSC signal is
for demodulating the signal x(t) = 5(1 + 2cos200pt)
(cos20000pt) (a) x(t)
(b) x(t)
(a) Envelope demodulator
(c) Only positive portion of x(t)
(b) Square-law demodulator
(d) x(t)cosq
(c) Synchronous demodulator
(2 Marks)
(d) None of the above
(1 Mark) 34. A modulated signal is given by g(t) = e−at cos[(w c + D
26. The image (second) channel selectivity of a g(t) = e−at cos[(w c + Dw )t] u(t), where a, wc and Dw are
superheterodyne communication receiver is deter- positive constants, and w c >> Dw. The complex
mined by envelope of s(t) is given by
−at [ j(wvc +Dw )t ]
(a) Antenna and pre-selector (a) e e u(t) (b) e−at e( jDwt)u(t)
(b) The pre-selector and RF amplifier (c) e( jDwt)u(t) (d) e( jw c +Dw )t
(c) The pre-selector and IF amplifier (2 Marks)
(d) The RF and IF amplifier
(1 Mark) 35. A band limited signal is sampled at the Nyquist
rate. The signal can be recovered by passing the
27. An FM signal having a modulation index of mf is samples through a (an)
passed through a frequency tripler. The FM signal
at the output of frequency tripler will have a mod- (a) RC filter
ulation index of (b) envelope detector
(c) PLL
(a) 3mf (b) mf (c) 2mf (d) mf/3 (d) ideal low-pass filter with the appropriate
(1 Mark) bandwidth
28. An FM signal is represented by x(t) = 15cos[2p(108)t (1 Mark)
+ 150cos2p(103)t]. The bandwidth of the signal is 36. In the following figure
(a) 298 kHz (b) 302 kHz 2 sin 2pt sin 199pt
(c) 405 kHz (d) 365 kHz m(t) = , s(t) = cos 200pt and n(t) =
t t
(2 Marks)
29. For the FM signal given in Question 28, the expres- m(t) + LPF y(t)
sion for instantaneous frequency of the signal is
(a) 2p(106) − 300p(103)sin2p(103)t Cut-off frequency = 1Hz
(b) 2p(108) − 500p(103)sin2p(103)t Passband gain = 1
(c) 2p(108) − 300p(103)sin2p(104)t s(t) n(t) s(t)
(d) 2p(108) − 300p(103)sin2p(103)t The output y(t) will be
(2 Marks)
sin 2pt
30. A PLL can be used to demodulate (a)
t
(a) PAM signals (b) PCM signals sin 2pt sin pt
(c) FM signals (d) DSBSC signals (b) + cos 3pt
(1 Mark) t t
sin 2pt sin 0.5pt
31. A PAM signal can be detected by using a (an) (c) + cos 1.5πt
t t
sin 2t sin 0.5pt cos 1.5pt
(a) ADC (b) integrator
(c) band-pass filter (d) high-pass filter (d) +
(1 Mark) 2t t (2 Marks)
2. (a) Modulation index, mf = Δf/fm = 400/1000 = 0.4 Therefore, frequency sensitivity kf = 1220 Hz/V
As mf < 1, therefore the FM type is narrow band. 4. (c) The carrier component will become zero for the
3. (b) The carrier component is zero, when its coef- second time when mf = 5.53
ficient J0(mf) = 0 Therefore, the value of Am for which the carrier
The Bessel function J0(x) = 0 for x = 2.44, 5.53, component is reduced to zero for the second time
8.65 and so on. mf 5.53 × 1000
Am = f m = = 4.53 V
Therefore, when the carrier component is reduced kf 1220
to zero for the first time, the value of mf = 2.44
5. (d) The term (30 sin 150t + 40 cos 150t) can be
We know that mf = Δf/fm = kfAm/fm.
rewritten as
2 sin 2pt
After passing from LPF with fc = 1 Hz,
36. (d) m(t) × s(t) = × cos 200p t
1
[sin 2pt + sin 2pt − sin pt ]
t
y(t) =
1
= (sin 202pt − sin 198pt)
2t
t Therefore,
sin 2t sin 0.5pt cos 1.5pt
y(t) = +
2t t
1. Comparing the given equation with the standard Converting the given equation in standard form,
form of equation for FM signal given by we get
v(t) = Vm sin(w c t + m sin w m t) 1
y(t) = cos[100(t − 10−8 )] cos[106 (t − 1.56 × 10−6 )]
100
Unmodulated carrier frequency, wc = 7.8 × 108
rad/s, or fc = (7.8 × 108)/2p = 124.14 MHz.
Therefore, phase delay tp = 1.56 × 10−6 s = 1560 ns
Therefore, group delay tg = 10−8 s = 10 ns
Ans. (124.14)
1
f = Hz = 10 kHz Df = 2p 302 + 402 mrad = 100p mrad = 314 mrad
m 100 × 10−6
The odd harmonic component is 3f = 30 kHz . Ans. (7.5, 314)
m
Note that the square wave has infinite odd har-
11. Given that Te = 21 K, Ta = 300 K, gain (G)dB = 13,
monic components of the fundamental harmonic
cable loss (L) = (3)dB
frequency. In the question, only the fundamen-
tal and second harmonic components need to be Therefore, (13)dB = 10 log10 (G) or G = 19.95
considered. Therefore, cable loss = (3)dB = 10 log10 (L) , or
Therefore, frequencies present at the output are L = 1.995
fc ± fm, fc ± 3fm = 990 kHz, 1010 kHz, 970 kHz For a cable, noise figure = F2 = cable loss = L
and 1030 kHz. Therefore, F2 = 1.995
Ans. (970, 990, 1010, 1030)
T 21
Noise figure of amplifier = F1 = 1 + e = 1 + = 1.07
10. Given that the angle-modulated signal T 300
T 21 a
F1 = 1 + e = 1 +
s(t) = cos(2p × 2 × 106 t + 2p × 30 sin 150
= 1.07
t Ta 300
+ 2p × 40 cos 150t) Noise figure of the cascaded amplifier
F −1 1.995 − 1
Therefore, F = F1 + 2 = 1.07 + = 1.12
G 19.95
f i = 2p × 2 × 10 t + 2p × 30 sin 150t + 2p × 40 cos 150t
6
1. The input to a coherent detector is DSBSC signal where a0 and a1 are constants. The output of the
plus noise. The noise at the detector output is non-linear device can be filtered by an appropriate
band-pass filter. Let vi = Ac¢ cos(2pfc¢t) + m(t) be
(a) in-phase component
(b) quadrature component the message signal. Then the value of fc¢ (in MHz) is
(c) zero (a) 1.0 (b) 0.333 (c) 0.5 (d) 3.0
(d) envelope (GATE 2003: 2 Marks)
(GATE 2003: 1 Mark)
Solution.
Solution. The coherent detector rejects the quadra-
ture component of noise and therefore noise at the vo = a Ac¢ cos(2pfc¢t) + a m(t)
0 0
output has in-phase component only. + a [Ac¢ cos(2pfc¢t) + m(t)]3
The in-phase component of noise and output are 1
additive at the output of the detector. After neglecting the carrier frequency term and
Ans. (a) m(t), we get
éAc¢3 cos3 (2pfc¢t) + 3Ac¢ cos2 (2pfc¢t)m(t)ù
2. A DSBSC signal is to be generated with a carrier
frequency fc = 1 MHz using a non-linear device with vo = a1 ê ú
the input−output characteristic vo = a0vi + a1vi3 êë+ 3Ac¢ cos(2pfc¢t)m2 (t) + m3 (t) úû
After neglecting the terms which will not come in Solution. FM signal is given by
the DSBSC signal, we get
∞
vo = a1[3Ac¢2 cos2 (2pfc¢t)m(t)] v
FM
(t) = Ac ∑ Jn (b ) cos(w c + nw m )t
n = −∞
3a
vo = 1 Ac¢2 [1 + cos(2p 2fc¢t)]m(t)
2 The bandwidth of the AM signal BW = 2wm
Given that the peak frequency deviation ∆w = 3 × BW.
After neglecting the modulating signal term, we get Therefore,
3a1 2 Dw = 6w m
vo = A¢ cos(2p 2fc¢t)m(t)
2 c
Dw
So the carrier frequency b= =6
wm
fc = 2fc¢ = 1 MHz
Given that (wc + nwm) = 2p × 1008 × 103.
Therefore fc¢ = 0.5 MHz Therefore,
Ans. (c)
2p × 106 + n × 4p × 103 = 2p × 1008 × 103
Common Data for Questions 3 and 4: Let
m(t) = cos[(4p × 103 )t] be the message signal and Solving the above equation, we get n = 4.
c (t ) = 5 cos[(2p × 106 )t] be the carrier.
Also, given that Ac = 5. Therefore, the coefficient
of the term cos[2p (1008 × 103 )t] is 5J4(6).
3. c(t) and m(t) are used to generate an AM signal. Ans. (d)
The modulation index of the generated AM signal 5. Choose the correct one from among the alterna-
is 0.5. Then the quantity (total sideband power)/ tives a, b, c, d after matching an item in Group 1
(carrier power) is with the most appropriate item in Group 2.
1
(a) 1 (b) 1 (c) (d) 1 Group 1
2 4 3 8
P. Ring modulator
(GATE 2003: 2 Marks)
Q. VCO
Solution. For an AM signal, R. Foster−Seeley discriminator
S. Mixer
Ac2
Carrier power is Pc =
2 Group 2
Total sideband power is 1. Clock recovery
2. Demodulation of FM
Ac2 ma2
×
Ps = 3. Frequency conversion
2 2 4. Summing the two inputs
Therefore, the ratio of total sideband power to car- 5. Generation of FM
rier power is 6. Generation of DSBSC
Ps A 2 m2 4 m2 − 1; Q − 3; R − 2; S − 4
= c 2a = a (a) P
Pc Ac 2 2 (b) P − 6; Q − 5; R − 2; S − 3
(c) P − 6; Q − 1; R − 3; S − 2
Given that ma = 0.5 Therefore,
(d) P − 5; Q − 6; R − 1; S − 3
Ps 1
= (GATE 2003: 2 Marks)
Pc 8
Ans. (b)
Ans. (d)
6. A superheterodyne receiver is to operate in the
4. c(t) and m(t) are used to generate an FM signal. If frequency range 550 kHz to 1650 kHz, with the
the peak frequency deviation of the generated FM intermediate frequency of 450 kHz. Let R = Cmax/
signal is three times the transmission bandwidth Cmin denote the required capacitance ratio of the
of the AM signal, then the coefficient of the term local oscillator and I denote the image frequency
cos[2p (1008 × 103 t)] in the FM signal (in terms of (in kHz) of the incoming signal. If the receiver is
the Bessel coefficients) is tuned to 700 kHz, then
5 5
(a) 5J 4 (3) (b) J (3) (c) J 8 (4) (d) 5J4(6) (a) R = 4.41, I = 1600 (b) R = 2.10, I = 1150
2 8 2 (c) R = 3.0, I = 1600 (d) R = 9.0, I = 1150
(GATE 2003: 2 Marks) (GATE 2003: 2 Marks)
1 1
<< RC <
fc fm (a) Constant (b) 1 + sin(2p × 106 t)
(Please note that condition 1/fc < RC is to avoid
5 5
fluctuations at recovered output and condition RC (c) − sin(2p × 106 t) (d) + cos(2p × 106 t)
≤ 1/fm is to avoid diagonal clipping.) 4 4
Therefore, 1/106 s << RC < 1/2 × 103 s or 1 µs << RC < 0.5 ms (GATE 2004: 2 Marks)
1 µs << RC < 0.5 ms. Hence, the value given in option (b) is the
correct value of RC. Solution. Let the carrier signal be denoted as
Ans. (b) coswct and the modulating signal as coswmt. The
output of balanced modulator is
8. An AM signal and a narrow-band FM signal with
identical carriers, modulating signals and modula- vBM (t) = [cos w c t][cos w m t]
tion indices of 0.1 are added together. The resul-
= [cos(w c + w m )t + cos(w c − w m )t]
tant signal can be closely approximated by 1
2
(a) broadband FM (b) SSB with carrier
(c) DSBSC (d) SSB without carrier 1
(GATE 2004: 1 Mark) = [cos 2p (101 × 106 )t + cos 2p (99 × 106 )t]
2
Solution. The narrow-band FM signal is If vBM(t) is passed through HPF with cut-off fre-
quency fH = 100 × 106Hz, then only (wc + wm)
Ac b
s(t)FM = Ac cos w c t + cos(w c + w m )t passes and output of HPF is
2
Ac b
− cos(w c − w m ) t vHP (t) =
1
cos(w c + w m )t
2 2
The AM signal is
cos[2p (101 × 106 )t]
1
=
Am
= Ac cos w c t + c a cos(w c + w m )t
2
s(t)AM
2
cos(2p × 100 × 106 t + 2p × 1 × 106 t)
1
Ac ma =
+ cos(w c − w m ) t 2
2
æ1 ö æ ö
2 2
1
= ç cos(2p 106 t)÷ + ç 1 − sin(2p 106 t)÷
è2 ø è 2 ø 10 kHz 13 kHz
1 1
= cos2 (2p 106 t) + 1 + sin2 (2p 106 t) − sin(2p 106 t) X(f)
4 4
1
= + 1 − sin(2p 106 t)
4
5 f (kHz)
= − sin(2p 106 t) -3 -1 1 3
4
Ans. (c) The positive frequencies where Y(f) has spectral
peaks are
10. Two sinusoidal signals of same amplitude and fre-
quencies 10 kHz and 10.1 kHz are added together. (a) 1 kHz and 24 kHz. (b) 2 kHz and 24 kHz.
The combined signal is given to an ideal frequency (c) 1 kHz and 14 kHz. (d) 2 kHz and 14 kHz.
detector. The output of the detector is (GATE 2004: 2 Marks)
(a) 0.1 kHz sinusoid. Solution. Peaks of X(f) are at 1 kHz and −1 kHz.
(b) 20.1 kHz sinusoid. Therefore, peaks of output of balanced modulator =
(fc ± 1) kHz and (fc ± −1) kHz.
(c) A linear function of time.
(d) A constant.
(GATE 2004: 2 Marks) Given that the carrier frequency fc = 10 kHz.
Therefore, peaks of output of the balanced modula-
Solution. The output of the adder is tor are (10 ± 1) kHz = 11 kHz, 9 kHz and (10 ± −1)
s(t) = A cos[2p × 10 × 103 t] + A cos[2p × 10.1 × 103 t]
kHz = 9 kHz, 11 kHz.
Output of HPF with fc = 10 kHz will be 11 kHz
where A is the amplitude of the two sinusoidal sig- frequency component.
nals being added. Therefore, the output of the second balanced modu-
Time period of the first signal lator Y(f) has peaks at (13 kHz ± 11 kHz) = 24 kHz
and 2 kHz.
1
T = = 100 ms Ans. (b)
1 10 × 103
12. Find the correct match between group 1 and group 2.
Time period of the second signal
Group 1
1
T = = 99 ms P. [1 + km(t)]Asin(wct)
2 10.1 × 103 Q. km(t)Asin(wct)
As the ratio R. A sin[w c t + km(t)]
T1 t
S. A sin w c t + k ∫ m(t)dt
= rational
T2
−∞
t
0 2 3 4
1
Therefore, h(t) = −s(t)
0 1 t
Ans. (c)
As, g(t) = p(t)*p(t) 19. An amplitude modulated signal is given as xAM(t)
Hence, g(t) is represented in the figure given below. = 100[p(t) + 0.5g(t)]coswct in the interval 0 ≤ t
≤ 1. One set of possible values of the modulating
g(t) signal and modulation index would be
(a) t, 0.5 (b) t, 1.0 (c) t, 2.0 (d) t2, 0.5
(GATE 2006: 2 Marks)
Solution. Given that
xAM(t) = 10[p(t) + 0.5g(t)]coswct.
where p(t)=u(t) − u(t−1) and g(t) = p(t)*p(t) =
0 1 2 t
r(t) −2r(t −1) + r(t −2).
For the desired interval, 0 ≤ t ≤ 1, p(t) = 1 and
Given that,
g(t) = t.
s(t) = g(t) − d (t − 2) * g(t)
Therefore, xAM(t) = 100(1 − 0.5t)coswct. Hence
Hence, s(t) = g(t) − g(t − 2) and g(t − 2) is as modulating signal = t and modulation index = 0.5
drawn below. Ans. (a)
g(t−2) Common Data for Questions 20 and 21:
Consider the following amplitude modulated (AM)
signal, where fm < B
1
xAM (t) = 10(1 + 0.5 sin 2pfm t) cos 2pfc t
N/2 f
−B 0 +B
f −2B −B 0 +B +2B
25 25 f f
−B −2B −B
(a) (b) 0 +B
8N o B 4N o B 0 +B +2B
25 25 (GATE 2007: 2 Mark)
(c) (d)
2N o B NoB Solution. The block diagram is redrawn as shown
(GATE 2006: 2 Marks) below.
cos(2πBt)
Solution. Form the solution of Question 20, aver-
age side band power, y1(t)
25 m(t)
Ps = 6.25W = W
4 +
Mean noise power is given by ∑ y(t)
+
PN = N o B Hilbert y2(t)
P 25 transform
SNR = s = m1(t)
PN 4N o B
sin(2πBt)
Ans. (b)
Here,
22. A Hilbert transformer is a ˆ (f )
M (f ) = M
1
(a) non-linear system (b) non-causal system æ e j2pB + e−j2pB ö
Y (f ) = M (f ) ç ÷÷
(c) time-varying system (d) low-pass system ç
è ø
1 2
(GATE 2007: 2 Marks)
Ans. (a) æ e j2pB − e−j2pB ö
Y (f ) = M (f ) ç ÷÷
ç
è ø
23. In the following scheme, if the spectrum M(f) of 2 1 2
m(t) is as shown in the following figure, then the Y (f ) = Y (f ) + Y (f )
spectrum Y(f) of y(t) will be 1 2
The waveforms are as shown below:
M(f )
M(f ) M(f )
B
−B
f f
B −B
f
−B 0 +B
Y1(f) Y2(f)
cos(2πBt)
m(t)
f f
−2B B B 2B −B B 2B
+
∑ y(t) −2B
Y(f)
+
Hilbert
transform f
−B B
sin(2πBt) Ans. (a)
24. Consider the amplitude modulated (AM) signal (a) FM only (b) AM only
Accoswct + 2coswmtcoswct. For demodulating the (c) Both AM and FM (d) Neither AM nor FM
signal using envelope detector, the minimum value (GATE 2008: 2 Marks)
of Ac should be
Solution. The signal [cos(wct) + 0.5cos(wmt) sin(wct)]
(a) 2 (b) 1
is either an AM or a narrow-band FM signal.
(c) 0.5 (d) 0
Ans. (c)
(GATE 2008: 1 Mark)
27. For a message signal m(t) = cos(2pfmt) and carrier
Solution. Modulated signal of frequency fc, which of the following represents a
(t) = Ac cos w c t + 2 cos w m t cos w c t
single side-band (SSB) signal?
v
AM
= (Ac + 2 cos w m t) cos w c t (a) cos(2pfmt)cos(2pfct)
(b) cos(2pfct)
Condition to be satisfied for envelope detection of (c) cos[2p(fc + fm)t]
an AM signal is (d) [1 + cos(2pfmt)]cos(2pfct)
(A + 2 cos w m t) ≥ 0 (GATE 2009: 1 Mark)
Minimum value of coswmt = −1. Therefore, the Solution. cos[2p(fc + fm)t] represents upper side
minimum value of (A + 2 cos w m t) is (Ac − 2). band single side band suppressed carrier AM signal.
Therefore, (Ac − 2) ≥ 0 or Ac ≥ 2 Ans. (c)
Therefore, minimum value of Ac should be 2.
Ans. (a) æ 1ö æ 1ö
28. A message signal given by m(t) = ç ÷ cos w 1t − ç ÷ sin w 2 t
è 2ø è 2ø
25. Consider the frequency modulated signal 10 cos[2p ×æ 1 ö æ 1ö
105t + 5 sin(2p × 1500t)+ 7.5 sin(2p × 1000t)] ) = ç ÷ cos w 1t − ç ÷ sin w 2 t is amplitude modulated with a carrier
m(twith
è 2ø è 2ø
5
carrier frequency of 10 Hz. The modulation index is of frequency wc to generate s(t) = [1 + m(t)] cos w c t
(a) 12.5 (b) 10 What is the power efficiency achieved by this mod-
(c) 7.5 (d) 5 ulation scheme?
(GATE 2008: 2 Marks)
(a) 8.33% (b) 11.11%
Solution. Modulation index, (mf) is given by (c) 20% (d) 25%
(GATE 2009: 2 Marks)
d
m =
f fm Solution. The message signal can be considered as
where, d is the maximum frequency deviation and
combination of two signal as follows:
æ1ö
m(t) = m1(t) − m2 (t) where m1(t) = ç ÷ cos w1(t) and m2 (t)= si
1
fm is the maximum frequency component.
è2ø 2
Given that fm = 1500 Hz æ1ö
w w
1
Phase, q (t) = 5 sin(2p × 1500t) + 7.5 sin(2p 1× 1000çèt2) ÷ø
m (t) = cos 1 (t) and m 2 (t)=
2
sin 2 t . Therefore, power efficiency
Therefore, (η) of the combined signal is
mf2
dq (t) h = × 100%
Dw = = 5 × 2p × 1500 cos(2p × 1500t) 2 + mf2
dt
+ 7.5 × 2p × 1000 cos(2p × 1000t) where m is the modulation index of combined signal
f
Maximum value of ∆w is given by m(t).
Dw max = 2p (7500 + 7500) = 2p (15000) rad/s 2
mf is given by mf1 2
+ mf2 , where
The ratio kp/kf (in radian/Hz) for the same maxi- Given, (f DPM )max = (f DFM)max
mum phase deviation is
Therefore,
8pkf = 2kp
(a) 8p (b) 4p
(c) 2p (d) p
(GATE 2012: 2 Marks) kp
or, = 4p
kf
Solution. For phase modulator, Ans. (b)
In this chapter, we discuss the fundamentals of information theory. The topics covered include measure of information,
source encoding, error-free communication over a noisy channel, channel capacity of discrete and continuous memory-
less channels and Shannon—Hartley theorem.
The value of entropy H(m) is maximum when P1 = P2 = descending probability. We repeat the procedure till the
= Pn = 1/n and is given by the following expression: final set reduces to r messages. Each of these messages
n are assigned one of the r numbers 0, 1, 2, …, (r − 1). We
H(m)max = − å
1 1
log2 = log2 n (44.4) regress in the same way as in the case of binary messages
i =1 n n till each of the original messages have been assigned a
code. For an r-ary code, we will have exactly r messages
44.2 SOURCE ENCODING left in the reduced set if and only if the total number of
original messages is r + k(r − 1), where k is an integer.
In case the original message does not satisfy this condi-
If all the messages of the source are equiprobable, then tion, then add dummy messages with zero probability of
the minimum number of bits required to encode a mes- occurrence, till this condition is fulfilled.
sage is equal to the source entropy log2(1/P), where P is
the message probability. For an arbitrary distribution of
non-equiprobable messages, the average number of binary 44.3 ERROR-FREE COMMUNICATION
digits per message required for encoding is H(m) (in bits). OVER A NOISY CHANNEL
The source encoding theorem says that to encode a
source with entropy H(m), a minimum of H(m) binary
Till now, the encoding we have discussed has zero
digits per message need to be transmitted. Therefore,
redundancy. However, there is absolutely no possibility
the average word length of an optimum code is H(m).
of error-free communication over a noisy channel when
To attain this length, we have to encode a sequence of
N messages (N → ∞) at a time. If we wish to encode
messages are encoded with zero redundancy. The use of
redundancy helps combat noise.
each message directly without using longer sequences,
then, in general, the average length of the code word per Let us consider a binary symmetric channel (BSC)
message will be greater than H(m). It is preferable to with an error probability Pe. Then for error-free commu-
encode messages directly, even if the price has to be paid nication over this channel, messages from a source with
in terms of increased word length. The optimum source entropy H(m) must be encoded by binary codes with a
code, referred to as the Huffman code, is arrived at by word length of at least H(m)/Cs, where Cs is the channel
the following sequence of steps. capacity and is given by the following expression:
The messages are arranged in order of descending é 1 1 ù
Cs = 1 − êPe log + (1 − Pe ) log
probability. The last two messages are added into one
ë Pe 1 − Pe úû
message with their probabilities added together. These
messages are then arranged in order of descending prob- Let us suppose we need to transmit a binary informa-
ability. The last two messages are again added and then tion bits per second. Then over a period of T seconds,
rearranged in order of descending probability. This is we have a block of aT binary information bits. If to this
done till the number of messages is reduced to two. block, we add (b − a)T check digits (i.e., b − a check
These two messages are assigned 0 and 1 as their first digits per second), then we need to transmit bT digits for
digits in the code sequence. We now go back and assign every aT information digits. Suppose, instead of trans-
numbers 0 and 1 to the second digit for the two mes- mitting one binary digit every 1/a second, we accumu-
sages that were added in the previous step. We keep late aT digits over T seconds. Let us consider this as a
proceeding like this till the first column is reached. The message to be transmitted. There are a total of 2aT such
code obtained is referred to as the optimum code or the super-messages, and every T seconds, we need to trans-
Huffman code or the compact code. The average length mit one of the 2aT possible super-messages. These super-
of the code is given by the following expression: messages are transmitted using a sequence of bT binary
n digits. Therefore, we have reduced the transmission rate
L= å PiLi (44.5) by a factor of a/b.
i =1
According to Shannon’s theorem, a/b must be less
The code efficiency h is defined as the ratio of H(m) (the than the channel capacity Cs.
average minimum length) to the average length L:
H(m)
h= (44.6) 44.4 CHANNEL CAPACITY OF A
L DISCRETE MEMORYLESS CHANNEL
Similar procedure can be used to find a compact r-ary
code. In this case, we arrange the messages in descending
order of probability and combine last r messages into Consider a source that generates a message that contains
one message and rearrange the new set in order of r symbols x1, x2, …, xr. The receiver receives symbols
y1, y2, …, ys. If the channel is noiseless, then the reception Channel capacity in bits per second is given by the
of some symbol yj uniquely determines the message following expression:
transmitted. Because of noise, however, there is a certain
amount of uncertainty regarding the transmitted symbol C = kCs (44.12)
when yj is received. If P (xi y j ) represents the condi- where k is the number of symbols transmitted per second.
tional probability that when xi is transmitted then yj is
received, then there is an uncertainty of log[1/P (xi y j )]
about xi when yj is received. 44.5 CHANNEL CAPACITY OF
When this uncertainty is averaged over all xi and yj,
A CONTINUOUS MEMORYLESS
we obtain H(x y) , which is the average uncertainty CHANNEL
about the transmitted symbol x when y is received and
is referred to as the conditional entropy of x given y.
H(x y) in bit/symbol is given by The entropy H(x) for a continuous random variable x is
given by the following expression:
+¥
H(x y) = å å P (xi , y j ) log
1
ò
(44.7) 1
P (xi y j ) H(x) = p(x) log dx (44.13)
i j p(x)
−¥
P (xi y j ) can be calculated using the following formulae: 2
For a given mean square value x , the entropy is maxi-
P (y j xi )P (xi ) mum for a Gaussian distribution of x given by
P (xi y j ) =
/ 2s 2
e−x
P (y j ) 1 2
p(x) =
P (y j xi )P (xi ) s 2p
å i P (xi , yj )
= (44.8) and the corresponding entropy is
log(2pes 2 )
1
P (y j xi )P (xi )
å i P (xi )P (yj
= 2
xi )
It may be mentioned here that when x is constrained to
some peak value M (−M < x < M), then the entropy is
where P (y j xi ) represents the priori probability that yj maximum when x is uniformly distributed.
is received when xi is transmitted. This is a characteristic
ì 1
of the channel and the receiver. ï , −M < x < M
p(x) = í 2M
The average information received by the receiver in ïî 0, otherwise
bits per symbol is given by the following expression:
I(x ;y) is also referred to as the mutual information 44.5.2 Mutual Information I (x ; y)
of x and y. It may be mentioned here that I(x ;y) is
symmetrical with respect to x and y, that is, The mutual information I(x ; y) of continuous random
variables x and y is given by the following expression:
I (x ; y) = I (y ; x) = H(y) − H(y | x) = H(x) − H(x | y)(44.10)
é 1 ù é 1 ù
I (x ; y) = log ê ú − log ê ú
ë p(x)Dx û ë p(x y) Dx û
The maximum value of I(x ; y) is referred to as the channel
capacity C s . C s in bits per symbol is given by the
following expression: p(x y)
= log (44.15)
Cs = max I (x ; y )
p(x)
(44.11)
P (xi ) = H(x) − H(x y) = H(y) − H(y x)
The maximum value of I(x ; y) is referred to as the channel where C is the channel capacity in bit/s, B is the channel
capacity Cs. Cs in terms of average information trans- bandwidth in hertz, S/N is the signal-to-noise ratio at
mitted per sample is given by the following expression: channel output or receiver input.
Cs = max I (x ; y) (44.16) For a system with infinite bandwidth, the channel
capacity C is given by the following expression:
Channel capacity in bits per second is given by the
following expression:
S
lim C = 1.44 bits/s (44.20)
C = kCs (44.17) B ®¥ z
where k is the number of values transmitted per second.
where N = zB.
H(x) − H(xïy)
Also, I(x ; y) = I(y ; x) = The Shannon−Hartley theorem underlines the
= H(y) − H(yïx) (44.18) fundamental importance of bandwidth and signal-to-noise
ratio in communication. It also shows that for a given
44.6 SHANNON−HARTLEY THEOREM channel capacity, we can exchange increased bandwidth
for decreased signal power. It may be mentioned that
increasing the channel bandwidth by a certain factor does
Shannon−Hartley theorem describes the capacity of not increase the channel capacity by the same factor in a
a noisy channel (assuming that the noise is random). noisy channel as would be suggested by Shannon−Hartley
According to this theorem, theorem apparently. This is because increasing the band-
width also increases noise, thus decreasing the S/N ratio.
S
C = B log2 1 + bits/s (44.19) However, channel capacity does increase with increase in
N bandwidth; the increase will not be in same proportion.
IMPORTANT FORMULAS
1. The information content Ii of message mi is 7. The average information received by the receiver
æ 1 ö
in bits per symbol is
I i = log2 ç ÷ I (x; y) = H(x) − H(x y)
è Pi ø
= å å P (xi , y j ) log
P (xi y j )
2. The entropy is
P (xi )
n n
æ 1 ö i j
H(m) = å P I bits = å Pi log2 çè P ÷ø bits
= å å P (xi , y j ) log
i i P (y j xi )
i =1 i =1 i
n P (y j )
= − å Pi log2 (Pi ) bits
i j
= å å P (xi , y j ) log
P (xi , y j )
i =1
i j P (xi )P (y j )
3. The value of entropy H(m) is maximum when
P1 = P2 = = Pn = 1/n and is given by 8. I(x ; y) = I(y ; x) = H(y) − H(yúx)
n
H(m)max = − å
1 1 9. The channel capacity for a discrete memoryless
log = log n channel is
i =1 n n
Cs = max I (x; y)
4. The average length of the code is P (xi )
∫
1
H(m) H(x) = p(x) log dx
h= −∞
p(x)
L
6. The channel capacity for a binary symmetric 12. For a given mean square value x2 , the entropy is
channel is maximum for a Gaussian distribution of x given by
é 1 ù / 2s 2
e−x
1
Cs = 1 − êPe log + (1 − Pe ) log
1 2
1 − Pe úû
p(x) =
ë Pe s 2p
1. A source produces four symbols with probability (a) 2.8876 (b) 1.8876
1/2, 1/4, 1/8 and 1/8. For this source, a practical (c) 3.8876 (d) 4.8876
coding scheme has an average code word length of
2 bits/symbol. The efficiency of the code is Solution. We have
7
(a) 1 (b) n
8 Entropy = H = − å Pi log2 (Pi )
1 1 i =1
= − [0.8 log2 (0.8) + 0.2 log2 (0.2)]
(c) (d)
2 4
= 0.72 bits/symbol
Solution. We have
n
Entropy = H = −å Pi log2 (Pi ) For the extended coding scheme, using blocks of
i =1 four symbols, the entropy is given by 4 × 0.72 =
é1 æ1ö 1 æ 1 öù 2.8876 bit/4 symbols
ê 2 log2 ç 2 ÷ + 4 log2 ç 4 ÷ ú
è ø è øú The bounds on the average code word length are
= −ê H ≤ L ≤ [H + 1]
ê 1 æ ö
1 æ 1 öú
ê + log2 ç ÷ log2 ç ÷ ú
ë 8 è8ø è 8øû Therefore, lower bound = 2.8876 bits.
7 Ans. (a)
=
4 3. For the data given in Question 2, the upper bound
7/4 7 on the average word length is
Code efficiency = h =
H
= =
L 2 8 (a) 2.8876 (b) 1.8876
Ans. (b) (c) 3.8876 (d) 4.8876
2. A binary source has symbol probabilities 0.8 and 0.2. Solution. From the solution of Question 2, upper
If extension coding (blocks of four symbols) is used, bound = 3.8876 bits
the lower bound on the average code word length is Ans. (c)
PRACTICE EXERCISE
1. Consider a binary symmetric communication channel 6. For the data given in Question 5, the conditional
whose input score is the alphabet X = {0, 1} with entropy of Y, given X, is
probabilities {0.5, 0.5}, and whose output alphabet (a) 0 bits
Y = {0. 1} and channel matrix is (b) 16 bits
é1 − e e ù (c) 8 bits
êë e 1 − e úû (d) Cannot be determined from given data
(2 Marks)
where e is the probability of transmission error.
The entropy of source H(X ) is 7. For the data given in Question 5, the conditional
(a) 1 bit (b) 2 bits (c) 0.5 bit (d) 0.7 bit entropy of X, given Y, is
(2 Marks) (a) 0 bits
(b) 16 bits
2. For the data given in Question 1, the entropy of
(c) 8 bits
the output distribution H(Y) is
(d) Cannot be determined from given data
(a) 1 bit (b) 2 bits (c) 0.5 bit (d) 0.7 bit (1 Mark)
(2 Marks) 8. For the data given in Question 5, the joint entropy
3. For the data given in Question 1, the joint entropy of X and Y is
for X and Y, H(X, Y), is (a) 0 bits
(a) 1 bit (b) 16 bits
(b) 1 − e log2 e − (1 − e ) log2 (1 − e ) (c) 8 bits
(d) Cannot be determined from given data
(c) (1 − e ) log2 (1 − e ) (1 Mark)
(d) 1 − e log2 e 9. Let us consider a continuous communication channel
(2 Marks) having bandwidth W Hz, perturbed by additive
4. For the data given in Question 1, the mutual white Gaussian noise of power spectral density No
information of the channel I(X ; Y) is and average transmitted power P. If the signal-
to-noise ratio (P/NoW ) is increased without limit,
(a) (1 − e ) log2 (1 − e ) then the channel capacity
(b) 1 + e log 2 e + (1 − e ) log2 (1 − e ) (a) increases monotonically without limit
(c) 1 + e log2 e (b) decreases monotonically without limit
(d) 1 + e log2 e + e log2 (1 − e ) (c) remains constant
(d) may increase or decrease
(2 Marks) (1 Mark)
5. Let us consider a random variable X whose entropy 10. For the data given in Question 9, the limit on
H(X) is 8 bits and a deterministic function Y(X) the capacity of the channel if the bandwidth W
that takes on a different value for each value of X. is increased without limit while noise power and
The entropy of Y is signal power are not changed
(a) 0 bits (a) increases monotonically without limit
(b) 16 bits (b) decreases monotonically without limit
(c) 8 bits (c) increases monotonically up to a limit (P/No)log2e
(d) Cannot be determined from given data (d) decreases monotonically up to a limit (P/No)log2e
(2 Marks) (2 Marks)
1. The entropy in bits of the following source 2. What is the shortest possible code length in bits per
alphabet whose letters have the following prob- symbol that could be achieved for a six-letter alpha-
abilities A (1/4), B (1/8), C (1/2) and D (1/8) is bet whose symbols have the following probability
. distribution (1/2, 1/4, 1/8, 1/16, 1/32, 1/32)?
(2 Marks) (2 Marks)
1. (a) Entropy of a source X is 6. (a) It is given that Y always takes a different value
n
H(X ) = − å Pi log2 (Pi ) bits
from X.
i =1 Therefore, conditional entropy of Y, given X,
Therefore, entropy H(X) = −[0.5log2(0.5) + 0.5log2(0.5)] H(YúX) = 0
= 1 bit 7. (a) It is given that Y always takes a different value
2. (a) Output probabilities are p(Y = 0) = 0.5(1 − e) from X.
+ 0.5e = 0.5 and p(Y = 1) = 0.5(1 − e) + 0.5e = 0.5 Therefore, conditional entropy of X, given Y,
Entropy of the source Y is H(XúY) = 0
n
H(Y ) = − å Pi log2 ( Pi ) bits
8. (c) Joint entropy H(X, Y) = H(X) + H(YúX)
= 8 bits
i =1
9. (a) Capacity of a channel in bits per second is
= −[0.5 log2 (0.5) + 0.5 log2 (0.5)] = 1 bit given by
3. (b) The joint probability distribution p(X, Y) is P
C = W log2 1 +
é0.5(1 − e ) 0.5e ù N o W
êë 0.5e 0.5(1 − e )úû
Increasing the quantity P/NoW inside the loga-
The entropy of the joint distribution is rithm without bounds causes the channel capacity
H(X , Y ) = −å p(x, y) log2 p(x, y) bits to increase monotonically without bounds.
x, y
10. (c) Capacity of a channel in bits per second is given by
= −(1 − e ) log2 [0.5(1 − e )] − e log2 (0.5e ) P
C = W log2 1 +
= 1 − e log2 e − (1 − e ) log2 (1 − e ) N o W
1. A source generates three symbols with probabilities Solution. Given that SNR >> 1. Therefore,
æ Sö æSö
0.25, 0.25, 0.50 at a rate of 3000 symbols/s.
Assuming independent generation of symbols, the C1 = B log2 ç 1 + ÷ ≅ B log2 ç ÷
most efficient source encoder would have an average è Nø èN ø
bit rate of When SNR is doubled,
(a) 6000 bit/s (b) 4500 bit/s æ 2S ö æSö
C ¢ ≅ B log2 ç ÷ = B log2 ç ÷ + B log2 2
(c) 3000 bit/s (d) 1500 bit/s èN ø èN ø
(GATE 2006: 2 Marks) = C1 + B
2. A memoryless source emits n symbols each with a Solution. We know that entropy is maximum
probability p. The entropy of the source as a func- when symbols are equal probable, so if probability
tion of n will change from equal to non-equal, entropy will
decrease.
(a) increases as log n Ans. (d)
(b) decreases as log(l/n)
(c) increases as n 5. Let U and V be two independent and identically
(d) increases as nlogn distributed random variables such that P (U = +1) = P (U = −1) =
(GATE 2008: 2 Marks)P (U = +1) = P (U = −1) = 1 . The entropy H(U + V) in bits is
2
Solution. Entropy, (a) 3/4 (b) 1
n (c) 3/2 (d) log23
H(m) = − å Pi log Pi bits
(GATE 2013: 2 Marks)
i =1
Solution. Given that
where Pi is the probability of individual symbol.
1
As probability of each symbol is same, therefore, P (U = +1) = P (U = −1) = .
2
1
P1 = P2 = = Pn = Therefore,
n 1 1 1
P (U + V = +2) = . =
n 2 2 4
H(m) = − å
1 1
log = log n 1 1 1
i =1 n n P (U + V = 0) = + =
Ans. (a) 4 4 2
1 1 1
3. A communication channel with AWGN operating and P (U + V = −2) = . =
at a signal-to-noise ratio (SNR) >> 1 and band-
2 2 4
width B has capacity Cl. If the SNR is doubled Therefore,
keeping B constant, the resulting capacity C2 is 1 1 1
given by H(U + V ) = log2 2 + log2 4 + log2 4
2 4 4
(a) C2 ≈ 2C1 (b) C2 ≈ C1 + B 1 3
= + 1 = bits
(c) C2 ≈ C1 + 2B (d) C2 ≈ C1 + 0.3B 2 2
Ans. (c)
(GATE 2009: 2 Marks)
In this chapter are discussed the topics related to digital communication systems, including the sampling theorem,
digital pulse communication techniques (such as pulse code modulation (PCM), differential PCM, delta modulation
and adaptive delta modulation) and digital modulation techniques (amplitude shift keying (ASK), frequency shift
keying (FSK), phase shift keying (PSK), differential PSK (DPSK), quadrature PSK (QPSK) and offset QPSK).
45.1 SAMPLING THEOREM for baseband or low-pass signals. The minimum sampling
rate of 2fM samples/s is called the Nyquist rate and its
reciprocal the Nyquist interval. For sampling of band-pass
Sampling is the process in which a continuous time signals, lower sampling rates can sometimes be used.
signal is sampled at discrete instants of time and its Sampling theorem for band-pass signals states that if
amplitudes at those discrete instants of time are mea- a band-pass message signal has a bandwidth of fB and an
sured. Quantization is the process by which the sampled upper frequency limit of fu, then the signal can be recov-
amplitudes are represented in the form of a finite set ered from the sampled signal by band-pass filtering if fs ≥
of levels. Encoding process designates each quantized 2fu/k, where k is the largest integer not exceeding fu/fB.
level by a code. Digital transmission of analog signals
has been made possible by sampling the continuous time
signal at a certain minimum rate, which is dictated by 45.2 DIGITAL PULSE COMMUNICATION
what we call as sampling theorem. SYSTEMS TECHNIQUES
Sampling theorem states that a band-limited base-
band signal with the highest frequency component as fM
Hz can be recovered completely from a set of samples Digital pulse communication techniques differ from
taken at the rate of fs samples/s provided that fs ≥ 2fM. analog pulse communication techniques in the sense that,
This theorem is also known as uniform sampling theorem in the case of analog pulse modulation, the sampling
The OOK has the disadvantage that appearance of shown in Fig. 45.6. Modulated carrier signal in this case
any noise during transmission of bit `0’ can be misinter- is represented by the following expression:
A sin ωc1t for bit `1’
preted as data. This problem can be overcome by switch-
ing the amplitude of the carrier between two amplitudes, xc (t) = (45.11)
one representing a `1’ and the other representing a `0’ A sin ωc 2 t for bit `0’
as shown in Fig. 45.5. Again, carrier can be suppressed
to have maximum power in information carrying signals
1 1 0 1 0 0 1
and also one of the sidebands can be filtered out to con-
serve the bandwidth.
1 1 0 1 0 0 1
0 wc2 wc1 w
words means that the bit rate equals the modulation there is no change, a `0’ is transmitted in the form of no
rate. Now, if the number of recognizable phase angles phase change in the carrier.
was increased to 4, then two bits of information could be A BPSK signal is detected using a coherent demod-
encoded into each signal element. ulator where a locally generated carrier component is
extracted from received carrier by a PLL circuit. This
1 1 0 1 0 0 1 locally generated carrier assists in the product demodu-
lation process where the product of the carrier and the
received modulated signals generate the demodulated
output. There could be a difficulty in successfully identi-
fying the correct phase of regenerated signal for demodu-
lation. Differential PSK takes care of this ambiguity to
a large extent.
+1
d0 d4
di(t) t
d2 d6
-1
0 wc w 2T 4T 6T 8T
1 p
x(t) = ⋅ di (t) ⋅ cos w c t +
As each symbol in case of QPSK comprises of two bits,
2 4 symbol transmission rate is half of bit transmission rate
of BPSK, and the bandwidth requirement is halved.
1 p
+ ⋅ dq (t) ⋅ sin w c t + The power spectrum for QPSK is the same as that for
2 4 BPSK.
cos[wct+θ(t)] −1
1 sin(wct+π/4) √2 sin(wct)
√2 0,1 0,0
dq(t)
IMPORTANT FORMULAS
1. In binary PCM, the number of bits to be transmit- 8. The SNR improvement in DPCM over PCM is at
ted per second = nfs least
where n = log2L and L is the number of standard levels. P
Gp = m
2. 2fPCM = nfs Pd
n
fPCM = fs 9. For delta modulation, to avoid slope overload
2
or,
D dm(t)
3. The mean square value or the power of the quanti- >
Ts dt max
zation noise for uniform quantization
(Dv)2 mp2 10. ASK signal is
Nq = = 2
12 3L
A sin w c t for bit `1’
4. The SNR (So/No) for uniform quantization is xc (t) =
0 for bit `0’
So 2
m
= 3L2 2
No mp 11. FSK signal is
5. The m-law for positive amplitude is A sin ωc1t for bit `1’
xc (t) =
mm A sin ωc 2t for bit `0’
ln 1 +
1 m
0 ≤ ≤1
mp
y=
ln(1 + m ) mp
12. BPSK signal is
6. The A-law for positive amplitudes is
xc0 (t) = A cos(w c t + q 0 )
m
for bit `0’
1 , 0 ≤ m ≤ 1
1 + ln A mp mp A xc1(t) = A cos(w c t + q1 ) for bit `1’
y =
1 Am 1
m
, ≤ ≤1
1 + ln A mp A mp
1 + ln 13. QPSK signal
1 p
7. The output SNR using m-law compander is x(t) = ⋅ di (t) ⋅ cos w c t +
2 4
So 3L2 1 p
≅ + ⋅ dq (t) ⋅ sin w c t +
No [ln(1 + m )]2 2 4
SOLVED EXAMPLES
dm(t)
= 12p × 103 cos[(2p × 103 )t] Input x(t) = d (t)
dt
Therefore, X(s) = 1
+16p × 103 cos[(4p × 103 )t]V
Output y(t) = u(t) − u(t − T)
dm(t) Therefore,
is maximum at t = 0
1 e−Ts 1 − e−Ts
dt
Therefore, Y (s) = − =
s s s
dm(t)
= 28p × 103 Transfer function
dt max
Y (s) 1 − e−Ts
dm(t)
To avoid slope overload, we require fsd ≥
H(s) = =
X(s) s
dt max
Therefore, Ans. (a)
5. A signal having uniformly distributed amplitude
28p × 103 in the interval −V to +V is to be encoded using
fs ≥ ≥ 280 × 103 Hz
0.314 PCM with uniform quantization. The signal-to-
The minimum pulse rate = 280 × 103/s quantizing-noise ratio (SQNR) is determined by the
Ans. (a) (a) dynamic range of the signal
3. An analog signal is sampled at 10 kHz. If the (b) sampling rate
number of quantizing levels is 128, the time dura- (c) number of quantizing levels
tion of 1 bit of binary encoded signal in nanosec- (d) power spectrum of signal
onds is
Solution. As the signal is uniformly distributed in
(a) 8124 (b) 10456 the interval −V to +V, therefore the probability
(c) 12109 (d) 14285 density function (PDF) of the signal is as shown
below.
Solution. Number of bits per sample (n) = log2L
= log2128 = 7
px(x)
where L is the number of quantizing levels.
Now, fs = 10 kHz
K
Therefore, time duration of 1 bit of binary encoded
1
signal = = 14285 ns x
7 × 10000 Ans. (d) −V V
Signal power
∞
1 x3
V V
∫ x2 rx (x)dx = ∫
1
x2
2V 3
s= dx =
2V
−∞ −V −V
Therefore,
V2
s=
3
In uniform quantization, quantization noise power
0 3Tb 6Tb 9Tb 12Tb 15Tb 18Tb t
∆ 2
Therefore, it is a rising staircase waveform.
QNP =
12 Ans. (a)
where step size 7. In a BPSK signal detector, the local oscillator has
a fixed phase error of 20°. This phase error deterio-
V p −p V p −p
∆=
rates the SNR at the output by a factor of
=
L 2n (a) cos20° (b) cos220°
(c) cos70° (d) cos270°
Vp2−p (2V )2 V2
QNP = = = Solution. In BPSK, if local oscillator in the detector
12 × 22n 12 × 22n 3 × 22n has a fixed phase error f, then the output power would
reduce by a factor cos2f. Therefore, the SNR deterio-
S V 2 3 × 22n rates by a factor cos2f. Given that f = 20°. Therefore,
SQNR = = × = 22n
QNP 3 V2 SNR at the output deteriorates by a factor of cos 2 20.
Ans. (b)
So, 8. For the bit stream 1101100010, the waveforms in
SQNR ∝ 2 2n the following figures (a), (b) and (c) correspond to
which of the keying techniques, respectively.
So, SQNR is determined by the number of quantiz-
ing levels. 1 1 0 1 1 0 0 0 1 0
Ans. (c)
6. The following figure shows a PCM waveform in (a)
which the amplitude levels of +1 V and −1 V are 1 1 0 1 1 0 0 0 1 0
used to represent binary symbols 1 and 0, respec-
tively. The code word used comprises of three bits.
The sampled version of the analog signal from
which this PCM is derived is (b)
1 0 0 1 1 0 0 0 1 0
001 0 10
+1 (c)
0 (a) ASK, BFSK, BPSK (b) BFSK, BPSK, ASK
t
−1 (c) ASK, BPSK, BFSK (d) BFSK, ASK, BPSK
9. The bit stream 01001 is differentially encoded using Given that the reference bit = 1, logic 0 → p and
`Delay and Ex-OR’ scheme for DPSK transmission. logic 1 → 0°
Assuming the reference bit as a `1’ and assigning
phases of `0’ and `p ’ for 1’s and 0’s, respectively, The transmitted phase sequence is
in the encoded sequence, the transmitted phase
0 1 0 0 1
sequence becomes
(a) p 0 p p 0 (b) 0 p p 0 0
(c) 0 p p p 0 (d) p p 0 p p
1 1 0 0 0 1
0 p p p 0
Solution. The delay and Ex-OR scheme is shown in
the following figure. Therefore, the transmitted phase sequence is 0 p p p 0
Ans. (c)
Input
MOD
10. Coherent demodulation of FSK signal can be
detected using
Delay (a) Correlation receiver
(b) Band-pass filters and envelope detectors
The truth table for an Ex-OR gate is given below, (c) Matched filter
where Y = A ⊕ B = AB + AB (d) Discriminator detection
A B Y
0 0 0 Solution. Coherent demodulation of FSK signal can
be detected using correlation receiver.
0 1 1
Ans. (a)
1 0 1
1 1 0
Numerical Answer Questions 3. Find the Nyquist sampling rate for the signal
x(t) = sinc (200t) sinc 2 (1000t) in samples/s.
1. A signal is quantized using 10-bit PCM. Find the
Solution. sinc(200t) has a rectangular spectrum in
the interval |f | ≤ 100 Hz.
signal-to-quantization-noise ratio in decibels.
sinc2(1000t) = sinc(1000t)sinc(1000t) has a triangu-
lar spectrum in the interval |f | ≤ 1000 Hz.
Solution. Signal-to-quantization-noise ratio in
decibels,
Therefore, X(f ) has a spectrum confined to the
SQNRdB = 6.02n + 1.76 range |f | ≤ 1100 Hz.
Therefore, the Nyquist rate is 2200 samples/s.
Given that n = 10, SQNRdB = 6.02 × 10 + 1.76 Ans. (2200)
= 61.96 dB
Ans. (61.96) 4. A signal has frequency components from 300 Hz
to 1.8 kHz. Find the minimum possible rate in
2. Find the maximum bit rates of an FSK signal in ksamples/s at which the signal has to be sampled.
bps, if the bandwidth of the medium is 12 kHz and
the difference between the two carriers is 2 kHz Solution. Given that fH = 1800 Hz and fL = 300 Hz
(given that transmission mode is full duplex). Therefore, BW = 1800 − 300 = 1500 Hz
f
k = Maximum numerical value of H
Solution. Given that the transmission is full duplex, BW
therefore only 6 kHz is allocated for each direction. 1800
= Maximum numerical value of =1
Bandwidth = Baud rate + fc1 − fc0
1500
Minimum sampling rate
2 × 1800
Therefore, 2fH
(fs )min = = = 3600 samples/s
Baud rate = Bandwidth − (fc1 − fc0) k 1
= 6000 − 2000 = 4000 bps = 3.6 ksamples/s
Ans. (4000) Ans. (3.6)
= a2−(R −1)
5. Let x(t) be modelled as a sample function of a zero
∆=
2a
Step size
mean stationary process with a uniform PDF, in 2R
the range (−a, a). Find the (SNR)0,q assuming a
2-bit code per sample. s x2 a2 3
(SNR)0, q = = 2 −2(R −1)
= 22R
∆ 12
2
a 2 12
Solution. As the signal x(t) has only a finite sup-
port, that is, xmax = a, its variance (SNR)0, q in dB = 6.02R
Given that R = 2
a2
s x2 = Therefore, (SNR)0, q in dB = 12.04.
3 Ans. (12.04)
PRACTICE EXERCISE
1. PCM represents 4. The bit rate required to digitize the human voice
assuming human voice frequencies to be in the
(a) each PCM-encoded sample as a whole
range of 0 to 4000 Hz and number of bits per
(b) first PCM-encoded sample as a whole and fol-
sample to be eight is
lowing samples as differences from the first
PCM-coded sample (a) 32 kbps (b) 64 kbps (c) 8 kbps (d) 4 kbps
(c) first PCM-encoded sample as a whole and fol- (1 Mark)
lowing samples as differences from the previous 5. Let m(t) be a sinusoidal signal with a peak value
PCM-coded sample A. m(t) is fed to a uniform quantizer. The value of
(d) None of the above (SNR)0,q in decibels for R-bit code word per sample is
(1 Mark) (a) 6.02R + 1.8 (b) 6.02R + 2.4
2. Quantization matrix in JPEG compression was (c) 6.02R + 1.2 (d) 6.02R
introduced because (1 Mark)
(a) it is computationally more efficient to work 6. Given a QPSK system with the following parameters
with matrix than with scalar quantization. C = 1 pW, Fb = 60 kbps, N = 1.2 × 10−14 W and
(b) it allows better entropy encoding due to DC B = 120 kHz. The value of energy per bit in dBJ is
and AC coefficient distribution in the 8 × 8 (a) −156.2 (b) −110.1 (c) −167.8 (d) −201.4
block matrix. (2 Marks)
(c) it allows better differentiation of DC and AC 7. For the QPSK system given in Question 6, the
coefficients in the 8 × 8 block matrix than carrier-to-noise power in decibel is
scalar quantization.
(d) None of the above. (a) 19.2 (b) 15.7 (c) 21.5 (d) 56.5
(1 Mark) (2 Marks)
3. A signal x(t) = 2 cos(800pt) + cos(1400p t) is sam- 8. For the QPSK system given in Question 6, the
energy per bit to noise density ratio is
pled with a rectangular pulse train xp(t) as shown in
the following figure. The spectral components of the (a) 13.4 (b) 16.7 (c) 40.5 (d) 22.2
sampled signal in the range of 2500 Hz to 3500 Hz are (2 Marks)
xp(t) 9. For an 8-PSK signal having a bandwidth of 5 kHz,
1 the baud rate and the bit rate, respectively, are
(a) 5000 bauds, 5000 bps (b) 5000 bauds, 15000 bps
(c) 5000 bauds, 40000 bps (d) None of the above
(2 Marks)
10. A constellation diagram consists of equally spaced
−1 −1 1 1 t(ms) points on a circle, 45° apart. If the bit rate is 6000 bps,
6 6 the baud rate is
(a) 2.5 and 3.5 kHz (b) 2.7 and 3.3 kHz (a) 6000 bauds (b) 18000 bauds
(c) 2.8 and 3.4 kHz (d) 2.6 and 3.2 kHz (c) 400/3 bauds (d) 2000 bauds
(2 Marks) (2 Marks)
+90°
fb I/Q C Output
3 coswct
0 0 −0.541 V
2-to-4-level PAM Product 0 1 −1.307 V
Q channel converter modulator 1 0 +0.541 V
1 1 +1.307 V
(a) (b)
18. Flat-top sampling of low-pass signals (a) lower frequency components only
(a) gives rise to aperture effect (b) higher frequency components only
(b) implies oversampling (c) lower amplitudes only
(c) leads to aliasing (d) higher amplitudes only
(d) introducing delay distortion (1 Mark)
(1 Mark) 26. A 4 GHz carrier is DSBSC modulated by a low-
19. For the signal constellation shown in the following pass message signal with maximum frequency of
figure, the type of modulation is 2 MHz. The resultant signal is to be ideally sam-
pled. The minimum frequency of the sampling
cos2π(n/T)t impulse train should be
(a) 4 MHz (b) 8 MHz
S1 S2 (c) 8 GHz (d) 8.004 GHz
(2 Marks)
sin2π(n/T)t
T = symbol duration 27. If the number of bits per sample in a PCM system
S3 S4 is increased from n to n + 1, the improvement in
signal-to-quantization-noise ratio will be
(a) 3 dB (b) 6 dB (c) 2n dB (d) n dB
(a) BPSK (b) QPSK (c) BFSK (d) 8PSK (1 Mark)
(1 Mark)
28. A PCM voice communication system uses bipolar
20. Source encoding in a data communication system return-to-zero pulses for transmission. The signal to
is done in order to be transmitted has a bandwidth of 3.5 kHz with
(a) enhance the information transmission rate peak-to-peak and RMS values of 4 V and 0.2 V,
(b) conserve the transmitted power respectively. If the channel bandwidth is limited to
(c) decrease probability of error 50 kHz, the maximum (SNR)0,q (in dB) of the system
(d) None of the above is (given that the system uses a 7-bit quantizer)
(1 Mark)
(a) 30.2 (b) 26.8 (c) 45.1 (d) 30.9
21. Increased pulse width in the flat-top sampling (2 Marks)
leads to
29. A 1.0 kHz signal is flat-top sampled at the rate
(a) attenuation of high frequencies in reproduction of 1800 samples/s and the samples are applied to
(b) attenuation of low frequencies in reproduction an ideal rectangular LPF with cut-off frequency of
(c) greater aliasing errors in reproduction 1100 Hz, then the output of the filter contains
(d) no harmful effects in reproduction
(1 Mark) (a) Only 800 Hz component
(b) 800 Hz and 900 Hz components
22. A BPSK modulator has a carrier frequency of (c) 800 Hz and 1000 Hz components
70 MHz and an input bit rate of 10 Mbps. The maxi (d) 800 Hz, 900 Hz and 100 Hz components
mum upper sideband frequency in MHz is (2 Marks)
(a) 75 (b) 65 (c) 70 (d) 5 30. The signal-to-quantization-noise ratio in an n-bit
(2 Marks) PCM system
23. For the modulator given in Question 22, the mini- (a) depends upon the sampling frequency employed
mum lower sideband frequency in MHz is (b) is independent of the value of `n’
(a) 75 (b) 65 (c) 70 (d) 5 (c) increases with increasing value of `n’
(1 Mark) (d) decreases with the increasing value of `n’
(1 Mark)
24. For the modulator given in Question 22, the mini-
mum required baud rate (in Mbauds) of the system is 31. A binary channel with a capacity of 48 kb/s is
used for PCM voice transmission. If the highest
(a) 75 (b) 65 (c) 20 (d) 10 frequency component in the message signal is taken
(1 Mark) as 3.2 kHz, the quantizing levels and number of
25. Companding in PCM systems lead to improved bits used are, respectively,
signal-to-quantization-noise ratio. This improve- (a) 256, 8 (b) 64, 6 (c) 32, 5 (d) 128, 7
ment is for (2 Marks)
32. For the given binary channel in Question 31, the 40. In a digital communication system employing FSK,
maximum possible sampling rate (fs) in kilohertz is the 0 and 1 bits are represented by sine waves of
10 kHz and 25 kHz, respectively. These waveforms
(a) 5.2 (b) 6.9 (c) 7.3 (d) 10.7
will be orthogonal for a bit interval of
(1 Mark)
(a) 45 ms (b) 200 ms (c) 50 ms (d) 250 ms
33. The Nyquist rate for message signal represented by (2 Marks)
m(t) = 10cos1000pt cos4000pt is
41. The Nyquist sampling interval for the signal
(a) 10 kHz (b) 2.5 kHz (c) 5 kHz (d) 2 kHz sinc(700t) + sinc(500t) is
(2 Marks) 1 p 1 p
(a) s (b) s (c) s (d) s
34. The number of bits in a binary PCM system is 350 350 700 175
increased from n to n + 1. As a result, the signal- (2 Marks)
+∞
y(t) = 5 × 10−6 x(t)∑ n =−∞ d (t − n
to-quantization-noise ratio will improve by a factor
42. Consider a sample signal
+∞
y(t) = 5 × 10−6 x(t)∑ n =−∞ d (t − nTs ) where x(t) = 10cos(8p × 103)t
(n + 1)/n
(a) (n + 1)/n (b) 2
(c) 22(n + 1)/n (d) which is independent of n
(2 Marks) and Ts = 100 ms. When y(t) is passed through an
ideal low-pass filter with a cut-off frequency of
35. The line code that has zero DC component for 5 kHz, the output of the filter is
(a) 5 × 10−6 cos(8p × 103)t
pulse transmission of random binary data is
(a) non-return to zero (NRZ) (b) 5 × 10−5 cos(8p × 103)t
(c) 5 × 10−1 cos(8p × 103)t
(b) return to zero (RZ)
(c) alternate mark inversion (AMI)
(d) 10cos (8p × 103)t
(d) None of the above
(2 Marks)
(1 Mark)
43. For a bit rate of 8 kbps, the best possible values
36. Given a message signal having a bandwidth of
of the transmitted frequencies in a coherent binary
100 kHz spanning over 200 kHz to 300 kHz. If
FSK system are
the signal is modulated using ASK with number
of bits per signal element equal to 1, the carrier (a) 16 kHz and 20 kHz (b) 20 kHz and 32 kHz
frequency and the bit rate required to transfer the (c) 20 kHz and 40 kHz (d) 32 kHz and 40 kHz
message is (2 Marks)
(a) 250 kHz, 50 kbps (b) 100 kHz, 100 kbps 44. A signal x(t) = 100cos(24p × 103)t is ideally sam-
(c) 200 kHz, 200 kbps (d) None of the above pled with a sampling period of 50 ms and then
(2 Marks) passed through an ideal low-pass filter with cut-off
frequency of 15 kHz. Which of the following fre-
37. Compression in PCM refers to relative compression of quencies is/are present at the filter output?
(a) higher signal amplitudes (a) 12 kHz only (b) 8 kHz only
(b) lower signal amplitudes (c) 12 kHz and 9 kHz (d) 12 kHz and 8 kHz
(c) lower signal frequencies (2 Marks)
(d) higher signal frequencies
(1 Mark) 45. Match List I with List II and select the correct
answer using the code given below the lists:
38. The Nyquist sampling frequency (in Hz) of a signal
given by 6 × 104sinc3(400t) × 106sinc3(100t) is List I List II
A. SSB 1. Envelope detector
(a) 200 (b) 300 (c) 1500 (d) 1000 B. AM 2. Integrate and dump
(2 Marks) C. BPSK 3. Hilbert transform
39. A message signal given by m(t) = Asinwmt is D. 4. Ratio detector
applied to a delta modulator having a step size 5. PLL
of D. Slope overload distortion will occur if (given Codes:
that the sampling frequency is fs). A B C
∆(fs fm ) ∆(fs fm ) (a) 3 1 2
(a) A > (b) A >
2p p (b) 3 2 1
2∆(fs fm )
(c) 2 1 3
(c) A > (d) None of the above (d) 1 2 3
p (1 Mark) (1 Mark)
46. In a modulation system, the modulating voltage 47. Which one of the following systems is an analog
remains the same and the modulation index is system?
halved when the modulating frequency is doubled,
the system is (a) PCM (b) Differential PCM
(c) Delta modulation (d) PAM
(a) AM (b) FM (c) PM (d) PCM (1 Mark)
(1 Mark)
1. The baud rate for a signal having bit rate of 9. A communication system employs ASK to trans-
1 Mbps being transmitted through QPSK using mit a 10 kbps binary signal. Find the baud rate
NRZ-L digital encoding technique is _________ required in bauds.
signal elements per second. (1 Mark)
(2 Marks) 10. For the data given in Question 9, find the mini-
2. An analog signal carries 4 bits in each signal unit. mum bandwidth required in hertz.
If 1000 signal units are sent per second, find the (1 Mark)
baud rate in bauds. 11. A deterministic signal has the power spectrum
(2 Marks) given in the following figure. Find the minimum
3. For the data given in Question 2, find the bit rate sampling rate in hertz needed to completely repre-
in bps. sent the signal.
(1 Mark) s(f)
5. (a) Step size 14. (d) The variance of the quantization noise for
x ≥ 0 is
∆=
2A
2R 2 4
s Q′2 = ∫ (x − 1) fx (x)dx + ∫ (x − 3) fx (x)dx
2 2
A2
Signal power = 0 2
2
Solving the above equation, we get
Signal power × 12 A2 22R12 3 2R
(SNR)0, q = = × = (2 )
s Q′2 =
1
∆2 2 4A2 2
6
(SNR)0, q in dB = 6.02R + 1.8
Total variance is twice that of variance for quan-
6. (c) The energy per bit Eb = 10log(C/Fb) = tization noise.
10log(10−12/60 × 103) = −167.8 dBJ Therefore,
sQ = 2s Q′2 =
7. (a) Carrier to noise power in decibel, (C/N) dB = 2 1
10log(10−12/1.2 × 10−14) = 19.2 dB 3
8. (d) Energy per bit to noise density ratio Eb/No = 15. (c) Quantization noise power (QNP) of a signal is
(C/N) dB + 10log(B/Fb) = 19.2 + 10log (120 × given by
103/60 × 103) = 22.2 dB
∆2
9. (b) For PSK scheme, the baud rate is the same QNP =
12
as the bandwidth, therefore the baud rate is 5000
Step size
bauds. In 8-PSK scheme, the bit rate is three times
V p −p
the baud rate. Therefore, bit rate = 3 × 5000 = ∆=
15000 bps. 2n
where Vp-p is the peak-to-peak signal and n is the 21. (a) Increased pulse width in the flat-top sampling
number of bits. leads to greater attenuation of high frequencies in
Therefore, reproduction. This effect is known as aperture effect.
Vp2−p 22. (a) The output of a BPSK modulator with carrier
QNP = frequency of 70 MHz and bit rate of 10 Mbps is
12 × 22n
(O/P)BPSK = [sin(2p × 5 × 106 )t]×
1
or, QNP ∝ [sin(2p × 70 × 106 )t]
22n
= 0.5 cos[2p × (70 × 106 − 5 × 106 )t]
Ratio of QNP for number of bits n1 and n2 is (Here,
n1 = 8 and n2 = 9.) − 0.5 cos[2p × (70 × 106 + 5 × 106 )t]
Therefore, the maximum upper sideband frequency
22n1 22×8 216
= 70 × 106 + 5 × 106 = 75 MHz
(QNP)2 1 1
= 2n = 2×9 = 18 = 2 =
(QNP)1 2 2
2 2 2 4
23. (b) Minimum lower sideband frequency = 70 × 106
(QNP)1 − 5 × 106 = 65 MHz
(QNP)2 =
4 24. (d) Minimum Nyquist sampling rate = 2 × (75 −
So, the QNP reduces by a factor of 4. 65) MHz = 20 MHz
16. (a) The inputs to the I channel 2-to 4-level con- Two bits are transmitted per symbol. Therefore,
verter are I = 0 and C = 0. From the figure, the baud rate = 10 Mbauds
output of the converter is −0.541 V. The inputs to 25. (c) Companding results in making the SNR uni-
the Q channel 2-to 4-level converter are Q = 0 and form, throughout the signal, irrespective of ampli-
C = 0. From the figure, the output is −0.541 V. tude levels. As in uniform quantization, step size is
Therefore, the two inputs to the I channel product same and the quantization noise power is uniform
modulator are −0.541 V and sinwct. The output throughout the signal.
is (−0.541sinwct). Also, the two inputs to the Q Thus, higher amplitudes of signal will have better
channel product modulator are −0.541 V and SNR than the lower amplitudes.
coswct. The output is −0.541coswct). Hence, companding is used for improving SNR at
The outputs from the I and Q channel product lower amplitudes.
modulators are combined in the linear summer
and produce a modulated output of −0.541sinwct 26. (b) Given that fc = 4 GHz and fm = 2 MHz
− 0.541coswct = 0.765sin(wct − 135°). Therefore, Upper sideband frequency
the amplitude of the transmitter is 0.765 V and the
phase of the transmitter is −135°. fH = fc + fm = 4000 × 106 + 2 × 106 = 4002 MHz
17. (c) Same procedure can be used to calculate the Lower sideband frequency
output for inputs Q = 0, I = 1 and C = 000.
fL = fc − fm = 4000 × 106 − 2 × 106 = 3998 MHz
Therefore, the amplitude is 0.765 V and the
phase is −45°. BW = fH − fL = 4002 × 106 − 3998 × 106 = 4 MHz
18. (a) Flat-top sampling of low-pass signals gives rise Minimum sampling frequency (fs )min = 2(BW) = 8 MHz
to aperture effect. (fs )min = 2(BW) = 8 MHz
19. (b) As the different signals are 90° apart with the
adjacent signal, therefore the modulation scheme is 27. (b) Signal-to-quantization-noise ratio in decibels
QPSK. for a PCM system is (SQNR)dB = (1.76 + 6n)
20. (a) The purpose of source encoding in a data com- For a PCM system with n number of bits per
munication system is to increase the information sample (SQNR)1 = 1.76 + 6n
transmission rate and purpose of channel encoding For a PCM system with (n + 1) number of bits
is to decrease the probability of error. Therefore, per sample (SQNR)2 = 1.76 + 6(n + 1) = 1.76 +
channel coding helps in detection and correction of 6n + 6
errors, and source encoding helps in enhancing the Therefore, (SQNR)2 − (SQNR)1 = (1.76 + 6n + 6) −
information transmission rate. (1.76 + 6n) = 6 dB
So, for every one bit increase in bits per sample will This gives nfs ≤ 48000, or n ≤ 48000/6400 = 7.5
result in 6 dB improvement in SQNR. This gives n = 7.
28. (b) The channel bandwidth requirements of return- Number of quantizing levels
to-zero bipolar pulse = 1/Tb. L = 2n = 27 = 128
Therefore, it is possible to send up to 50000 pulses/s 32. (b) (fs)max = 48000/7 = 6.9 kHz
on this channel. 33. (c) 10cos1000pt cos4000pt = 5 × 2cos1000pt
Bit rate = Sampling rate × Number of bits/sample cos4000pt = 5 × (cos5000pt + cos3000pt)
This is a band-limited signal with the highest fre-
Number of bits/sample is maximum when the sam-
quency component equal to (5000p) radians/s or
pling rate is taken as the minimum value permitted.
2500 Hz.
The signal has a bandwidth of 3.5 kHz. Therefore, Therefore, Nyquist rate = 2 × 2500 = 5000 Hz
minimum sampling rate = 7000 samples/s. = 5 kHz
Given that the system uses a 7-bit quantizer. 34. (d) Signal-to-quantization-noise ratio of a binary
Therefore, the number of quantization levels = 27. PCM system
(2 ⋅ 2 ) 22
3 n 2
29. (c) Given that fm = 1 kHz and fs = 1.8 ksamples/s
(SQNR)2
= 2 = =4
The frequency components in the sampled signal
⋅2
(SQNR)1 3 n 1
are nfs ± fm. 2
For n = 0, the frequency component of the sampled or, (SQNR)2 = 4(SQNR)1
signal is 1000 Hz.
For n = 1, the frequency components of the sam- SQNR increases by a factor of 4. As we can see,
pled signal are 800 Hz and 2800 Hz. this improvement in SQNR is independent of the
For n = 2, the frequency components of the sam- value of `n’.
pled signal are 2600 Hz and 4600 Hz. 35. (c) Alternate mark inversion (AMI) code has zero
For n > 2, the sampled signal contains higher-fre- DC component for pulse transmission of random
quency components. binary data.
Given that the cut-off frequency of LPF is 1100 Hz. 36. (a) The middle of the bandwidth is located at 250 kHz.
Therefore, the output of the filter has 800 Hz and
1000 Hz components. Therefore, the carrier frequency is at 250 kHz
30. (c) The signal-to-quantization noise ratio in an Bandwidth = 300 × 103 − 200 × 103 = 100 kHz
n-bit PCM system is given by
2× Bit rate
Bandwidth =
(SQNR)dB = 1.76 + 6n Number of bits per signal element
From the above equation it is clear that SQNR Therefore,
increases with increase in value of `n’.
31. (d) As per Nyquist criterion, fs ≥ 2fM 100 × 103
Bit rate = = 50 kbps
This gives fs ≥ 6400 samples/s 2
Also, nfs ≤ bit transmission capability of the chan- 37. (a) Compression in PCM refers to relative com-
nel, where n is the number of bits used. pression of higher signal amplitudes.
38. (c) Given signal is 41. (c) Given that the input signal is sinc(700t) +
[6 × 10 sinc (400t)]× [10 sinc (100t)]
4 3 6 3 sinc(500t)
The signal has two frequency components of 500 Hz
Therefore, the sampling frequency is and 700 Hz.
fs = 3fm1 + 3fm2 Therefore, Nyquist sampling frequency = 2fm = 700 Hz
where fm1 = 400 Hz and fm2 = 100 Hz 1 1
Sampling rate = = s
Therefore, fs = 3 × 400 + 3 × 100 = 1500 Hz Sampling frequency 700
39. (a) Given that message signal m(t) = Asinwmt 42. (c) Output of the filter
Therefore, dm(t)/dt = Awmcoswmt
y(t) ⋅ x(t) 5 × 10−6 × 10 cos(8p × 103 )t
The condition for avoiding slope overload is given by = =
∆ ∆ Ts 100 × 10−6
≥ Aw m
dm(t)
≥ or
Ts dt max Ts = 5 × 10−1 cos(8p × 103 )t
∆ ∆fs ∆fs fm
or, A≤ ≤ ≤ 43. (d) As bit rate is 8 kbps, transmitted frequencies
Tsw m 2pfm 2p in coherent BFSK should be integral multiple of
This is the condition for avoiding slope overload. 8 kbps, that is, 32 kHz and 40 kHz.
Thus, slope overload will occur when 44. (b) Given that the sampling period is 50 ms.
∆fs fm Therefore, the sampling frequency is (1/50) × 106
A>
2p = 20 kHz
40. (b) For orthogonality of two sine waves in Tb dura- Given that the signal frequency fm = 12 kHz
tion there should be integral multiple of cycles of The frequency components present after sampling
both the sine waves. are (20 ± 12) kHz or 8 kHz and 32 kHz.
Time period of first sine wave Therefore, frequency at filter output with cut-off
frequency of 15 kHz is 8 kHz.
1
= 100 ms
45. (a) SSB → Hilbert transform
Tb1 =
10 ´ 103
Time period of the second sine wave AM → Envelope detector
1 BPSK → Integrate and dump
Tb2 = = 40 ms
25 ´ 103 46. (b)
Therefore, 200 ms is the integral multiple of both 47. (d)
Tb1 and Tb2 . Hence, the two given waveforms will
be orthogonal for a bit interval of 200 ms.
Therefore, B/fb = antilog 3 = 2 10. For an ASK system, the minimum bandwidth is
Therefore, B = 2 × 10 × 10 Hz = 20 MHz 6 the same as the bit rate of the signal. Therefore,
Ans. (20) minimum bandwidth = 10000 Hz
Ans. (10000)
7. The bandwidth required for the transmission of a
PCM signal 11. Given that fm = 1 kHz
(BW)PCM = nfs From the figure, we can see that approximately
90% of the total signal strength lies in the major
where fs is the sampling frequency and n = log2 L lobe.
(L being the number of levels) Minimum sampling rate required is (fs )min = 2fm
Given that quantization levels = 4. Therefore, Therefore,
n1 = log24 = 2
(fs )min = 2 ×(1×103 ) = 2000 Hz
Given that quantization levels = 64. Therefore,
n2 = log264 = 6 Ans. (2000)
Therefore, (BW)1 = n1fs = 2fs and (BW)2 = n2fs = 6fs 12. Since the peak-to-peak signal amplitude is 1 V,
Hence, therefore, only half of the quantization levels are
utilized.
(BW)2 6f
= s = 3 times SNR = 1.76 + 6 × 7 = 1.76 + 42 = 43.8 dB
(BW)1 2fs
Ans. (43.8)
Therefore, the bandwidth requirement increases by 13. Maximum amplitude of matched filter output is
a factor of 3 when the transmission levels increase
A2 T 102
from 4 to 64. = × 10−4 = 5 mV
Ans. (3) 2 2 Ans. (5)
8. With fs = 45 × 10 samples/s and assuming 8 bits/
3
14. Frames/s = 625
sample the transmitted bit rate Pixels/frame = 400 × 400
= 45 × 103 × 8 = 36 × 104 bps 64 intensity levels per pixel can be represented by
−6
6 bits/pixel.
With fs = =355 ×
×10103 × 10 cos(8p ×number
103 )t of bits/
samples/s, Therefore, date rate = 625 × 400 × 400 × 6
sample that can be used × 10−6
100are = 600 Mbps = 600000 kbps
= 5 × 10−36
1
× 1084p × 10 )t bps
cos( 3
Ans. (6000)
R= = 10.28
35 × 103 Bit rate
15. Baud rate =
As R has to be an integer, therefore R can be taken Number of bits per baud
as 10.
Therefore,
As R has improved by 2 bits, the improvement in
SNR is 12 dB. Bit rate 8000
Number of bits per baud = = =8
Ans. (12) Baud rate 1000
9. For an ASK system, the baud rate required is the Number of bits per baud = log2(Number of signal
same as the bit rate of the signal. The baud rate elements)
= 10000 bauds Therefore, number of signal elements = 28 = 256
Ans. (10000) Ans. (256)
1. At a given probability of error, binary coherent 2. Let x(t) = 2cos(800pt) + cos(1400pt). x(t) is sam-
FSK is inferior to binary coherent PSK by pled with the rectangular pulse train shown in the
figure. The only spectral components (in kHz) pres-
(a) 6 dB (b) 3 dB ent in the sampled signal in the frequency range
(c) 2 dB (d) 0 dB 2.5 kHz to 3.5 kHz are
(GATE 2003: 1 Mark) (a) 2.7, 3.4 (b) 3.3, 3.6
Ans. (b) (c) 2.6, 2.7, 3.3, 3.4, 3.6 (d) 2.7, 3.3
Solution. Bit rate = nfs = 8 × 8 × 103 = 64 kbps The source output is transmitted using two modu-
(SNRq)dB = 1.76 + 6n = 1.76 + 48 = 49.8 dB lation schemes, namely, binary PSK (BPSK) and
Ans. (b) quadrature PSK (QPSK). Let B1 and B2 be the
bandwidth requirements of BPSK and QPSK,
8. In a PCM system, if the code word length is increased respectively. Assuming that the bandwidth of the
from 6 to 8 bits, the signal-to-quantization-noise above rectangular pulses is 10 kHz, B1 and B2 are
ratio improves by the factor
(a) B1 = 20 kHz, B2 = 20 kHz
(a) 8/6 (b) 12 (b) B1 = 10 kHz, B2 = 20 kHz
(c) 16 (d) 8 (c) B1 = 20 kHz, B2 = 10 kHz
(GATE 2004: 1 Mark) (d) B1 = 10 kHz, B2 = 10 kHz
(GATE 2004: 2 Marks)
Solution. Signal-to-quantization noise, SNR ∝ 22n,
where n is the code length. Solution. Given that the bandwidth of the rectan-
When the code length increases from 6 to 8 bits, the gular pulses Rb = 10 kHz
improvement in SNR is by a factor of (22×8/22×6) Therefore, bandwidth of the BPSK system B1 = 2Rb
= 24 = 16 times = 20 kHz
Ans. (c) Therefore, bandwidth of the QPSK system B2 = Rb
9. In the output of a DM speech encoder, the con- = 10 kHz
secutive pulses are of opposite polarity during time Ans. (c)
interval t1 ≤ t ≤ t2. This indicates that during this 11. Consider a binary digital communication system
interval with equally likely 0’s and 1’s. When binary 0 is
(a) the input to the modulator is essentially constant. transmitted, the detector input can lie between the
(b) the modulator is going through slope overload. levels −0.25 V and +0.25 V with equal probability.
(c) the accumulator is in saturation. When binary 1 is transmitted, the voltage at the
(d) the speech signal is being sampled at the detector can have any value between 0 and 1 V
Nyquist rate. with equal probability. If the detector has a thresh-
(GATE 2004: 1 Mark) old of 0.2 V (i.e., if the received signal is greater
than 0.2 V, the bit is taken as 1), the average bit
Solution. In between the two adjacent sampled error probability is
values, if the baseband signal changes by an amount
less than the step size, the output of the DM is a (a) 0.15 (b) 0.2
sequence of alternate positive and negative pulses. (c) 0.05 (d) 0.5
This small change in baseband signal indicates that (GATE 2004: 2 Marks)
the baseband signal is almost constant.
Solution. In the following figure, shaded areas in
Ans. (a)
the figure show the probability of error.
10. A source produces binary data at the rate of 10 kbps.
The binary symbols are represented as shown in 0.25
the figure. Threshold
0.2 1
Binary 1
0
1V
0.2
−0.25 0
0 0.1 t(ms)
For 0 For 1
Let the probability of error when bit `0' is trans- Solution. Impulse response h(t) = x(4 − t)
mitted to Pe(0) and the probability of error when
bit `1' is transmitted be Pe(1) The following figure shows the waveform of h(t)
been drawn from the waveform of x(t).
Error 0.25 − 2 x(−t)
Therefore, Pe (0) = = = 0.1
Total 0.25 − (−0.25)
0.2 − 0
−4 −3 −2
and Pe (1) = = 0. 2
1− 0 1 0
t
As probability of occurrence of 0 and 1 are equal,
therefore average bit error probability
Pe (0) + Pe (1)
Pe (avg) = = 0.15 x(−t+4)
2
Ans. (a)
12. Choose the correct one from among the alterna-
tives a, b, c, d after matching an item from Group 1
0 1 2 3 4
with the most appropriate item in Group 2. t
Group 1 Group 2
1. FM P. Slope overload
2. DM Q. m-law From the figure, we can see that slope in region t
3. PSK R. Envelope detector = 3 to 4 = −1
4. PCM S. Capture effect Ans. (b)
T. Hilbert transform
U. Matched filter 14. A 1 kHz sinusoidal signal is ideally sampled at 1500
samples/s and the sampled signal is passed through
(a) 1-T, 2-P, 3-U, 4-S (b) 1-S, 2-U, 3-P, 4-T an ideal low-pass filter with cut-off frequency
(c) 1-S, 2-P, 3-U, 4-Q (d) 1-U, 2-R, 3-S, 4-Q 800 Hz. The output signals has the frequency
(GATE 2004: 2 Marks)
(a) zero Hz (b) 0.75 kHz
Ans. (c)
(c) 0.5 kHz (d) 0.25 KHz
13. Consider the signal x(t) shown in the figure. Let (GATE 2004: 2 Mark)
h(t) denote the impulse response of the filter
matched to x(t), with h(t) being non-zero only Solution. Given that sampling frequency (fs) =
in the interval 0 to 4 s. The slope of h(t) in the 1500 samples/s and highest frequency component
interval 3 < t < 4 s is is fm = 1 kHz. So, the sampled frequencies are
2.5 kHz and 0.5 kHz.
x(t) However, as the low-pass filter has a cut-off
frequency 800 Hz, so only the output signal of
1 frequency 0.5 kHz would pass through it.
15. Refractive index of glass is 1.5. Find the wave-
t(s) length of a beam of light with frequency of 1014 Hz
1 2 3 4 in glass. Assume velocity of light is 3 × 108 m/s in
vacuum
-1
(a) 3 mm (b) 3 mm
(c) 2 mm (d) 1 mm
(GATE 2005: 1 Mark)
1 −1
(a) s (b) −1 s−1
2 Solution.
1 −1 c = fl (in vacuum)
(c) − s (d) 1 s−1
2 3 × 108
l= = 3 × 10−6 m
c
=
(GATE 2004: 2 Marks) f 1014
In glass would be
−6
l 3 × 10 (a) 2 × 103 (b) 4 × 103
lg = = 2 mm (c) 6 × 103 (d) 8 × 103
=
m 1.5
Ans. (c) (GATE 2006: 2 Marks)
16. A signal as shown in the following figure is applied Solution. Minimum sampling frequency fs = (2fm)
to a matched filter. Which of the following options × 3 = (2 × 1000) × 3 = 6 × 103 Hz
does represent the output of this matched filter? Ans. (c)
Input 18. The minimum step size required for a delta
modulator operating at 32 k samples/s to track
1
the signal (here u(t) is the unit-step function)
1 2 3 x(t) = 125t[u(t) − u(t − 1)] + (250 − 125t) [u(t − 1)
t
− u(t − 2)] so that slope overload is avoided would be
−1
(a) 2−10 (b) 2−8
(a) Output (b) Output (c) 2−6 (d) 2−4
(GATE 2006: 2 Marks)
1
1
2 1 2 3 Solution. To avoid slope overload,
t
∆
1 3 t
−1 ≥ x ′(t)
−1
Ts
(c) Output (d) Output
Given that 1/Ts = 32 × 103 samples/s
2 2
Given that x(t) = 125t[u(t) − u(t − 1)] + (250 −
1 2 3 125t) [u(t − 1) − u(t − 2)]. Therefore, x′(t) = 125
t 1 2 3 t
Therefore,
−1
∆× 32 × 103 ≥ 125
(GATE 2005: 2 Marks) Now, 32 × 103 ≅ 215 and 125 ≅ 27
Therefore,
Solution. Output of the matched filter is computed
as shown in the following figure ∆× 215 ≥ 27
h(t) Output
y(t) Quantizer Q with L y1(t)
A2 T = (1)2 × 2 = 2 levels, step size ∆
Same
1 + quantizer y2(t)
2 allowable signal Q
1 3 dynamic range(−V,V ) +
1 2 t -A2 T
-1
2
-2 x(t) with −V V
, C
range 2 2
Ans. (c)
∆
17. The minimum sampling frequency (in samples/s) (a) ∆ (b)
2
required to reconstruct the following signal from its
samples without distortion ∆2 ∆
(c) (d)
12 L
sin 2p 1000t 3 sin 2p 1000t 2
x(t) = 5
+ 7
pt pt
(GATE 2006: 2 Marks)
Solution. From the following figure, we can see that As it comes in the form of 0/0, so applying LH rule
for y1(t) and y2(t) to be different, minimum step we get
size of ∆/2 is needed else they will be same. 1 d dt (sin 4p Wt)
p
4W d dt [4p Wt(1 − 16W 2 t2 )]
=
(cos 4p Wt)4pW
W
=
∆ (1 − 48W 2 t2 ) 4p W
2 Putting t = 1 4W , we get
∆
∆ 1 cos p
p
4W 1 − 3
2 = = 0.5
Ans. (c)
22. During transmission over a certain binary com-
munication channel, bit errors occur independently
with probability p. The probability of AT MOST
Ans. (b) one bit in error in a block of n bits is given by
20. In delta modulation, the slope overload distortion (a) pn (b) 1 − pn
n − 1
can be reduced by (c) np(1 − p) + (1 − p) n
(d) 1 − (1 − p)n
(a) decreasing the step size (GATE 2007: 2 Marks)
Solution. Given that 23. The ratio of the average energy of constellation 1
to the average energy of constellation 2 is
sin 4pWt
p(t) = (a) 4a2
4p Wt(1 − 16 W 2 t2 ) (b) 4
(c) 2 (d) 8
Putting t = 1 4W , we get (GATE 2007: 2 Marks)
sin 4p W × (1 4W )
p (t ) =
Solution. Average energy of constellation 1 is
4p W ×
1
[1 − 16W 2 (1 16W 2 )] 0 + 4a2 + 4a2 + 8a2
4W E1 = = 4a2
4
∫ f(x) dx = 3
a2 + a2 + a2 + a2 1
E2 = = a2
4 1
5
∫ bdx = 3
Therefore, 1
or,
2
E1 4a 1
= 2 = 4 Ans. (b)
E2 a Solving the above equation, we get
24. If these constellations are used for digital commu- 5 1
b x1 =
nications over an AWGN channel, then which of 3
the following statements is true?
1
or, b=
(a) Probability of symbol error for constellation 1 12
is lower.
(b) Probability of symbol error for constellation 1 Also,
1
is higher.
∫ f(x) dx = 3
1
(c) Probability of symbol error is equal for both the
constellations. −1
(d) The value of No will determine which of the 1
∫ adx = 3
1
two constellations has a lower probability of or,
symbol error. −1
(GATE 2007: 2 Marks)
Solving the above equation, we get
Solution. The probability of symbol error decreases 1 1
with increase in average energy. a x −1 =
3
As constellation 1 has more average energy than
that of constellation 2, therefore the probability of 1
or, a=
symbol error for constellation 1 is lower. 6
Ans. (a)
Ans. (a)
26. Assuming that the reconstruction levels of the
Statement for Linked Answer Questions 25
quantizer are the mid-points of the decision bound-
and 26: An input to a 6-level quantizer has the
aries, the ratio of signal power to quantization
probability density function f(x) as shown in the
noise power is
given figure. Decision boundaries of the quantizer
are chosen so as to maximize the entropy of the 152 64
(a) (b)
quantizer output. It is given that three consecutive 9 3
decision boundaries are `−1’, `0’ and `1’. 76
(c) (d) 28
3 (GATE 2007: 2 Marks)
f(x)
Solution. There will be error if all the three received (a) 1024 (b) 512
bits are 0 or two of the three received bits are 0. (c) 256 (d) 64
Therefore, probability of error in output (GATE 2008: 2 Marks)
S′ =
(a) 64 kHz (b) 32 kHz S
or
(c) 8 kHz (d) 4 kHz 2
(GATE 2008: 2 Marks) Vp − p
Now, S=
2n
Solution. While using the bipolar pulses to transmit
the bits 0 and 1, the minimum bandwidth required where Vp-p is the peak-to peak value and n is the
for distortion-free transmission is four times the no. of bits/sample.
theoretical bandwidth (Nyquist bandwidth). Therefore,
Given that fm = 4 kHz Vp − p Vp − p Vp − p
= =
n′
Therefore, Nyquist bandwidth, 2 2⋅2 n
2 ⋅ 28
′
fs(min) = 2fm = 8 kHz or 2n = 29 = 512
Minimum bandwidth in bipolar signalling is Hence, the number of quantization levels required
to reduce the quantization noise by a factor 4
BW = 4fs (min) = 4 × 8 kHz = 32 kHz would be 512.
Ans. (b) Ans. (b)
VH − VL 5 − (−5)
= n
= = 0. 07 V t t
2 27 0 T 0 T 2T
Hence, closest answer is 0.0667 V, so correct option (GATE 2010: 1 Mark)
is (c).
Ans. (c) Solution. Impulse response of the matched filter for
signal s(t) is h(t) = s(T − t)
32. If the positive values of the signal are uniformly
quantized with a step size of 0.05 V, and the nega- s(T + t) is the left-side shifted version of s(t) by T
tive values are uniformly quantized with a step size as shown in the figure below.
of 0.1 V, the resulting signal-to-quantization-noise s(T + t)
ratio is approximately 1
(a) 46 dB (b) 43.8 dB
(c) 42 dB (d) 40 dB
(GATE 2009: 2 Marks)
V(p−p)2 5
and L2 = = = 50
∆2 0.1
L = L1 + L2 = 150 = 2n
0 T t
Therefore, n = 7 Ans. (c)
34. The Nyquist sampling rate for the signal
(SNR) dB = 1 .72 + 6.02n = 1.72 + 6.02 × 7
sin(500p t) sin(700p t)
= 43.86 dB s(t) = ×
pt pt
Ans. (b)
is given by
33. Consider the pulse shape s(t) as shown. The (a) 400 Hz (b) 600 Hz
impulse response h(t) of the filter matched to this (c) 1200 Hz (d) 1400 Hz
pulse is (GATE 2010: 2 Marks)
So,
Q d Es1 r1 0.707d
= =
r2 Es2 r2 1.307d
d r1
r r Therefore,
Es2 1.307 2
= = 3.42
Es1 0.707
To achieve same error probability, 8-PSK signal
must have 3.42 times the average transmitted
36. For the constraint that the minimum distance signal energy than 4-PSK signal.
between pairs of signal points be d for both con- The value in decibels = 10log(3.42) = 5.33 dB
stellations, the radii r1 and r2 of the circles are Ans. (d)
(a) r1 = 0.707d, r2 = 2.782d 38. In a baseband communications link, frequencies up
(b) r1 = 0.707d, r2 = 1.932d to 3500 Hz are used for signalling. Using a raised
(c) r1 = 0.707d, r2 = 1.545d cosine pulse with 75% excess bandwidth and for
(d) r1 = 0.707d, r2 = 1.307d no intersymbol interference, the maximum possible
(GATE 2011: 2 Marks) signalling rate in symbols/s is
T ∫
sin(w c t) sin(w c t + 45°)dt
2E
Rb =
function of the local oscillator
≤B 0
2
f 1(t) = sin(w c t + 45T°)
2
that is, if Rb is data rate then Rb/2 is the minimum
∫ 2 [sinare
45° + sin(2w c t + 45°)]dt
T 2E 2 1
bandwidth required for transmission. =
T T points
Therefore, Rb ≤ 2B, or Rb ≤ 5250 The coordinates of the message 0
T T T
So (Rb)max = 5250 symbols/s
∫ s1(t)f1(t)=dtT E ∫ 2 dt + T E ∫ sin(2w c t + 45°)dt
s11 = 1 1 1
Ans. (d)
39. A binary symmetric channel (BSC) has a transition
0 0
0
Solving the above integral we get, 0
probability of 1/8. If the binary transmit symbol X
is such that P(X = 0) = 9/10, then the probability E
s11 =
of error for an optimum receiver will be 2
1 E
+ Q E
= Q E
Q
2 No N N
1/4
=
o o -1 0 4 1 2 4
z
5
Ans. (b)
∞
Y = 1 z
41. The bit rate of digital communication system is
R kbps. The modulation used is 32 QAM. The
P
X = 0
= ∫ f 0 dz = 0
1
minimum bandwidth required for ISI-free trans-
Y = 0 z
1
P ∫ f 1 dz = 4 × 4 × 1 = 8
1 1 1
X = 1
mission is =
(a) R/10 Hz (b) R/10 kHz 0
(c) R/5 Hz (d) R/5 kHz Therefore,
(GATE 2013: 1 Mark) 1 1 1 1
n Pe = ×0 + × = Ans. (d)
Solution. We know that 2 = M 2 2 8 16
Therefore, n = log2(M) = log2(32) = 5
43. The optimum threshold to achieve minimum bit
Therefore, baud rate
error rate is
Rb′ =
R R
= kbps 1 4
n 5 (a) (b)
Therefore, the minimum bandwidth required for 2 5
ISI-free transmission is 3
(c) 1 (d)
Rb′
2
R (GATE 2013: 2 Marks)
(BW)min = = kHz
2 10
Ans. (b) Solution. Optimum threshold is given by the point
of intersection of two PDF curves.
Common Data for Questions 42 and 43: Bits
z
1 and 0 are transmitted with equal probability. At f = 1 − z ; z ≤ 1
the receiver, the PDF of the respective received 0
signals for both bits are as shown in the following z
f = z / 4 0<z<2
1
figure: ;
PDF of received
1
signal for bit 0
The point of intersection which decides optimum
threshold is
z
0.5 PDF of received 1−z =
signal for bit 1 4
Therefore,
4
−1 0 1 2 4 z= Ans. (b)
5
In this chapter different multiplexing and multiple access techniques are discussed.
46.1 MULTIPLEXING TECHNIQUES the form of a bit stream. The two techniques are briefly
described in the following paragraphs.
wc1 wc1
m2(t)
LPF
Synchronized m2(t)
LPF
m2(t) m2(t)
∑ Communication BPF LPF Communication
channel channel
wc2 wc2 Commutator Commutator
m3(t) m3(t) mn(t)
BPF LPF mn(t) LPF
LPF
Transmitter
wc3 wc3
Figure 46.2 | Time-division multiplexing.
Figure 46.1 | Frequency-division multiplexing.
different message signals after they have modulated rates of 2.4, 2.4 and 4.8 kHz, then each cycle of com-
their respective carrier signals may be used to modulate mutation will have one sample each from the first two
another high-frequency carrier before it is transmitted messages and two samples from the third message.
over the common link. In that case, these individual car- At the receiving end, the composite signal is demul-
rier signals are known as sub-carrier signals. tiplexed using a similar electronic switching circuitry
FDM is used in telephony, commercial radio broadcast that is synchronized with the one used at the transmit-
(both AM and FM), television broadcast, communica- ter. TDM is widely used in telephony, telemetry, radio
tion networks and telemetry. In case of commercial AM broadcast and data processing.
broadcast, the carrier frequencies for different signals are
If T is the sampling time interval of the time-multiplexed
spaced 10 kHz apart. This separation is definitely not
signal of n different signals each having a sampling inter-
adequate if we consider a high-fidelity voice signal with
a spectral coverage of 50 Hz to -15 kHz. Because of this
val of Ts, then
reason, AM broadcast stations using adjacent carrier Ts
frequencies are usually geographically far apart to mini- T = (46.1)
n
mize interference. In case of FM broadcast, the carrier
frequencies are 200 kHz apart. In case of long-distance Also, if time-multiplexed signal is considered as a low-
telephony, 600 or more voice channels each with a spec- pass signal having a bandwidth of fTDM and fm is the
tral band of 200 Hz to 3.2 kHz can be transmitted over bandwidth of individual signals, then
a coaxial or microwave link using SSB modulation and a
carrier frequency separation of 4 kHz. fTDM = nfm (46.2)
Guard time
f1 f2 fn Carrier Carrier
and clock Unique Signalling and clock Unique
word
Signalling
Transponder recovery word channel recovery channel
bandwidth
Figure 46.3 | Basic concept of FDMA. Figure 46.5 | Typical TDMA frame structure.
time between various traffic bursts from different sta- the frequency of transmitter. The transmitter transmits
tions. The traffic bursts are synchronized to the refer- a short burst of data on a narrowband, then tunes to
ence burst to fix their timing reference. another frequency and transmits again. The transmit-
ter thus hops its frequency over a given bandwidth sev-
eral times per second, transmitting on one frequency
46.2.3 Code-Division Multiple Access
for a certain period of time, then hopping to another
frequency and transmitting again. This is achieved by
In case of code-division multiple access (CDMA), the
using a frequency synthesizer whose output is controlled
entire bandwidth of the transponder is used simultane-
by a pseudorandom code sequence. The pseudorandom
ously by multiple earth stations at all times. CDMA
code sequence decides the instantaneous transmission
therefore allows multiple earth stations to access the same
frequency. On the receiver side, the data can be recov-
carrier frequency and bandwidth at the same time. Each
ered by using a similar frequency synthesizer controlled
transmitter spreads its signal over the entire bandwidth,
by an identical pseudorandom sequence.
which is much wider than that required by the signal oth-
erwise. One of the techniques to do this is to multiply In case of TH-CDMA system, the pseudorandom bit
the information signal, which has a relatively lower bit sequence determines the time instant of transmission of
rate, by a pseudorandom bit sequence with a much higher information. In fact, the signal is transmitted by a user
bit rate. Interference between multiple channels is avoided in rapid bursts during time intervals, which are deter-
as each transmitter uses a unique pseudorandom code mined by the pseudorandom code assigned to the user.
sequence. Receiving stations recover the desired informa- A given user transmits during only one of the M time
tion by using a matched decoder that works on the same slots each frame has been divided into. However, the
unique code sequence used during transmission. CDMA is time slot used by a given user for transmission of data in
also referred to as spread spectrum multiple access because successive frames depends upon the code assigned to it.
of the reason that the carrier spectrum is spread over a As each user transmits its data only during one of the M
much larger bandwidth as compared to the information time slots in each frame, the bandwidth available to it
rate. The spread spectrum signal is inherently immune to increases by a factor of M.
jamming as it forces the jammer to deploy its transmitted It may be mentioned here that DS-CDMA uses the
jamming power over a much wider bandwidth than would entire bandwidth all the time, FH-CDMA uses a small
have been necessary for a conventional system. part of bandwidth at a given time instant but the chosen
CDMA can be of three types, namely, the direct- frequency slot varies with time so as to cover the entire
sequence CDMA (DS-CDMA), frequency-hopping bandwidth and TH-CDMA uses the entire bandwidth
CDMA (FH-CDMA) and time-hopping CDMA for short periods of time.
(TH-CDMA). DS-CDMA uses direct-sequence tech-
niques to achieve the multiple access capability. In 46.2.4 Space Domain Multiple Access
this, each of the N users is allocated its own PN code
sequence. PN code sequences fall in the category of Space domain multiple access (SDMA) is a technique that
orthogonal codes. Cross-correlation of two orthogonal primarily allows frequency reuse where adjacent earth sta-
codes is zero, while their auto-correlation is unity. This tions within the footprint of the satellite can use the same
forms the basis of each of the N stations being able to carrier transmission frequency and still avoid co-channel
extract its intended message signal from a bit sequence interference by using orthogonal antenna beam polariza-
that looks like white noise. tion. Also, transmissions from/to a satellite to/from mul-
In case of a frequency-hop spread spectrum system, the tiple earth stations can use the same carrier frequency
carrier is sequentially hopped into a series of frequency by using narrow antenna beam patterns. As mentioned
slots spread over the entire bandwidth of the satellite earlier also, in an overall satellite link, SDMA is usu-
transponder. The transmitter operates in synchroniza- ally achieved in conjunction with other types of multiple
tion with the receiver, which remains always tuned to access techniques such as FDMA, TDMA and CDMA.
IMPORTANT FORMULAS
1. If T is the sampling time interval of the time- 2. If time-multiplexed signal is considered as a low-
multiplexed signal of n different signals each having pass signal having a bandwidth of fTDM and fm is
a sampling interval of Ts, then the bandwidth of individual signals, then
Ts fTDM = nfm
T =
n
SOLVED EXAMPLES
1. Three message signals m1(t), m2(t) and m3(t) with and using an 8-bit PCM) are transmitted over a
respective bandwidths of 2.4, 3.2 and 3.4 kHz are common communication channel using the TDM
to be transmitted over a common channel in a approach. If the signal is sampled at 1.2 times the
time-multiplexed manner. The minimum sampling Nyquist rate and a single synchronization bit is
rate for each of the three signals if a uniform sam- added at the end of each frame, the duration of
pling rate is to be chosen is each bit in microseconds is
(a) 10.2 kHz (b) 6.8 kHz
(a) 0.125 (b) 0.301
(c) 7.5 kHz (d) 3.4 kHz
(c) 0.678 (d) 0.914
Solution. Sampling rate = 1.2 × 2 × 3.2 × 103 Hz
Solution. As the sampling has to be uniform for the
three signals, minimum sampling rate for each of
= 7.64 kHz
the signals would be twice the highest frequency
component, that is, 2 × 3.4 = 6.8 kHz. Therefore, time period of each multiplexed frame
Ans. (b) = (1/7.64) × 103 s = 130.9 ms
2. For the data given in Question 1, the sampling Now, number of bits in each frame = 24 × 8 + 1 = 193
Therefore, bit duration = 130.9 × 10−6/193 = 0.678 ms
interval of the composite signal in microseconds is
(a) 51 (b) 50 Ans. (c)
(c) 45 (d) 49
Solution. Sampling rate of the composite signal 4. For the data given in Question 3, the transmission
= 3 × 6.8 × 103 Hz = 20.4 kHz
rate is
Therefore, sampling interval of the composite (a) 8 Mb/s (b) 3.32 Mb/s
signal = 1/(20.4 × 103) s = 49 ms (c) 1.475 Mb/s (d) 1.09 Mb/s
Ans. (d) Solution. Transmission rate = 1/0.678 × 10−6 bits/s
3. In a certain digital telephony system, 24 voice chan- = 1.475 Mb/s
nels (each voice channel band-limited to 3.2 kHz Ans. (c)
1. A geostationary satellite has a round-trip propa- This gives maximum chip rate = 1/56 × 10−9 bps
gation delay variation of 40 ns/s due to station- = 17.857 Mbps
keeping errors. If the time synchronization of
DS-CDMA signals from different earth stations is Ans. (17.857)
not to exceed 20% of the chip duration, find the
maximum allowable chip rate (in Mbps) so that 2. If in Question 1, maximum chip rate is to be
a station can make a correction once per satellite 25 Mbps, what is the maximum permissible
round-trip delay. Assume satellite round-trip delay Doppler effect variation due to station-keeping
to be 280 ms. errors in ns/s?
Solution. Chip rate = 25 Mbps
Solution. Doppler effect variation due to station-
keeping errors = 40 ns/s Therefore, chip duration = 1/(25 × 106) s = 40 ns
Satellite round-trip delay = 280 ms Maximum allowable timing error per satellite round
trip = 0.2 × 40 × 10−9 = 8 ns
Therefore, time error due to Doppler effect in one sat-
ellite round trip = 40 × 10−9 × 280 × 10−3 = 11.2 ns
This 8 ns error is to occur in 280 ms.
Therefore, maximum permissible Doppler effect
Let Tc = Chip duration variation = 8 × 10−9/280 × 10−3 = 28.57 ns/s
Therefore, 0.2 × Tc = 11.2 × 10−9 or Tc = 56 ns Ans. (28.57)
PRACTICE EXERCISE
Frequency
1. Multiple access technique that is suitable only for
digital transmission is
(a) FDMA (b) TDMA
(c) CDMA (d) SDMA
(1 Mark)
13. 64-kbps PCM voice-encoded voice channel is trans- (a) FM-FDMA (b) FDM-FM-FDMA
mitted in TDMA mode with the channel allotted (c) PCM-TDM/PSK/FDMA (d) Both (b) and (c)
a time slot of 2 ms in every frame. The length of (1 Mark)
information sub-burst in every frame would be
(a) 128 bits (b) 64 bits 15. Quadrature multiplexing is
(c) 256 bits (d) Indeterminate from given data (a) the same as FDM
(2 Marks) (b) the same as TDM
(c) a combination of FDM and TDM
14. One of the following is a designation of a multi- (d) quite different from FDM and TDM
channel per carrier FDMA system. (1 Mark)
Numerical Answer Questions
1. A certain TDMA transmission has frame efficiency 4. For the data given in Question 6, find the noise
of about 98.5%. If the TDMA frame length and reduction (in dB) achievable in this system.
burst bit rate are, respectively, 15 ms and 80 Mbps, (1 Mark)
determine the number of overhead bits that do not
5. In a certain TDMA system, the TDMA frame
carry any traffic information.
length and burst bit rate are, respectively, 20ms
(2 Marks)
and 90 Mbps. If the total number of overhead bits
2. Four independent messages have bandwidths of per frame is 25000 bits, find the TDMA frame
100, 100, 200 and 400 Hz, respectively. Each is efficiency.
sampled at the Nyquist rate, and the samples are (2 Marks)
time-division multiplexed (TDM) and transmitted.
6. Find the number of 64-kbps PCM-encoded voice
Find the transmitted sample rate (in Hz).
channels that the TDMA system of Question 5 is
(2 Marks)
able to support.
3. In a DS-CDMA system, the information bit
(2 Marks)
rate and chip rate are, respectively, 20 kbps and
20 Mbps. Find the processing gain in decibels.
(2 Marks)
= 20 dB
It is also known as quadrature amplitude modula-
tion (QAM). So, quadrature multiplexing is quite
13. (a) Length of information sub-burst in every frame different from FDM and TDM.
1. Total number of bits = TDMA frame length × burst 2. Sampling rate = 2(100 + 100 + 200 + 400) Hz
bit rate = 1600 Hz
−3
= 15 × 10 × 80 × 10 = 120000 bits
6 Ans. (1600)
Number of overhead bits 3. Chip rate = 20 Mbps
= (1 − frame efficiency) × total number of bits Information bit rate = 20 kbps
= (1 − 0.985) × 120000 = 18000 bits
Ans. (18000)
1. Three analog signals having bandwidths 1200, Assuming a frequency reuse factor of 1/5, that is,
600 and 600 Hz are sampled at their respective a five-cell repeat pattern, the maximum number of
Nyquist rates, encoded with 12-bit words and simultaneous channels that can exist in one cell is
time-division multiplexed. The bit rate for the (a) 200 (b) 40
multiplexed signal is (c) 25 (d) 5
(a) 115.2 kbps (b) 28.8 kbps (GATE 2007: 2 Marks)
(c) 57.6 kpbs (d) 38.4 kbps Solution. Given that allocated bandwidth = 5 MHz
(GATE 2004: 2 Marks) and frequency reuse factor = 1/5
Solution. Overall sampling frequency
Bandwidth allocated for one cell
fs = fs1 + fs2 + fs3 = 2 × 1200 + 2 × 600 + 2 ×
1
600 = 4.8 kHz = 5 × = 1 MHz
Bit rate, Rb = nfs = 12 × 48 × 103 = 57.6 kbps 5
Ans. (c) Number of simultaneous channels
1× 106
2. In a direct-sequence CDMA system, the chip rate = × 8 = 40
is 1.2288 × 106 chips/s. If the processing gain is 200 × 103
Ans. (b)
desired to be at least 100, the data rate
(a) must be less than or equal to 12.288 × 103 bits/s. 4. Four messages band-limited to W, W, 2W and
(b) must be greater than 12.288 × 103 bits/s. 3W, respectively, are to be multiplexed using time-
(c) must be exactly equal to 12.288 × 103 bits/s. division multiplexing (TDM). The minimum band-
(d) can take any value less than 122.88 × 103 bits/s. width required for transmission of this TDM signal is
(GATE 2007: 2 Marks) (a) W (b) 3W
Solution. Given that processing gain ≥ 100 (c) 6W (d) 7W
(GATE 2008: 2 Marks)
Rc
We know that processing gain =
Rb Solution. We have
where Rc is the chip rate and Rb is the data rate. fs1 = 2 × W = 2W , fs2 = 2 × W = 2W ,
Therefore,
fs3 = 2 × 2W = 4W and fs4 = 2 × 3W = 6W
Rc
≥ 100 fs = fs1 + fs2 + fs3 + fs 4 = 14W
Rb
Rc Bit rate Rb = nfs
or, Rb ≤
100 For minimum bandwidth n = 1.
or, Rb ≤ 12.228 × 103 bits/s Therefore,
Ans. (a) Rb = 1 × 14W = 14W
3. In a GSM system, eight channels can co-exist in Rb 14W
200 kHz bandwidth using TDMA. A GSM-based (BW)min = = = 7W
2 2
cellular operator is allocated 5-MHz bandwidth. Ans. (d)