Signal to Noise Ratio

1
 Definition

Noise is the static you hear in the speaker when you tune any AM or FM
receiver to any position between stations. It is also the “snow” or
“confetti” that is visible on a TV screen.
2
 Definition
• Noise: is the unwanted signals that tend to disturb the desired
signal. The presence of noise degrades the performance of analog
and digital communication systems when signal is small.

• The affects of noise is measured by the signal-to-noise power ratio

(SNR)
• Classification of noise: The source of noise may be from external or
internal. In general it is divided to Shot noise (from current pulse)
and Thermal noise (also called AWGN- additional white Gauss
Noise, from temperature)

3
 Sources of noise
Noise

Internal Noise External Noise

• Due to random movement of • Man-made noise: Power lines,
electrons in electronic circuit. motors, or machinery
• Electronic components in a • Natural resources: Atmospheric
receiver such as resistors, noise.
diodes, and transistors are • External noise comes from
major sources of internal noise sources over which we have
• Thermal (agitation) noise little or no control:
• Shot noise • Electromagnetic interference
• Transit time noise (EMI)
• Quantization noise

4
 Gaussian (Normal) Distribution

• The Normal or Gaussian distribution, is an important family

of continuous probability distributions
• The mean ("average", μ) and variance (standard deviation
squared, σ2) are the defining parameters
• The standard normal distribution is the normal distribution
with zero mean (μ=0) and unity variance (σ2 =1)
• Many measurements, from psychological to thermal noise can
be approximated by the Gaussian distribution.
 Gaussian RV
A Gaussian RV, with mean value 𝑚, variance 𝜎 2 , denoted as 𝑁 𝑚, 𝜎 2 . PDF is:

1 (𝑥−𝑚)2
−
𝑓𝑋 (𝑥) = 𝑒 2𝜎2
2𝜋𝜎
Also known as normal distribution.
• 𝑁(0,1) : standard normal distribution, pdf and CDF are given respectively
as:
𝑥
1 −𝑥2 1 −
𝑥2
𝑓𝑋 (𝑥) = 𝑒 2 , 𝐹𝑋 (𝑥) = න 𝑒 2 𝑑𝑥
2𝜋 2𝜋 −∞
• Define function 𝑄(𝑥) as the tail integration of normal Gaussian:
∞ 𝑥2
1 −2
𝑄(𝑥) = න 𝑒 𝑑𝑥
2𝜋 𝑥
• From definition of 𝑄-function, we can see that
𝐹𝑋 (𝑥) = 1 − 𝑄(𝑥) = 𝑄(−𝑥)
 General Gaussian RV

• For general Gaussian RV, 𝑁 𝑚, 𝜎 2 ,

1 𝑥 (𝑡−𝑚)2
−
𝐹𝑋 (𝑥) = න 𝑒 2𝜎2 𝑑𝑡
2𝜋𝜎 −∞

Letting 𝑧 = (𝑡 − 𝑚)/𝜎, we have

(𝑥−𝑚)/𝜎 𝑧2
1 − 𝑥−𝑚
𝐹𝑋 (𝑥) = න 𝑒 2 𝑑𝑧 =1−𝑄
2𝜋 −∞ 𝜎
• Approximation: for 𝑥 > 3,
1 2 /2
𝑄(𝑥) = 𝑒 −𝑥
2𝜋𝑥
PDF of Gaussian Distribution

Standard Normal Distribution

CDF of Gaussian Distribution

Standard Normal Distribution

 Area under Gaussian PDF

The area within +/- σ is ≈ 68% (dark blue)

The area within +/- 2σ is ≈ 95% (medium and dark blue)
The area within +/- 2σ is ≈ 99.7% (light, medium, and dark blue)
 Area under Gaussian PDF
Q. Suppose that the received signal in a communication system is a
Gaussian random variable with a mean of 10 and a variance of 4.
Determine the probability that the received signal is less than 12?

We know that the probability density function (PDF) of a Gaussian

random variable is given by:
1 (𝑥−𝜇)2
−
𝑓𝑋 (𝑥) = 𝑒 2𝜎2
2𝜋𝜎 2
where 𝜇 is the mean and 𝜎 2 is the variance.
(𝑥−10)2
1 −
Substituting the given values, we have: 𝑓𝑋 (𝑥) = 𝑒 2(4)
2𝜋(4)

(𝑥−10) 2
1 − 8
Simplifying, we get: 𝑓𝑋 (𝑥) = 𝑒
2 𝜋
 Area under Gaussian PDF
The cumulative distribution function (CDF) is:
𝑥
𝐹𝑋 (𝑥) = න 𝑓𝑋 (𝑡)𝑑𝑡
−∞

We need to find 𝑃(𝑋 < 12), which is the same as 𝑃 𝑋 ≤ 12 have:

12
𝑃(𝑋 < 12) = 𝐹𝑋 (12) = න 𝑓𝑋 (𝑡)𝑑𝑡
−∞

Substituting the PDF expression and simplifying, we get:

12 (𝑡−10)2
1 −
𝑃(𝑋 < 12) = 𝐹𝑋 (12) = න 𝑒 8 𝑑𝑡
−∞ 2 𝜋

This integral does not have a closed-form solution, but we can use
numerical methods to approximate its value:
𝑃(𝑋 < 12) ≈ 0.6915
➔ Probability that received signal is less than 12 is approximately 0.6915
 Area under Gaussian PDF
Q. Suppose the received signal 𝑦 in a communication system is the
sum of the transmitted signal 𝑥 and additive Gaussian noise 𝑛, with
𝑦 = 𝑥 + 𝑛. The transmitted signal 𝑥 is a binary signal with equal
probability of being either 1 or -1. The noise 𝑛 is a zero-mean
Gaussian random variable with variance 𝜎 2 .
- Derive an expression for the probability density function (PDF) of
𝑦.
- Calculate the probability of error, assuming that the receiver
makes a decision based on whether 𝑦 is positive or negative.
 Area under Gaussian PDF
To find the PDF of 𝑦: Sum of two independent Gaussian random
variables is also Gaussian with mean equal to the sum of the
means and variance equal to the sum of the variances.

Since the transmitted signal 𝑥 takes on the values ±1 with equal

probability, its mean is 0 and variance is 1, i.e., 𝐸[𝑥] = 0 and
𝑉𝑎𝑟[𝑥] = 1. Therefore, the mean and variance of 𝑦 are:
𝐸[𝑦] = 𝐸[𝑥] + 𝐸[𝑛] = 0
𝑉𝑎𝑟[𝑦] = 𝑉𝑎𝑟[𝑥] + 𝑉𝑎𝑟[𝑛] = 1 + 𝜎 2

Thus, we can write the PDF of 𝑦 as:

1 𝑦2
𝑓𝑌 (𝑦) = exp − 2)
2
2𝜋(1 + 𝜎 ) 2(1 + 𝜎
 Area under Gaussian PDF
To calculate the probability of error ➔ determine the probability that 𝑦 is
positive when 𝑥 = −1 and the probability that 𝑦 is negative when 𝑥 = 1.
Let 𝑃𝑒 denote the probability of error.
Then we have:
𝑃𝑒 = 𝑃(𝑦 > 0|𝑥 = −1)𝑃(𝑥 = −1) + 𝑃(𝑦 < 0|𝑥 = 1)𝑃(𝑥 = 1)
∞ 1 0 1
= ‫׬‬0 𝑓𝑌 (𝑦|𝑥 = −1)𝑑𝑦 + ‫׬‬−∞ 𝑓𝑌 (𝑦|𝑥 = 1)𝑑𝑦
2 2
1 ∞ 1 (𝑦 + 1)2 0 1 (𝑦 − 1)2
= ‫׬‬ exp − 𝑑𝑦+ ‫׬‬−∞ exp − 2 ) 𝑑𝑦
2 0 2𝜋(1 + 𝜎 2 ) 2(1 + 𝜎 2 ) 2
2𝜋(1 + 𝜎 ) 2(1 + 𝜎
1 1 1 1
= 𝑄 +𝑄 =𝑄
2 1 + 𝜎2 1 + 𝜎2 1 + 𝜎2

where 𝑄(𝑥) is the complementary cumulative distribution function (CCDF)

of the standard normal distribution.
 Thermal Noise
• Thermal noise is the result of the random motion of charged particles
(usually electrons) in a conducting medium such as a resistor.
• Because the number of electrons is very large and their motions are
statistically independent of each other, the central limit theorem indicates that
thermal noise is Gaussian distributed with zero mean.

Noise Type Source Characteristics Effect on Signal Typical Appearance

Random motion of
Continuous, Gaussian- Causes slight, constant Smooth, "fuzzy" fluctuation
Thermal Noise electrons in resistive
distributed, wideband distortion over entire signal throughout
components
Generated by multiplying
Nonlinear mixing of Introduces unwanted
signals; frequencies at
Intermodulation Noise multiple frequencies (e.g., frequencies; distorts Complex, wavy distortions
sums/differences of input
in amplifiers) original waveform
signals

Signal leakage from Coherent signal from Adds interfering signal;

Looks like an overlay of
Crosstalk adjacent channels or another source, often reduces clarity or causes
another sine wave
circuits sinusoidal misinterpretation

Sudden external Causes bit errors, data

Short-duration, high- Sharp vertical spikes at
Impulse Noise disturbances (e.g., power loss, and burst noise in
amplitude spikes random positions
lines, lightning, switching) digital systems

16
Thermal Noise

17
Thermal Noise (Johnson Noise /white noise)

Movement of the electrons

will form kinetic energy in the
conductor related to the
temperature of the conductor.

When the temperature increases the movement of free electrons will

increases and the current flows through the conductor.

18
Thermal Noise (Johnson Noise /white noise)

Experimental results (by Johnson) and theoretical studies (by

Nyquist) give the mean square noise voltage as

Where k = Boltzmann’s constant = 1.38 x 10-23 Joules per K

T = absolute temperature (Kelvin)
B = bandwidth noise measured in (Hz)
R = resistance (ohms)

19
 Thermal Noise (Johnson Noise)
Example: A wireless communication with 50kΩ resistor at a
temperature of 290K, 3 kHz bandwidth. Find Vrms value of noise:

Vrms = √ 4 x 1.38 x 10-23 x 290 x 3000 x 50 = 49 nV

Example: One operational amplifier with a frequency range of (18-20)

MHz has input resistance 10 k. Calculate noise voltage at the input
if the amplifier operate at ambient temperature of 270C:

Vn2 = 4KTBR = 4 x 1.38 x 10-23 x (273+ 27) x 2 x 106 x 104

Vn = 18 volt

from Kelvin to Kelvin

Celsius [°C] = [K] − 273.15 [K] = [°C] + 273.15
Fahrenheit [°F] = [K] × 9⁄5 − 459.67 [K] = ([°F] + 459.67) × 5⁄9

20
 Signal to Noise
The signal to noise ratio is given by

S Signal Power , Ps ( watts )

=
N Noise Power , Pn ( watts )
The signal to noise in dB is expressed by
S and N
S S  S 
= 10 log 10 dB   dB =S dBm − N dBm measured in
N N N mW.

Si N i
Noise Figure F =
So N o
Noise Factor NF = 10 log F
S N
= 10 log i i dB
So N o

21
 Signal to Noise
Example : For an amplifier with an output signal power of 10 W and an
output noise power of 0.01 w, determine the signal to noise power ratio

S Ps 10
= = = 1000
N Pn 0.01
To express in dB;

S S 10
  dB =10 log 10   = 10 log = 30dB
N N 0.01
Example: For an amplifier with an output signal voltage of 4V, an output
noise voltage of 0.005 V, and an input and output resistance of 50 ohm,
determine the signal to noise power ratio.

S  Vs  4
 
  dB = 20 log 10   = 20 log = 58.06dB
N  Vn  0.005 22
 Transmission Loss
Transmission Medium Frequency Loss dB/km
Twisted-pair Cable 10kHz 2
100kHz 3
300kHz 6
Coaxial Cable 100kHz 1
1MHz 2
3MHz 4
Rectangular Waveguide
10GHz 5

Fiber Optic Cable 3.6 x 1014Hz 2.5

2.4 x 1014Hz 0.5
1.8 x 1014Hz 0.2
23
 Noise - Bit Energy
The signal also measured in terms of the bit energy
in joules (J), Eb.

The energy per bit is simply the energy of a single bit

of information, Eb.
It is defined as below:
Eb = CTb (J/bit)
Eb = energy of a single bit (joules per bit)
Tb = time of a single bit (seconds)
C = carrier power (watts)
 White process
• White process is processes in which all frequency components
appear with equal power, i.e., the power spectral density (PSD),
Sx(f), is a constant for all frequencies.
• The PSD of thermal noise, Sn(f), is usually given as S n ( f ) = kT2
(where k is Boltzrnann's constant and T is the temperature)
• The value kT is usually denoted by N0, Then S n ( f ) = 20
N

25
Effect of Noise
Effect of Noise on a Baseband System
• No carrier demodulation to be performed.
• The receiver consists only of an ideal lowpass filter with the
bandwidth W.
• The noise power at the output of the receiver, for a white noise
input, is W N0
Pn0 =  df = N 0W
−W 2
• If we denote the received power by PR, the baseband SNR is
given by
S PR
  =
 N b N 0W

27
Effect of Noise on a Baseband System
• For comparing, SNRo of LPF used to as a reference value
(W: bandwidth)

28
 Effect of Noise on DSB-SC AM
• Transmitted signal :
u (t ) = Ac m(t ) cos(2 f c t )
• The received signal at the output of the receiver noise-limiting filter:
Sum of this signal and filtered noise.
• A filtered noise process can be expressed in terms of its in-phase
and quadrature components as

(where nc(t) is in-phase component and ns(t) is quadrature

component)

29
 Effect of Noise on DSB-SC AM
• Received signal (Adding the filtered noise to the modulated signal)

r (t ) = u (t ) + n(t )
= Ac m(t ) cos(2 f c t ) + nc (t ) cos(2 f c t ) − ns (t ) sin (2 f c t )

• Demodulate the received signal by first multiplying r(t) by a locally

generated sinusoid cos(2fct + ϕ), where ϕ is the phase of the
sinusoid.
• Then passing the product signal through an ideal lowpass filter
having a bandwidth W.
 Effect of Noise on DSB-SC AM
• The multiplication of r(t) with cos(2fct +ϕ) yields

Lowpass filter rejects the double frequency, passes only lowpass

components.

31
 Effect of Noise on DSB-SC AM

• Assume that we are employing a coherent demodulator.

– With this assumption, we assume that ϕ = 0

y (t ) = 1
2
Ac m(t ) + nc (t )

32
 Effect of Noise on DSB-SC AM
• Therefore, at the receiver output, the message signal and the noise
components are additive and we are able to define a meaningful
SNR. The message signal power is given by
Ac2
Po = PM
4
– power PM is the content of the message signal
• The noise power is given by
1 1
Pn0 = Pnc = Pn
4 4
• The power content of n(t) can be found by noting that it is the result
of passing nw(t) through a filter with bandwidth Bc.

33
 Effect of Noise on DSB-SC AM
• Therefore, the power spectral density of n(t) is given by
 N20 | f − f c | W
Sn ( f ) = 
0 otherwise

• The noise power is

 N0
Pn =  S n( f )df =  4W = 2WN 0
− 2
• Now we can find the output SNR as
Ac2
S P0 PM Ac2 PM
  = =1 = 4

 N 0 Pn0 4 2WN 0 2WN 0

• In this case, the received signal power is
PR = Ac2PM /2.

34
 Effect of Noise on DSB-SC AM
• The output SNR for DSB-SC AM may be expressed as

S PR
  =
 N  0 DSB N 0W

which is identical to baseband SNR.

• In DSB-SC AM, the output SNR is the same as the SNR for a baseband
system
 DSB-SC AM does not provide any SNR improvement over a
simple baseband communication system

35
SNR of DSB systems

36
 Effect of Noise on SSB AM
• SSB modulated signal :

• Input to the demodulator

• Assumption : Demodulation with an ideal phase reference.

• Output of the lowpass filter is the in-phase component (with a
coefficient of ½) of the preceding signal.

37
 Effect of Noise on SSB AM
• We have
Ac2
Po = PM
4
1 1 S P0 Ac2 PM
Pn0 = Pnc = Pn   = =
4 4  N  0 Pn0 WN 0
 N0
Pn =  S n( f )df =  2W = WN 0
− 2

PR = PU = Ac2 PM

S PR S
  = =  
  0 SSB
N N 0W  N b

• The signal-to-noise ratio in an SSB system is equivalent

to that of a DSB system.
38
SNR of the SSB system

39
 The coherent receiver

• Narrowband noise is produced when white noise go pass a narrowband

filter (2W<<fc , so noise from –W to W).
(no(t) and Φ(t) are random processes in bellow formular)

40
 Effect of Noise on Conventional/
coherent AM
DSB AM signal : u (t ) = Ac [1 + amn (t )] cos(2 f c t )
• Received signal at the input to the demodulator
r (t ) = Ac [1 + amn (t )] cos(2 f c t ) + n(t )
= Ac [1 + amn (t )] cos(2 f c t ) + nc (t ) cos(2 f c t ) − ns (t ) sin (2 f c t )
= Ac [1 + amn (t )] + nc (t )cos(2 f c t ) − ns (t ) sin (2 f c t )
– a is the modulation index
– mn(t) is normalized so that its minimum value is -1
– If a synchronous demodulator is employed, the situation is
basically similar to the DSB case, except that we have
1 + amn(t) instead of m(t).
• After mixing and lowpass filtering y (t ) = 1
2
Ac amn (t ) + nc (t )
41
 Effect of Noise on coherent AM
Received signal power
𝐴2𝑐
𝑃𝑅 = 1 + 𝑎2 𝑃𝑀𝑛
2
Assumed that the message process is zero mean.
Now we can derive the output SNR as
1 2 2 2 2 2
𝐴2𝑐 2𝑃
𝑆 𝐴 𝑐 𝑎 𝑃𝑀 𝐴 𝑎 𝑃 𝑎 𝑃 1 + 𝑎 𝑀𝑛
=4 2
𝑛 𝑐 𝑀𝑛 𝑀𝑛
= =
𝑁 1 2𝑁0 𝑊 1 + 𝑎2 𝑃𝑀𝑛 𝑁0 𝑊
0𝐴𝑀 𝑃 𝑛
4 𝑐
𝑎2 𝑃𝑀𝑛 𝑃𝑅 𝑎2 𝑃𝑀𝑛 𝑆 𝑆
= = = 𝜂
1 + 𝑎2 𝑃𝑀𝑛 𝑁0 𝑊 1 + 𝑎2 𝑃𝑀𝑛 𝑁 𝑏 𝑁 𝑏

𝑎2 𝑃𝑀𝑛
• 𝜂= denotes the modulation efficiency
1+𝑎2 𝑃𝑀𝑛
Since 𝑎2 𝑃𝑀𝑛 < 1 + 𝑎2 𝑃𝑀𝑛 , the SNR in conventional AM is always smaller
than the SNR in a baseband system.
42
 Effect of Noise on coherent AM
– In practical, modulation index about 0.8-0.9.
– Power content of the normalized message process depends on
the message source.
– Speech signals : Large dynamic range, PM is about 0.1.
• The overall loss in SNR, when compared to a baseband
system, is a factor of 0.075 or equivalent to a loss of 11 dB.
– The reason for this loss is that a large part of the transmitter
power is used to send the carrier component of the modulated
signal and not the desired signal.
• To analyze the envelope-detector performance in the presence of
noise, ➔ approximations (due to nonlinear structure of an envelope
detector ➔ exact analysis is difficult).

43
 Effect of Noise on coherent AM
The input to the envelope detector is
𝑟(𝑡) = 𝐴𝑐 1 + 𝑎𝑚𝑛 (𝑡) + 𝑛𝑐 (𝑡) cos 2𝜋𝑓𝑐 𝑡 − 𝑛𝑠 (𝑡)sin 2𝜋𝑓𝑐 𝑡
- Therefore, the envelope of 𝑟(𝑡) is given by

𝑉𝑟 (𝑡) = 𝐴𝑐 1 + 𝑎𝑚𝑛 (𝑡) + 𝑛𝑐 (𝑡) 2 + 𝑛𝑠2 (𝑡)

- Now we assume that the signal component in 𝒓(𝒕) is much

stronger than the noise component. Then
𝑃 𝑛𝑐 (𝑡) << 𝐴𝑐 1 + 𝑎𝑚𝑛 (𝑡) ≈ 1
- Therefore, we have a high probability that
𝑉𝑟 (𝑡) ≈ 𝐴𝑐 1 + 𝑎𝑚𝑛 (𝑡) + 𝑛𝑐 (𝑡)

44
 Effect of Noise on coherent AM
• After removing the DC component, we obtain
y (t ) = Ac amn (t ) + nc (t )

which is basically the same as y(t) for the synchronous

demodulation without the ½ coefficient.
– This coefficient, of course, has no effect on the final SNR.
– So we conclude that, under the assumption of high SNR at
the receiver input, the performance of synchronous and
envelope demodulators is the same.
• However, if the preceding assumption is not true, that is, if we
assume that, at the receiver input, the noise power is much
stronger than the signal power ➔

45
 Effect of Noise on coherent AM
𝑉𝑟 (𝑡) = 𝐴𝑐 1 + 𝑎𝑚𝑛 (𝑡) + 𝑛𝑐 (𝑡) 2 + 𝑛𝑠2 (𝑡)

= 𝐴2𝑐 1 + 𝑎𝑚𝑛 (𝑡) 2 + 𝑛𝑐2 (𝑡) + 𝑛𝑠2 (𝑡) + 2𝐴𝑐 𝑛𝑐 (𝑡) 1 + 𝑎𝑚𝑛 (𝑡)

𝑎 2𝐴𝑐 𝑛𝑐 (𝑡)
⟶ ቆ𝑛𝑐2 (𝑡) + 𝑛𝑠2 (𝑡) 1 + 1 + 𝑎𝑚𝑛 (𝑡)
𝑛𝑐2 (𝑡) + 𝑛𝑠2 (𝑡)
𝑏 𝐴𝑐 𝑛𝑐 (𝑡)
⟶ 𝑉𝑛 (𝑡) 1 + 2 1 + 𝑎𝑚𝑛 (𝑡)
𝑉𝑛 (𝑡)
𝐴𝑐 𝑛𝑐 (𝑡)
= 𝑉𝑛 (𝑡) + 1 + 𝑎𝑚𝑛 (𝑡)
𝑉𝑛 (𝑡)
- 𝐴2𝑐 1 + 𝑎𝑚𝑛 (𝑡) 2
is small compared with other components
- 𝑛𝑐2 (𝑡) + 𝑛𝑠2 (𝑡) = 𝑉𝑛 (𝑡); the envelope of the noise process
2𝐴𝑐 𝑛𝑐 (𝑡)
• Use the approximation where 𝜀 = 2 1 + 𝑎𝑚𝑛 (𝑡)
𝑛𝑐 (𝑡)+𝑛𝑠2 (𝑡)
𝜀
1 + 𝜀 ≈ 1 + , for small 𝜀
2
46
 Effect of Noise on coherent AM
• Then Vr (t ) = Vn (t ) +
Ac nc (t )
(1 + amn (t ) )
Vn (t )

– We observe that, at the demodulator output, the signal and

the noise components are no longer additive.
– In fact, the signal component is multiplied by noise and is
no longer distinguishable.
– In this case, no meaningful SNR can be defined.
– We say that this system is operating below the threshold.
– There is threshold effect in AM envelope detector, results
from the nonlinear action, occurred when (S/N)i<10dB

47
Q. An AM system operates with a modulation index of 0.5 , and the
power in the normalized message signal is 0.1𝑊. Is this system in
ideal condition?

(0.5)2 (0.1)
The efficiency is 𝜂 = = 0.0244
1+(0.5)2 (0.1)
𝑃𝑇
and the post-detection SNR is (SNR)𝐷 = 0.0244
𝑁0 𝑊

The detection gain is

(SNR)𝐷
= 2𝜂 = 0.0488
(SNR) 𝑇
This is more than 16 dB inferior to the ideal system requiring the
same bandwidth. The motivation for using AM is not noise
performance but rather that AM allows the use of simple envelope
detectors for demodulation. Reason for the poor efficiency of AM is?

48
 Effect of Noise on AgM
• Recall that in AM, the message is contained in the amplitude of the
modulated signal
• Since noise is additive, the noise is directly added to the signal.
• However, in a frequency-modulated signal, the noise is added to the
amplitude and the message is contained in the frequency of the
modulated signal.
• Therefore, the message is contaminated by the noise to the extent
that the added noise changes the frequency of the modulated signal.
• The frequency of a signal can be described by its zero crossings.
• So the effect of additive noise on the demodulated FM signal can be
described by the changes that it produces in the zero crossings of
the modulated FM signal.

49
 Effect of Noise on AgM
• The effect of noise in a low-power FM system is more severe than in
a high-power FM system.
• In a low power signal, noise causes more changes in the zero
crossings.
• In a low-power FM system, the transmitted signal has a lower
amplitude, which makes it more susceptible to noise.
• In contrast, in a high-power FM system, the transmitted signal has a
higher amplitude, which makes it less susceptible to noise. The
noise-induced fluctuations in amplitude are typically small compared
to the signal amplitude, which results in fewer changes in the zero
crossings..

Effect of noise in50FM

 Effect of Noise on AgM
• The angle-modulated signal is represented as
 A cos 2f t + 2k t m( )d  FM
  c f  
u (t ) = Ac cos(2f c t +  (t ) ) =  c  − 
 Ac cos(2f c t + k p m(t ) ) PM
• The AWGN nw(t) is added to u(t), and the result is passed through a
noise-limiting filter whose role is to remove the out-of-band noise.
• The bandwidth of this filter is equal to that of the modulated signal
• Therefore, it passes the modulated signal without distortion.
• However, it eliminates the out-of-band noise.
• Hence, noise output of the filter is a filtered noise denoted by n(t).

51
 Effect of Noise on AgM
• The output of this filter is
r (t ) = u (t ) + n(t ) = u (t ) + nc (t ) cos(2 f c t ) − ns (t ) sin (2 f c t )
• A precise analysis is complicate due to the nonlinearity of
demodulation .
• Let us assume that the signal power is much higher than the
noise power.
• Then, the bandpass noise is represented as
 ns (t ) 
2
c
2
s

n(t ) = n (t ) + n (t ) cos 2f c t + arctan  = Vn (t ) cos(2f c t +  n (t ) )
 nc (t ) 
– where Vn(t) and n(t) represent the envelope and the phase of
the bandpass noise process.
52
 Effect of Noise on AgM
• The phasor diagram of signal and noise are shown, we can write

53
 Effect of Noise on AgM
 k p m(t ), PM
• Noting that  (t ) =  t
2k f − m( )d , FM

• We see that the output of the demodulator is given by


sin ( n (t ) −  (t ) )
Vn (t )
  (t ) + PM
 Ac
y (t ) = 
 
 1 d  (t ) + Vn (t ) sin ( n (t ) −  (t ) ) FM
 2 dt  Ac 

sin ( n (t ) −  (t ) )
Vn (t )
 k m ( t ) + PM  k p m(t ) + Yn (t ) PM

p
Ac
= = 1 d
+
sin ( n (t ) −  (t ) ) FM  f
1 d Vn (t ) k m (t ) Yn (t ) FM
k f m(t ) + 2 dt
 2 dt Ac
– where we define Yn (t ) def n sin ( n (t ) −  (t ) )
V (t )
Ac
54
 Effect of Noise on AgM

sin ( n (t ) −  (t ) )
Vn (t )
 k m (t ) + PM  k p m(t ) + Yn (t ) PM

p
Ac
y (t ) =  = 1 d
+
sin ( n (t ) −  (t ) ) FM  f
1 d Vn (t ) k m (t ) Yn (t ) FM
k f m(t ) + 2 dt
 2 dt Ac

– The first term is the desired signal component.

– The second term is the noise component.
– The noise component is inversely proportional to the signal
amplitude Ac.
– Hence, the higher the signal level, the lower the noise level.
– This is not the case with amplitude modulation.
– In AM systems, the noise component is independent of the
signal component, and a scaling of the signal power does not
affect the received noise power.

55
 Effect of Noise on AgM
• The properties of the noise component
𝑉𝑛 (𝑡)
𝑌𝑛 (𝑡) = sin Φ𝑛 (𝑡) − 𝜙(𝑡)
𝐴𝑐
1
= 𝑉𝑛 (𝑡)sin Φ𝑛 (𝑡)cos 𝜙(𝑡) − 𝑉𝑛 (𝑡)cos Φ𝑛 (𝑡)sin 𝜙(𝑡)
𝐴𝑐
1
= 𝑛 (𝑡)cos 𝜙(𝑡) − 𝑛𝑐 (𝑡)sin 𝜙(𝑡)
𝐴𝑐 𝑠
• when we compare variations in 𝑛𝑐 (𝑡) and 𝑛𝑠 (𝑡), we can assume
that 𝜙 (𝑡) is almost constant:
1
𝑌𝑛 (𝑡) = 𝑛𝑠 (𝑡)cos 𝜙 − 𝑛𝑐 (𝑡)sin 𝜙
𝐴𝑐
cos 𝜙 sin 𝜙
= 𝑛 (𝑡) − 𝑛 (𝑡), where 𝑎 = cos 𝜙/𝐴𝑐 and 𝑏 = sin 𝜙/𝐴𝑐
𝐴𝑐 𝑠 𝐴𝑐 𝑐
= 𝑎𝑛𝑠 (𝑡) − 𝑏𝑛𝑐 (𝑡)
56
 Effect of Noise on AgM
• We have
( )
SYn ( f ) = a + b S nc ( f ) =
2 2
S nc ( f )
Ac2
• Snc(f) is the power spectral density (psd) of the in-phase component
of the filtered noise:
N0 | f | B2c
S nc ( f ) = 
0 otherwise

– Note that bandwidth of the filtered noise extends from fc – Bc/2 to

fc + Bc/2 . Hence, spectrum of nc(t) extends from –Bc/2 to +Bc/2.
• Therefore
 20
N
| f | 2c
B
SYn ( f ) =  Ac (PM)
0 otherwise
57
 Effect of Noise on AgM
• This equation provides an expression for the power spectral density
of the filtered noise at the front end of the receiver.
• After demodulation, another filtering is applied; this reduces the
noise bandwidth to W, which is the bandwidth of the message signal.
• Note that in the case of FM modulation, the process Yn(t) is
differentiated and scaled by 1/2.
• The PSD of the process (1/2) (dYn(t)/dt) is given by

(FM)

58
 Effect of Noise on AgM
• In both cases, Bc/2 must be replaced by W after Lowpass filter.
• Hence, for |f|<W

• Power spectrum of the noise component at the output of the

demodulator for PM and FM.

Noise power spectrum at demodulator output for |f|<W in (a) PM (b) and (b) FM.
59
 Effect of Noise on AgM
• PM has a flat noise spectrum and FM has a parabolic noise
spectrum.
• Therefore, effect of noise in FM for higher frequency components is
much higher than the effect of noise on lower frequency components.
• The noise power at the output of the lowpass filter is the noise
power in the frequency range [W, +W].

 W N 0 df 
 −W Ac2
2WN 0
W  Ac2 PM
Pn0 =  S n0 ( f )df =  W =  2 N 0W 3

−W

N0
2 f
2
df 
FM
 −W c
A 3 Ac2

60
 Effect of Noise on AgM
k p2 PM PM
• First, we have the output signal power PSO =  2
k f PM FM
 k p2 Ac2 PM
 PM
• Then the SNR, which is defined as S  2 NW
  = 2 2 0
 N O  3k f Ac PM
S PSO
 2W 2 N 0W
FM
  def
 N O PnO

• Noting that Ac2/2 is the received signal power, denoted by PR , and

  p
2
 PM
  p = k p max m(t )  PM  PR  
  max m(t )   N 0W
 PM

 k f max m(t )  S
  =

 f = FM  N O   f
2
 PM
 W  
61 

3PR 
 max  m (t )   NW
 0
FM
 Effect of Noise on AgM
S PM
• Denote   =
 N b N 0W
is SNR of a baseband system with the same received power, we
obtain
 PM  p2 S
 2   PM
S  (max m(t ) )  N b
  =
 N O  PM  f2 S
3   FM
 (max m(t ) )2  N b


• PM / (max m(t ) ) : The average-to-peak-power-ratio of the message

2

signal (or equivalently, the power content of the normalized

message, PMn ).  2 S
  p Mn 
P  PM
 
S   b
N
  =
 N O 3 2 P  S  FM
 f M n  N b
62
 Effect of Noise on AgM
• Now using Carson's rule Bc = 2( + 1)W, we can express the output
SNR in terms of the bandwidth expansion factor, which is defined as
the ratio of the channel bandwidth to the message bandwidth and is
denoted by  :
= 2( + 1)
Bc
 def
W
• From this relationship, we have  =/2 – 1.
• Therefore,
  p = k p max m(t )  PM

 k f max m(t ) 

 f = FM    −1 2 S

  
W
 PM  2 PM
 PM  p2 S S   max m(t )   N b
 2     =
PM
 (max m(t ) )  N b
2
S  O 
N  2 −1   S 

  =
 N O  PM  f2 S 3PM  max m(t )   N  FM
3   FM     b
 (max m(t ) )2  N b
63

 Effect of Noise on AgM
  p  PM
2
   −1 2 S
 PR    PM  2   
  max m(t )   N 0W
 PM PM
S S   max m(t )   N b
  = 2   =
 N O   f  PM  N O 
2
 2 − 1   S 
3PR  max m(t )   N W FM 3PM  max m(t )   N  FM
   0     b

• Observations
– In both PM and FM, the output SNR is proportional to 2.
Therefore, increasing  increases the output SNR.
– Increasing  increase the bandwidth (from Carson’s rule).
So angle modulation provides a way to trade off
bandwidth for transmitted power.

64
 Effect of Noise on AgM
– Although we can increase the output SNR by increasing ,
having a large  means having a large Bc (by Carson's rule).
– Having a large Bc means having a large noise power at the
input of the demodulator. This means that the approximation
P(Vn(t) << Ac)  1 will no longer apply and that the preceding
analysis will not hold.
– In fact, if we increase  such that the preceding
approximation does not hold, a phenomenon known as the
threshold effect will occur and the signal will be lost in the
noise.
– This means that although increasing the modulation index,
, up to a certain value improves the performance of the
system, this cannot continue indefinitely.
– After a certain point, increasing  will be harmful and
deteriorates the performance of the system.
65
 Effect of Noise on AgM
– A comparison of the preceding result with the SNR in AM
shows that, in both cases (AM and angle modulation),
increasing the transmitter power (and consequently the
received power) will increase the output SNR
– But the mechanisms are totally different. In AM, any
increase in the received power directly increases the
signal power at the output of the demodulator.
– This is basically because the message is in the amplitude of
the transmitted signal and an increase in the transmitted
power directly affects the demodulated signal power.
– However, in angle modulation, the message is in the phase
of the modulated signal and increasing the transmitter power
does not increase the demodulated message power.
– In angle modulation, the output SNR is increased by a
decrease in the received noise power.

66
 Effect of Noise on AgM
– In FM, the effect of noise is higher at higher
frequencies. This means that signal components at higher
frequencies will suffer more from noise than signal
components at lower frequencies.
– To compensate for this effect, preemphasis and
deemphasis filtering are used.

67