Noise in Analog Communication Systems

 Noise is unwanted signal that affects wanted signal

 Noise is random signal that exists in communication systems
 Noise
 Internal
 External
Internal:
 It is due to random movement of electrons in electronic circuits
 Major sources are resistors, diodes, transistors etc.
 Thermal noise or Johnson noise and shot noise are examples.
External:
 Man- made and natural resources
 Sources over which we have no control
 Examples are Motors, generators, atmospheric sources.
Noise level in system is proportional to
 Temperature and bandwidth
 Amount of current
 Gain of circuit
  Resistance of circuit
Effect of noise
 Degrades system performance (Analog and digital)
 Receiver cannot distinguish signal from noise
 Efficiency of communication system reduces

Types of noise
 Thermal noise/white noise/Johnson noise or fluctuation noise
 Shot noise
 Noise temperature
 Quantization noise

Thermal Noise: This noise is generated due to thermal motion (Brownian motion) of electrons
inside resistor. This noise is zero at absolute zero degree Kelvin and generated when temperature
rises, also called thermal noise. Also called Johnson noise who invented it.
Thermal noise also referred as ‘White noise’ since it has uniform spectral density across the EM
Spectrum.
PSD of thermal noise 𝑆𝑛 (𝑓) is

𝑘𝑇
𝑆𝑛 (𝑓) =
2
( where k is Boltzman’s constant and T is temperature)
kT is denoted by N0
Then
𝑁0
𝑆𝑛 (𝑓) =
2
Work of Johnson and Nyquist gave the expression for noise power

𝑃𝑛 = 𝑣̅𝑛2 = 4𝑘𝑇𝐵𝑅 𝑣𝑜𝑙𝑡 2

Where,
k = Boltzman constant
T = Absolute temp. (Kelvin)
B = Bandwidth (Hz)
R = Resistance (ohms)
Where 𝑣̅𝑛 is mean noise voltage.
SHOT NOISE
It is electronic noise that occurs when there are finite number of particles that carry energy such as
electrons or photons.
Due to analogy of lead shots called shot noise.
It has uniform spectral density like thermal noise.
Determination of Noise level
Signal to noise level ratio (SNR)
 𝑆 𝑃𝑜𝑤𝑒𝑟 𝑜𝑓 𝑢𝑠𝑒𝑓𝑢𝑙 𝑠𝑖𝑔𝑛𝑎𝑙 𝑣𝑜𝑙𝑡𝑎𝑔𝑒
=
𝑁 𝑃𝑜𝑤𝑒𝑟 𝑜𝑓 𝑛𝑜𝑖𝑠𝑒 𝑣𝑜𝑙𝑡𝑎𝑔𝑒

𝑆𝑖
𝑆𝑖𝑔𝑛𝑎𝑙 𝑡𝑜 𝑛𝑜𝑖𝑠𝑒 𝑟𝑎𝑡𝑖𝑜 𝑎𝑡 𝑖𝑛𝑝𝑢𝑡 ⁄𝑁
𝑖
𝑁𝑜𝑖𝑠𝑒 𝐹𝑖𝑔𝑢𝑟𝑒 = =
𝑆𝑖𝑔𝑛𝑎𝑙 𝑡𝑜 𝑛𝑜𝑖𝑠𝑒 𝑟𝑎𝑡𝑖𝑜 𝑎𝑡 𝑜𝑢𝑡𝑝𝑢𝑡 𝑆0⁄
𝑁0
Noise figure is always > 1
Noise temperature
Equivalent noise temperature is not the physical temperature of amplifier, but a theoretical construct,
that is an equivalent temperature that produces that amount of noise power
𝑇𝑒 = 𝑇(𝐹 − 1)
Noise figure of cascaded stages
FRIIS formula for calculating total noise factor of several cascaded amplifiers
𝐹2 − 1 𝐹3 − 1
𝐹 = 𝐹1 + + + −−−−−
𝐺1 𝐺1 𝐺2
𝐹1 , 𝐹2 , − − & 𝐺1 , 𝐺2 , 𝑒𝑡𝑐 𝑎𝑟𝑒 Noise figure and gains of different stages in cascade.
Note that noise figure is mainly dominated by first two stages.
Effect of Noise on AM Systems:
The channel introduces additive noise in message and thus message received becomes corrupted.
Figure of merit is defined
𝑆0
⁄𝑁
0
𝛾=
𝑆𝑖
⁄𝑁
𝑖

Modulation system with higher γ has better noise performance.

Total noise = Noise power spectral density × Bandwidth

𝑁0
𝑁= ×𝑊
2
Noise amplitude is having Gaussian distribution i.e. additive white Gaussian Noise (AWGN)
For AM-SC For both DSB and SSB γ = 1
Thus S/N ratios at input and output of detector are identical i.e. no improvement is S/N ratio
For conventional AM using envelope detector

𝛾 = 1⁄3 𝑓𝑜𝑟 𝜇 = 1

The maximum values of γ is 1/3 for modulation index of unity

Noise in Angle Modulated Systems
Like AM, noise performance of angle modulated systems is characterized by parameter γ
3 2
𝛾𝐹𝑀 = 𝛽
2
If it is compared with AM
𝛾𝐹𝑀 1 𝜔𝐹𝑀 2
= ( )
𝛾𝐴𝑀 2 𝜔𝐴𝑀
𝛾𝐹𝑀
Note if bandwidth ratio is increased by a factor 2, then ⁄𝛾𝐴𝑀 increases by a factor 4

This exchange of bandwidth and noise performance is an important feature of FM.

FM Threshold Effect
 In FM Systems, where signal level is well above noise ratio and demodulated signal to signal
noise ratio are related by
𝑆⁄ = 3. 𝛽 2 . 𝐶⁄
𝑁 𝑁
S/N – Signal to noise ratio at output of demodulator
β – Modulation index
 C/N – Carrier to noise ratio at input of demodulator

 Does not apply when S/N decreases below certain point called Threshold

FM system is more susceptible to noise threshold as compared to AM.

Threshold improvement can be obtained by Pre-emphasis and De-emphasis

  Pre-emphasis: Improving signal to noise ratio by increasing the magnitude of high frequency
signal with respect to lower frequency signals.
 A simple high pass filter can serve as transmitter’s pre-emphasis circuit
 De-emphasis: Improving signal to noise ratio by decreasing the magnitude of high frequency
signals with respect to lower frequency signals
 A simple low pass filter can operate as de-emphasis circuit in receiver.
The combined effect of these circuit is to increase signal to noise ratio for high frequency
components during transmission, so that they are not masked by noise
Analog Communications Unit 5

Noise in analog communication system model

Accordingly, for the performance analysis of different modulation types in the

presence of noise, we will be limiting the transmitter to the modulator only and the
receiver to a filter followed by the demodulator. To accommodate the channel noise, the
new communication system model is shown in Fig 1
Let ST be the transmitted signal power. We assume that the noise corrupting the
signal in the channel is additive and, unless otherwise specified, is white.

Fig 1 communication system model

We assume that the channel is allpass and of unity gain so that the input signal to
the bandpass filter at the receiver is the transmitted (modulated) signal plus channel
noise. The receiver bandpass filter is also assumed to be of unity gain and having a pass
band corresponding to the frequency band of the received modulated signal. Hence, the
modulated signal passes through the filter unchanged while the channel noise gets
converted to bandpass noise.
Let us denote
Si = useful signal power at the input of the demodulator
So = useful signal power at the output of the demodulator
Ni = noise power at the input of the demodulator and
N0 = noise power at the output of the demodulator

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

In analog communication systems, the quality of the received signal is determined

So
by the output SNR, for a given transmitted power, ST. However, since Si is more
No
accessible, and also given the fact that it is proportional to ST, it is common to determine
So S S
for a given Si. A useful of merit is the difference (in dB) between o and i
No No Ni

The simplest communication system one can think of is the baseband system.
This serves as the basis for comparing the performance of different systems. In baseband
communication, no modulation is involved. Both the transmitter and the receiver are
simple ideal lowpass filters with bandwidth fm. Again, the channel is assumed to be
distortionless and having unity gain. The baseband system is shown in Fig 2

Fig 2 The baseband system

Since channel gain is unity, we have: Si = ST

To find the output noise power No, we observe that white noise from the channel passes
through an ideal lowpass filter (receiver) to yield an output noise with a lowpass PSD as
given in Fig 3

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Fig 3 lowpass PSD

The output noise power, No, is simply the area under the PSD. Hence,

fM fM
n
NO = ∫
− fM
S no ( f )df = 2 ∫
0
2
df = nf m

Therefore,

So S
= i =γ
N o nf m

Si
The parameter γ = denotes the baseband signal-t-noise ratio.In this system, the out-
nf m
of-band noise has been removed by filtering. The result will serve as the standard for
comparing the performance of the other systems

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Noise in DSBSC

A DSB-SC receiver can be modeled as a bandpass filter followed by a coherent

(synchronous) detector as shown in Fig 1. Assume that the bandpass filter is ideal with
unity gain and that its passband corresponds to the modulated signal frequency range.

Fig.1 coherent (synchronous) detector

The (modulated ) signal plus white noise enters the receiver input. The signal is
unchanged upon filtering whereas the white noise nw(t), gets converted to bandpass noise
n(t). The output of the bandpass filter is the input to the demodulator yi(t), and is given
by;

yi(t) = Acm(t)cosωct + n(t) (1)

Clearly, from Eq (1), the input signal component is Si(t) = Acm(t)cosωct. Hence,
the signal power at the demodulator input is given by the mean square value of Si(t);

1 2 2
S i = {s i (t )} = [ Ac m(t ) cos ω c t ] =
2 2
Ac m (t ) (2)
2

Note that, the factor is due to the mean square value of the cosine term.

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

From Eq (1), the demodulator input noise power Ni, is the mean square value of the
bandpass noise n(t). Since the pass band of the bandpass filter is from ωc - ωm to ωc + ωm,
the PSD of n(t) is the same as in Fig .2

Fig.2 PSD of bandpass white noise

Hence;

fc + fm fc + fm
η
N i = n 2 (t ) = 2 ∫ S n ( f )df = 2
fc − fm
∫
fc − fm
2
df = 2ηf m (3)

To find the output signal power So, and output noise power No, we have to obtain
the expression for the output of the lowpass filter, y0(t). Clearly,

Yo(t) = LPF {yi(t) Χ cosωct} (4)

Where LPF has been used to denote the lowpass filtering operation
For yi(t), we use Eq (1) and write the bandpass noise n(t) in terms of its
quadrature representation. Thus, Eq (4) becomes:

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

y o (t ) = LPF {[ Ac m(t ) cos ω c t + n c (t ) cos ω c t − n s (t ) sin ω c t ] × [cos ω c t ]}

{[
y o (t ) = LPF Ac m(t ) cos 2 ω c t + nc (t ) cos 2 ω c t − n s (t ) sin ω c t cos ω c t ]}
A A n (t ) n c (t ) n (t ) 
y o (t ) = LPF  c m(t ) + c m(t ) cos 2ω c t + c + cos 2ω c t − c sin 2ω c t 
 2 2 2 2 2 

The output of the low pass filter is

Ac n (t )
y o (t ) = m(t ) + c (5)
2 2
Ac
From Eq (5), the useful output signal component is S o (t ) = m(t ) and the noise
2
n c (t )
component is no (t ) = . We observe that the noise at the detector output involves the
2
in-phase component only.
Hence, the output signal power is:

2
A   A2 
S i = s (t ) =  c m(t ) =  c m 2 (t )
2
o
2   4  (6)

2
n  1 1
N o = n (t ) =  c  = n c2 (t ) = n 2 (t )
2
o (7)
2 4 4

Where we have used Eq.(8)

n(t)= nc (t ) cos ωct − ns (t ) sin ωc t (8)

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

1
From Eqs 2 and (6) we see that S o (t ) = S i . Similarly, from Eqs (3) and (7) we
2
1 ηf
have N o =   N i = m . Hence,
4 2

 So  s
  = I =γ (9)
 No  DBS − SC ηf m

So S
= i =γ (10)
N o nf m

Comparing Eqs (9) and (10) we see that, for a given demodulator input signal power and
identical noise conditions in the channel, the noise performance of a DSB-SC system is
identical to that of the reference baseband system.

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Noise in SSB

To obtain the noise performance of an SSB-SC system, we have to first draw the
model for the SSB receiver. We recall that an SSB-SC signal is demodulated similar to
DSB-SC, by using a coherent (synchronous) detector. Hence, the model of an SSB
receiver in the presence of additive channel noise will be the same as that for the DSB-SC
receiver with the difference that the input bandpass filter will now have a bandwidth of
ωm corresponding to the single sideband and not 2ωm as in the case of DSB-SC. Hence,
the passband of the receiver input bandpass filter for SSB-SC demodulation will be ωc –
ωm to ωc for LSB-SSC and ωc to ωc + ωm USB-SSC. Fig 1 shows the SSB-Sc receiver for
noise calculation.

Fig-1 SSB-SC demodulation

The input to the SSB-SC receiver is the SSB-SC signal plus additive white noise. The
signal passes through the (ideal) unity gain bandpass filter unchanged whereas the white
noise, upon filtering, yields bandpass noise. Hence, the input to the demodulator is :

y i (t ) = s i (t ) + ni (t ) = AC [ m(t ) cos ω c t ± m(t ) sin ω c t ] + n(t ) (1)

The input signal power is the mean square value of

s i (t ) = AC [m(t ) cos ω c t ± m(t ) sin ω c t ] and is given by:

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

s i = {AC [ m(t ) cos ω c t ± m(t ) sin ω c t ]}

2

This expression can be simplified by recalling that a function and its Hilbert
transform are orthogonal. If m(t) = 0 then m(t) m(t) = 0 .Thus,

 A2 
[
S i =  c  m 2 (t) + m 2 (t) ]
 2 
Another useful property is that a function and its Hilbert transform have equal

power (verify). Thus is, m 2 (t) = m 2 (t) . Hence, the demodulator input signal power is
given by,

S i = Ac2 m 2 (t) (2)

In order to compute the demodulator input noise power, we need to find out the
area under the bandpass noise PSD. For SSB-SC lower sideband (LSB) transmission, the
bandpass noise PSD is shown in Fig 2.

Fig-2 PSD of bandpass white noise at SSB demodulator input

The demodulator input noise power:

Analog Communications Unit 5

N i = n 2 (t ) = Area under the PSD

fc
η (3)
=2 ∫
fc − fm
2
df = ηf m

In order to find out the output signal and noise powers, we have to obtain the
expression for the demodulator output, y0(t) . From Fig 1

y 0 (t ) = LPF {y i (t ) × cos ω c t }
= LPF {([Ac m(t ) cos ω c t ± Ac m(t ) cos ω c t ] + n(t )) × cos ω c t}

By using the quadrature representation for n(t), expanding and then eliminating
the terms rejected by the lowpass filter, we obtain :

A  n (t )
y 0 (t ) = s 0 (t ) + n 0 (t ) =  c m(t ) + c (3)
 2  2

Hence, the demodulator output signal power is the mean square value of
A 
s 0 (t ) =  c m(t ) and is given by:
 2 

2
 A   A2c 1
s 0 =  c m(t ) + m(t ) = S i (4)
 2   4 4

The output noise power is given by

2
 n (t )  1 1 1
s 0 = n (t ) =  0  + n c2 (t ) = n c2 (t ) = ηf m
2
0 (5)
 2  4 4 4

From Eq 4 and 5, we obtain the output SNR for SSB-SC as:

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

 S0  S
  = i =γ (6)
 N0  SSB− SC ηf m

Hence, we see that, for the same demodulator input signal power, the noise
performance of SSB-Sc system as well as DSB-SC system are identical and equal to that
of the reference baseband system.

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Synchronous Detection

The model of the AM receiver using synchronous detection is the same as that
used for the detection of DSB-SC signal as shown in Fig 1

Fig 1 AM receiver using synchronous detection

The input to the receiver is the conventional AM signal plus noise. yi(t), the output of the
bandpass filter is the signal plus bandpass noise:

y i (t ) = s i (t ) + ni (t ) = Ac [1 + k a m(t )]cos ω c t + n(t ) (1)

Hence, the demodulator input signal power:

S i = s i (t ) = {Ac [[1 + k a m(t )]cos ω c t + n(t )]} = Ac [1 + k a m(t )]

2 2 2 2

A2   2 (2)
=  c 
] 2 
[
 1 + k a2 m(t ) + 2k a m(t ) =  Ac  1 + k a2 m2 (t )
[ ]
 2   

Since m(t) is assumed to have zero mean.

The noise power at the demodulator input:

N i = n 2 (t ) = Area under thePSD from ω c − ω m to ω c + ω m

η  (3)
= 2 (2 f m ) = 2ηf m
2

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

In order to find out the output signal and noise powers, we have to obtain the
expression for the demodulator output y0(t). From Fig 1

y 0 (t ) = LPF {y i (t ) × cos ω c t}
= LPF {( Ac [1 + k a m(t )]cos ω c t + nc (t ) cos ω c t − n s (t ) sin ω c t ) × cos ω c t }

By Expanding and discarding the terms rejected by the lowpass filter, we obtain:

A  A  n (t )
y 0 (t ) =  c  +  c  k a m(t ) + c (4)
 2   2  2

The first term on the right-hand side of Eq 5 is a DC term which can be removed by using
A 
a blocking capacitor. The desired signal component in y0(t) is s 0 (t ) =  c k a m(t ) and the
 2 
nc (t )
noise component in the demodulator output is n0 (t ) = . Therefore, the output signal
2
and noise power are:

2
 A   A 
S 0 =  c  k a m(t ) =  c k 2 a m 2 (t )
 2    2 
(5)
2
 n (t )  n c (t ) n (t ) N i
2 2
S0 =  c  = = =
 2  4 4 4

Hence, the output signal-to-noise power ratio for a conventional AM system using
synchronous detection is:

S o  Ac2 2 2   N i 
=  k a m (t ) /   (6)
N0  4   4 

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Eq (6) can be expressed in terms of the input signal power, Si, as:

So k 2 m 2 (t )  S i  k 2 m 2 (t )
= a   = a ×γ (7)
N 0 1 + k a2 m 2 (t )  ηf m  1 + k a2 m 2 (t )

 k 2 m 2 (t ) 
The ratio  a  is called the AM transmission efficiency . It is the ratio of the
1 + k 2
a m 2
(t ) 
sideband power to the total power in the transmitted signal. Since this ratio is less than
unity, we see from Eq (7) that the SNR in conventional AM using synchronous detection
is always smaller than γ which is the output SNR in the case of the baseband ( and DSB-
SC and SSB-SC) systems.
In order to get a feel for the above result, let us consider an AM system operating

with a ka of 0.8 and having a normalized message power m 2 (t ) = 0.1 which is the typical

value for speech signals. Substituting the values of ka and m 2 (t ) in Eq 7, we obtain

So 0.8 2 (0.1)
= γ = 0.06γ
N 0 1 + 0.8 2 (0.1)

101og10(0.06)=-12dB. Hence, this system is about 12dB inferior to the ideal system
requiring the same bandwidth.

Envelope Detection:-

We recall that what kept conventional AM going inspite of its poor noise
performance is its simple receiver advantage ( envelope detector). The noise calculation
model of an AM receiver using envelope detection is shown in Fig 2
Analog Communications Unit 5

Fig 2 AM receiver using Envelope detector

The only difference between this model and the earlier one using synchronous
detection is in the demodulator. Hence, the expressions for yi(t), Si, and Ni are the same
as before and are given below for convenience:

y i (t ) = Ac [1 + k a m(t )]cos ω c t + n(t )

 A2 
[
S i =  c  1 + k a m 2 (t )
2
]
 2 
N i = 2ηf m

And
To obtain the signal and noise powers at the demodulator output, we obtain the
envelope of yi(t) as follows:

y i (t ) = Ac [1 + k a m(t )]cos ω c t + n(t )

= Ac [1 + k a m(t )]cos ω c t + nc (t ) cos ω c t − n s (t ) sin ω c t (8)
= {Ac [1 + k a m(t )]cos ω c t + n(t )}cos ω c t − n s (t )sin ω c t

Equation (8) is of the form A(t)cosθ(t) + B(t)sinθ(t), where

A(t ) = Ac [1 + k a m(t )]cos ω c t + n(t ) , B(t) = -ns(t) and θ(t) = ωct. Hence yi(t) can be written
in terms of the envelope and phase angle as yi(t) = E(t)cos[ωct. + θ(t)], where the

2 2  B(t ) 
envelope E (t ) = A(t ) + B(t ) and the phase angle is φ (t ) = tan −1   . The envelope
 A(t ) 
detector is insensitive to the carrier phase and detects only the envelope of the input
signal. Hence, the ideal envelope detector output is:

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

E (t ) = {Ac [1 + k a m(t )] + n(t )}2 + {− ns (t )}2

(9)
Ac [1 + k a m(t )] + n (t ) + 2 Ac [1 + k a m(t )]nc (t ) + n s (t )
2 2 2
= c

In order to interpret Eq 9, let us consider two kinds of noise situation: low noise
and high noise. For low noise in the demodulator input yi(t), the signal component si(t), is
large compared to the noise component ni(t). That is Ac [1 + k a m(t )] >> ns (t ) . For high

noise, the noise component in yi(t) is large compared to the signal component. That is;
Ac [1 + k a m(t )] << nc (t ) , ns (t ) . Let us consider the two cases separately.

Low Noise (High SNR):-

For Ac [1 + k a m(t )] >> nc (t ) , ns (t ) , Eq (9) reduces to:

Ac [1 + k a m(t )] + nc2 (t ) + 2 Ac [1 + k a m(t )]nc (t )

2
E (t ) =
2nc (t ) (10)
= Ac [1 + k a m(t )] 1 +
Ac [1 + k a m(t )]

 x
Since 1 = x = 1 +  for x << 1 Eq (10) reduces to:
 2

 nc (t ) 
E (t ) = Ac [1 + k a m(t )]1 + 
 Ac [1 + k a m(t )] (11)
= Ac + Ac k a m(t ) + nc (t )

The output of the envelope detector is given by Eq (11). The first term Ac on the right-
hand side is a DC term which can be blocked by a capacitor. The second term Ac k a m(t )
is the useful signal component in the demodulator output and the third term nc(t)
represents the noise component. Hence, the output signal and noise powers are given by

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

S 0 = [ Ac k a m(t )] = Ac2 k a m(t ) and

2 2

(12,13)
N 0 = nc2 (t ) = n 2 (t ) = 2ηf m

Therefore, the output signal-to-noise power ratio is given by

2
So Ac2 k a m(t )
= (14)
N0 2ηf m

The right-hand side of Eq (!4) can be expressed in terms of the input signal power

 A2 
[
S i =  c  1 + k a m 2 (t )
2
]
as follows:
 2 

k m 2 (t )  S i 
2
So
= a 2  
N 0 1 + k a m(t )  ηf m 
(15)
2
k a m 2 (t )
= 2
γ
1 + k a m(t )

This result is identical to Eq (8) obtained for synchronous detection of AM.

Hence, for low noise, the performance of the envelope detector is identical to that of the
synchronous detector.

High-noise (Low SNR):-

For Ac [1 + k a m(t )] << nc (t ), ns (t ) , Eq (11) reduces to:

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

E (t ) = nc2 (t ) + 2 Ac [1 + k a m(t )]nc (t ) + ns (t )

2

2 A [1 + k a m(t )] (16)
=  nc (t ) + n s (t )  × 1 + 2c
2 2

  2
nc (t ) + n s (t )

 x
Using the approximation 1 = x = 1 +  for x << 1 in the second square-
 2
root on the right-hand side, Eq (!6) reduces to:

 A [1 + k a m(t )]
E (t ) =  nc (t ) + ns (t )  × 1 + c2
2 2

   nc (t ) + n s 2 (t ) 

Clearly, at the demodulator output, the signal and noise components are no longer
additive. Since the signal component is multiplied by noise, it can no longer be
recovered. The phenomenon of distortion of the message signal at low input SNR is
called the threshold effect.

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

NOISE IN PHASE MODULATED SYSTEMS

If an FM demodulator is used to receive a PM signal, it yields the output

dm(t )
. By passing this output through an ideal integrator, we obtain the desired signal
dt
m(t) as shown in Fig 1.

Fig 1 Generation of Phase modulated signal using Frequency modulator

Fig 2.PM demodulator using FM receiver

Recall the expression for the PM signal

[ ]
s PM (t ) = Ac cos ω c + k p m(t ) (1)

Hence, the normalized input signal power,

Ac2
Si = (2)
2

The input noise power is the noise power within a bandwidth of and is given by

η 
N i = 2 (2∆f ) = 2η∆f (3)
2

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

The instantaneous frequency of the PM signal is:

d dm(t )
ω i (t ) =
dt
[ ]
ω i (t ) + k p m(t ) = ω i + k p
dt
(4)
Recall the FM receiver output is proportional to the instantaneous frequency.
dm(t )
Hence, the FM receiver output = αω c + αk p
dt

Note that the first term above is a DC component. Hence, the signal component at
dm(t )
the output of the FM receiver is αk p . This component is integrated ( see Fig 2) to
dt
obtain the FM receiver output s 0 (t ) = αk p m(t ) . Hence, the PM receiver output signal
power is:
2
s 0 (t ) = α 2 k p m 2 (t ) (5)

 α  dn (t )
Recall from Eq nd(t)= −   s that, for the small-noise case, the noise at the output
 Ac  dt
 α  dn (t )
of the FM receiver is: −   s
 Ac  dt

Upon integration, this gives the noise at the output of the PM receiver

α 
n0 (t ) = −  n s (t ) (6)
 Ac 

Hence, the PSD of this output noise is given by:

α 
S n0 (ω ) = −  S n0 (ω )
 Ac 
 α 2
[ ]
S n0 (ω ) =  2 S n0 (ω − ω C ) + S n0 (ω + ω C ) for ω < ω m
 Ac
0, otherwise

Assuming white noise (PSD), we obtain

Analog Communications Unit 5

 α 2η
S n0 (ω ) =  2 for ω < ω m
 Ac (7)
0, otherwise

Notice from Eq (7) that the output noise for PM has a uniform PSD. This is in
contrast to FM where the output noise PSD has a parabolic spectrum. Hence, there is no
need for pre-emphasis and de-emphasis in PM systems.
The output noise power is obtained as the area under the output noise PSD:

fm
α 2η 2α 2ηf m
N0 = 2 ∫ 2
df = (8)
0 Ac 2ηf m

Using Eqs (5) and (8), the output SNR for the PM receiver is given by:

 S0  k 2 A 2 m 2 (t )
  = P c (9)
 N0  PM 2ηf m

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

Pre-emphasis & De-emphasis

The noise triangle showed that noise has a greater effect on the higher modulating
frequencies than on the lower ones. Thus, if the higher frequencies were artificially
boosted at the transmitter and correspondingly cut at the receiver, improvement in noise
immunity could be excepted. This boosting of the higher modulating frequencies, in
accordance with a prearranged curve, is termed pre-emphasis, and the compensation at
the receiver is called de-emphasis. An example of a circuit used for each function is
shown in Fig 1 a)Pre-emphasis b)De-emphasis circuit.

If two modulating signals have the same initial amplitude, and one of them is pre-
emphasized to (say) twice this amplitude, whereas the other is unaffected (being at a
much lower frequency), then the receiver will naturally have to de-emphasize the first

Fig 1 a)Pre-emphasis Fig 1 b)De-emphasis circuit

signal by a factor of 2, to ensure that both signals have the same amplitude in the output
of the receiver. Before demodulation i.e, while susceptible to noise interference the
emphasized signal had twice the deviation it would have had without pre-emphasis, and
was thus more immune to noise. Alternatively, it is seen that when this signal is de-

G. Kiran Kumar, Assistant Professor, ECE@NITs

Analog Communications Unit 5

emphasized any noise sideband voltages are de-emphasized with it, and therefore have a
correspondingly lower amplitude than they would have had without emphasis; again,
their effect on the output is reduced.
The amount of pre-emphasis in U.S. FM broadcasting, and in the sound
transmissions accompanying television, has been standardized at 75 µs, whereas a
number of other services, notably CCIR and Australian TV sound transmission, use 50
µs. The usage of microseconds for defining emphasis is standard. A 75-µs de-emphasis
corresponds to a frequency response curve that is 3dB down at the frequency whose time
constant RC is 75 µs. This frequency is given by f = 1 2πRC and is therefore 2120 Hz;
with 50-µs de-emphasis it would have been 3180 Hz. Fig 2 shows pre-emphasis and de-
emphasis curves for a 75-µs emphasis, an used in the United States.
It is a little more difficult to estimate the benefits of emphasis than it is to evaluate
the other FM advantages, but subjective BBC tests with 50 µs give a figure of about 4.5
dB; American tests have shown an even higher figure with 75 µs. However, there is a
danger that must be considered; the higher modulating frequencies must not be
overemphasized. The curves of Fig 2 show that a 15-KHz signal is pre-emphasized by
about 17 dB; with 50 µs this figure would have been 12.6 dB. IT must be made certain
that when such boosting is applied, the resulting signal cannot overmodulate the carrier
by exceeding the 75-KHZ deviation, or distortion will be introduced. It is thus seen that a
limit for pre-emphasis exits, and any practical value used is always a compromise
between protection for high modulating frequencies on the one hand, and the risk of
overmodulation on the other.
If emphasis is applied to amplitude modulation, some improvement will also
result, but it is not as great as in FM because the highest modulating frequencies in AM
are no more affected by noise than any others. A part from that, it would be difficult to
introduce pre-emphasis and de-emphasis in existing AM services since extensive
modification would be needed, particularly in view of the huge number of receivers in
use.

G. Kiran Kumar, Assistant Professor, ECE@NITs