USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in Telecommunications

2. Additive White Gaussian Noise (AWGN)

Chapter 4: Noise in Telecommunications
2.1. Definition

An additive white Gaussian noise (AWGN) is

a noise signal with the following characteristics:
1. Definition and sources of noise
A) Additive: it adds to the useful signal.
1.1. Definition of noise B) White: its spectral power density is constant for all frequencies.
Noise is the collection of unwanted random signals that are added to the signal. of bandwidth (Figure 1). The term 'White' draws an analogy with light.
useful during its transmission. These signals are caused by random movements white that mixes all light frequencies.
charges in the circuits.

1.2. Sources of noise

It is possible to classify sources of noise in the following way:

A) Sounds of natural origin:
The terrestrial thermal noise. Figure 1: The power spectral density of white noise.
Atmospheric parasites (lightning).
C) Gaussian: A Gaussian noise is a random quantity with unpredictable amplitude,
Sound of solar and cosmic radiation. but follows a Gaussian distribution (Figure 2).
B) Artificially created noises related to human and industrial activity, for example
disturbances generated by devices and machines.
C) Interference signals from other telecommunications systems: The transmitters
television, radio, mobile telephony, WIFI, Bluetooth...
D) Noise originating from within the telecommunications system, created by the circuits
electronics of the transmitter and receiver. We mention:
Thermal noise that arises from the random motion of electrons Figure 2: The amplitude of a noise as a function of time and its Gaussian probability density.
free in a dissipative environment under the effect of temperature.
2.2. Band-limited noise
Quantum noise (or grain noise) caused by fluctuations of
Let it be an additive white Gaussian noise with a power spectral density N0/2
charge carriers passing through a potential barrier. For example
which goes through a bandpass filter with a bandwidth B between1etf2.
in semiconductor junctions.
Scintillation noise: originates from existing defects in the crystal lattice. AWGN noise Band-limited noise
Filter
of a semiconductor, it varies in 1/f.
Figure 3: filtering of additive Gaussian white noise.

S. Labandji – [Link] 1 2 See: 12/31/19

USTHB / L3 Telecom B / Course on analog communications Chapter 4: Noise in Telecommunications

The filter only allows the power contained in the band [f1, f2At the exit, 3. Thermal noise
we recover a signal whose spectral distribution is shown in figure 4.
3.1. Tension and power of thermal noise

Let there be a resistance R at the temperature T (Kelvin). The electrons in this resistance
are subjected to thermal agitation (random motion) which produces a low
random Gaussian voltage at the terminals of the resistor.
The effective value denoted Vn is given by: V = 4kTB/R √
Figure 4 :The narrowband thermal noise power spectral density. With:
The noise power becomes: (SNis a DSP) k= 1.3810-23JK-1it is the Boltzmann constant.
T: conductor temperature in (°K).
2 2 B: bandwidth of noise in (Hz).
At the output of the filter, the noise is no longer white, we say it is banded noise. R: the value of the resistance in (Ω).
limited pass and limited power. The noisy resistance can be modeled by a Thévenin equivalent circuit.
The temporal expression of the narrowband noise 'B' centered aroundf0, is given which contains a voltage source Vn and a resistance R without noise.
by:
With:
is a low-frequency random process.
se calcule à partir de by applying a phase shift of 90° to this
last (Hilbert transform).
It is noted that: and have the same power spectral density:
Figure 6: A resistor R generating a random voltage v(t) at a temperature of T°K, and its
modeling by a Thévenin generator.

It is noted that: and contains the same amount of power contained in A resistance R' is connected as a load across R. To obtain the
n(t) : maximum possible power in R', it must be that R' = R.
So the maximum thermal noise power provided by a resistor R at
the temperature Test given by:

4R
With:
N: noise power [W].

Figure 5 :The spectral power density of and . It is the spectral power density [W/Hz].

S. Labandji – [Link] 3 4 See: 31/12/19

USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in Telecommunications

According to the expression of thermal noise power, it is possible to say that: The noise factor of a noisy quadripole is greater than 1. A very quadripole
A system with a narrower bandwidth collects power from noisy has a higher noise factor. If F=1, the quadrupole is non-noisy.
quieter noise.
3.4. Noise factor of a cascaded four-port network
A system that operates at a lower temperature generates less
Let quadripoles Q arranged in cascade. For the quadripole Qi, we denote the factor
noise.
of noise by F and the gain by Gi.
The spectral power density is constant (SX=0/2).
Thermal noise is an AWGN noise generated by electronic circuits.
the transmitter, the receiver, and in the channel (conductors and even for example through the ground Q1 Q2 Qn
or the atmosphere). It is the noise that is the most dominant in telecommunications.
Figure 8: Cascading of 'n' quadripoles.
3.2. Thermal noise in a quadriport So the noise figure of the system is written as:
Let there be a four-pole Q with a power gain G. If a signal is applied to it F 1F 1F 1 F 1
FF ⋯
Power At the input, we receive a signal with a power Ps = G · Pe at the output. G G G G G G G G G …G
But, if we apply a power noise at the input, we get at the output a The noise figure of the system approaches the noise figure of the first two-port network.
noise with power: Ns = G · NeꞋNp.
So logically we have an interest in placing a four-pole with the minimum possible
Nprepresents the thermal noise power produced by the quadripole (own noise of noise factor and the maximum possible gain as the first stage of a chain
at the quadripole).
cascading, thus its gain masks the noise brought by the following stages.
It is possible to consider Q as a non-noisy four-terminal network powered at the input.
by an equivalent noise source. This source is the equivalent noise resistance. 4. Signal to Noise Ratio
R at the equivalent temperature Teand delivers a power of kTB. So: N 4.1. Principle of signal transmission in the presence of noise

Ne=0 NS= Np Ne= kTeB Ns= GkTeB During the transmission of useful signal from the transmitter to the receiver through the

Quadrupole Quadrupole channel, it undergoes an attenuation from this channel. If we set PE the emitted power
R
noisy not noisy and RP the received power, then: P

Figure 7: Modeling of thermal noise for a quadrupole.

With: A is the attenuation in the channel in decibels, it depends on the distance
between the sender and the receiver.
3.3. Noise factor of a quadripole
Total noise
The noise factor of a quadripole is defined at T=290°K by:
PE PR
à N Emitter Transmission channel + Receiver
à GN Distance
In decibels : F 10 logF. Figure 9: Schematic diagram of an analog telecommunications system.

S. Labandji – [Link] 5 6 See: 31/12/19

USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in Telecommunications

Several noise signals combine with the signal: at the level of the transmitter, in 4.3. Improvement of the signal-to-noise ratio
the channel and in the receiver to form the total noise in relation to the system. The
There are several methods to improve the signal-to-noise ratio, including:
The transmission system can be schematized as shown in Figure 9.
Lower the temperature of the devices (transmitters).
At the receiver input, we retrieve a fraction of useful signal and noise that
Optimize the use of components that generate noise in circuits.
deform this signal. If it is very deformed, it becomes incomprehensible.
Blind the transmission equipment to reduce the effects of
external parasites.
Use a low-noise amplifier at the input of a receiver.
Limiting the noise power at reception also helps improve the SNR. For
For this, we use a bandpass filter with the same bandwidth as the received signal.

Signal received Useful signal

Input filter Demodulation

Figure 11: Use of a filter to limit noise power in a receiver.

At the output of the input filter, the noise is narrowband and thus of limited power.
Figure 10: Example of a sinusoidal signal with noise.
This filter is located at the input of a reception system, it is also called a channel filter.
4.2. Definition of signal-to-noise ratio (SNR)

The signal-to-noise ratio (SNR) is the ratio of

5. Evaluation of modulation performance in the presence of noise
power between the useful signal that contains the information and the noise. The useful signal can The calculation of SNR at the input and output of a given demodulator allows for the study
to be a baseband signal or modulated in AM, FM, or PM. the performances of the different modulations against noise and to know if the
demodulation improves or degrades the SNR. We use this information to
build telecommunications links.

5.1. Signal to noise ratio report for a baseband link

To compare the performances of the different modulations against noise, we

In dBm: SNR 10 Log
uses a reference system without modulation (baseband transmission). The
The SNR allows for the evaluation of the quality of a transmission. For a reception The receiver contains only a filter whose bandwidth (B) is equal to the bandwidth
comfortable, the SNR must exceed a certain value that varies from one system to of the modulating signal.
telecommunication to another.
Baseband signal Useful signal: Sbb
For example, from an SNR=25dB, an audio or even video signal becomes Band-pass filter B
AWGN noise Noise: Nbb
exploitable.
Figure 12: Baseband reception system.

S. Labandji – [Link] 7 8 See: 12/31/19

USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in Telecommunications

At the entrance of the system, we have the unmodulated useful signal and AWGN noise at Signal to noise ratio at the input:
unlimited bandwidth. At the output of the filter, we retrieve the signal with the same
power Sbb and a narrowband noise with power Nbb=2(N0/2)B. 4N
The signal-to-noise ratio in baseband (SNR) is defined bbbetween power Signal power at the output of the demodulator:
baseband signal Sbband the noise power Nbbat the exit by: At point B: ∙

2 1 1
S t A b t cos 2ω t 2φ
2 2
5.2. Effect of noise on AM modulation
The low-pass filter only allows the low-frequency component to pass. Therefore, at the
The figure below shows the block diagram of a synchronous AM receiver. output 's' from the receiver retrieves the information signal:
At the input of the receiver, we have an amplitude-modulated signal [Link] by a noise

AWGN noted nWwhich represents the total noise that can exist around a system. 2
This is an additive noise so the signal at the input is written as: SAM+nW. The signal power at the output of the demodulator:

Noise power at the output of the demodulator (NS):

SAM(t) + n(t)
SAM(t) + nW(t) At point B: ∙
Filter of Filter b(t)
e canal B low pass s
A
BP=2B L(t) BP=B
Figure 13: Block diagram of a synchronous AM demodulator.

The bandwidth of the channel filter is 2B centered on the AM signal carrier. Thus, cos2ω t + 2ϕ sin2ω t + 2φ I sin0
2 2 2 2
At point (A), we have a narrowband noise:
The low-pass filter eliminates the components at 2ωp.

At the exit, we retrieve:

The multiplier multiplies the signal (SAM(t)+n(t)) by a signal L(t) provided by
The power of this signal is written as:
a local oscillator: L t = cos(ω t) + φ.
Finally, the bandwidth of the low-pass filter is B. We know that the power of is equal to the power of n(t):

Signal power at the input of the demodulator: 1

∶ ∙ 2N
4
On a S t F A b t cos ω t F
Finally, the noise power at the output of the demodulator:
The power of this signal is written as: S . is the signal power b(t). Signal to noise ratio at the output:
Noise power at the input of the demodulator:
The noise power is: N L df F2 df F2· 2B 2N 2N

S. Labandji – [Link] 9 10 See: 12/31/19

USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in telecommunications

Signal to noise ratio in AM: Noise power at the input of the DSB demodulator:
The noise power at the input of the demodulator: 2N
2P
Signal-to-noise ratio at the input:

During AM demodulation, the signal-to-noise ratio at the output is always lower than the 4N
signal to noise ratio at the input. The demodulator degrades the SNR, this degradation is due Signal power at the output of the DSB demodulator:

to the power losses in the transmission of the carrier. ∙ ∙

Signal Report on Noise in AM / Baseband: 1 1
2 2
To be able to compare, we take the baseband signal power Sbbequal 2 2

to the power of modulated signal. At the output of the low-pass filter:

Hence, the output signal power of the demodulator:

2 2N Noise power at DSB demodulator output:

Like in AM: ∙ .

So:

Signal to noise ratio at the output:

Compared to a baseband reception, the SNR is degraded in AM.
Example of a sinusoidal modulating signal:
2N
For a signal: the power is:
Signal to noise ratio in AM DSB:
2m
1 2
2

The SNR between the output and the input of the modulator is equal to 2, therefore the demodulator
1 ❌
DSB improves the SNR.
5.3. Effect of noise on DSB AM modulation Signal to Noise Report in DSB / baseband:

We use the same synchronous AM demodulator (Figure 13) and with the same
1
filters, but with a DSB signal at the input.
Signal power at the input of DSB demodulator:
The SNR remains at the same level as that of a signal transmitted in baseband. The
On a :
DSB has no impact on the SNR, because the gain provided by the demodulator
The signal power at the input of the demodulator: is accompanied by doubled bandwidth.

S. Labandji – [Link] 11 12 See: 31/12/19

USTHB / L3 Telecom B / Course on Analog Communications Chapter 4: Noise in Telecommunications

5.4. Effect of noise on SSB AM modulation Signal to noise ratio at the output:

We use the synchronous demodulator from figure (13) but this time we take a
input filter with a bandwidth B (band occupied by an SSB signal) 4N

Useful signal power at the input of SSB demodulator: Signal to Noise Ratio Report in AM SSB:
SSB Power = ½ DSB Power.
L1 1
So:

Noise power at the input of SSB demodulator:

In SSB AM modulation, the SNR at the output of the demodulator is the same as the
The noise power n(t) at the input of the demodulator:N 2
The SNR at the input of the demodulator is the same as the SNR in baseband. The AM SSB
Signal to noise ratio at the entrance: has no impact on the SNR.

5.5. Effect of noise on angular modulation

B4N
Useful signal power at the output of the SSB demodulator: It is assumed that the FM/PM demodulator used is a discriminator.

Let's take the example of a LSB signal. The power spectral density of the noise for FM/PM modulation is given:
1 1
2 2
At the output of the multiplier (mixer):
∙ ⎩
With: APit is the amplitude of the carrier.
1
∙ The SNR is given at the output of the FM/PM demodulator:
2
3P
cos2ωt + 2φ = I sin2ω t + 2φ = A sin0
4 4 4 4 2N
The low-pass filter only lets through the term:
⎩ 2N
The power of the useful signal:
With: Pbis the power of the modulating signal.
Noise power at the output of the SSB demodulator: The ratio between the SNR is [Link] the baseband SNR:
As in AM: ∙ 3P

The noise power:

It is noted that the SNRs vary linearly with respect to [Link].
Like so: ∙ 2 Figure (14) shows the variations of depending on for

Noise power at the output of the demodulator: some values of m, for FM demodulation at discriminator.

S. Labandji – [Link] 13 14 See: 31/12/19

USTHB / L3 Telecom B / Analog Communications Course Chapter 4: Noise in Telecommunications

Figure 15: Spectrum of the message to be transmitted compared to the noise spectrum.

To solve this problem, a sharpening filter (high pass) is used which amplifies
the highest frequency components relative to the noise.
At reception, a de-accentuation filter (low-pass) is used to give the
signal its original state just after demodulation.

Figure 14: Variation of SNRS(dB) as a function of SNRbb(dB) for some values of m.

Here we note that the variation is linear only beyond a certain threshold which Figure 16: Preaccentuation and deaccentuation in an FM system.
depends on Mr. 5.6. Comparison of SNR between the different modulations
The use of a high modulation index in FM/PM improves the quality of
In amplitude modulation, the information signal is contained in the amplitude of
signal at the reception. But the gap between SNRS and SNRbbdoes not increase in a way
the carrier and as the noise is additive it adds to the signal it carries
unlimited with the modulation index. Because, if we increase the modulation index it
the information.
returns to widening the bandwidth of the FM signal, thus increasing the noise power,
In angular modulations, the information signal is contained in the phase of the
and ultimately deteriorate the signal-to-noise ratio.
therefore the additive noise cannot reach it. For this, the SNR in FM/PM is
By using plots similar to that of figure (14), it is possible to choose
better than in amplitude modulations, and the quality of the transmission is
the appropriate modulation index. For example for SNRbb15dB is better
best.
take m=2 than m=5.
Bibliography:
Pre-accentuation and de-accentuation
1. John G. Proakis, Masoud Salehi; Communication systems engineering, 2nd Ed.; Prentice-Hall,
Inc. 2002.
In FM, the noise power increases with the square of the frequency. Thus, for a signal
2. Louis E. Frenzel Jr ; Principles of Electronic Communication Systems, Fourth Edition;
spread over a band B, the highest frequency components suffer McGraw-Hill Education, 2016.
more noise than the low-frequency components (Figure 15). Therefore, the SNR 3. Leon W. Couch II; Digital and analog communication systems, Eighth Edition; Pearson
Education, Inc, 2013.
is not constant and it weakens for the highest components of the signal. 4. Michael P. Fitz; Fundamentals of Communications Systems; McGraw-Hill, 2017.

S. Labandji – [Link] 15 16 See: 12/31/19