Chapter 3: Transmission Signal & Noise in System

 Components of Communication Systems

 Types of errors (noise) in data transmission:

1- White, or Gaussian noise is continuous and is dependent on the temperature of the medium.
White noise might change the content of data. White noise can be removed by passing the noisy
signal through a set of filters.
2- Spike noise is not continuous but may completely obliterate the data, so that it cannot be
recovered.
3- Cross talk is a coupling between two active links. Coupling can be electrical or electromagnetic.
4- Echo is the reflecting impact of a transmitted signal. A signal can hit the end of a cable and
bounce back through the wire, interfering with the original signal.
5- Jitter is a timing irregularity that shows up at the rises and falls of a signal, causing errors. Jitter
can result from electromagnetic interference or cross talk and can be reduced by proper system
shielding.
6- Bit attenuation is the loss of a bit's strength as it travels though a medium. This type of error
can be eliminated with the use of amplifiers and repeaters for digital systems.
 Types of distortion:
o Amplitude distortion: amplitude not constant in desired frequency band.
o Phase distortion: phase not linear through origin in desired frequency band.
o Nonlinear distortion: system is nonlinear.
 Distortionless Transmission:
o Transmission is said to be distortionless if the input and the output have identical wave shapes
within a multiplicative constant. A delayed output that keeps the input waveform is distortionless.
o Given input x(t) and output y(t), a distortionless transmission satisfies:
y (t )  k x (t  t d )
Where td = delay time and k = constant.
Take the Fourier transform of both sides
Y (f )  F y (t )  F k x (t  t d )  k F x (t  t d )
Using Time shifting property of Fourier transform
Y (f )  k X (f )e  j 2 ftd
Since Y (f )  H (f )X (f ) then, H (f )  ke  j 2 ftd
Then, H (f )  k  (f )  2 ft d
o Thus, for distortionless transmission, amplitude response |H(f)| must be a constant and phase
response θh(f) must be linear function of f going through the origin.
1- Amplitude response, H (f )  k (must be constant, all pass system)
2- Phase response  (f )  2 ft d must be linear with frequency

o The slope of θh(f) with respect to ω = 2πf is –td (delay of output w.r.t input.
o The time delay:
d  (f )
 (f )  2 ft d then,  2 t d
df
1 d  (f )
Then, the time delay is td  
2 df
 Type of filters in communication system:
 o Filters types: Low-Pass Filter (LPF), High-Pass Filter (HPF), Band-Pass Filter (BPF), Band-
Stop Filter (BSF).

Important Example: if g(t) and y(t) are the input and output respectively of a simple RC low-
pass filter as illustrated in figure. Determine the transfer function H(f) and sketch |H(f)|, θh(f) and
td(f). For distortionless transmission through this filter, what is the requirement on the bandwidth
of g(t) if amplitude response variation within 2 % and time delay variation within 5 % are
tolerable? What is the transmission delay? Find the output y(t).
Solution:
Applying voltage division rule:
(1 / j 2 fC ) 1 a
H (f )   
R  (1 / j 2 fC ) 1  j 2 fRC a  j 2 f
Where, a is the cut-off frequency
1 1
a  6 12
 106
RC 10  10
a
Hence H (f ) 
a 2  ( 2 f )2
Where, H (f ) 1 for f  a
2 f
 (f )   tan1
a
2 f
Where,  (f )  for f  a
a
1 d  (f ) a
t d (f )   
2 df (2 f )2  a 2
1
Where, t d (f ) for f  a
a
 The maximum amplitude response and time delay occur at f=0
1
H (0) max  H (0)  1 t d (f ) max  t d (0)   106
a
Let fo be the highest bandwidth of a signal that can be transmitted within these specifications:
1
Therefore, H (f o )  0.98 and t d (f o )  0.95
a
a
H (f o )   0.98  2 f o  0.203a then, f o  32.3 kHz
a  ( 2 f o )
2 2

a 1
t d (f o )   0.95  2 f o  0.2294a then, f o  36.5 kHz
(2 f o )  a
2 2
a
The smaller value of the two values is fo = 32.3 kHz or ωo = 203,000 rad/sec is the highest
bandwidth which satisfy both conditions on |H(f)| and td(f).

The bandwidth fo is f o  32.3 kHz

1
o Within this bandwidth td =106 and H (f ) 1
a
o Then, the output y(t) is y (t ) g (t 106 )
 Ideal versus Practical filters:
o Ideal filter: Signal g(t) is transmitted without distortion but with delay td. For this filter
 f   j 2 ftd
H (f )     e
 2B 
 f 
H (f )     and h (f )   j 2 ft d
 2B 
Unit impulse response of this filter is
  f   j 2 ftd 
h (t )  F 1   e   2B sinc  2B (t  t d )
  2 B  
o For a physically realizable system h(t) must be causal
h (t )  0 for t  0
In frequency domain
 ln H (f )
 1  ( 2 f ) 2
df  
o The impulse response of previous filter is not realizable. To make it realizable (causal), the tail
could be cut off

hˆ (t )  h (t )u (t )

o Ideally a delay of td = ∞ is needed for ideal filter

o The half-power bandwidth of a filter is defined as the bandwidth over which the amplitude
response |H(f)| remains constant within a 3 dB or (ratio of 0.707)
o Half-power bandwidth of a low-pass filter is called the cutoff frequency.

Example : In the illustration system whose response is h(t) is a rect pluse and the input signal is
cos(2πfct). Sketch the amplitude spectrum of the output signal.
Solution:
o The output is
y (t )  h t  * x t 
Using convolution theory,
Y(f )=X(f ) . H(f )
Using Fourier transform:

1
X (f )  F cos  2 f ct  =  (f  f c )   (f  f c ) 
2
  t  
H (f )  F  rect 
   Tsinc(fT )
  
T
o Then, the Amplitude Spectrum of the output is
T
Y(f )  sinc(T (f  f c ))  sinc(T (f  f c )) 
2

Practice: A wide sense stationary noise process N(t) has an autocorrelation

3 
R NN ( )  Pe
Find and sketch the PSD.
Solution:
Since R ( ) 
 S (f )
Then,
 
S NN (f )  F  R NN ( )   R NN ( )e d    Pe
2 j  f  3 
e 2 j  f  d 
 

  3 2 j  f  0
    (3 2 j  f ) 0

 P e e d   e e
3 2 j  f 
d    P e d    e (32 j  f ) d  
0   0  
 e  (3 2 j  f ) 
  e (3 2 j  f )
 
0
 1 1 
 P       
 (3  2 j  f ) (3  2 j  f ) 
P
 (3  2 j  f ) 0  (3  2 j  f )  
 
6P

9  (2 f ) 2
Other Solution:
a t 2a
Using the Fourier transform of e 

a 2  (2 f )2
3 6
Then, e 

9  (2 f )2
6P
The PSD is S NN (f ) 
9  (2 f )2
 Effects of Transmission Losses and Noise in Analog Communication System
o In any communication system, there are usually two dominant factors that limit the performance
of the system
1) Additive noise 2) Signal attenuation
o Basically all physical channels, including wireline and radio channels, are lossy. Hence the
Signal is (reduced in amplitude) as it travels through the channel.
o A simple mathematical model of the attenuation may be constructed, as shown in the figure
below, by multiplied the transmitted signal by the factor α < 1. Consequently, if the transmitted
signal s(t), the received signal is r(t)= α s(t)+n(t).

o Clearly, the effect of signal attenuation is to reduce the amplitude of the desired signal s(t) and,
thus, render the communication signal more vulnerable to additive noise. In many channels,
such as wire lines and microwave line-of-sight channels, signal attenuation can be reduced by
using amplifiers to boost the level of the signal during transmission. However, an amplifier also
introduces additive noise in the process of amplification and, thus, corrupts the signal. This
additional noise must be taken into consideration in the design of the communication system.
 Characterization of Thermal Noise Sources
o Thermal noise is produced by the random movement of electrons due to thermal agitation. The
power spectral density of thermal noise is generally expressed as
kT N o
S n (f )   W/Hz
2 2
Where, No=kT k: Boltzmann's constant (equal to 1.38×10−23 J/K)
For example, at room temperature (To = 290◦ K), No = 4 × 10−21 W/Hz.
 Effect of Noise on a Baseband System
o Baseband system: The receiver consists only of an ideal low-pass filter with the bandwidth
W. The noise power at the output of the receiver for a white noise input:
W
No
Pno  
W
2
df  N oW Watts

Then, the baseband signal-to-noise ratio (SNR) is

S  PR
   Where, PR: the received power
 b
N N oW
Example 1: Find the SNR in a baseband system with a bandwidth of 5 kHz and with No/2= 10−14
W/Hz. The transmitter power is one kilowatt and the channel attenuation is 10 −12.
Solution:
We have PR  1012 PT  1012 103  109 Watts
 S  PR 109 S 
Therefore,     14  20 ,    10log(20)  13 dB
 N b N oW 10  5000  N b (dB )
Example 2: A white noise has a 2-sided power spectral density of 6 kW/MHz. It is passed
through a low pass filter having a bandwidth of 1 kHz. Compute the output noise power.
Solution:

6 kW 6 103 W No
The power spectral density in W/Hz is   0.006 W/Hz 
MHz 106 Hz 2
W
Therefore, the noise power is N o  N o  BW  0.012 1 kHz  12 W
Hz
Example 3: A composite TV signal has a bandwidth of 40 MHz and is transmitted with a power
of 0.1 W through a satellite channel. The white noise has a 2 sided PSD is 10 -15 W/Hz. What is
the noise power for the link?
Solution:
The noise is going to exist over the range of the frequencies over which the signal is transmitted.
Hence the signal bandwidth is equal to the noise bandwidth.
No
The power spectral density in W/Hz is  1015 W/Hz  N o  2 1015 W/Hz
2
Therefore, the noise power is N o  N o  BW  2 1015  40 106  80 nW
 Effective Noise Temperature and Noise Figure
o When we employ amplifiers in communication systems to boost the level of a signal, we are also
amplifying the noise corrupting the signal. We may model an amplifier as a filter with the
frequency response characteristic H(f).
o The output noise power from an ideal amplifier
may be expressed as
Pno  GN o B neq
G = |H(f)|2max is the maximum available power gain of the amplifier.
Bneq: is the noise equivalent bandwidth of the filter.
o Any practical amplifier introduces additional noise (Pni) at its output due to internally generated
noise. Hence, the noise power at its output may be expressed as
Pno  GN o B neq  Pni  GkTB neq  Pni
Pni
Therefore, Pno  GkB neq (T  )  GkB neq (T T e )
GkB neq
Where, Te is called the effective noise temperature of the two-port network (amplifier).
 Pni
Te 
GkB neq
o A signal source at the input to the amplifier with power Psi will produce an output with power
Pso  GPsi
o Hence, the output SNR from the two-port network is
S  Pso GPsi Psi 1 S 
       
 N o Pno GkTB (1  T e ) N B (1  T e ) (1  T e )  N i
neq o neq
T T T
S  Psi
Where,    the input SNR of the two-port network.
 i
N N B
o neq
o The SNR at the output of the amplifier is degraded (reduced) by the factor (1+Te/T). An ideal
amplifier is one for which Te = 0.
o When T is taken as room temperature (To = 290◦ K), the factor F = (1+Te/ T0) is called the
noise figure of the amplifier.
S  1S 
Consequently,     
 N o F  N i
o By taking the logarithm of both sides,
S  S 
10log    10log(F )  10log  
 N o  N i
10log(F ) : is the loss in SNR due to the additional noise introduced by the amplifier.
o Specifically, the noise figure of the two port network is defined as the ratio of the output noise
power Pno to the output noise power of an ideal (lossless) two port network for which the thermal
noise source is at room temperature (To = 290◦ K).
 Te 
F  1 
 T 
o The overall noise figure of a cascade of K amplifiers with gains Gk and corresponding noise
figures Fk, 1 ≤ k ≤ K is
F2  1 F3  1 FK  1
F  F1    ..........  .
G1 G1G 2 G1G 2 .....G K 1
This expression is known as Fries’ formula. We observe that the dominant term is F1, which is
the noise figure of the first amplifier stage.
 Transmission Losses
o As we indicated previously, any physical channel attenuates the signal transmitted through it.
o The amount of signal attenuation generally depends on the physical medium, the frequency of
operation, and the distance between the transmitter and the receiver. We define the loss L in
signal transmission as the ratio of the input (transmitted) power, PT to the output (received)
power, PR of the channel, i.e.,
PT
L
PR
 or, in decibels, LdB  10log(L )  10log(PT ) 10log( PR )
o In wireline channels, the transmission loss is usually given in terms of decibels per unit length,
e.g., dB/km. For example, the transmission loss in coaxial cable of 1 cm diameter is about 2
dB/km at a frequency of 1 MHz. This loss generally increases with an increase in frequency.
Example: Determine the transmission loss for a 10 km and a 20 km coaxial cable if the loss per
kilometer is 2 dB at the frequency operation.
Solution
i) The loss for the 10 km channel is LdB  2 10  20 dB
P P 2
Hence, the output (received) power is PR  T  T20  10 PT
L
10 10
ii) The loss for the 20 km channel is LdB  2  20  40 dB
P P 4
Hence, the output (received) power is PR  T  T40  10 PT
L
10 10
Note that doubling the cable length increases the attenuation by two orders of magnitude.
o In line-of-sight radio systems, the transmission loss in free space is given as

 4 d 
2
L  
  
where λ = c/f is the wavelength of the transmitted signal, c is the speed of light (3×108 m/sec),
f is the frequency of the transmitted signal, and d is the distance between the transmitter and
the receiver in meters. In radio transmission, L is called the free-space path loss.
Example: Determine the free-space path loss for a signal transmitted at f = 1 MHz over
distances of 10 km and 20 km.
Solution:
c 3 108
The wavelength is     300 m
f 1 10 6

 4 d
2

The free-space path loss is L  
  
 4 10 103 
i) For the 10 km path, LdB  20log    52.44 dB
 300 
 4  20 103 
ii) For the 20 km path, LdB  20log    58.44 dB
 300 
It is interesting to note that doubling the distance in radio transmission increases the free-space
path loss by 6 dB.
 Additive White Gaussian Noise:
o The thermal noise is described by a zero-mean Gaussian random process, n(t) the adds on to the
signal (Additive).
o Its PSD is flat, hence it is called white noise.
Autocorrelation is a spike at 0: uncorrelated at any non-zero lag.

Power Spectral density Autocorrelation Function Probability density Function

Example1: Consider a white noise process W(t) of zero mean and power spectral density No/2
applied to the low-pass RC filter, shown in the figure. The transfer function of the filter is
1
H (f ) 
1  j 2 fRC
Determine the following:
1) The power spectral density of the noise N(t)
2) Sketch the output power spectral density against frequency
3) The autocorrelation function of the output noise
4) Sketch the autocorrelation against time-difference (τ)
5) The mean power of the output noise.
Solution:
1) The power spectral density of the noise N(t) appearing at the low-pass RC filter is therefore,
No / 2
S N (f )  H (f ) SW (f ) 
2

1  (2 fRC )2
2) The sketch the output power spectral density against
frequency shown in figure.
3) We have the Fourier-transform pair
2a
exp(a t )  2
a  (2 f )2
No / 2 No 1
S N (f )  2=
1  (2 fRC ) 2(RC ) ( 1 )2  (2 f )2
2

RC
2
N RC
S N (f )  o
4RC ( 1
)2  (2 f )2
RC
Therefore, the autocorrelation of the filtered noise process
N(t) is
 2 
 N 
R N ( )  F-1 S N (f )  -1  o RC 
 4RC ( 1 )2  (2 f )2 
 RC 
 No 
R N ( )  exp( )
4RC RC

No
(5) The mean power of the output noise P   S N (f )df  R N (0) 

4RC
Example 2: Suppose that a random signal is a white noise signal. This means that all frequencies
are present in the signal and its power spectral density will be a straight line as shown in Figure
below. The level of the power spectral density is given as 25 Watts/Hertz. This signal is passed
through a low-pass filter whose frequency response magnitude is shown in Figure (b). Calculate
the total signal power of the signal coming out of the low-pass filter.

Solution:
o The power spectral density of the output signal is:
y (f )  G ( j 2 f ) x (f )
2

o The total power of the filter output signal is the

integration of this function from 0 to ∞
 10 20
Signal power   y (f )df   2500df  25  (20  f ) 2 df
0 0 10

= 25,000  1,250  26,250 Watts

Example 3: We are given that x(t) is a stationary random process with Rx (τ)= e-α|τ| where α > 0.
It is the input to an LIT system with impulse response h(t) = e-βt u(t), where β > 0 and β≠α.
Determine the output power spectral density.
Solution:
 
The autocorrelation function is R x ( )  e
The input spectral density is S x (f )  F  R x ( )  F e
 
  2 2
    ( 2 f )2
1
The amplitude response of the system H (f )  F  h (t )  F e  t u (t ) 
  j 2 f
1   j 2 f  2 f 1
H (f )    2  j 
  j 2 f   j 2 f   ( 2 f )2  2  ( 2 f )2  2  ( 2 f )2
2
Thus, the output power spectral density is S y (f )  S x (f ) H (f )
2 1
S y (f )  2 
  (2 f )2  2  (2 f )2
Example 4: A zero-mean white Gaussian noise with power-spectral density No/2 passes through
an ideal low pass filter with bandwidth B.
a. Find the autocorrelation function of the output process Y(t).
b. Mean power of the noise at the output.
Solution:
N
The input spectral density is S i (f )  o
2
The amplitude response of the ideal LPF  f 
H (f )  rect  2B 
 
Thus, the output power spectral density is Ideal Low pass filter
2
S y (f )  S i (f ) H (f )
2
N  f  N  f 
S y (f )  o rect    o rect  
2  2B  2  2B 
Then, the autocorrelation function of the output process y(t):
N  f 
R y ( )  F 1 S y (f )  F 1  o rect  
 2  2B 
t 
We know that, A rect   AT sinc(fT )
T 
 f  f 
Using Duality theorem, AT sinc(tT ) A rect     A rect  
 T  T 
N  f  N o
a) Therefore, R y ( )  F 1  o rect     2 ( 2B ) sinc( 2B  )  N o B sinc( 2B  )
 2  2B 
b) Mean power of the noise at the output
  B
No  f  No
PN   S y (f )df   2
rect 
 2B
 df 
 2  df  NoB
  B

Or PN   S y (f )df  R y (0)  N o B
