COE 341: Data & Computer

Communications (T071)
Dr. Radwan E. Abdel-Aal

Chapter 3: Data Transmission

Remaining Six Chapters:
Chapter 7: Data Link:
Flow and Error control,
Link management
Data Link
Chapter 8: Improved
utilization: Multiplexing

Chapter 6: Data
Physical Layer Communication:
Chapter 4: Synchronization,
Transmission Media Error detection and
correction

Transmission Medium
Chapter 5: Encoding:
Chapter 3: Signals, their From data to signals
representations, their
transmission over media,
Resulting impairments
2
Agenda
 Concepts & Terminology
 Signal representation:
Time and Frequency domains
 Bandwidth and data rate
 Decibels and Signal Strength (Appendix 3A )
 Fourier Analysis (Appendix B )
 Analog & Digital Data Transmission
 Transmission Impairments
 Channel Capacity

3
Terminology (1)
Transmission system: Components
 Transmitter

 Receiver

 Medium
 Guided media
 e.g. twisted pair, coaxial cable, optical fiber
 Unguided media
 e.g. air, water, vacuum

4
 Terminology (2)
Link Configurations:
 Direct link
 No intermediate ‘communication’ devices A
(these exclude repeaters/amplifiers)
Amplifier

Two types: C

 Point-to-point
B
 Only 2 devices share link
 Multi-point
 More than two devices share the same link, e.g.
Ethernet bus segment

5
Terminology (3)
Transmission Types (ANSI Definitions)
 Simplex
 Information flows in one direction only all the time
e.g. Television, Radio broadcasting
 Duplex
 Information flows in both directions
 Two types:
 Half duplex
 Only one direction at a time
e.g. Walki-Talki
 Full duplex
 In both directions at the same time
e.g. telephone

6
 Frequency, Spectrum and
Bandwidth
 Time domain concepts
 Analog signal
 Varies in a smooth, continuous way in both time and
amplitude
 Digital signal
 Maintains a constant level for sometime and then changes to
another constant level
(i.e. amplitude takes only a finite number of discrete levels)
 Periodic signal
 Same pattern repeated over time
 Aperiodic signal
 Pattern not repeated over time

7
Analogue & Digital Signals
All values on the time and amplitude axes
are allowed

Only a few amplitude levels allowed

- Binary signal: 2 levels

8
 T
Periodic Temporal Period

…
Signals
t t+1T t+2T

S (t+nT) = S (t); 0  t  T
Where:
t is time over first period
T is the waveform period
n is an integer

Signal behavior over one period

describes behavior at all times
9
Aperiodic (non periodic)
Signals in time

s(t)

0
+ X/2 t
- X/2

10
 Continuous and Discrete
Representations
Availability of the signal over the horizontal axis
(Time or Frequency)

Continuous:

Signal is defined
at all points on
the horizontal axis

Sampling with
Discrete: a train of delta function

Signal is defined
Only at certain points
on the horizontal axis

11
 T (Period)
Sine Wave = A sin ()
A (Amplitude)
s(t) = A sin(2f t +)

 Peak Amplitude (A)
 Peak strength of signal, volts
 Repetition Frequency (f)
 Measures how fast the signal varies with time
 Number of waveform cycles per second (Hz)
 f = 1/ T(xx sec/cycle) = yy cycles/sec = yy Hz
 Angular Frequency ()
= radians per second = 2 f = 2 /T
 Temporal (time) Period, T = 1/f
 Phase ()
 Determines relative position in time, radians (how
12 to calculate?)
Varying one of the three parameters of a
sine wave carrier Can be used to convey information…!
s(t) = A sin(2ft +) ModulatIon

= A sin(t+)
AM
Varying A

Va
ryi
ng

Varying f

FM
PM

13
 Sine Wave Traveling in the +ive x
direction
s(t) = A sin (k
k = Wave Number
x -  t]
 = Angular Frequency  Spatial Period
= 2 /  = 2 f = 2 / T = Wavelength

For point p on the wave: x
p
Total phase at t = 0: kx -  (0) = kx
x
Total phase at t = t: k(x+ x) -  (t)
Distance,
Same total phase, x
kx = k(x+ x) -  (t) t=0
k x =  t
t = t + ive x direction
Wave propagation velocity v = x / t
v = /k = /T = f Direction of wave travel, at velocity v
 Show that the wave
s(t) = A sin (k x +  t] v = f V is constant for a given
14 type and medium
wave
travels in negative x direction
 Wave Propagation Velocity, v
 m/s for:
Constant
 A given wave type (e.g. electromagnetic, seismic, ultrasound, ..)
 and a given propagation medium (air, water, optical fiber)
 For all types of waves:
 v=f
 For a given wave type and medium (given v): higher frequencies
correspond to shorter wavelengths and vise versa:
Electromagnetic waves:
Shorter wavelength .. Higher frequency

long wave radio (km), short wave radio (m), microwave (cm)… light (nm)
 For electromagnetic waves:
 In free space, v  speed of light in vacuum

v = c = 3x108 m/sec
 Over other guided media (coaxial cable, optical fiber, twisted

pairs): v is lower than c

15
Wavelength, meters)
 Is the Spatial period of the wave:
i.e. distance between two points in space on
the wave propagation path where the wave has
the same total phase
 Also: Distance traveled by the wave during one
temporal (time) cycle:
dT = v T = (f) T = 

16
 Frequency Domain Concepts
 Response of systems to a sine waves is easy to analyze
 But signals we deal with in practice are not all sine waves,
e.g. Square waves
 Can we relate waves we deal with in practice to sine
waves? YES!
 Fourier analysis shows that any signal can be treated as
the sum of many sine wave components having different
frequencies, amplitudes, and phases (Fourier Analysis:
Appendix B)
 This forms the basis for frequency domain analysis
 For a linear system, its response to a complex signal will
be the sum of its response to the individual sine wave
components of the signal.
 Dealing with functions in the frequency domain is simpler
than in the time domain 17
 Addition of
Fundamental

A = 1*(4/)
Two frequency = f

frequency +
1/3 rd the Amplitude
Components 3rd harmonic 3 times the frequency

A = (1/3)*(4/)
frequency = 3f
Frequency Spectrum

=
Approaching
a square wave
Fourier Series
3 t

f
Frequency Domain: S(f) vs f Time Domain: s(t) vs t
Fourier Series
18
Discrete Function in f Periodic function in t
Asymptotically approaching a square
wave by combining the fundamental
+ an infinite number of odd
harmonics
1.5
at prescribed amplitudes
1
Topic for a
square signal, sw(t)

programming
0.5 assignment
Adding more
higher What is the highest
0 harmonics Harmonic added?

-0.5

-1

-1.5
0 2 4 6 8 10
t 19
 More Frequency Domain
Representations:
(Aperiodic
s(t)

A single square pulse

Sinc(f) = sin(f)/f Fourier 1

signal) Transform

To  To 
0

+ X/2 t
To 
- X/2

frequency time
Fourier
1/X Frequency Domain: S(f) vs f Time Domain: s(t) vs t
Transform
Continuous Function in f Aperiodic function in t

• What happens to the spectrum as the pulse gets broader … DC ?

• What happens to the spectrum as the pulse gets narrower … spike ?

20
 Spectrum & Bandwidth of a
 signalof a signal
Spectrum
 Range of frequencies contained in a signal
 Absolute (theoretical) Bandwidth (BW)
 Is the width of spectrum = fmax- fmin S(f)

 But in many situations, fmax = !

(e.g. a square wave), so: f 3f 5f 7f …. f

 Effective Bandwidth
 Often called bandwidth

 Narrow band of frequencies containing most of

the signal energy

 Somewhat arbitrary: what is “most”?

21
Signals with a DC Component
NO DC Component,
Signal average over a period = 0

+
_ t

+
1V DC Level

+
t
1V DC
Component
DC Component
Component at zero frequency
Determines if fmin = 0 or not 22
 = (fmax- fmin)
Bandwidth for these signals:
fmin fmax Absolute Effective
BW BW

1f 3f 2f 2f

0 3f 3f 3f

0   1/X ?

23
Bandwidth of a transmission
system
 Is the Range of signal frequencies that are
S(f)
adequately passed by the system
 Effectively, the transmission system ….
f 3f 5f 7f f
(TX, medium, RX) acts as a filter
 Poor transmission media, e.g. twisted pairs, have
a narrow filter bandwidth
 This cuts off higher frequency signal components
 poor signal quality at receiver
 And limits the signal frequencies (Hz) that can be
used for transmission
 limits the data rates used (bps) 24
 Received Waveform

Limiting Effect of
System Bandwidth 1,3
BW = 2f f 3f
Better reception requires larger BW 1
More difficult reception with smaller BW

Varying System BW

1,3,5
BW = 4f f 3f 5f

1,3,5,7
BW = 6f f 3f 5f 7f
3
…
BW = 
1,3,5,7
4  1
s(t )  
 k odd, k  1 k
sin(2kft ) ,9,…
f 3f 5f 7f …… 
4 25
Fourier Series for a Square Wave
 System Bandwidth and Achievable
Data Rates
 Any transmission system supports only a limited range of
frequencies (bandwidth) for satisfactory transmission
 For example, this bandwidth is largest for expensive
optical fibers and smallest for cheap twisted pair wires
 So, bandwidth is money  Economize in its use
 Limited system bandwidth degrades higher frequency
components of the signal transmitted  poorer received
waveforms  more difficult to interpret the signal at the
receiver (especially with noise)  Data Errors
 More degradation occurs when higher data rates are used
(signal will have more components at higher frequency )
 This puts a limit on the data rate that can be used with a
given signal to noise requirement, receiver type, and a
specified error performance  Channel capacity issues
26
 Data Element =
Bandwidth and Data Rates Signal Element

Data rate = 1/(T/2) = (2/T) bits per sec = 2f bps Period T = 1/f T/2

B
1 0 1 0 Data
B = 4f
Given a bandwidth B,
f 3f 5f
Data rate = 2f = B/2
To double the data rate you need to double f: Two ways to do this…
1. Double the bandwidth with same received waveform (same RX conditions & error rate)
2B = 4f’
2B
New bandwidth: 2B, X2 1 0 1 0 1 0 1 0
Data rate = 2f’ = 2(2f)= 4f = B
f’ 3f’ 5f’

Same bandwidth, B, but tolerate poorer received waveform (needs better receiver,
gher S/N ratio, or tolerating more errors in data)
B = 2f’ 1 0 1 0 1 0 1 0
B
X2
Bandwidth: B,
Data rate = 2f’ = 2(2f) = 4f = B
f’ 3f’ 5f’ 27
 Bandwidth & Data Rates: Tradeoffs…
Compromises
Increasing the data rate (bps) while keeping BW the
same (to economize) means working with inferior
(poorer) waveforms at the receiver, which may require:
 Ensuring higher signal to noise ratio at RX (larger signal
relative to noise):
 Shorter link distances
 Use of more en-route repeaters/amplifiers
 Better shielding of cables to reduce noise, etc.
 More sensitive (& costly!) receiver
 Suffering from higher bit error rates
 Tolerate them?
 Add more efficient means for error detection and correction- this also
increases overhead!.

28
Appendix 3A: Decibels and
Signal
 Strength
The decibel notation (dB) is a logarithmic measure of the ratio
between two signal power levels
 N = number of decibels
dB P2
 P = input power level (Watts)
1 N dB  10 log10
 P = output power level (Watts)
2 P1
 e.g.  Amplifier gain P1 P2 P3
 Signal loss over a link
Lossy Link
Amplifier
 Example:
 A signal with power level of 10mW is inserted into a transmission line

 Measured power some distance away is 5mW

 Power loss in dBs is expressed as

NdB =10 log (5/10)=10(-0.3)= -3 dB

 - ive dBs: P2 < P1 (Loss),

 +ive dBs: P2 > P1 (Gain)
29
Relationship Between dB
Values and Power ratio
(P 2/PRatio
Power 1) dB Power Ratio dB
1 0
101 10 10-1 -10
102 20 10-2 -20
103 30 10-3 -30
104 40 10-4 -40
105 50 10-5 -50
106 60 10-6 -60
2 3 1/2 -3
30
 Decibels and Signal Strength

 Decibel notation is a relative, not absolute, measure:

 A loss of 3 dB halves the power (could be 100 to 50, 16 to 8, …)
 A gain of 3 dB doubles the power (could be 5 to 10, 7.5 to 15, …)
 Will see shortly how we can handle absolute levels
 Advantage:
 The “log” allows replacing:
 Multiplication with Addition

C=A*B
Log C = Log A + Log B
 and Division with Subtraction

A=C/B
Log A = Log C - Log B 31
 Decibels and Signal Strength
 Example: Transmission line with an intermediate amplifier
Amplifier
4 mW ?
Transmitted Gain: 35 dB Received
Loss: 12 dB Loss: 10 dB
Signal Signal

 Net power gain over transmission path:

+ 35 – 12 – 10 = + 13 dB (+ ive means there is net gain)

Received Signal Power

 13  10 log10 [
4mW
]

Received Signal Power 13

[ ]  log 110 [ ]
4mW 10
 Received signal power = (4 mW) log10-1(13/10) = 4 x 101.3
32
= 4 x 10 1.3 mW = 79.8 mW Still we use some multiplication!
WK 4

How to represent absolute power

levels?
 Decibel-Watt
As (dBW)
a ratio relative to a fixed reference and Decibel-
power level

mW
With 1 W(dBm)
used as a reference  dBW
Power
Power _ LeveldBW  10 log10 W
Caution!: Must be same units at
1W top and bottom
 With 1 mW used as a reference  dBm
PowermW
Power _ LeveldBm  10 log10 Caution!: Must be same units at
1 mW top and bottom

 Examples:
 Power of 1000 W is 30 dBW, 1 W = ? dBW
 –10 dBm represents a power of 0.1 mW,
 1 mW = ? dBm 33
X dBW = (X + ?) dBm
dBs & dBms are added
algebraically G P1 P2

P 2mW G is  Positive for gain

G = power ratio =
P1mW  Negative for attenuation

G dBs = 10 log10 G = 10 log10 P 2mW  10 log10 P 2mW / 1mW

P1mW P1mW / 1mW
 10 [log10 ( P 2mW / 1mW )  log10 ( P1mW / 1mW )]
 10 [log10 ( P 2mW / 1mW )]  10[log10 ( P1mW / 1mW )]
 P 2dBm  P1dBm

Similarly for dBs & dBWs

34
 Decibels and Signal Strength
 Example: Transmission line with an intermediate amplifier
• If all ratios are in dBs and all levels are in dBm  solve by algebraic addition
Same for {dBs and dBWs} (No need for any
multiplication/division)
4 mW ?
Transmitted Gain: 35 dB Received
Loss: 12 dB Loss: 10 dB
Signal Signal
Amplifier
 Net power gain over transmission path:
+ 35 – 12 – 10 = + 13 dB (+ ive means actual net gain)

TX Signal Power in dBm = 4 mW = 10 log (4/1) = 6.02 dBm

 RX signal power (dBm) = 6.02 + 13 =19.02 dBm
 Check: 19.02 dBm = 10 log (RX signal in mW/1 mW)
 RX signal = log-1 (19.02/10) = 79.8 mW
As in
35
Slide 31
 Decibels & Voltage ratios
 Power decibels can also be expressed in terms
of voltage ratios
 Power P = V2/R, assuming same R
Relative:
2
P2 V2 / R V2
N dB  10 log  10 log 2  20 log Note that this is still a power ratio…
But expressed in terms of voltages
P1 V1 / R V1

Absolute: dBV and dBmV

 Decibel-millivolt (dBmV) is an absolute unit,
with 0 dBmV being equivalent to 1mV. Also dBV
Voltage mV
N dBmV  20 log Caution!: Must be same units at
36
top and bottom
1 mV
 Appendix B: Fourier Analysis
Signals in Time

Periodic Aperiodic
…

Discrete Continuous Discrete Continuous

DFS FS DFT FT
Use Fourier Series Use Fourier Transform

FS : Fourier Series
DFS : Discrete Fourier Series
FT : Fourier Transform
DFT : Discrete Fourier Transform 37
 Fourier Series for periodic

continuous
Any periodic signalsignals
x(t) of period T and repetition
frequency f0 (f0 = 1/T) can be represented as an infinite
sum of sinusoids of different frequencies and
amplitudes – its Fourier Series. Expressed in Two forms:
Frequencies
 1. The sine/cosine form: are multiples

A0
   An cos( 2nf 0t )  Bn sin( 2nf 0t )
of the fundamental
x(t )  frequency f0
2 n 1
f0 = fundamental
Where: DC frequency = 1/T
Component T
2
T 2
A0   x(t )dt An   x(t ) cos(2nf 0t )dt = f(n)
T 0 T 0
All integrals over Two components
at each frequency
If A0 is not 0, one period only T
2
x(t) has a DC Bn   x(t ) sin( 2nf 0t )dt = f’(n)
component T 0
38
 Fourier Series: 2. The Amplitude-
 Phase
Previousform:
form had two components at each frequency
(sine, cosine i.e. in quadrature) : An, Bn coefficients
 The equivalent Amplitude-Phase representation
has only one component at each frequency: Cn, n
 Derived from the previous form using trigonometry:
cos (a) cos (b) - sin (a) sin (b) = cos [a +b]
 Now we have
C0
x(t )     Cn cos(2nf 0t   n ) Only one component
at each frequency nf
2 n 1 0

The C’s and ’s Now components

are obtained C0  A0 Cn  A B
2
n
2
n have different
from the previous amplitudes,
A’s and B’s using frequencies,
the equations:   Bn 
 n  tan 
1
 and phases

 An 39

 Fourier Series: General

A Fourier Series Expansion
   A cos( 2nf t )  B sin( 2nf t )
Observations
x(t )  0
n 0 n 0
2 n 1
Function Odd Function
DC Even Function

Function Series
No DC T
A0 = 0
1

T 0
x (t )dt  0

Even Function x(t) = x(-t) Bn = 0;

Symmetric about Y axis for all n

Odd Function x(t) = - x(-t) An = 0;

Symmetric about the origin for all n
40
 Correction

41
 Fourier Series Example
x(t)
1

-3/2 -1 -1/2 1/2 1 3/2 2

-1
T

Note: (1) x(– t)=x(t)  x(t) is an even function

(2) f0 = 1 / T = ½ Hz
T 2 1 1/ 2 1
2 2
A0   x (t )dt   x (t )dt  2  x (t )dt  2  1dt  2   1dt
T 0 20 0 0 1/ 2
Note: A0 by definition
is 2 x the DC content  2{[t ]10/ 2  [t ]11 / 2 }  0
42
 Contd…
1
-3/2 -1 -1/2 1/2 1 3/2 2
-1
T T /2 T 1
2 4
An   x(t ) cos( 2nf 0t )dt   x(t ) cos( 2nf 0t )dt  2 x(t ) cos( 2nf 0t )dt
T0 T 0 0

= 0 for n even

f0 =1/2 = (4/n) sin (n/2) for n odd

T T /2
2 2
Bn   x(t ) sin( 2nf 0t )dt   x(t ) sin( 2nf 0t )dt a function of n only
T0 T T / 2
2
0
2
T /2  Replace t by –t
T T/ 2 0 x(t ) sin(2nf0t )dt  Swap limits
 x (t ) sin( 2nf 0 t ) dt 
T
in the first integral
T /2 T /2
2 2
   x ( t ) sin(2nf0t )d ( t )   x (t ) sin(2nf0t )dt  0
T 0 T 0
- sin(2nf t) dt Then Bn = 0 for all n
x(t), since x(t) is an even function 43
 Contd…

A0
x(t )     An cos( 2nf 0t )  Bn sin( 2nf 0t )
2 n 1
A0 = 0, f0 = ½, so
2 f0 = 
Bn = 0 for all n,
An = 0 for n even: 2, 4, … Original x(t)
= (4/n) sin (n/2) for n odd: 1, 3, … is an even

4 n function!
x (t )   sin cos nt
n 1,odd n 2 Amplitudes,
n odd
4 4 4 4
x (t )  cos  t  cos 3 t  cos 5 t  cos 7 t  ...
 3 5 7
4� 1 1 1 �
x (t )  �cos  t  cos 3 t  cos 5 t  cos 7 t  ...�
� 3 5 7 �
Cosine is an
2  (1/2) t 2  3 (1/2) t
even
44 function
f0 = ½  Fundamental 3rd Harmonic
 Another Example
Previous
Example
x1(t)
1

-2 -1 1 2

-1
T

Note that x1(-t)= -x1(t)  so, x(t) is an odd function

Also, x1(t)=x(t-1/2) This waveform is the previous waveform shifted right by 1/2

4  1 1  1 1  1 1  1 
x1(t )  cos   t    cos 3  t    cos 5  t    cos 7  t  
  2 3  2 5  2 7  2 
45
 Another Example, Contd…

4   1  3  1  5  1  7 
x1(t )  cos  t    cos  3t    cos  5t    cos  7t  
  2 3  2 5  2  7  2 
4 1 1 1 
x1(t )   sint  sin 3t  sin 5t  sin 7t  ...  Sine is an
 3 5 7  odd function
4  1
x1(t )  
 k odd, k  1 k
sin( 2kf 0t ) As given before for the square wave
on slide 25.

   3 
Because: cos  t    sin t cos  3t     sin 3t
 2  2 
 5   7 
cos  5t    sin 5t cos  7t     sin 7t
 2   46
2 
 Fourier Transform
 For aperiodic (non-periodic) signals in time, the spectrum
consists of a continuum of frequencies (not discrete
components)
 This spectrum is defined by the Fourier Transform

 For a signal x(t) and a corresponding spectrum X(f), the

following relations hold

 
X ( f )   x(t ) e  j 2ft dt x(t )   X ( f ) e j 2ft
df Imaginary
e j
 
1
T/2   nf0  f
Forward FT (from time to frequency) Inverse FT (from frequency to time )
Real

e  j  cos  j sin  e j  cos   j sin 

 Express sin and cos e  j
 X(f) is always complex (Has both real & Imaginary
47
parts), even for x(t) real.
(non-periodic (Continuous
in time) in Frequency)

Sinc function

Sinc2 function

48
 Fourier Transform Example
 e j  e  j  x(t)
sin    
 2 j  A

 e j  e  j 
cos      2  2
 2 



 j 2ft
X(f )  x (t ) e dt

 /2
A  j 2ft  / 2 Sin (x) / x

 j 2ft
X( f )  Ae dt   e i.e. “sinc”
 /2
j 2f  / 2 function

A  e j 2f / 2  e  j 2f / 2  A  f  sin(f )   sin(f ) 

     
 
  A   
f  2j  f  1  f Area  of pulse f 
49
In time domain
 Fourier Transform Example,
contd.
sin(f )
X ( f )  A
f
Sin (x) / x
Lim x0 (sin x)/x = (cos x)x=0/1 =1 “sinc” function

First zero in the

Frequency spectrum:
A
sin f = 0 = A
f
f =  
2/ f

pulse width 
Study the effect of the50
 The narrower a function is
in one domain,
the wider its transform is
in the other domain

The Extreme Cases

51 0
 Power Spectral Density (PSD) &
Bandwidth
 Absolute bandwidth of any time-limited signal is infinite
 But luckily, most of the signal power will be
concentrated in a finite band of lower frequencies
 Power spectral density (PSD) describes the
distribution of the power content of a signal as a
function of frequency
 Effective bandwidth is the width of the spectrum
portion containing most of the total signal power
 We estimate the total signal power in the time domain

52
 Signal Power in the time domain
 Signal is specified as a function s(t) representing signal voltage or
current
 Assuming resistance R = 1 
Instantaneous signal power (t) = v(t)2/1= i(t)2*1 = |s(t)|2
 Signal power can be obtained as the average of the instantaneous
signal power over a given interval of time = constant
t2
1
PTotal 
2

t 2 t1 �x
s (t )
t1
dt (1)

 For periodic signals, this averaging is taken over one period, i.e.
T
1

2
PTotal  s ( t ) dt
T 0

 This measure in the time domain gives the total signal power
 Effective BW is then determined such that it contains a specified
portion (percentage) of this total signal power
53
Signal Power in the Frequency Domain:
Periodic signals
 For periodic signal we have a discrete spectrum (the F Series):
  (f-nf0)
C0
x(t )     Cn cos(2nf 0t   n ) 1
2 n 1 …
0 f0 2f0 3f0 f
 For a DC component, Power = Vdc2
 For AC components Power = Vrms2= 1Vpeak 2 (use eqn. 1 on prev. slide)
2
 Power spectral density (PSD) is a discrete function of
frequency: C 02 1 �
(A function
PSD 
4
� (f ) 
2

2
�C
n 1
n � (f  nf o )
of frequency)
Where (f) is the Dirac delta function:  (f )  {
1 f =0
0f � 0

 Total signal power (watts) up to the j th harmonic is:

j
1 2 1
PUpto the j th Component  C0 
4 2
 n
C 2

n 1

(A quantity, summation of PSD components- not a 54

function of a frequency)
Example
 Consider the following signal
 1 1 1 
x (t )  1sint  sin 3t  sin 5t  sin7t 
3 5 7 (No DC)
 
 The PSD is: (A function of Frequency)
1 2 12 12 1 2
PSD( f )  [1  ( f  0.5)  [ ]  ( f  1.5)  [ ]  ( f  2.5)  [ ]  ( f  3.5)]
2 3 5 7

 The signal power is: (A quantity)

1 1 1 1 
Power ]1,3,5,7 th harmonics  1      0.586 watt
2 9 25 49 

55
 Signal Power in the Frequency Domain:
Aperiodic signals Watts/Hz

 Continuous (not discrete) frequency spectrum

 PSD (Power spectrum density) function, in Watts/Hz,
is a continuous function of frequency: S(f),
 Total signal power contained in the frequency band
f1< f < f2 (in Watts) is given by:
(Integration, instead of summation, over frequency)
f2
Components exist in both negative P  2  S ( f ) df
and positive frequencies 56
f1
 Complete Fourier Analysis
 Example
Consider the half-wave rectified cosine signal, Figure
B.1 on page 793:

1. Write a mathematical expression for s(t) over its period T

2. Compute the Fourier series for s(t) (Amplitude & Phase form)
3. Get an expression for the power spectral density function for s(t)
4. Find the total power of s(t) from the time domain
5. Find the order of the highest harmonic n such that the Fourier
series for s(t) contains at least 95% of the total signal power
6. Determine the corresponding effective bandwidth for the signal

57
Example (Cont.)
1. Mathematical expression for s(t):

s (t )  { A cos(2 f o t )
0
, -T/4 �t �T/4
, T/4 �t �3T/4
T/2

Where f0 is the fundamental frequency,

f0 = (1/T)

-3T/4 -T/4 +T/4 +3T/4

58
Example (Cont.)
2. Fourier series
Before we start… what to expect?

-3T/4 -T/4 +T/4 +3T/4

• DC Component?
• Even or odd function? Sine/cosine form of the Fourier Series
• A0 ?
A0 
• An ? s (t )     An cos( 2nf 0t )  Bn sin( 2nf 0t )
2 n 1
• Bn ?
To get to the amplitude-phase form of the Fourier series,
we must first obtain the sine-cosine form 59
 Example (Cont.) T
2
A0   x(t )dt
Fourier Analysis: T 0
f0 = (1/T)
T /4 T /4
2 2A
A0 
T  s(t )dt 
T / 4
 cos(2f o t )dt
T T / 4
t T / 4
2 A  sin(2t / T )  A
      sin( / 2)  sin( / 2)
T  2 / T  t  T / 4 
A A
   sin( / 2)  sin( / 2)    2  sin( / 2)
 
2A
 , as sin( / 2)  1

DC = ?
-3T/4 -T/4 +T/4 +3T/4
60
 T
Example (Cont.) 2
An   x(t ) cos(2nf 0t )dt
T 0
f0 = (1/T)
2 Fourier Analysis
T /4 T /4
2. 2A
An  � s (t ) cos(2 nf o t )dt  �cos(2 f t ) cos(2 nf ot )dt
T (cont.):
o
T / 4
T T / 4
T /4
sin(2 (n  1)f o t ) sin(2 ( n  1)f o t ) �
2A �
 ��  � , for n �1
T � 4 (n  1)f o 4 (n  1)f o � T / 4

cos( n  / 2)  cos(n  / 2) �
A � n = 1 will be treated
 ��  �, for n �1 Separately later
 � (n  1) ( n  1) �

sin(ax  bx ) sin(ax  bx )
Note: � cos(ax ) cos(bx )dx   , and
2(a  b ) 2(a  b )
From integral )  cos( x   2 )
sin( xtables
61
Example (Cont.)
Fourier Analysis (cont.):
n1
An  0 , for n odd and n �1
A �(1) ( 1) � , for n even
(n 2) (n 2)
A n  ��  �
 �( n  1) ( n  1) �
A �
(1) (n 2)
( n  1)  ( 1)( 1) (n 2)
( n  1) �
 �� �
 � ( n  1)( n  1) �
A (1) (n 2)
 � ( n  1)  ( 1)( n  1) 
 (n  1)
2

(1 n 2 )
2A (1)
 , for n even
 (n 2  1) 62
 Example (Cont.)
Fourier Analysis (cont.):
For n = 1, A1 is obtained separately

Note: cos2 = ½(1 + cos 2)

63
 Example (Cont.) T
2
Bn   x(t ) sin( 2nf 0t )dt
T 0
Fourier Analysis (cont.):
T /4 T /4
2 2A
Bn  � s (t ) sin(2 nf o t )dt  �cos(2 f o t ) sin(2 nf o t )dt
T T / 4
T T / 4
T /4
- 2Acos(2 (n  1)f o t ) cos(2 ( n  1)f o t ) �
�
 ��  � , for n �1
T � 4 (n  1)f o 4 (n  1)f o �
T / 4

0 , for n �1

 cos(ax  bx ) cos(ax  bx )
Note: �
sin(ax ) cos(bx )dx  
2(a  b ) 2(a  b )

64
 Example (Cont.)
Fourier Analysis (cont.):
For n = 1, B1 is obtained separately
T /4 T /4
2 2A
B n 1  � s (t ) sin(2 ��
1 f o t )dt  �cos(2 f o t ) sin(2 f o t )dt
T T / 4
T T / 4
T /4
A

T �sin(4 f
T / 4
o t )dt

A A
� cos(4 f o t )  T / 4  � cos( )  cos(  ) 
T /4

4 4
0

i.e. Bn = 0 for all n (our function is even!)

65
Example (Cont.)
Fourier Analysis (cont.):

C0  A0 Cn  An2  Bn2  An , since Bn  0 for all n

66
Note: n are not required for PSD and power calculations
 Example (Cont.)
3. Power Spectral Density function (PSD):
C0 
x (t )     Cn cos(2nf 0t   n )
2 n 1

C 02 1 � 2
PSD  � (f )  � C n � (f  nf o )
4 2 n 1
A2 A2 2A 2 �
 (f  nf o )
 2 � (f ) 
 8
� (f  f o )  2

�
n  2,4,6,... ( n 2
 1) 2

n = 0 (DC) n=1 n = Even

For large n, power decays  (1/n4)… Good or bad?

67
Example (Cont.)
4. Total Power:
(From the time domain) -3T/4 -T/4 +T/4 +3T/4

3T / 4 2 T /4
1 2 A
Ps  � s (t ) dt  � � cos 2
(2 f ot )dt
T T / 4
T T / 4
T / 4 Note: cos  = ½(1 + cos 2)
2

A2 �t sin(4 f o t ) �
 ��  �
T � 2 8 f ot � T / 4

A2 = 0.25 A2

4 = Half the
Zero
power of a full
sine wave
68
Example (Cont.)
5. Finding n such that we get at least 95% of the
total power:
For n  0 (Only the DC component)
2 2 2
C 4A A
� PSD n 0 
Power  0
 2  0.1014A 2

4 4 2

2
0.1014A
� Power %  2
 40.5%
0.25A
% of total
power in this
component
69
Example (Cont.)
Finding n such that we get at least 95% of the
total power, contd.:

For n  1 (DC + first harmonic)

C 02 C 12 A 2 A 2
PSD n 1 
�Power   2   0.226A 2
4 2  8
2
0.226A
� Power %  2
 90.5%
0.25A
% of total power
in these two
components
70
 Example (Cont.)
Finding n such that we get at least 95% of the
total power, Contd.:
For n  2 (DC + first harmonic + second harmonic)
C 02 C 12 C 22 A 2 A 2 2A 2
PSD n  2 
� Power    2    0.2485A 2

4 2 2  8 9 2
0.2485A 2
� Power %  2
 99.41% OK!  95%
0.25A

n = 2, and
Beff
6. the effective bandwidth is:
…
Beff = fmax – fmin
0 f0 2f0 3f0 f
Beff = 2f0 – 0 = 2f0 71
DC
 Bandwidth about a Center

Frequency
So far we have considered signals in their base band form
(without modulation)
 Data is often sent as variations in a high frequency carrier
signal having a frequency fc (modulation)
 So, bandwidth (BW) of this signal occupies a range of
frequencies centered about fc
Carrier With Amplitude Modulation,
BW For each component of the modulating signal:
Modulating
Carrier Signal 0 fc f

 2The
cos(2f c t ) cos(
larger 2f mlarger
f , the t )  cos[2BW
the f m )t ]  cos[2 ( f c  f m )t ]
( f c obtainable
c

 Largest BW obtainable for a given center frequency fc is 2 fc

72
 Analog and Digital Data
WK5
Transmission
 Data
 Entities that convey meaning
 Signals
 Electric or electromagnetic representations of data

 Data Transmission
 Communication of data
through propagation and processing of signals
that represent them

73
 Data types in nature: Analog and
 Digital Data
Analog Data
 Continuous values within some interval
 Examples: audio, video
 Typical bandwidths:
 Speech: 100Hz to 7kHz
 Voice over telephone: 300Hz to 3400Hz
 Video: 4MHz
 Digital Data
 Discrete values (not necessarily binary)
 Examples: integers, text characters, mixture:
2347, “text”, SDR054
74
Analog and Digital Signals
 Means by which data get transmitted over
various media, e.g. wire, fiber optic, space

 Analog signal:
 Continuously variable in time and amplitude

 Digital signal:
 Uses a few (two or more) DC levels

75
 Analog Signal Example 1:
Speech Data
 Frequency range for human hearing: 20Hz-20kHz
 Almost fully utilized by music
 Human speech: 100Hz-7kHz
 Telephone voice channel: Spectrum is further limited to 300-
3400Hz (why?)
 Mechanical sound waves (data) are easily converted into
electromagnetic signal for processing and transmission:
 Mechanical waves (Sound) of varying pitch and loudness (Data)
is represented as:
Electromagnetic signals of different frequencies and amplitudes
(Signal)

76
Analog Example: 1. The Acoustic
Spectrum
Dynamic range of the human ear
can be as high as 120 dBs!

dBs

Dynamic Range of Signal Power

Source Data

Hearing Spectrum

SPEECH

Frequency Range

-70

77
Log Scale
Conventional Telephony:
Analog data – Analog Signal
 Telephone mouthpiece converts mechanical voice
analog data into electromagnetic analog electrical
signal
 Signal travels on telephone lines
 At receiver, speaker re-converts received electrical
signal to voice

78
 Analog Signal Example 2.
Video Data
 Electrical signal proportional to the brightness of

image spot on a raster-scanned phosphor screen

Interlaced Scan

11 s 52.5 s (Active)

Line Scan

Frame Scan
79
 Bandwidth of a Black & White
 Video
USA Signal
Specification: 525 lines per frame scanned at the rate
of 30 frames per second
 525 lines = 483 active scan lines + 42 lost during vertical retrace
 So 525 lines x 30 frames/second = 15750 lines per second
 Line scan interval = 1/15750 = 63.5s
 11s go for horizontal retrace, so 52.5 s for active video per line
 Effective vertical resolution = 0.7 x 483 = 338 lines
 Horizontal resolution = 338 x aspect ratio
= 338 x (4/3) = 450 dots
 Max frequency is when black and white dots alternate
 450 picture dots correspond to 225 cycles in 52.5 s 
Time period = 52.5/225 s  fmax = 1/Period = 4.2 MHz
 fmin (DC) = 0  Bandwidth = fmax - fmin = 4.2 MHz
80
 Digital Signals
Advantages:
 Cheaper and easier to generate: No extra processing
needed
 Less susceptible to noise
(The threshold effect)

Disadvantages:
 When noise is above threshold  Total data reversal (Bit
error) (1 0, 0 1)
 Greater attenuation
 Line capacitances make pulses rounded and smaller in amplitude,
leading to loss of information
 More so at higher data rates and longer distances
 So, use at low data rates over short distances
81
Attenuation of Digital Signals

1111...

0000...

Pulse shaping Worse at higher

Due to line capacitances: Effect of line
capacitances data rates (narrower pulses)
Worse over longer distances

82
Digital Binary Signal
 Example: Between keyboard and computer
 Two bipolar dc levels (+ and – : Why?)
 Bandwidth required depends on the signal
frequency, which depends on:
 The data rate (bps) and
 The actual data sequence transmitted

_
Data

-
-Data rate = ?
1
- Maximum f = ? Data element
- Minimum f = ? Signal
83
 Data and Signal
 combinations
We have seen above: (data and signal of same type)
 Analog signals carrying analog data: Telephony, Video
 Digital signals carrying digital data: Keyboard to PC
 Simple- one only needs a transducer/transceiver
 But we may also have: (data and signal of different types)
 Analog signal representing digital data: Data over
telephone wires (using a modem)
 Digital signal representing analog data: CD Audio, PCM
(pulse code modulation) (using a codec)
 More complex- We Need a converter
 So, all the four data-signal combinations are
possible!
84
Analog Signals can carry
Analog Data or Digital Data

(Base band)
i.e. in its original
(Transducer) form

(Converter)
We need a converter when the signal type is different from 85
the data type
Digital Signals can carry
Analog Data or Digital Data
Digitized Analog Samples

e.g. using PCM

(Pulse Code Modulation)
Coder-Decoder
(Converter)

Transmitter-Receiver

We need a converter when the signal type is different from the data type
86
 Four Data/Signal Combinations
Signal
Analog Digital
Two ways: Signal has:
Analog - Same spectrum as Use a (converter):
data (base band): e.g. codec, e.g. for PCM
Telephony to exchange (pulse code modulation)
Data - Different spectrum
(through modulation):
e.g. AM Radio, FDM

Simple two signal

-

Digital Use a (converter): levels: e.g. NRZ code

modem e.g. the V.90 - Special Encoding:
standard e.g. Manchester
code (Chapter
87 5)
 Two Modes of Transmitting
Signals:
1. The Analog Mode (associated
 Treats the signal as “analog” regardless of what it represents
(Not interested in the data content of signal)
 with
FollowingFDM)
attenuation over distance, signal level is boosted
using “amplifiers”
 Unfortunately, this also amplifies in-band noise
 With cascaded amplifiers (i.e. one after the other at locations
along the link), effect on noise and distortion is cumulative,
i.e. they get amplified again and again
 Effect of noise and distortion on analog systems may be
tolerated, e.g. with telephony you can still manage to get it!
(Humans are good at filling-in gaps!)
 But digital systems are more sensitive to the effects of
excessive noise and distortion  unacceptable errors
 So… Do not transmit digital signals the analog 88 way!
 Two Modes of Transmitting
Signals:
 2. The Digital
Concerned with the dataMode
content of(Associated
the signal
 with TDM)
It assumes that the signal carries digital data
 Uses “repeaters” (not amplifiers), which:
 Receive the signal
 Extract the data bit stream from it
 Retransmit a fresh, strong signal representing the
extracted bit stream
 This way:
 Effect of attenuation is overcome
 Noise and distortion are not cumulative
89
Four Signal/Transmission Mode
Combinations Transmission mode
Which transmission
mode is more versatile
Analog Digital
and useful for integrating - Uses amplifiers -Uses repeaters
different signal types? - Not concerned with - Assumes signal represents

what data the signal digital data, recovers this data

represents and represents it as a new
- Noise and distortion
outbound signal
FDM: Frequency Division
- This way, noise and distortion
Multiplexing are cumulative
TDM: Time Division -- Associated with FDM
are not cumulative
Multiplexing - Associated with TDM

Makes sense only if the analog

signal represents digital data!
Analog OK (Ask yourself: What data is the
Signal repeater going to extract?!)

Digital Avoid OK
90
 Advantages of Digital Mode of
 Transmission
Use of digital technology
 Lower cost, smaller size, and high speed VLSI technology
 Higher data integrity (reliability) as noise effects are not cumulative (fresh
signal restoration en-route)
 Cover longer distances, at higher data rate, at low error rates, over lower

quality lines:
 Easier to implement multiplexing for improved utilization of link capacity
 High bandwidth links are now economical (Fiber, Satellite…)

 To utilize them efficiently we need to do a lot of multiplexing

 This is done more efficiently using digital (TDM) rather than analog (FDM)

(Chapter 8)
 Encryption for data security Frequency
Time Division Division
and confidentiality is digital Multiplexing Multiplexing
 Easier to integrate different data types
 Convert analog data to digital signals…and use one system to handle all

voice, video, and data, e.g. one network for all types of traffic
91
Transmission Impairments
 Signal received is often a degraded form of the signal
transmitted
 Why? What happens en-route?... Impairments:
 Attenuation:
 Limits the bandwidth of the received signal
 In-band signals arrive weaker
 Attenuation distortion (Attenuation is not uniform over bandwidth)
 Delay
 Delay distortion
 Noise and interference (including crosstalk)
 Effect:
 On analog data - Some degradation in signal quality
 On digital data – Fatal bit errors (total bit reversals)
92
 Attenuation
 Signal strength falls off with distance traveled
 Nature of loss in signal power depends on medium:
 Guided (Wires, etc.): Signal power after traveling distance d
 Exponential drop is signal power with distance: Pd = P0 e-d
10 ln (Pd/P0) = -d
10 log (Pd/P0) = -’d
 Loss: ’ dBs per km (’ depends on medium type e.g. fiber,
twisted pair, cable)
 Unguided (Open space):
 Inverse square law spread with distance: P  P0 /d2
 Loss: 6 dBs for each distance doubling
 Absorption, scattering
 May also depend on weather, e.g. rain, sunspots,
93
 Effects of Attenuation
 Received signal strength must be:
 Sufficiently Large enough to be detected
 Sufficiently higher than noise to be interpreted correctly
(without error)
 To overcome these problems:
 Use amplifiers (analog transmission mode)
or repeaters (digital transmission mode) en-route
 Amplifier gains should not be too large as this may
cause signal distortion due to saturation (nonlinearities)
 Problem with networks: distance actually traveled
(hence attenuation) will depend on actual route taken
through the network!
94
Attenuation Distortion
 Attenuation usually increases with frequency
 This causes bandwidth limitation (understood)
 Moreover, over the transmitted bandwidth itself:
 Different frequency components of the signal get
attenuated differently  Signal distortion
 Affects analog signals more
 To overcome this problem:
 Use Equalizers that reverse the effect of frequency-
dependent attenuation distortion:
 Passive: e.g. loading coils in telephone circuits
 Active: Amplifier gain designed specifically for this purpose
95
Attenuation Distortion
Equalization
To Reduce
Attenuation
Distortion

Q. What is
the signal ?

96
Delay Distortion
 Happens only on guided media
 Wave propagation velocity varies with frequency:
 Highest at the center frequency (minimum delay)
 Lower at both ends of the bandwidth (larger delay)
 Effect: Different frequency components of the signal
arrive at slightly different times! (Dispersion in time)
 Affects digital data more: due to bit spill-over
(timing is more critical here than for analog data)
 Again, equalization can help overcome the problem

97
Delay Distortion
Equalization
To Reduce
Without Delay
Equalizer Distortion

With
Equalizer

98
 Noise (1)
 Definition: Any additional unwanted signal inserted
between transmitter and receiver
 The most limiting factor in communication systems
 Noise Types:
 Thermal Noise
 Inter-modulation Noise
 Crosstalk Noise
 Impulse Noise

99
Noise (2)
PSD

 Thermal (White) Noise

 Due to thermal agitation of electrons
(Increases with temperature) f

 Uniformly distributed over frequency (White noise)

 Difficult to eliminate
(exists even in the same bandwidth as your signal!)
 Effect is more significant on weak received signals,
e.g. from satellites

100
Thermal Noise, Contd.
 Thermal noise power density in 1 Hz of bandwidth, N0
(Constant, Independent of frequency): PSD
N 0  kT ( W / Hz) N0

 k Boltzmann’s constant = 1.3810-23 J/K B

 T temperature in degrees Kelvin (= 273 + t C) 1 Hz f

 Thermal noise power in a bandwidth of B Hz:

Can you see some
10 log k disadvantage now
in having a larger
BW?

Example: at t = 21 C (T = 294 K) and for a bandwidth of 10 MHz:

N = -228.6 + 10 log 294 + 10 log 107

= - 133.9 dBW 101

WK 6
Noise (3)
 Inter-modulation Noise
 Signals having the sum and difference (frequency mixing)
of original frequencies sharing a transmission system
(e.g. in FDM systems)
f1, f2  (f1+f2) and (f1-f2)
 Caused by nonlinearities in the medium and equipment,
e.g. due to overdrive and saturation of amplifiers
 Danger: Resulting new frequency components may fall
within valid signal bands, thus causing interference
Output

A cos 1 + B cos 2 Linear System A’ cos 1 + B’ cos 2

K(A cos 1 + B cos 2)

A’ cos 1 + B’ cos 2
Input
A cos 1 + B cos 2 Non-Linear System + f(21)+f(22)+f(1-2)+f(1+2)

K(A cos 1 + B cos 2) + K(A cos 1 + B cos 2)2 Inter-modulation

Input components
102
New spurious components can fall within genuine signal bands causing interference
Noise (4)
 Crosstalk Noise
 A signal from one channel picked up by another channel in
close proximity
 Examples:
 Physical proximity: coupling between adjacent twisted pair
channels
 Shield cables properly
 Directional proximity: antenna pick up from other directions
 Use directional antennas
 Spectral proximity: leakage between adjacent channels in
frequency division multiplexing (FDM) systems
 Use guard bands between adjacent channels

103
Noise (5)
 Impulse Noise
 Pulses (spikes) of irregular shape and high amplitude lasting
short durations
 Causes: External electromagnetic interference due to
switching large currents, car ignition, lightning, …
 Minor effect on analog signals (e.g. crackling noise in voice
channels)
 Major effect on digital signals- Bit reversal error!
 More damage at higher data rates
(a noise pulse of a given width can destroy a larger block of
bits)

104
 Effect of Impulse Noise on a
Digital Signal

+
Impulse

RX

Q: What is the effect of the same noise at 10 times the data rate?
105
 Channel Capacity
 Channel capacity: Maximum data rate usable under a
given set of communication conditions
 How channel BW (B), signal level, noise and impairments,
and the amount of data error that can be tolerated limit the
channel capacity?
 In general, Max possible data rate, C, on a given channel
= Function (B, Signal wrt noise, Bit error rate allowed)
 Max data rate: Max rate at which data can be communicated on the
channel, bits per second (bps)
 Bandwidth: BW of the transmitted signal as constrained by the
transmission system, cycles per second (Hz)
 Signal relative to Noise, SNR = signal power/noise power ratio
(Higher SNR  better communication conditions  higher C)
 Bit error rate (BER) allowed: in (bits received in error)/(total bits
transmitted). Equal to the bit error probability.
106
e.g. Higher allowed  higher usable data rates  higher C
 Channel Capacity, C:
 So, in general: C bps = F(B, SNR, BER)
 Three Formulations under different assumptions:
Idealistic

Assumptions Formulation
Ideal: Noise-free, Error-free: C = F(B) Nyquist

Noisy, Error-free: C = F(B, SNR) Shannon

Practical: Noisy, Error: C = F(B, SNR, BER) Eb/N0 Vs
Error Rate
Realistic

107
 Bandwidth (or Spectral)
Efficiency (BE):
Channel Capacity C
BE  , bps / Hz
Bandwidth B
 Measures how well we are utilizing a given bandwidth
to send data at a high rate….
 Can be greater than 1 (not like engineering efficiencies)
 The larger the better

108
1. Nyquist Channel capacity: (Noise-free,
Error-free)
 Idealized, theoretical
 Assumes a noise-free  error-free channel
 Nyquist showed that (without noise, without errors): If rate of signal transmission
is 2B then a signal with frequency components up to B Hz is sufficient to carry
that signalling rate
 In other words: Given bandwidth B, highest signalling rate possible is 2B signal
elements/s
 How much data rate does this represent?
(depends on how many bits are represented by each signal element!)
 Given a binary signal (1,0), data rate is same as signal rate 
Data rate supported by a BW of B Hz is 2B bps  C = 2B
 For the same B, data rate can be increased by sending one of M
different signals (symbols): as each signal level now represents
log2M bits
 Generalized Nyquist Channel Capacity, C = 2B log2M bits/s (bps)

 Bandwidth efficiency = C/B = 2 log2M bits/signal

(bits/s)/Hz : Dimensionless quantity Signals/s 109
 Nyquist Bandwidth: Example
 C = 2B log2M bits/s
 C = Nyquist Channel Capacity Signal
Element
 B = Bandwidth
1
 M = Number of discrete signal levels (symbols) used
 Data on telephone Channel:
B = 3400-300 = 3100 Hz
 With a binary signal (M = 2 symbols, e.g. 2 amplitudes): 0
C = 2B log2 2 = 2B x 1 = 6200 bps 11
 With a quadnary signal (M = 4 symbols): 10

C = 2B log2 4 = 2B x 2 = 4B = 12,400 bps 01 2 bits/Symbol

i.e. 2 bits /signal
 Channel capacity increased, but 00 element
disadvantage: Larger number of signal levels (M) makes
it more difficult for the receiver to determine data correctly
in the presence of noise 110
 2. Shannon Capacity Formula: (Noisy,
Error-Free)
 Highest error-free data rate in the presence of noise
 Signal to noise ratio SNR = signal / noise levels
SNRdB= 10 log10 (SNR ratio)
 Errors are less likely with lower noise (larger SNR ratios).
This allows higher error-free data rates i.e. larger
Shannon channel capacities
Caution! Log2 Not Log10

 Shannon Capacity C = B log2(1+SNR):

Caution! Ratio- Not dBs
Highest data rate transmitted error-free with a given noise level
 For a given BW, the larger the SNR the higher the data
rate I can use without introducing errors
 C/B: Spectral (bandwidth) efficiency, BE, (bps/Hz) (>1)
 Larger BEs mean better utilization of a given bandwidth
B for transmitting data fast. 111
 Shannon Capacity Formula:
 Comments
Formula says: for data rates  calculated C, it is
theoretically possible to find an encoding scheme
that achieves error-free transmission at the given
SNR… But it does not say how!
Also:
 It is a theoretical approach based on thermal (white)
noise only. But in practice, we also have impulse
noise, attenuation and delay distortions, etc…
 So, maximum error-free data rates measured in practice
are expected to be lower than the C predicted by the
Shannon formula due to the greater noise
 However, maximum error-free data rates can be
used to compare practical systems: The higher that
rate the better the system… 112
Shannon Capacity Formula:
Comments Contd.
Formula suggests that changes in B and SNR can
be done arbitrarily and independently… but
 In practice, this may not be the case!
 Higher SNR obtained through excessive

amplification may also introduce nonlinearities 

increased distortion and inter-modulation noise …
which reduces SNR!
 High Bandwidth B opens the system up for more

thermal noise (kTB), and therefore reduces SNR!

113
 Shannon Capacity Formula:
Example
 Spectrum of communication channel extends from 3 MHz to 4 MHz
 SNR = 24dB
 Then B = 4MHz – 3MHz = 1MHz
SNRdB = 24dB = 10 log10 (SNR)
SNR (ratio) = log-110 (24/10) = 1024/10 = 251
 Using Shannon’s formula: C = B log2 (1+ SNR)
C = 106 * log2(1+251) ~ 106 * 8 = 8 Mbps
 Based on Nyquist’s formula, determine M that gives the above
channel capacity:
C = 2B log2 M
8 * 106 = 2 * (106) * log2 M
4 = log2 M
M = 16 114
 3. Eb/N0 Vs Error Rate Formulation
(Noise and Error are both specified

Together)
Handling both noise and a quantified error rate
simultaneously
 We introduce Eb/N0: A standard quality measure of three
channel parameters (B, SNR, R) and can also be
independently related to the error rate
R is the data rate. Max value of R is the channel capacity C
 It expresses SNR in a manner related to the data rate, R
 Eb = Signal energy in one bit interval (Joules)
T = 1/R
= Signal power (Watts) x bit interval Tb (second) b

= S x (1/R) = S/R = SNR/BE

 N0 = Noise power (watts) in 1 Hz = kT. Two formulations:
E b ST b S / R S E b S / R S BT �BT �
      SNR � �
N0 N0 kT kTR N0 N0 N115R �R �
 Eb/N0 (Cont.) BER vs Eb/N0 curve for a given encoding scheme

Lower Error Rate: larger E b/N 0

 Bit error rate for digital data is a B
decreasing function of Eb/N0 for a A
Better
given signal encoding scheme
Encoding

 Analysis:  For a given system

(SNR, B, R)  (Eb/N0), determine
error rate BER

 Design:  Given a desired error

rate BER, get Eb/N0 to achieve it,
then determine other parameters
from formula, e.g. S, SNR, R, etc.
�E b �
� �  S dBW  10 log R  10log k  10 logT
 Effect of S, R, T on error �N 0 �
dB
performance  S dBW  10 log R  228.6 dBW  10logT

 Which encoding scheme Eb S /R S BT �BT � SNR

   SNR � �= BE
is better: A or B? N0 N0 N R �R �
116
Max R = C, BE = C/B
 Example:
�E b �
� �  S dBW  10 log R  10 log k  10 logT
�N 0 �
dB

 S dBW  10 log R  228.6 dBW  10 logT

 Given:
 The effective noise temperature, T, is 290 oK

 The data rate, R, is 2400 bps

 Would like to operate with a bit error rate of 10-4 (e.g. 1 error in 104 bits)

What is the minimum signal level required for the received signal?

 From curve, a minimum Eb/No needed to achieve a bit error rate of 10-4 =
8.4 dB
 8.4 = S(dBW) – 10 log 2400 + 228.6 dBW – 10 log290
= S(dBW) – (10)(3.38) + 228.6 – (10)(2.46)

S = -161.8 dBW Design or Analysis?

117
Eb/N0 in terms of BE, assuming
Shannon channel capacity
 From Shannon’s formula:

C = B log2(1+SNR)

We have:
SNR  (2C / B  1)  (2 BE  1)
 From the Eb/N0 formula:

Eb SNR 1
  ( 2 BE  1)
N0 BE BE
C/B (bps/Hz) is the spectral (bandwidth) efficiency BE based on
Shannon channel capacity 118
Example

 Find the minimum Eb/N0 required to achieve a

Shannon bandwidth efficiency (BE=CShannon/B) of
6 bps/Hz:
Eb 1
 ( 2 BE  1)
N 0 BE
 Substituting in the equation above:
Eb/N0 = (1/6) (26 - 1) = 10.5 = 10.21 dB

119