Professional Documents
Culture Documents
Chulantha Kulasekere
Department of Electronic and Telecommunication Engineering
University of Moratuwa
1 / 70
Course Outline
Course Aims
2 / 70
Course Outline
Course Content
3 / 70
Course Outline
Reading Material
Main Text:
Data Communications, Computer Networks and Open
Systems/4e Fred Halsall, Pearson Education Inc. 1996.
Supplementary Text:
Communication Networks: Fundamental concepts and key
architectures/2e, Alberto Leon-Garcia and Indra Widjaja, Tata
McGraw-Hill 2004.
4 / 70
Course Outline
Grading Policy
5 / 70
Network Background
We are concerned with the issues that have to be considered
when two or more computers communicate with each other.
Computer A
Computer B
AP
User-to-user communication
AP
Communication
subsystem
Computer-to-computer
Communication
subsystem
communication
Computer-to-network communication
6 / 70
Our Concentration
Application A
Application B
Application
layer
Application
layer
Presentation
layer
Presentation
layer
Session
layer
Session
layer
Transport
layer
Communication network
Transport
layer
Network
layer
Network
layer
Network
layer
Network
layer
Data link
layer
Data Link
layer
Data Link
layer
Data link
layer
Physical
layer
Physical
layer
Physical
layer
Physical
layer
Communication Networks
Figure 2.4
7 / 70
Application B
Data
Application
layer
Application
layer
ah
Presentation
layer
Presentation
layer
ph
Session
layer
Transport
layer
Physical
layer
Leon-Garcia/Widjaja
Transport
layer
th
Network
layer
Data link
layer
Session
layer
sh
Network
layer
nh
dt
dh
Bits
Communication Networks
Data link
layer
Physical
layer
Figure 2.7
8 / 70
The Justification
Signals converted to electrical signals travel in the physical layer
as a bit stream.
The transmitted signals undergo attenuation and distortion. This
can be acceptable untill the receiver is unable to discriminate
between the binary 1 and 0 in the bit stream.
The above is influenced by
Transmission medium.
Bit rate of transmission.
Physical separation of the source and destination.
9 / 70
t = d/c
z of
Amplitude
response & is
bandwidth
The speed
propagation
dependent on the relative permittivity
z dependence on distance
of the channel
(v
=
c/
).
It
is 2.3E8 m/s for Cu and 2E8 m/s for
z Susceptibility to noise & interference
optical fiber.z Error rates & SNRs
10 / 70
11 / 70
12 / 70
13 / 70
26 gauge
30
24 gauge
27
Attenuation (dB/mi)
24
22 gauge
21
18
19 gauge
15
12
9
6
3
f (kHz)
1
Copyright 2000 The McGraw Hill Companies
10
100
Leon-Garcia & Widjaja: Communication Networks
1000
Figure 3.37
14 / 70
15 / 70
16 / 70
Attenuation (dB/km)
35
0.7/2.9 mm
30
25
1.2/4.4 mm
20
15
2.6/9.5 mm
10
5
0.01
0.1
1.0
10
100
f (MHz)
Figure 3.40
17 / 70
18 / 70
100
50
10
Loss (dB/km)
Infrared absorption
1
0.5
Rayleigh scattering
0.1
0.05
0.01
0.8
1.0
1.2
1.4
1.6
1.8
Wavelength (m)
19 / 70
20 / 70
21 / 70
Low frequency radio waves can also be used for replacement for
guided media.
Wireless distributed connections.
Usable data rate is generally lower than the fixed wiring system.
22 / 70
Assignment
Use the attenuation vs. frequency graphs for twisted pair and
coaxial cables and explain why the bandwidth decreases with
distance.
Suppose we wish to delay an optical signal by 1ns. How long a
length of optical fiber is needed to do this? How much is the signal
attenuated? Repeat for 1ms.
23 / 70
Transmitted data
+V
Time
Transmitted signal
V
Attenuation
Limited bandwidth
Attenuation and
distortion effects
Delay distortion
Line noise
Chulantha Kulasekere (ENT)
24 / 70
Attenuation
Delay distortion
Line noise
Combined effect
Received signal
Sampling signal
Received data
Bit error
25 / 70
Attenuation
Decrease of amplitude oer a distance during transmission is known
as attenuation.
Both hardware and media will determine the correct transfer of
information. Repeaters/amplifiers may have to be used.
Signal attenuation increases as a function of frequency. Hence a
signal with a large frequency content will be distorted due to
disparate nature of attenuation for each frequency.
Quantitatively
Attenuation = 10 log10
26 / 70
Limited Bandwidth
Each type of cable used for data transmission has a particular
bandwidth associated with it. Generally one has to choose the
appropriate type of medium depending on the maximum
frequency of transmission.
Fourier analysis can be used to decompose a binary waveform
into its frequencies.
Even though the binary waveform can consists of randomly
varying sequences, the worst case sequence of 101010... is
considered for computations.
Example: If a signal 500bps is transmitted over a communication
channel, what is the minimum bandwidth required if it should not
attenuate up to the fifth harmonic.
Definitions: The absolute BW of a signal is the width of the
spectrum while the effective BW is the band within which most of
the signal energy is concentrated.
Chulantha Kulasekere (ENT)
27 / 70
1
1
cos 6f0 + cos 10f0 ...}
3
5
28 / 70
29 / 70
30 / 70
Delay Distortion
Similar to amplitude distortion, the delay experienced by different
frequencies vary in a channel. Hence delay distortion occurs at the
receiver.
Can be reduced by using an equalization process.
For higher bit rates the delay distortion increases leading to inter
symbol interference and thus limits the bit rate. Measured using eye
diagrams.
The transmission delay is different from the propagation delay. For
example if the input signal x(t) to a channel with amplitude
response A(kf0 ) and phase response (kf0 ) is given by
X
x(t) =
ak cos(2kf0 t).
The output from channel
X
y (t) =
ak A(kf0 ) cos(2kf0 t + (kf0 )).
If A(kf0 ) is constant no amplitude distortion will occur. If (kf0 ) is
constant no delay distortion will occur.
Chulantha Kulasekere (ENT)
31 / 70
tprop
ttr
32 / 70
33 / 70
34 / 70
Assignment
35 / 70
Noise Analysis
In general we look at the following quantity. It is important to find the
minimum signal level relative to the noise level. This is
characterized by bit error rate (BER); The Energy measured in joule
per bit in a signal, Eb is given by Eb = STb where Tb = R 1 .
The level of thermal noise in a BW of 1Hz is given by N0 = kT
watts/Hz; This is a noise power density where k = 1.3803e 3
Joule/K (Boltzmann constant) and T the temperature in kelvin.
By normalizing
Eb
S
= 10 log10 (S/R) 10 log10 (kT ) dB
=
N0
RkT
Explain the above w.r.t parameters and actual communication
frequency.
36 / 70
37 / 70
38 / 70
2W samples / sec
Analog
source
Sampling
(A/D)
Quantization
Original x(t)
Bandwidth W
2W m bits/sec
Transmission
or storage
Approximation y(t)
Display
or
playout
Interpolation
filter
Pulse
generator
2W samples / sec
Figure 3.20
Leon-Garcia/Widjaja
Communication Networks
39 / 70
40 / 70
Quantizer Characteristics
Uniform
quantizer
3.5
output y(nT)
2.5
1.5
0.5
4 3 2
-0.5
-1.5
input x(nT)
-2.5
-3.5
3.5
2.5
1.5
0.5
-0.5
-1.5
-2.5
-3.5
ChulanthaLeon-Garcia/Widjaja
Kulasekere (ENT)
Communication
Networks
EN 5350 Data
Communications
Quantization
The issue with quantization is you have to compromise between
M = 2m with m levels and the corresponding bit rate 2Wm.
For uniform quantization of the range V to V = V /2m1 .
The quantization error is in the range [/2, /2]. This is for the
round function, for twos complement representation etc, this range
can change.
The quantizer SNR is given by
SNR =
42 / 70
= 2V/M
error = y(nT)-x(nT)=e(nT)
...
-V
e2
input
...
x(nT)
Communication Networks
1
e2 =
Figure 3.22
/2
2
x 2 dx =
12
/2
43 / 70
Quantizer SNR
The previous equation is written as
SNR dB = 6m + 10 log10 3
2
x
44 / 70
45 / 70
vi
vi'
Linear
ADC
Network
vi'
(b)
11
Segment
codes
(+)
Linear
quantization
intervals
10
01
00
Linear
DAC
Expander
circuit
vo '
vo
+V '
11
10
01
00
11
10
01
00
11
10
01
00
11
10
01
00
00
01
10
11
00
01
10
11
00
01
10
11
00
01
10
11
00
01
10
+V
vi
Segment
codes
()
11
V '
(c)
vo '
+V '
46 / 70
00
01
10
11
00
01
10
11
10
()
V '
vo '
(c)
+V '
vo
+V
Quantization codes
Segment codes
Polarity bit
V '
Figure 2.25
Chulantha Kulasekere (ENT)
47 / 70
48 / 70
bits
samples
16
2 channels = 1.4 Mbps.
seconds
sample
If we sample more than the Nyquist rate the bit rates are even
higher than this. For example surround sound will require 4 times
as more bit rates.
Now we have to use compression techniques to get the same
information across in lesser time.
Chulantha Kulasekere (ENT)
49 / 70
Consequences of Limited BW
50 / 70
vASK (t)
Bandpass
filter
vd (t)
(Data signal)
Lowpass
filter
PSTN
vc(t)
(Carrier signal)
v'd (t)
vc(t)
(b)
1
1
vd (t)
Time, t
0
1
vc(t)
1
1
vASK (t)
0
1
(c)
6f0
The amplitude
of a single audio tone
is switched between two
2f
Signal
power
levels.
0
51 / 70
ASK ...
1
3
cos 30 t +
1
5
cos 50 t . . .} we have
1
1
cos(c 30 )t cos(c + 30 )t + . . .}
3
3
We see that how close the signal representation to the original
signal is dependent on the number of sidebands that we are
willing to accommodate in the channel.
52 / 70
ASK ...
6f0
(c)
2f0
Signal
power
Frequency
fc 3f0
fc f0
fc
f c + f0
fc + 3 f0
We also note that the carried signal is present even though it does
not have any information.
53 / 70
ASK ...
To transmit the worst case bit rate (it has a fundamental frequency
f0 which leads to a bit rate of 2f0 bps) we require 2f0 .
Conversely, assuming that the BW is 2f0 , the Nyquist rate of
transmission is twice what is computed above.
To avoid duplicity of information (f0 ) we can use a BPF to isolate
only one side band.
Then the BW is f0 and the Nyquist rate is now satisfied.
However the signal side band power is reduced by half. This
power reduction in turn reduces the S/N ratio and hence results in
an increased BER ratio.
If more than the fundamental is received, will the system violate
the Nyquist rate? What are the advantages if any?
54 / 70
55 / 70
(a)
1
Data
signal
vd (t)
1
Time, t
0
+1
Carrier 1
v1(t)
0
1
+1
Carrier 2
v2(t)
0
1
vFSK (t)
(b)
6f0
6f0
2f
Here the difference
between 2fthe two carriers
play a role in
Signal
power
isolating spectra.
56 / 70
FSK ...
Explain the significance of VFSK (t) and the two signals vd (t) and
0
vd (t).
To understand the principle we have to plot the following spectra.
(1) Signal, (2) carrier, (3) modulated signal (4) demodulated.
What is the significance of the frequency shift fs and how is it
chosen?
The BW of the FSK signal is f2 f1 + 2f0 for the first harmonic.
(Halsall Figure 2.19[c])
Another way of looking at this is considering that each carried is
modulating half the bit rate (1s or 0s). That is it has two ASK
modulators each with a carried frequency and one of them will be
active half the time.
At the receiver, the frequency of the signal is identified using the
ASK demodulator technique and hence will be getting data half
the time.
Chulantha Kulasekere (ENT)
57 / 70
FSK ...
vFSK (t)
6f0
(b)
Signal
power
6f0
2f0
2f0
Frequency
0
f13f0 f1f0 f1+ f0 f1+3 f0 f23f0 f2f0 f2+ f0 f2+3 f0
f2
f1
fs
1
f0 = Fundamental frequency component = 1/4 bit rate (Hz)
0
fs = Frequency shift
Each ASK component will have half the original bit rate. Hence
the fundamental for each component will be half of what we used
2.19
for Figure
the ASK
in the previous section.
58 / 70
FSK ...
For each ASK component you need 2f0 BW and in total you will
require 4f0 plus the separation of the two frequencies as BW for
FSK.
However since the fundamental frequency in this case is half of
that used in the original ASK system, the BW requirement is still 2
times the fundamental frequency plus the frequency shift.
In terms of BW, ASK and FSK are almost the same. However in
terms of noise performance, FSK is better.
eg. a signal of 600 bps over analog PSTN.
59 / 70
60 / 70
What are the general goals of digital modulation? Why, typically is,
digital modulation not enough for radio transmission?
In general the goal of digital modulation is to translate the digital
data into an analog signal (baseband) suitable for transmission.
However another analog modulation scheme will be applied
during radio transmission to shift the center frequency to the radio
carrier. The radio carrier is required so that at that frequency the
antenna sizes (/4) becomes manageable and so that the signal
can be conditioned to suit the medium characteristics such as
path loss,reflections etc.
61 / 70
Question
A B&W TV picture may be considered as composed of
approximately 3E5 picture elements. Assume that each picture
element is equi probable among 10 distinguishable brightness
levels. If 30 frames/sec are transmitted. Calculate the minimum
BW required to transmit the video signal assuming that a 30dB
S/N is necessary for satisfactory picture reproduction.
Note that a commercial TV transmission actually uses a video BW
of approximately 4 MHz.
If its for HDTV, what would the minimum BW be?
Can this bit rate be transmitted over a digital communication link
and how?
Can coding techniques increase BW utilization?
Also try Example 2.7 and its interpretation.
Chulantha Kulasekere (ENT)
62 / 70
(a)
+1
Data
signal
vd (t)
1
Time, t
0
1
+1
Carrier
vc(t)
0
1
Phase
coherent
vPSK (t)
180
Differential
180
180
180
v'PSK (t)
90
(b)
Signal
power
270
270
90
90
270
6f0
2f0
63 / 70
PSK ...
90
270
270
(b)
90
90
270
6f0
2f0
Signal
power
Frequency
fc 3f0
fc f0 fc fc + f0
fc + 3 f0
(c)
180= 0
0= 1
I (In phase)
Phase coherent PSK uses the same carrier with 180 degree
shift between binary 1 and 0. A reference carrier signal
should be at the receiver to match against the incoming phase
disadvantageous.
Figure
2.21
phase
64 / 70
PSK ...
Differential PSK (DPSK) is better since the detection is only a
phase transition of 900 relative to the current indicates a 0 in the
next bit and a phase transition of 2700 relative to the current
indicates a 1 in the next bit. Only the magnitude of the shift in
phase is required and not the absolute value as in the previous
case
A bipolar periodic data stream can be represented by
vd (t) =
1
1
4
(cos 0 t cos 30 t + cos 50 t . . .)
3
5
65 / 70
PSK ...
vd is a Fourier representation of m(t). Since the quality of the
bipolar signal is already incorporated in vd one can simply multiply
this signal by vc (t) to obtain the PSK signal.
As the spectrum show this is nothing other than the ASK spectrum
without the carrier signal.
If only the fundamental component of the worst case signal is
received, the minimum BW required would be 2f0 . This is equal to
the bit rate.
Since the carrier is not present, there will be more power in the side
bands (discussed later). This makes PSK more reliable in the
presence of noise when compared to ASK or FSK.
Again the Nyquist rate can be achieved if you limit the signal to the
a single side band.
The situation will be similar to ASK, however now all the signal
power will be embedded in the side band which is advantageous. In
the ASK when a side band is lost the carrier has more power
compared to the data signal. Here the situation is different.
Chulantha Kulasekere (ENT)
66 / 70
Assignment
67 / 70
Example
We have seen that the channel capacity in the presence of white
Gaussian noise is controlled by Shannon-Hartley law.
For a fixed value of N and W , find C in the limit as the received
signal power S .
How can you achieve this limiting value in the channel?
In practice what are the limitations you observe that do not allow
us to achieve this theoretical value.
It is seen that C increases linearly with W . Hence it appears that
an increase in channel bandwidth results in a linear increase in
channel capacity. Can this be practically achieved?
Find the channel capacity C in the limit W .
Comment on the above.
68 / 70
Example
Here is an illustration of the trade off between data rate and BER.
DPSK is a binary modulation technique. In channels with WGN
assume that we can design our modulation scheme so that the
channel BW B = 1/T where T is the pulse duration. Find the
maximum data rate that can be transmitted using DPSK for a bit
error probability Pb = 10E 3 and Pb = 10E 7 assuming
S/N0 = 60 dB.
69 / 70
70 / 70