CS422 Computer Networks

1
CS422 Computer Networks

• Gustavo Rodriguez-Rivera
• Purdue University
• (Part 1)

2
Introduction

• Instructor
• Gustavo Rodriguez-Rivera
• E-mail grr_at_cs.purdue.edu
• Office LWSN1169
• Webpage
• http//www.cs.purdue.edu/homes/cs422
• The syllabus and everything you need for the
  class is there. Add the link to your list of
  favorites.

3
Mailing List

• Add yourself to the mailing list.
• From your CS account type
• mailer add me to cs422-pso1 or
• mailer add me to cs422-pso2 or
• mailer add me to cs422-pso3
• depending on your PSO
• To verify type
• mailer list cs422-psoN where N is the number of
  your PSO.
• If you do not have a CS account send e-mail to
  grr_at_cs.purdue.edu

4
Textbook and Grading

• Textbook
• Comer, Computer Networks and Internets
• Grading
• 50 Projects
• 50 Midterm and Final
• PSOs
• I encourage you to attend PSOs. Projects will be
  explained there.

5
Syllabus

• 1. Low Level Details
• Signals and Media
• Frequency Spectrum
• Bandwidth, Throughput and Noise
• Digital vs. Analog signals
• Synchronous/Asynchronous Communication
• Modulation and Modems
• Time and Frequency Division Multiplexing

6
Syllabus

• 2. Packet Transmission
• Frames
• Error detection Techniques
• LAN/WAN topologies
• Bus Topology, CSMA/CD CSMA/CA
• Ethernet and 802.11 wireless networks
• Repeaters, Bridges, Hubs, and Switches
• Routing Algorithms Distance Vector and Link
  State. Shortest Path Algorithm.

7
Syllabus

• 3.Internetworking
• Protocol Layering
• IP Addressing
• ARP
• IP fragmentation/reassembly
• ICMP, UDP
• TCP, reliability, flow control, congestion
  control, port multiplexing
• NAT, Firewalls

8
Syllabus

• 4.Applications
• DNS
• Telnet, FTP, HTTP, CGI
• RPC, Java RMI, CORBA
• IPV6

9
History of the Internet

• In the late 1960s the Department of Defense
  Advance Research Project Agency (ARPA) created a
  nationwide network to allow computer access to
  the different research centers.
• The alternative was to give a computer to each
  center/university but this was very- expensive.
• Research in the Internet continued in the 70s and
  80s
• The Internet became a commercial success in the
  90s.

10
History of the Internet

• The Internet has been doubling in size every nine
  to twelve months.
• Some people attributes the increase in
  productivity in the last 10 years to the
  existence of the Internet. People produce more in
  less time.

11
What is a Network?

• A network is a group of interconnected computers
• A network can provide
• Resource Sharing Files, Data Bases, Printers
• Remote Execution ssh, telnet
• Communication E-mail
• PCs started as autonomous units. Now access to
  the Internet is very important.

12
Transmission Media

• Forms of energy used for transmission of
  information
• Electricity
• Radio Waves
• Light (Laser, Infrared)
• Sound

13
Transmission Media

• Copper Wire
• Coaxial Cable (Tolerant to electromagnetic noise)
• Twisted Pair (not as tolerant to noise but
  cheaper).
• Optical Fiber
• Less susceptible to noise than copper wire
• Higher throughput
• Longer distances
• Difficult to handle and expensive

14
Transmission Media

• Radio. Also called RF
• It uses electromagnetic waves.
• Different sources of information can be sent
  using multiple channels.
• Each channel will use a different non-overlapping
  section of the frequency spectrum.
• For example, the AM band is 540KHZ-1630KHZ. Each
  radio station uses 10KHZ of the spectrum. Max
  number of radio stations is (1630-540)/10109
• Microwaves and high radio frequencies need a
  clear path

15
Radio Frequency Spectrum
16
Transmission Media

• Infrared
• Cheap
• Used in Small Areas
• Used to communicate computers to small devices.

17
Limits of a Transmission System

• Propagation Delay
• Time required for a signal to travel across
  media.
• Light in vacuum c3x108m/s
• Electricity in cooper .7c
• Bandwidth
• Maximum number of times per second a signal can
  change. This limits the maximum throughput.

18
Analog and Digital Encoding

• Analog Encoding
• The signal is sent through the transmission
  medium as is.
• This makes the signal susceptible to noise.
• At the receiving edge, it is difficult to tell
  the signal from the noise.

19
Analog Encoding

Transmission channel
transmitter
receiver
Noise (Power Lines, Thunderstorms, Thermal Noise,
Blenders etc.)
Input Signal
Output Signal
20
Analog and Digital Encoding

• Digital Encoding
• The signal is sampled at a fixed interval.
• Each sample is quantized to a finite number of
  levels where the corresponding level is
  translated into a number
• This number is transmitted in 0s and 1s
• At the destination the 0s and 1s (bits) in the
  signal are reconstructed even in the presence of
  noise and translated back to a signal level.
• Digital encoding is tolerant to noise.

21
Digital Encoding
Transmission channel

transmitter
receiver
Input Signal
15
12
8
Noise From Transmission Channel (Power Lines,
Thunderstorms, Thermal Noise, Blenders etc.)
4
0
Output Signal
Signal Samples
8,12,13,13,14,14,14,13,12,11
Transmitted Signal
Received Signal
Reconstructed Signal
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
8
12
8
12
13
13
8
12
13
22
Quantization Noise

• One of the problems of digital encoding is that
  the number of levels is finite. For example,
  there are only 16 levels in the last example.
• This means that 12.6 and 13 are rounded to level
  13.
• This rounding error is called quantization
  noise.
• The more levels the smaller the quantization
  noise.
• Usually this quantization noise is negligible.

23
Standards

• To ensure the interoperability among hardware of
  different vendors, standards are created.
• ITU International Telecommunications Union
• EIA Electronic Industries Association
• IEEE Institute of Electrical and Electronic
  Engineers

24
Example The RS232 Standard

• Called Serial Line
• Produced by EIA
• Used initially to connect dumb terminals
• It is currently used by keyboard and mouse.
• It is a predecessor of USB (Universal Serial Bus)
• RS232 is a very simple standard for the
  transmission of characters.

25
RS232 Standard

• It is called serial because one bit is sent
  after another.
• Communication is asynchronous because the
  receiver does not know when the next character
  will be received.
• A start bit and a stop bit are required.
• Uses 15V for 0 and 15V for 1.
• When idle the level is always 15V

26
RS232 Standard

• It uses one start bit to always signal start of
  transmission.
• Sender and receiver must agree on the of bits
  per character (7 or 8) and speed in bauds
  (bits/sec).

15
0
0
0
0
0
1
1
1
1
-15
Idle
Stop bit (same as 1)
Start Bit (same as 0)
27
RS232 Standard

• Typical Speeds
• 9.6 Kbps
• 14.4 Kbps
• 28.8 Kbps
• Disagreement between transmitter and receiver
  results in framing error
• RS232 can be
• One way (half duplex)
• Two way (full duplex)

28
RS232 Standard

• Serial lines usually use only three cables

T
T
R
R
G
G

• Because of its simplicity, RS232 is still in use
  in embedded devices.

29
Transmission Problems

• Resistance of cable makes the signal decrease as
  it travels.
• Intrinsic capacitance and inductance of cable
  distorts signal
• Distortion results in loss of information

Transmitted Signal
Received Signal
1
1
1
1
1
1
0
0
0
0
0
0
8
12
8
12
13
13
30
Transmission Problems

• RS232 deals with these problems by
• Tolerating a lower voltage
• Takes multiple samples
• Direct current (two levels) cannot be used for
  long distances.
• For long distances, an sine signal (called
  carrier wave) is used
• Some of its properties are changed (modulated) to
  encode the data.

31
Characteristics of a Carrier Wave
Frequency (Hertz) of cycles per sec

Amplitude (volts)
Phase (Radians)
32
Types of Modulation

• Amplitude Modulation (AM)
• Frequency Modulation (FM)
• Phase Modulation (PM)

33
Amplitude Modulation (AM)

• Data is encoded by changing the amplitude of the
  carrier wave.
• This encoding is simple to implement but it is
  susceptible to noise.
• Noise affects the amplitude of the carrier wave.

Carrier Wave
Modulated Wave
Amplitude Modulator
Input Data
34
Frequency Modulation (FM)

• Data is encoded by changing the frequency of the
  carrier.
• More tolerant to noise than AM since noise
  affects little the frequency of the carrier.
• However, FM is more difficult to implement than
  AM and uses more bandwidth.

Carrier Wave
Modulated Wave
Frequency Modulator
Input Data
1
1
0
0
35
Phase Shift Modulation (PM)

• Data is encoded by changing the phase (angle or
  displacement) of the signal.
• This encoding is also tolerant to noise but also
  it is more difficult to implement.
• The carrier is displaced to encode the data

(datajump) 00pi/4,01pi/2, 103pi/4,11pi
Carrier Wave
Jump of pi/201
Modulated Wave
Phase Modulator
Input Data
1
1
1
0
Jump of pi 11
36
Modems

• A modem is a hardware device used for
  long-distance communication
• A modem is formed by a modulator that modulates
  the data in a carrier signal and a demodulator
  that converts the modulated signal back to data.
• Each endpoint in the communication has a modem.
• Modulator in one side connects to demodulator in
  the other side.

37
Modems
Transmitted Data
Transmitted Modulated Data

Modulator
Received Data
Received Modulated Data
Demodulator
Modem
Phone Modem, Cable Modem, DSL Modem etc.
Remote PC, or Internet hub
PC at home
M
D
Modem
Modem
D
M
RS232, Ethernet, etc
RS232, Ethernet etc.
Phone Line, cable, optical fiber etc.
38
Modem Types

• Telephone modem
• Optical Modem (Optical fiber)
• Cable modem
• DSL Modem
• etc

39
Modem Types

• Full-duplex modem
• 2 way communication
• Allows simultaneous transmission in both ways
• Needs 4 wires
• Half duplex
• Can provide 2 way communication but it transmits
  in only one direction at a time.
• It uses only two wires.

40
Bandwidth and Throughput

• Bandwidth
• the number of times per second a signal can
  change in the transmission medium.
• Throughput
• The number of bits per second that can be
  transmitted in the transmission medium
• The higher the bandwidth, the higher the
  throughput.
• There are two equations that relate bandwidth and
  throughput Nyquist theorem and Shannons theorem.

41
Nyquist Theorem

• It relates network bandwidth and throughput but
  it does not include noise.
• It is very optimistic. It just gives a
  theoretical limit
• D2Blog2K
• Dthroughput,
• BBandwidth,
• KNumber of levels used to encode the data.
• Example for RS232 with two levels (K2) with
  B4KHz
• D2(4K)log228Kbits/sec

42
Shannons theorem

• It gives a more practical limit since it includes
  noise in the equation.
• CBlog2(1S/N)
• CThroughput in bits/sec,
• Bbandwidth,
• Spower of the signal,
• Npower of the noise.
• S/N is called signal to noise ratio.

43
Shannons Theorem

• Example Telephone System
• Bandwidth 3000Hz
• S/N1000
• C3000log2(11000)30,000bps

44
Signal to Noise Ratio

• Signal to noise ratio is usually given in
  decibels(dB)
• S/N in decibels 10log10(S/N)
• Example
• S/N1000
• S/N in decibels 10log10(1000)10(3)30dB
• Example
• S/N20dB
• 20 10log10(S/N) -gt2 log10(S/N) -gtS/N102100

45
Nyquist and Shannons Theorems

• Nyquist theorem implies that finding new ways to
  encode more bits per cycle improves data rate.
• Shannons theorem implies that no amount of
  clever encoding can overcome the limits of real
  transmission lines.

46
Multiplexing

• Multiplexing allows transmitting two sources of
  information using a shared channel.
• Each destination receives the data that
  corresponds to that source.

IN1
OUT1
Shared Channel
IN2
OUT2
Multiplexer
Demultiplexer
OUT3
IN3
47
Multiplexing

• Multiplexor
• Accepts data from different sources and sends
  combined data through shared channel.
• Demultiplexor
• Receives combined data and extracts the data that
  corresponds to each destination.
• Types of multiplexing
• Time Division Multiplexing
• Frequency Division Multiplexing

48
Time Division Multiplexing (TDM)

• The channel is multiplexed by interleaving data
  from one source or the other at different times.
• The item is marked to identify the source.
• The demultiplexer uses the mark to deliver the
  data to the corresponding destination.

A 0100100
A01
Shared Channel
B 1011010
Multiplexer
Demultiplexer
B10
A0, B1, C1, A1, B0, C1
C11
C 1100111
49
Time Division Multiplexing (TDM)

• If there is no mark to identify the source, then
  the transmitter need to synchronize to be able to
  tell which data item belongs to what input
  signal.
• An example of TDM is the ethernet, where data
  from different sources share the ethernet cable.
• Each ethernet packet is labeled with the source
  and destination hardware address.

50
Frequency Division Multiplexing (FDM)

• Data from multiple sources is sent simultaneously
  using different sections of the frequency
  spectrum.
• Two or more signals that use different carrier
  frequencies can be transmitted over the same
  medium without interference.

AM Frequency Spectrum
Station 1
WBAA
Station 3
Station 4
540KHz
1600KHz
920KHz
10KHz
51
Frequency Division Multiplexing (FDM)

• Example AM/FM Radio and Cable TV use multiple
  frequency channels
• When applied to light FDM is called Wave Division
  Multiplexing or WDM.

52
Packet Switching

• Instead of connecting computers 1 to 1, computers
  are connected to a shared network.
• Connecting computers 1 to 1 would require
  n(n-1)/2 connections.
• It is less expensive to have one shared network
  with many computers connected.

C
A
Shared Network
D
B
53
Problems with shared networks

• They may have a lot of traffic
• Some applications may transfer large amounts of
  information (music, videoconferencing) and may
  monopolize the shared network for a long time.
• Some applications cannot wait, like interactive
  applications or VoIP.
• We need fair access for all computers

54
Solution for Fairness

• In shared networks we use Packet Switching.
• In Packet Switching the data is divided into
  small units called packets.
• One station can send only one packet before
  another station waiting to send sends another
  packet.
• This is a form of time division multiplexing.

55
Packet Switching

• The packets have a maximum length and a format.
• The structure/format of the packet is called
  frame.
• Packet Switching Steps
• Acquire Shared Medium
• Send one packet
• Allow other stations to send packets before
  sending again

56
Frame Format Used by RS232

• RS232 transmissions consist in a sequence of
  characters.
• SOH and EOT at mark are special characters that
  mark the beginning and end of the message
• SOH - Start of Header
• EOT End of Text

57
Byte Stuffing

• Problem If the data contains special characters
  like SOH or EOT, the receiver may confuse them as
  start/end of transmission.
• Solution For these characters use an alternative
  representation called byte stuffing.
• Byte stuffing consists in having the sender
  substitute the special characters by an escape
  sequence before the message is sent.
• The receiver translates the stuffed message back
  to the original message

58
Byte Stuffing

• Escape sequences
• SOH ESC X
• EOT ESC Y
• ESC ESC Z

Original Message
EOT
A
B
C
SOH
EOT
SOH
D
MessageSent
SOH
A
C
D
ESC
X
Y
EOT
ESC
B
59
Handling Errors

• Data sent can be corrupted during transmission.
• Bits can be lost
• Bit values can be changed
• The frame includes additional information to
  detect and correct errors.
• This information is set by the sender and checked
  by the receiver.
• There is a very small probability of having
  undetected errors but it works well in practice.

60
Error Detection Techniques

• There are three main error detection techniques
• Parity Bit
• Checksum
• Cyclic Redundancy Check (CRC)

61
Parity Bit

• One additional bit per character makes sure that
  the number of 1s in the character is even or
  odd.
• Even parity Total number of 1s (including
  parity bit) is an even number
• Odd parity Total number of 1s (including the
  parity bit) is an odd number
• Transmitter and receiver agree on what kind of
  parity to use (even or odd).
• Receiver verifies that the character received has
  the correct number of 1s.

62
Parity Bit

• Example 1
• Assume parity is even ( of 1s including parity
  bit is even)
• The character is 1001001.
• Parity bit is the highest bit
• The character sent by the sender is 11001001 with
  the parity bit set to 1 (highest bit) to have
  even parity.\
• Receiver verifies that number of 1s in character
  is even.

63
Parity Bit

• Example 2
• Assume parity is odd ( of of 1s including parity
  bit is odd)
• The character is 1111001.
• Parity bit is the highest bit
• The character sent by the sender is 01111001 with
  the parity bit (highest bit) set to 1 to have odd
  parity.
• Receiver verifies that number of 1s in character
  is odd.

64
Parity Bit

• The sender and the receiver agree before the data
  transmission of the type of parity (even or odd)
• The sender sets the parity bit.
• The receiver checks that the parity of the bit is
  correct and signals an error if it is not correct.

65
Checksum

• In checksum the data is treated as a sequence of
  integers.
• The sender computes the arithmetic sum of the
  data and adds it into the message.
• The receiver receives the message and computes
  the checksum again.
• If the checksum computed by the receiver is
  different than the checksum in the message, the
  receiver signals an error and discard the
  message.

66
Checksum Example

• Message

00010010
01010010
10011010
01010011
01010001
Checksum
Data
00010010-gt 18

01010010-gt 82

10011010-gt 154

01010011-gt 83
101010001-gt 337
Checksum
Only take the lower 8 bits for the checksum In
other cases the checksum is computed using words
of 4 bytes
67
Undetected Errors

• It is possible to have undetected errors using
  checksum.
• For instance, an error that happens twice in the
  same bit position in two of the data bytes can
  cause an error that will be undetected.
• These errors that happen in the same bit is
  called a positional error and may happen in bad
  hardware.
• Checksum works well in practice.

68
Undetected error
Error
Error

• Message

0
1
00010010
01010010
10011010
01010011
01010001
Checksum
00010010-gt 18

Data
00010010-gt 18

11011010-gt 218

01010011-gt 83
101010001-gt 337
Checksum

• The data changed but the checksum is the same.
• The receiver will not be able to detect the
  error

69
Cyclic Redundancy Check (CRC)

• It is implemented in hardware.
• Implementing it in software is too slow
• Basic Operations
• XOR
• Shift Operator

0
0
1
1
0
1
0
1
1
0 In
1 out
Before
70
Cyclic Redundancy Check (CRC)

• Hardware to compute the 16 bit CRC
• The CRC registers are initialized with 0s
• The data bits are fed one by one in the input.
• The final CRC will be stored in the CRC
  registers.

1
1
1
0
1
0
1
0
0
0
0
1
1
0
1
0
1
0
0
1
1
0
0
0
Input 1
71
Cyclic Redundancy Check (CRC)

• Example of an input of 101
• .

1
1
1
0
1
0
1
0
0
0
1
1
1
0
1
0
1
0
1
0
1
1
0
0
Input 1
0
0
0
1
0
1
1
0
1
1
1
0
0
1
0
0
0
0
0
0
1
0
1
0
Input 0
1
1
1
0
1
0
1
1
0
1
0
0
0
0
1
1
1
0
0
0
1
1
0
0
CRC
Input 1
72
Cyclic Redundancy Check (CRC)

• CRC detects the positional errors that checksum
  does not detect.
• The CRC is used in Ethernet and is computed in
  hardware.
• Also disks use CRC to detect errors in disk.

73
Network Classification

• Local Area Network (LAN)
• Room/Building
• Metropolitan Area Network (MAN)
• City
• Wide Area Network (WAN)
• Group of cities

74
Local Area Networks

• Many technologies exist
• High Throughput
• Relatively low cost
• Limited to short distances
• Often rely on shared media
• Wide Area networks tend to be slower than local
  area networks.

75
Scientific Justification for LANs

• Why Local Area Networks work well in practice?
• A computer is more likely to communicate with
  computers that are nearby than with computers
  that are distant.
• This is known as the locality principle.

76
Scientific Justification for LANs

• In the cases that computer communicate mostly
  with distant computers, the communication will be
  slow.
• In some cases, the interaction with distant
  computers can be reduced by adding mirror
  servers or web caches.
• Web caches cache frequently used pages.

77
Lab2 HTTP Server Part Iterative Server

• Iterative Server
• while (1)
• ss accept()
• if (sslt0)
• continue
• dispatchRequest(ss) //closes ss

78
Create process for each request

• -f   Create a new process for each request
• while(1)
• ss accept()
• if (sslt0) continue
• ret fork()
• if (ret0)
• //child
• dispatchRequest(ss) // close ss
• //parent
• close(ss)

79
Create thread for each request

• -t Create a new thread for each request
• while(1)
• ss accept()
• if (sslt0) continue
• ret pthread_create(dispatchRequest, ss)

80
Pool of threads

• Create a pool of threads to serve the requests.
• Main()
• .
• //if pool of threads
• for (i- iltnumofthreads-1 i)
• pthread_create(server_thread,
  ms)
• server_thread(ms)
• void server_thread( int ms)
• while(1)
• ss accept( ms, )
• if (sslt0) continue
• dispatchRequest(ss)

81
Network Topology

• Specifies the general shape of a network.
• Primarily refers to interconnections between
  computers.
• Hides the actual devices.
• Types of Network Topologies
• Star
• Ring
• Bus

82
Star Topology

• Central component of the network is known as hub.
• Each computer has a separate connection to the
  hub.

83
Ring Topology

• No central facility.
• Connections go from one computer to another.

84
Bus Topology

• Shared Medium connects the computers.

Shared Cable
85
Ethernet Network

• This is an example of a bus network
• Most popular LAN
• IEEE standard is 802.3
• There are several generations of the ethernet
  technology
• Thicknet (10Base5)
• Thinnet (10Base2)
• Twisted Pair (10Base-T)

86
Thicknet Ethernet (10Base5)

• Thicknet Ethernet was the first generation of
  ethernet
• It used a heavy coaxial cable. Difficult to
  handle
• The formal name is 10Base5
• The computers attached to the cable through a
  tranceiver that transmitted/received the data to
  the cable.
• It could reach 500m

Tranceivers
87
ThinNet Ethernet (10Base2)

• Thicknet Ethernet was the second generation of
  ethernet
• It used a thiner coaxial cable. Easier to handle
• The formal name is 10Base2
• The computers attached to the cable through a T-
  connector
• It could reach 185m

T-connectors
88
Twisted Pair (1OBase-T)

• It is the modern Ethernet wiring
• It uses a hub or a switch. It could reach 100m.
• A hub simulates the coaxial cable with
  electronics.
• A switch provides 1 to 1 conection (it will be
  covered later)
• The modern ethernet is a physical star and a
  logical bus or a star-shaped bus

Hub or Switch
89
Ethernet Speeds

• Also it has different rates
• 10Mbps
• 100Mbps (Fast Ethernet uses 100Base-T)
• 1Gbps (Gigabit Ethernet uses 1000Base-T)
• It has always been the same packet format
  independent of the speed.

90
Ethernet Transmission

• Only one station transmits at a time
• Signal propagates across the entire cable.
• All stations including itself receive the
  transmission

91
Ethernet Transmission

• The ethernet interface will filter the packets
  and it will pass to the OS only the packets
  destined to the machine
• A technique called CSMA/CD controls the access to
  the shared bus.

92
CSMA/CD

• Carrier Sense Multiple Access/Collision Detection
• Carrier Sense (CS)
• Computers will wait until the medium is idle to
  begin transmitting the frame.
• Multiple Access (MA)
• Multiple computers share the same media
• All of them use the same access algorithm
• Collision Detection (CD)
• Simultaneous transmissions are possible and will
  interfere with each other. This is called a
  collision.

93
Collision Detection Mechanism

• A computer transmitting a message will listen to
  the medium during transmission
• If what it listens is different than what it
  transmits, the computer has detected a collision.
• If a collision has been detected, the computer
  will back off and it will try later.

94
Back Off After Collision

• After a collision has been detected, the computer
  will wait for a random time t1 before
  transmitting again where 0 lt t1 lt d where d is
  a constant.
• If a second collision occurs, the computer will
  wait for a random time t2 where 0 lt t2 lt 2d
• The range will be doubled after every successive
  collision.
• This is called exponential back off.

95
Ethernet Addressing

• All the stations in the shared medium receive all
  transmissions.
• Each station has an assigned 48 bit ethernet
  address.
• The address of an ethernet Network Interface
  Card (NIC) is assigned by the manufacturer.
• IEEE gives ranges of ethernet addresses to the
  different manufacturers.

96
Ethernet Addressing

• Each frame contains a destination address
• All stations receive the transmissions.
• The stations discard any frames addressed to
  another station.

MSG B-gtD
D
C
B
A
97
Ethernet Addressing

• The check of the destination address is done by
  the ethernet card itself and not by the OS.
• When an ethernet packet is received by a
  computer, the OS receives an interrupt so the
  packet is processed.
• If the checking of the destination address were
  done by the OS instead of by the hardware, the
  CPU would have unnecessary overhead due to
  packets not directed to the machine.

98
Ethernet Addressing

• Packets can be sent to
• A single destination (Unicast)
• All the computers on the ethernet (Broadcast)
• A subset of the computers in the ethernet
  (Multicast)
• Ethernet Unicast
• Sent from one computer to another computer
• Most ethernet packets are unicast.

99
Ethernet Broadcast

• All 1s in the destination address specifies a
  broadcast FFFFFFFFFFFF
• The sender places a packet with the broadcast
  address as destination
• All computers in the Ethernet will receive a
  copy, including the computer that transmitted the
  packet.

100
Ethernet Multicast

• A Ethernet interface card can be configured by
  software which multicast address to accept.
• If a packet has as destination this multicast
  address, all the computers that have joined the
  multicast address receive a copy.
• The check for the destination multicast address
  is done by the Ethernet interface card.
• Half of the Ethernet addresses are reserved for
  multicast
• Multicast is useful for teleconferencing in the
  same net to eliminate duplicate packets.

101
Ethernet Card Filtering

• An Ethernet card will pass to the OS the packets
  that have as destination
• The computers unicast address
• The broadcast address
• Any of the multicast addresses that the computer
  has joined.
• It is recommended not to use broadcast too
  frequently since the CPUs in all the computers
  in the network will be interrupted even if they
  are not going to use the packet.

102
Ethernet Promiscuous Mode

• This is a special Ethernet interface modes
• It is mode allows the interface to pass all the
  packets to the OS even when the destination is
  not that computer.
• It is available in most Ethernet interfaces
• This mode is useful for network monitor programs
  like snoop or tcpdump.
• A program needs to run as root to be able to
  configure an interface into promiscuous mode.

103
Ethernet Packet Format

• This the Ethernet packet format
• The frame type fields tells the recipient the
  type of data being carried
• Possible Ethernet Types
• 0x800 -gtIP
• 0x806 -gtARP
• 0x8137-gtIPX (Novell) etc.

Destination Address
Source Address
Frame Type
Preamble
Data in Frame
CRC
8 bytes
6 bytes
6 bytes
2 bytes
Variable 46 to 1500 bytes
4 bytes
104
Ethernet Packet Format

• The maximum number of data bytes that can be sent
  in an Ethernet packet is 1500 bytes.
• This is called Maximum Transfer Unit or MTU.
• Different networks have different MTU sizes.

105
Wireless Networks

• Two kinds of wireless networks
• Infrastructure Mode Networks Hosts in the
  network use a base station to communicate.
• Ad-Hoc Networks Hosts can communicate with each
  other without the need of a base station.

106
Wireless Networks Characteristics

• Decreasing signal strength.
• Signal will attenuate as it travels in the space
  or through matter.
• Interference with other sources.
• Radio sources transmitting at the same frequency
  will interfere with each other. Example 2.4GHz
  wireless phones and 802.11b
• Multipath propagation.
• Radio signals may bounce causing reflections of
  the same signal being received simultaneously.

107
Wireless Networks Characteristics

• All this characteristics make wireless networks
  to have more bit errors than wired networks.
• This is why wireless networks employ more
  powerful CRCs to detect errors as well as
  link-level ARQ (Automatic Repeat Request)
  protocols that retransmit corrupt packets.

Data packet
Acknowledgement
A
B
108
Wireless Networks and 802.11b

• 802.11 (WiFi or Wireless Fidelity) denotes a
  family of wireless network protocols
• 802.11b - 2.4 GHz band - 11 Mbps
• 802.11g - 2.4 GHz band - 54 Mbps
• 802.11a - 5 GHz band - 54 Mbps
• 802.11b and 802.11g are the most common and both
  are compatible.

109
802.11 Architecture

• 802.11 may work in both ad hoc mode and
  infrastructure mode. Being infrastructure
  mode more common.

Internet
Had Hoc Network
Infrastructure Mode Network
110
802.11 Architecture

• When the Administrator installs the access point,
  it needs to assign an SSID (Service Set
  Identifier) and a channel number.
• There are 11 partially overlapping channels in
  the band of 2.4GHz-2.485GHz
• A base station periodically sends beacon frames.
• A computer scans the 11 channels to be able to
  list the available networks.

111
802.11 Architecture

• Once the user chooses one of the networks, it
  sends a DHCP discovery message to request its IP
  address and other parameters.
• The Access Point may require the the wireless
  station to authenticate itself.
• The authentication may require a user name and
  password.

112
Hidden Terminal Problem

• Given three computers A, B, C in a wireless
  network, it is possible that B can receive the
  messages from both A and C but A and C cannot
  receive messages from each other.
• This is due to the restricted range of
  transmission (d).
• This is problematic because stations will not be
  able to tell in some cases when collisions
  happen, for example if A sends a message to B at
  the same time C sends a message to B, so CSMA/CD
  is impractical.

d
d
A
B
C
113
CSMA/CA (Collision Avoidance)

• Instead of using CSMA/CD, 802.11 uses CSMA/CA
• CSMA/CA- Carrier Sense Multiple Access Collision
  Avoidance
• The stations reserve the channel by first sending
  a small message RTS/CTS Request to Send / Clear
  To Send

114
CSMA/CA (Collision Avoidance)

• Assume computer X will send a message to computer
  Y.
• X sends an RTS (Request to Send) message to Y.
  All machines reachable from X receive it.
• Y replies with a CTS message (Clear to Send). All
  machines reachable from Y receive it.
• For the requested time no machine reachable from
  X and Y will use the shared medium.

115
Collision Avoidance

RTS
d
CTS
X
2
3
1
d
Y
7
5
4
6
116
Collision Avoidance

• Even with Collision Avoidance, collisions may
  still be possible if computers move or the radio
  propagation conditions change.
• Still, Collision Avoidance reduces the
  probability of collisions.

117
802.11 Multiple Access Protocol

• Since wireless networks are more prone to errors,
  they use also acknowledgments at the link level
• After a packet is sent, an acknowledgment the
  transmitter expects an acknowledgment from the
  receiver.
• If the acknowledgment is not received after some
  time, the frame is retransmitted again.
• The retransmissions are independent from the
  TCP/IP retransmissions and they are called
  link-layer acknowledgment

Data packet
Acknowledgement
A
B
118
802.11 Multiple Access Protocol

• A packet transmission in 802.11 will look like
  this

Source
Destination
RTS
CTS
DATA
ACK
119
802.11 Frame Format

• The 802.11 Frame was derived from the Ethernet
  frame
• The MTU of 802.11 is 2,312 bytes

2
2
6
6
6
2
6
0-2312
2
Frame Control
Address 2
Address 1
Address 3
SequenceControl
Address 4
Duration
Payload
CRC
Protocol Version
More Frag
Power Mgmt.
More Data
Type
Subtype
To AP
From AP
WEP
RSRVD
2
2
4
1
1
1
1
1
1
1
120
802.11 Forwarding

• Address1 is the immediate destination address
• Address2 is the source address
• Address3 is used as the final destination in case
  the packet is destined to a computer/router in
  one of the ethernet ports in the base station.

Address1 B
Address2 A
A
Address3 R
B
R
Ethernet
Router
Access Point
C
121
WEPWired Equivalent Privacy

• It is a security feature in 802.11 that uses
  encryption at the link level.
• An ASCII or hexadecimal key (up to 128 bits) is
  configured in the base station and in all the
  computers in the same SSID.
• The packets are encrypted with this key when
  transmitted and decrypted when received.
• The method used for encryption stream cypher
  RC4 has some weaknesses and potentially packets
  can be sniffed.

122
WEPWired Equivalent Privacy

• In general it is better to use point-to-point
  security like VPN or SSH.
• WPA or WPA2 can be used instead of WEP to offer
  more security.
• Also some users do not use WEP altogether since
  WEP is disabled by default in base stations to
  simplify the installation.

123
Ring Topology

• It is the second most popular LAN topology
• Bits flow in a single direction
• A special short message called token is passed
  from computer to computer.
• Only the machine that has the token is able to
  transmit.
• This guarantees fair access

token
124
Token Passing

• A computer x that is going to transmit waits for
  the token.
• The computer x removes the token and sends the
  message to y.
• The computers transmit the token around the ring.

token
x
y
125
Token Passing

• The destination computer receives the message and
  makes a copy.
• The sender receives its own transmission and it
  restores the token to the network.
• When there is no data to send, the token
  circulates continuously.

x
y
126
Advantages of Token Ring

• Easy detection of
• Broken Ring
• Hardware Failures
• Interference
• Weaknesses
• Broken ring disables the entire ring
• Difficult to add stations
• Examples of Token Ring
• Pronet-10 (10 Mbps)
• IBM Token Ring (4Mbps/16Mbps)
• Fiber Distributed Data Interconnect (FDDI) 100Mbps

127
FDDI

• Fiber Distributed Data Interconnect
• It uses optical fiber
• High Reliability
• Immune to interference
• There is also CDDI that is FDDI over copper with
  the same data rate and frame format but less
  noise immunity.

128
FDDI

• The stations can also be attached to a hub that
  provides a logical ring
• It uses also the same frame format and rdata rate
  than a physical FDDI.
• FDDI with a hub is called star-shaped ring

x
y
FDDI Hub
129
FDDI Failure Recovery

• Dual Attached - FDDI uses two rings instead of
  one
• Self Healing FDDI has an automatic failure
  recovery
• Counter Rotation The token in the second ring
  rotates in the opposite direction.

Loopback
Failed Station
Loopback
After Failure. Stations adjacent to failure loop
back.
Before Failure. Outer Ring is used for data.
Inner Ring is only used in failure mode.
130
ATM Example of Star Topology

• ATM Asynchronous Transfer Mode
• Designed by Telephone companies
• Intended to accommodate
• Voice
• Video
• Data
• The building block is known as ATM Switch
• Each station is connected to a switch
• Switches can be interconnected with each other.

ATM Switch
131
ATM

• ATM uses full-duplex connections
• Two fibers are required.
• It has high data rates (E.g. 155 Mbps)
• Fixed size packets called cells of 53 bytes
• 48 bytes for data
• 5 bytes for header
• The small size of the packet decreases the
  network latency. The smaller the packet, the
  smaller the waiting time to send a packet because
  the shared network is in use.

132
ATM

• Before transmitting, it is necessary to create a
  virtual circuit across all the switches,
  similar to a telephone long distance call.
• The switches negotiate the maximum throughput
  that will be used. This bandwidth will be
  reserved while the circuit is open.

A
B
133
ATM

• ATM is a connection oriented network in the
  sense that a connection setup is necessary to be
  done in hardware before a transmission. The
  bandwidth will be respected by all the switches
  in the circuit while the connection is open.
• In contrast, the Internet is connection less in
  the sense that no connection is necessary for the
  communication of two machines. However, bandwidth
  will be used first come first served.
• ATM is designed for real-time traffic where
  Quality of Service (QoS) can be assured like
  voice and video. The internet is not designed for
  real-time data but it works well for non
  real-time data like e-mail, file transfer,
  browsing etc.

134
ATM

• This connection setup overhead in ATM was one of
  the big disadvantages that motivated users not to
  use ATM in LANs.
• Also, ATM was too expensive. Switches and
  interfaces cost thousands of dollars
• ATM is still in use for high speed network
  backbones where a virtual connection can be used
  permanently
• There are other cheaper alternatives to ATM like
  the Gigabit ethernet.

135
Extending Networks

• Each LAN has a distance limitation. For example
  the Ethernet has a distance limit of 185 m.
• However sometimes it is necessary to extend a
  LAN.
• Options to extend a LAN
• A repeater Analog device that connects two LAN
  segments into a single LAN. It forwards all
  packets including noise. Works at the Link Level.
• A bridge Digital device that connects two LAN
  segments into a single LAN. It forward packets
  only when necessary. Works at the Link Level.
• A router It connects two LANs keeping them
  separate. Works at the Internet Level.(We will
  cover it when we see TCP/IP protocols)

136
Repeaters

• It is a hardware device used to extend connect
  two ethernet segments.
• It connects two LAN segments
• It copies blindly packets from one segment from
  the other.

Ethernet 1
Ethernet 2
Repeater
137
Repeater

• The repeater amplifies signals from one segment
  before sending them to the other segment.
• The repeater operates in both directions
  simultaneously.
• It propagates noise and collisions.
• Repeaters also may be used to join more than one
  ethernet.

Ethernet 1
R
Ethernet 2
Main Ethernet
R
Ethernet 3
R
138
Bridge

• A bridge is also used to connect two Ethernet
  segments to form one single Ethernet segment.
• It is a digital device.
• It is smarter than a repeater. It forwards
  packets but it does not forward noise or
  collisions.
• It only forwards packets when they are destined
  to the other segment.

139
Bridge Algorithm

• Listen in promiscuous mode
• Watch source address in incoming packet.
• Listen to the source address of the packets and
  make a list of computers attached to each
  segment.
• Only forward packets when necessary.
• Always forward broadcast and multicast.

140
Bridge Algorithm

• A message from u to w is not forwarded
• A message from v to y is forwarded.
• The bridge learns from the source addresses of
  the packets.
• When the destination is not in the list, the
  bridge forwards the packet.

Ethernet 1
Ethernet 2
u
v
w
x
y
z
bridge
List1 u,v,w
List2 x,y,z
141
Ethernet Hubs and Switches

• Hubs and switches allow building a LAN with
  ethernet using a single box.
• Ethernet Hub
• It is an analog device that operates on signals
• It simulates a common shared cable.
• It propagates incoming signals to all
  connections. It does not understand packets.

142
Ethernet Hub

• Ethernet Hub (cont.)
• It can be modeled as a box where each port is
  connected with each other through a repeater
• The repeaters forward the signals to all the
  ports.
• It is inexpensive.
• Hubs may be connected in a daisy chain to form a
  large Ethernet.
• The cost of a hub is O(n) with the number of
  ports.

Port1
Port2
Port3
Port4
Port5
R
R
R
R
143
Ethernet Switch

• Ethernet Switch
• It also allows building an ethernet using a
  single box.
• It is a digital device that understands packets
  and addresses.
• It only forwards when necessary.
• It also allows two separate pairs of computers to
  communicate simultaneously.
• It is higher in cost than a hub.

144
Ethernet Switches

• It can be modeled as a box where there is a
  bridge connecting each pair of ports.
• In general a bridge with n ports will need
  n(n-1)/2 bridges to be able to connect all ports
  one-to-one.
• This is why the cost of an Ethernet switch grows
  O(n2) with the number of ports.

Port1
Port2
Port3
Port4
B
B
B
B
B
B
145
Ethernet Switches

• Since a switch allows simultaneous traffic across
  two separate pairs of ports, the total throughput
  in a switch will be higher than in a hub.
• For example, if a 100Mbps switch has 4 ports
  a,b,c,d and there is traffic between a to b
  and c to d, then the total traffic will be
  2 100Mbps 200Mbps.

100Mbps
100Mbps
Total Throughput A-gtB 100Mbps C-gtD
100Mbps Total 200Mbps
A
B
C
D
B
B
B
B
B
B
146
Ethernet Hub

• In contrast, a 100Mbps Ethernet hub with ports
  a,b,c,d where there is traffic between a to b and
  c to d the total throughput will not exceed
  100Mbps
• Additionally, switches enhance security since
  packet sniffing is limited. A port will receive
  only packets destined to the port and broadcasts.

Total Throughput A-gtB 50Mbps C-gtD 50Mbps Total
100Mbps
50Mbps
50Mbps
D
A
B
C
R
R
R
147
Wide Area Networks

• They span a large geographic distance
• Generally these connections are leased from a
  telephone company.
• They provide a point-to-point connection.
• They are not part of the conventional telephone
  system.
• Copper, fiber, microwave, or satellite channels
  are available.

148
Long-Distance Connections

• The customer