Title: Packet Transmission
1Packet Transmission
- Computers use data grouped into packets for
transmission - Local Area Networks
- Wide Area Networks
- Addressing and Routing
2Concepts of Packets
- Computer networks divide data into small blocks
called packets - Packets are send individually
- Often called packet networks and packet switching
networks - Motivation for using packets
- Sender and receiver needs to coordinate
transmission to ensure that data arrives
correctly - Helps determine which blocks arrive intact and
which do not - Computers often share underlying connections and
hardware - Packet switching helps ensure fairness to access
3Shared Resources
- The first networks
- A 5 MB file at 56 Kbps will take 12 min to
transfer from A to D - B C must wait
- Packet networks
- Divide data into packets of 1000 bytes each
- A sends a packet to D taking only 143 ms
- B transmits data to C
- A continues
- No long delays
4Packets and TDM
- Time Division Multiplexing
- Many resources take turns accessing the shared
communication resources - All sources receive prompt service
- The source with less data finishes early
5Packets and Frames
- Packet refers to a small block of data
- Each hardware technology uses different packet
format - Frame denotes packet used with specific type of
network - EX RS-232 mechanism
- Does not include a mechanism that allows a sender
to signal the end of a block of characters - Sending and receiving computers must agree on
such details
6Packets and Frames (Cont.)
- Network systems can choose two unused values to
define format - EX RS-232 can use frame delimiters
- Soh start of header
- Eot end of transmission
- Overhead is an disadvantage
- An extra, unnecessary character between blocks of
data - Advantageous when large delays or computer
crashes - Missing eot indicates sending computer crashed
- Missing soh indicates receiver missed beginning
of frame
7Byte Stuffing
- Data and control information must be
distinguished - Network system change the data slightly before it
is sent - Termed data stuffing
- Insert extra bits or bytes to change data
- Byte stuffing and character stuffing
- Data stuffing used with character oriented
hardware - Bit stuffing
- Data stuffing used with bit oriented hardware
8Byte Stuffing (Cont)
- EX RS-232
- soh and eot must not appear in the data
- Byte stuffing reserves a third character esc
- Marks occurrences of reserved characters
9Implementing Byte Stuffing
- Sender must scan and perform mapping before any
data is sent - Sender replaces characters
- Receiver looks for a combination of esc
followed by a x, y or z - Replaces combination by appropriate single
characters - Receiver is sure that soh and eot are frame
delimiters
10Transmission Errors
- Interference can introduce unwanted electric
currents in wires - Interference can cause
- The receiver to misinterpret the data
- The receiver to lose the data sent by sender
- The receiver to detect data, although sender did
not send any data - Termed transmission errors
- The problem of lost ,changed or spuriously
appearing data
11Parity
- Even or odd
- Sender and receiver must agree in which form to
use - Even parity the total number of 1 bits
(including parity bit) must be even - EX parity bit for 0100101 is 1
- Parity bit for 0101101 is 0
- Odd parity total number of 1 bits (including
parity bit) must be odd - EX parity bit for 0100101 is 0
- Receivers computation of parity must agree to
senders - Else receiver reports parity error
12Parity Checking
- Parity check mechanism requires the sender to
compute an additional bit, called parity bit - RS-232 circuits uses parity check to ensure that
each character arrives intact - Attach parity bit to each character before
sending - Receiver removes the parity bit and performs the
same operation as the sender - Verifies the result with the value of the parity
bit - If one of the bits is damaged, receiver reports
error
13Error Detection
- Parity cannot detect error involving an even
number of bits - EX Two 0 bits changed to 1
- Two 1 bits changed to 0
- One 0 bit changed to 1 and vice versa
- Parity is preserved even with errors
- Alternative mechanisms used depending on
- The size of the additional information
- The computational complexity of the algorithm
- The number of bit errors that can be detected
-
14Checksums
- Checksum sender treats the data as a sequence
of binary integers and computes their sum - Carry bits, if any, are added into the final sum
- Advantages size and ease of computation and
cost of transmission - Disadvantages cannot detect all common errors
15Cyclic Redundancy Checks (CRC)
- CRC hardware uses
- A shift register
- An exclusive or (xor) unit
- To compute a CRC
- Values in shift registers initialized to 0
- Bits of message shifted once at a time
- One bit of message applied at input
- All shift register perform shift operation
Shift registers
X-or unit
16CRC (Cont.)
- Shift registers contain CRC after entire message
has been shifted - Receiver uses identical hardware and compares CRC
- To simplify checking CRC
- Append and additional 16 bits of zeroes to
message - Receiver computes CRC over incoming message plus
incoming CRC - If no errors,value should be zero
- Uses a polynomial expressed as a power of X
- P( X ) X 16 X 12 X 5 1
17Burst Errors
- CRC is especially useful with
- Vertical errors
- Burst errors
- Vertical errors appear in a vertical column when
characters are arranged in rows - EX Damaged character oriented I/O device
- Burst errors involve changes to a small set of
bits near a single location - Caused by electric interference from lighting ,
electric motor, etc.
18Frame Format and Error Detection
- Networks usually associate error detection with
each frame - If no characters are lost , byte stuffing of CRC
is not required - If CRC is not byte stuffed, a single character
loss causes the receiver to discard two frames - Individual standards specify whether CRC is
computed on the message or the encoded frame
19Local Area Network (LAN)
- Most networks are local i.e. the network fits
inside a building or a single room - Permits multiple computers to share resources
- Exa printer accessed by two computers in a
network - No separate modems and cables
- Computers must take turns using the shared medium
20Direct Point to Point Communication
- Point to point network or mesh network
- Each communication channel connects and is
availabel to two computers - Advantages
- Independent installation facilitates use of
appropriate hardware - Connected computers decide how to communicate
- Easy to enforce security and privacy
- Disadvantages
- Must provide a separate communication
- channel for each pair of computers
- Number of connections grows quickly
- as the size of set increases
-
21Direct Point to Point Communication (Cont.)
- Number of connections needed for N computers is
(N2 N) /2 - Adding Nth computer requires N-1 connections
- Expenses are high because many connections follow
same physical path - Ex In fig., 6 connections pass between two
locations - If one computer is added to location1 , number of
connection become 9
22Shared Communication Channels
- LAN developed during the late 1960s and early
1970s - Consists of a single shared medium
- Computers take turns using the medium to send
packets - Reduces cost
- Shared network used for only local commnucation
- Large geographic separation introduces longer
delays - Shred network with long delays are inefficient
- Providing high bandwidth communication channel
over long distances is expensive
23Locality of Reference
- LANs now connect more computers than any other
type of network - Locality of reference computer communication
follows two patterns - Temporal locality of reference a computer is
more likely to communicate with the same set of
computer repeatedly - Physical locality of reference a computer tends
to communicate with computers that are physically
nearby
24LAN Topologies
- Star topology - All computers attach to a
central point - The center of the star network often called hub
- Hub accepts and delivers data
- In practice, star networks seldom have a
symmetric shape - A hub often resides in a location separate from
the computers attached to it
25LAN Topologies (Cont.)
- Ring topology- arranges for computers to be
connected in a closed loop - A cable connects first computer to second,another
cable connects second to third and so on - A cable connects the final computer to the first
- Refers to logical connection not physical
orientation
26LAN Topologies (Cont.)
- Bus topology- Consists of a single ,long cable
to which computers attach - Any computer can send data to any computer
- Coordination is necessary to ensure that only one
computer sends a single at any time
27Why Multiple Topologies?
- Each topology has advantages and disadvantages
- Advantages
- Ring makes it easy to coordinate access and
detect operation - Star protects network from damage by single wire
- Bus requires fewer wires than star
- Disadvantages
- Entire ring network is disabled if one of the
cables is cut - Bus network is disabled if main wire is damaged
28Ethernet
- Widely used network topology that employs bus
topology - Invented at Xerox corporations Palo Alto
Research center in early 1970s - Consists of a single coaxial cable, called the
ether, to which multiple computers connect - Ethernet coaxial cable also termed segment
- Length limited to 500 m , minimum separation
between pairs is 3 m
29Ethernet Operation
- Original Ethernet hardware operated at 10 Mbps
- Fast Ethernet operates at 100 Mbps
- Gigabit Ethernet operates at 1000 Mbps or 1 Gbps
- The Ethernet standard specifies all details
- Multiple computers share access to a single
medium - Sending computer has exclusive use of the entire
cable
30Carrier Sense on Multiple Access(CSMA) Networks
- Ethernet network does not have a centralized
controller - Ethernet employs CSMA to coordinate transmission
among multiple attached computers - CSMA - Idea of using the presence of a signal to
determine when to transmit - Uses electrical activity on the cable to
determine status - Signals informally called a carrier
- If no carrier present , transmit
- Carrier present, must wait for the sender to
finish - Technically , Carrier Sense is checking for a
carrier wave
31Collision Detect
- CSMA cannot prevent all possible conflicts
- Two computers send a frame at the same time
finding the cable idle - Interference between two signals is called a
collision - No hardware damage but produces garbled value
- Ethernet standards require sending station to
monitor signals - Technically termed as collision detect
- Ethernet mechanism known as Carrier Sense
Multiple Access with Collision Detect
32Back Off With CSMA/CD
- While detecting collisions , CSMA/CD recovers
from them - To avoid multiple collisions , each computer
delays retransmission - Computer choose random delay, between 0 and
maximum delay , d - If choice of delay is nearly same , collisions
occur - Random delay doubled at each successive
collisions 0-d , 0-2d, 0-4d , - Binary exponential back off
- Doubling the range of random delay after each
collision
33Wireless LAN
- Uses antenna to broadcast RF signals
- Data send at 2 Mbps using 900 Mhz frequency
- All computers configured to the same frequency
- Transmitters use low power
- Enough power to travel a short distance
- Metallic obstructions can block the signal
- Cannot use CSMA/CD mechanism
34CSMA/CA
- Wireless LANs use Carrier Sense Multiple Access
with Collision Avoidance - Operation
- Computer 1 transmits brief control message
- Computer 2 receives and responds
- Computer 1 receives response and begins
transmission - Control message collide
- Sending station apply random
- back-off before retransmission
35Local Talk
- A LAN that employs bus topology
- Invented by apple computer corporation
- Designed for apple Macintosh which includes all
required hardware - Uses a version of CSMA/CD
- Disadvantages
- Lower bandwidth (230.4 kbps)
- Distance limitations
- Advantages
- Almost free
- Easy to install
- Available on many computers
36IBM Token Ring
- LANs employing ring topology use token passing
mechanism - Token Ring operates as a single shared medium
- A special, short message called token coordinates
use of the ring - A token permits transmission of one frame
37IBM Token Ring (Cont.)
- One token exists on the ring t any time
- Each computer sends one frame before passing
token - Token cycles around when no data to send
- Time taken is brief (milliseconds) because
- Token is small
- Handled by ring hardware , not CPU
- IMB Token Ring is best known token passing
network - Operates at 16 million bps
- Used with computers from IBM , other vendors and
printers
38Fiber distributed data Interconnect (FDDI)
- Token Ring technology
- Transmission rate of 100 million bps
- Uses optical fibers to interconnect computers
- Contains two complete rings ( counter rotating)
to overcome failures - Self healing network
- Hardware detects
- a catastrophic failure
- and recovers
- automatically
39Asynchronous Transfer Mode (ATM)
- A star topology developed by telephone companies
- One or more interconnected switches form a
central hub to which all computers attach - Designed to provide high bandwidth
- ATM switch operates at 155 Mbps or faster
- Each connection uses a pair of optical fibers
40Hardware Addressing
- Any signal sent across a shared network reaches
all attached stations - Each station on the LAN is assigned a unique
numeric value - Called physical/hardware/media access address
- Sender includes hardware address of intended
recipients - Each frame begins with a header consisting of
- Destination address fields
- Source address fields
- Network interface hardware examines address
fields in frames - Accepts only those frames where destinations
address matches stations address
41LAN Hardware
- Handles details of sending and receiving frames
- Operates without using the stations CPU
- Uses physical addressing to prevent receiving all
packets
42Addressing Schemes
- Static addressing scheme
- Hardware manufacturers assign unique physical
address - Address does not change unless hardware is
replaced - Easy to use and permanent
- Configurable addressing scheme
- Mechanism to set a physical address
- Used by most network administrators because
- Address are permanent
- No large addresses because unique only to a
single network - Interface can be replaced without changing
computers physical address
43Addressing Schemes (Cont.)
- Dynamic addressing scheme
- Mechanism that automatically assigns a physical
address to the station when the station first
boots - Tries random numbers until a unique address is
found - Advantages
- No need for manufacturers to coordinate in
assigning addresses - Allows each address to be smaller
- Uniqueness is only important within a single LAN
- Disadvantages
- Lack of permanence
- Potential conflict
44Broadcasting
- Refers to transmissions available to a large
audience - All stations receive a copy of the signal each
time a frame is transmitted - To make broadcasting efficient, most LANs use
broadcast address - Hardware interface recognizes both the special
broadcast address and the stations physical
address - A frame with either of the two addresses is
accepted and delivered to the computers
operating system - Ex Finding a printer by its name
45Multicasting
- Broadcasting is extremely inefficient because
- Processing and discarding a frame requires
computational resources - Multicasting operates like broadcasting
- Single copy of the frame travels across the
network - All network interfaces receives a copy
- Interface hardware must be programmed with
specifications - Accepts or rejects frames according to the
specifications
46Multicast Addressing
- Some addresses reserved for multicast
- Interface is programmed to recognize only the
computers address and the broadcast address - Application wishing to receive multicast frames
must inform interface - Multicast address must be chosen for an
application - Application must be configured to use the address
- Passes multicast address to the interface
- Interface adds the address to the set it
recognizes
47Identifying Packets Contents
- The address does not specify what the packet
contains - Each frame contains additional information
specifying the type of the contents - Two methods used to identify contents of the
frame - Explicit frame type
- Network hardware designers specify how type
information is included in the frame - Different values used to identify various frame
types - Also called self identifying frame
- Implicit frame type
- Frame carries only data
- Sender and receiver must agree on the contents of
the frame
48Frame Headers Format
- Frame format is defined by LAN technology
- Most LAN technologies define a frame consisting
of two parts - Frame header
- Contains information such as source and
destination addresses - Data area or payload
- Contains the information being sent
- All frames have same header size but different
data area
49Ethernet Frame Format
- Begins with a header with three fields
- 64-bit preamble contains alternating 1s and 0s
for synchronization - First two fields contains physical address
- Ethernet uses 48-bit static addressing scheme
- Third field contains16-bit frame type
- Ethernet types have been standardized
50Ethernet Frame Format(Cont.)
- DIX standard specifies the values used in the
header fields and their meanings
51Networks Without Self Identifying Frames
- Some technologies do not include type field
- Type of data is specified by two approaches
- To use a single format of data
- To use first few octets of the data field to
store type information
52Type Information Standard
- IEEE standard includes a field to specify
standards organization and individual field types - Known as Logical Link Control(LLC) Sub Network
Attachment Point(SNAP) - LLS specifies that a type field follows
- SNAP contains two fields
- Organizationally Unique Identifier (OUI)
identifying organization - Second contains a type value defined by
organization - LLC/SNAP type field makes it possible to
broadcast frames
53Network Analyzer
- A device used to determine how well a network
system is performing - Most analyzers are portable flexible
- Consists of a standard portable computer and LAN
interface - User configures parameters used by analyzers
- Network interface hardware is in promiscuous mode
i.e. accepts all frames - Can be used to debug problems on a network
- Network analyzers can be configured for specific
analysis
54Network Interface Hardware
- Networks operates at a much higher speed than a
CPU - Network adapter card/ network Interface Card
(NIC) - Connects computer to a network and handles all
details of packet transmission and reception
55NIC
- Operates independent of the CPU
- Handles the details of accessing the medium and
transmitting bits - Ex Receiving a packet
- CPU allocates buffer space in memory
- Instructs NIC to read incoming packets
- NIC copies, verifies and checks the frame
- If address matches, NIC stores a copy
- Interrupts the CPU
56Thick Ethernet Wiring
- Informally called thick wire Ethernet or Thicknet
- Consists of a large coaxial cable
- Digital hardware
- NIC handles digital aspects including error
detection and address recognition - Analog hardware
- Transceiver handles analog signals
- Must for each computer
- Attaches directly to the Ethernet cable
- A separate called Attachment Unit Interface(AUI)
connects the transceiver and the NIC
57Thick Ethernet Wiring (Cont.)
- AUI cables contains many wires
- Two for data
- One each for providing power to and controlling
transceiver - Cable terminated by a terminator
- It is a resistor connecting center wire in a
cable to the shield - Prevents reflection of the signal from the end
58Connection Multiplexing
- Connection multiplexor allows multiple computers
to attach to a single transceiver - Provides exactly the same signal as a transceiver
- Cable from each computer connects to a port on
multiplexor - A single AUI cable connects the multiplexor to
the Ethernet
59Thin Ethernet Wiring
- Informally called thin wire Ethernet or Thinnet
- Uses a thinner, more flexible coaxial cable
- Advantages
- Costs less to install and operate
- No external transceivers are needed
- Uses BNC connectors instead of AUI cable
- Both thick and thin cables are coaxial, requires
termination and use the bus topology
60Twisted Pair Ethernet
- Formally called 10 Base T
- Also twisted pair Ethernet or simply TP Ethernet
- An electronic device called an
- Ethernet hub serves as a
- center of the network
- Connection from NIC to
- the hub uses twisted pair
- wiring with RJ-45
- connectors
61Office Wiring Schemes
Thicknet
Thinnet
10Base-T
62Topology Paradox
- Network technology can use a variety of wiring
schemes - Technology determines logical topology
- Wiring scheme determines the physical topology
- Physical topology can be different can be
different from logical topology - Ex A twisted pair Ethernet forms a star but
functions like a bus
63NIC and Wiring Schemes
- Network interface supports multiple wiring
schemes - A single Ethernet NIC has three connectors
- Can use only one wiring scheme at a time
- Wiring can be changed without changing NIC
64Other Network Technologies
- Different technologies accommodate a variety of
wiring scheme - Ex The original Local Talk uses transceivers
like thicknet - Uses point-to-point connection between pairs of
transceivers - Although Local Talk is a bus technology it
sometimes uses hub technology
65LAN Design
- Distance limitation is a fundamental point
- LANs use a shared communication media
- CSMA/CD or Token passing is used to guarantee
fair access to medium - LAN is designed with a fixed maximum cable length
to minimize delays - An electrical signal gradually becomes weaker as
it travels along a copper wire - This puts a limitation on the maximum length of
the wire allowed
66Fiber Optic Extensions
- LAN extension mechanisms insert additional
hardware components that can relay signals across
longer distances - Ex Optical fibers and a pair of fiber modems
- Fiber has low density and high bandwidth
- Provides a connection between a computer and a
distant Ethernet - Inserted between the network interface on a
computer and a remote transceiver
67Repeaters
- An analog electronic device that continuously
monitors signals on each cable - Used to extend LAN
- Connects two Ethernet cables called segments
- When it senses a signal on one cable, it
transmits an amplified copy on another - A repeater can double the effective length
- Any pair of computers on the extended LAN can
communicate
Repeater
68Repeaters(Cont.)
- Each repeater and segment along the path increase
delay - Ethernet standards limits that no more than four
repeaters separate any pairs of stations - The connection can be extended by using fiber
modems and Fiber Optic Intra Repeater Link (
FOIRL) - Along with valid transmissions, the repeaters
propagates a collision or electrical interference
69Bridges
- An electronic device that connects and extends
two LAN segments - Handles complete frames and uses same network
interface as a conventional computer - Helps isolate problems by forwarding only
complete and correct frames - Any pair of computers can communicate on extended
LAN
Bridge
70Frame Filtering
- A typical bridge consists of a conventional
computer with a CPU, memory and two network
interfaces - A bridge performs frame filtering
- Does not forward a frame
- unless necessary
- Uses physical address to
- determine whether to forward a frame
- Called adaptive or learning bridges because they
learn the locations of computers automatically - Uses source address to list computers
71Bridged Networks
- Bridged networks running for a long time
restricts frames to the fewest segments necessary - Propagation principle
- In the steady state, a bridge forwards each frame
only as far as necessary - Permits communication on separate segments at the
same time(parallelism) - To optimize performance, a set of computers that
interact frequently should be attached to the
same segment
72Bridging Between Buildings
- An optical fiber and pair of fiber modems are
used to extend one of the connections between a
bridge and a LAN segment - The use of a bridge has following advantages
- Single fiber connection makes it less expensive
- Individual computer can be added or removed
without installing or changing the wiring - Communication in buildings is independent
73Bridges Across Longer Distances
- Involves a long distance point-to-point
connection and special bridge hardware - Leased serial line used because it is less
expensive - Leased satellite channel used for communication
across an arbitrary distance - Bridge hardware has
- two main functions
- Filtering frames
- Buffering
74Cycle Of Bridges
- A bridge network can span many segments
- Not all bridges allowed to broadcast frames
- A cycle of bridges causes infinite number of
frames
Eight segment bridged network
Bridges connected in a cycle
75Distributed Spanning Tree(DST)
- To prevent infinite loops, a bridged network
cannot allow - All bridges forwarding all frames
- A cycle of bridged segments
- To prevent loops, bridges configure themselves
automatically - When a bridge first boots, it communicates with
other bridges - Computes Distributed Spanning Tree algorithm
- To decide which bridges will not forward frames
- DST prevents bridges from introducing a cycle
- After DST completes, bridges are arranged in a
form of a tree
76Switching
- A switched LAN consists
- of a single electronic
- device that transfers frame among many
computers - A switch simulates a bridged LAN with one
computer per segment - Consists of multiple ports each attached to a
computer - One-half of the computers can send data at the
same time
- A switched LAN consists
- of a single electronic
- device that transfers frame among many
computers - A switch simulates a bridged LAN with one
computer per segment - Consists of multiple ports each attached to a
computer - One-half of the computers can send data at the
same time
77Switches And Hubs
- Switches cost more per connection than a hub
because it provides higher aggregate data rates - Combination is used to reduce cost
- A hub connects to each port on switch
- Each computer connects to one of the hub
- Each hub appears to be a single LAN segment
- Switch makes it appears that bridges connect all
segments - Communication can occur in parallel
78Digital Telephony
- Digitization is performed by an analog-to-digital
converter(A-to-D converter - Takes analog input a signal
- Samples the signal regularly
- Computes a corresponding value at time of the
sample - Known as Pulse Code Modulation (PCM)
- Samples once every 125 µ sec and converts into an
integer between 0 and 255
79Synchronous Communication
- Telephone industry have devise complex digital
communication systems - Voice system use synchronous or clocked
technology - Most data networks use asynchronous technology
- Data moves at a precise rate in synchronous
network - Network does not slow down as traffic increases
- Telephone systems transmits additional
information along with digitized data to ensure
continuous transmission
80Digital Circuits and DSU/CSU
- leased digital circuits from common carriers form
the fundamental building blocks for long distance
computer networks - Standards differ between computer and telephone
industry - Data Service Unit/Channel Service Unit (DSU/CSU)
- Hardware needed to interface a computer to a
digital circuit - CSU portion
- Handles line termination and diagnostics
- Helps in installing and testing circuits
- Uses bit stuffing
- DSU portion
- Translates data between
- two digital formats
81Telephone Standards
82DS Standards
- A single voice channel requires 64 Kbps(8000 8
bit samples/sec) - Digital circuits are classified according to a
set of telephone standards - Most popular circuit types in North America
- T1 and T3
- Digital signal level standards or DS standards
- Specify how to multiplex phone calls onto a
single connection - 28 T1 circuits can be multiplexed over single T3
circuit
83Lower Capacity Circuits
- T1 circuit is too expensive
- Fractional T1 circuits
- Capacity much less than 1.544 Mbps
- Most popular fractional T1 rate is 56 Kbps
- Time Division Multiplexing (TDM)
- Concept of subdividing T1 circuits
84Intermediate Capacity Digital Circuits
- Slightly more than T1 and less than T3
- Inverse multiplexing is used
- Allows one to lease multiple T1 circuits
- Multiple circuits acts like a single higher
capacity circuit - Inverse multiplexor is needed at each end of line
- DSU/CSU may be required if not built in inverse
mux
85Highest Capacity Circuits
- Also termed as trunk
- Synchronous Transport Signal (STS) standards
- Specifies details of high speed connections
- Serves connections across country or between
countries
86Optical Carrier
- Higher data rates associated with the STS
standards require optical fiber - STS referred to electrical signals
- OC refers to optical signals
- Both can be concatenated (suffix C)
- C denotes a circuit with no inverse multiplexing
- OC-3 consists of 3 OC-1 operating at 51.840 Mbps
each - OC-3C (STS-3C) is a single circuit operating at
155.520 Mbps - Single circuit is more flexible
87Synchronous Optical NETwork (SONET)
- Used in North America
- Known as Synchronous Digital Hierarchy(SDH) in
Europe - Specifies details about framing, multiplexing and
synchronization - Size of the SONET frame depends on the bit rate
- Can be used to build a high capacity ring network
with multiple data circuits - Mostly used to define framing and encoding
STS-1 SONET frame
88Local Subscriber Loop
- Termed local loop or local subscriber line
- Connection between the phone company Central
Office and individual subscriber residence - Uses analog signals
- Most subscribers use a telephone to dial a local
service provider - Voice bandwidth and signal-to-noise ratio of
telephone lines limit the rate at which bits are
sent
89ISDN
- Integrated Services Digital Network
- Provides digitized voice and data over local loop
wiring - Uses twisted pair copper wiring
- Offers three separate digital channels
- B, B and D (2B D)
- The two B channels
- Operate at 64 Kbps each
- Carries digitized voice, data or compressed video
- The D channel
- Operates at 16 Kbps
- Intended as a control channel
- Manages or terminates a session
- Both B channels bonded as a single channel
90Asymmetric Digital Subscriber Line
- Asymmetric service, termed ADSL
- Bit rate in one direction is much higher
- Typical users receive more information than they
send - ADSL provides higher bit rate downstream (to the
subscriber) than upstream (from subscriber to the
provider) - Maximum downstream rate is 6.144 Mbps
- Maximum upstream rate is 640 Kbps
- Operates on local loop wiring
91ADSL (Cont.)
- ADSL is adaptive
- Modems probe the line and agree to communicate
using techniques to optimize line - Uses Discrete Multitone Modulation (DMT)
- Combination of frequency
- division and inverse multiplexing
- Divides bandwidth in 286
- separate frequencies or
- sub channels
- Selects the best frequencies
- and modulation techniques
92Other DSL Technologies
- Symmetric Digital Subscriber Line (SDSL)
- Provides symmetric rates in both direction
- Businesses prefer SDSL
- Can operate over local loops
- High-rate Digital Subscriber Line (HDSL)
- Provides DS-1 (1.544 Mbps) in two directions
- Requires two independent twisted pairs
- Able to tolerate failure
- Very-high bit rate Digital Subscriber Line(VDSL)
- Data rate of up to 52 Mbps
- Requires intermediate concentration parts
93Cable Modem Technology
- Uses cable TV wiring
- Offers higher speed and less susceptibility to
electromagnetic interference - Consists of high capacity coaxial cable
- Uses broadband signaling (frequency division
multiplexing) - One pair of cable modems is required for each
subscriber - When subscribers are more, Time Division
Multiplexing is used - One frequency for a set of subscribers
94Upstream Communication
- CATV is designed for downstream direction only
- Dual path approach
- Cable system handles only downstream traffic
- Upstream traffic travels across a dial-up
telephone connection - Needs hardware interface device to connect cable
modem and dial-up modem - Hybrid Fiber Coax (HFC)
- Combination of optical fibers and coaxial cables
- HFC can only be used with modified infrastructure
- Trunk lines replaced by optical fibers
- All amplifier modified to be bi-directional
95Large Networks
- Local Area Networks (LAN)
- Spans a single building or campus
- Metropolitan Area Networks (MAN)
- Spans a single city
- Wide Area Networks (WAN)
- Spans sites in multiple cities, countries or
continents - WAN differs from LAN
- Must be able to grow (scalability)
- Must deliver reasonable performance to large
sized networks - Must provide capacity for simultaneous
communication
96Packet Switches
- WAN is constructed from many switches to which
individual computers connect - Called a packet switch
- Moves complete packet from one connection to
another - Consists of a small computer with processor,
memory and I/O devices - Two types of I/O connectors
- One operates at high speed and connects to other
packet switches - Second operates at low speed and connects switch
to computers
97Forming a WAN
- A set of packet switches are interconnected
- A switch has multiple I/O connectors
- Forms many different topologies
- Can connect multiple computers
98Store And Forward
- WAN permits many computers to send packets
simultaneously - Uses store and forward switching
- A packet switch must buffer packets in memory
- The store operation
- Occurs when a packet arrives
- Copies the packet in memory
- Informs the processor
- The forward operation
- Processor examines the packet
- Determines the destination path
- Start the output device
- Buffers a short burst of packets that arrives
simultaneously
99Physical Addressing
- Each computer assigned a physical address
- For efficient forwarding hierarchical addressing
scheme is used - Divides an address into multiple parts
- First part indicates a packet switch
- Second part identifies computer attached to that
packet switch - An address is represented as a single binary value
100Next-hop Forwarding
- A packet switch uses destination address to
forward each packet - Next-hop forwarding
- Switch contains information about the next place
(hop) - Depends on the packets destination and not on
the source - Called source independent
101Hierarchical Addresses To Routing
- Routing
- Process of forwarding a packet to its next hop
- Routing table
- Table used to store next-hop information
- All destination addresses have an identical first
part - Using only the first part helps in
- Reducing computation time
- Shortening routing table
- The final packet switch
- uses the second part
102Routing in a WAN
- A WAN with large capacity can be build by
increasing switching capacity - Interior switches
- Handles load, but need not have computers
attached - Exterior switches
- Packet switches to which computers attach
- Both switches have routing tables
- Universal routing
- Routing table contains next-hop route for each
possible destination - Optimal routes
- The next-hop value points to the shortest path to
the destination
103Routing in a WAN (Cont.)
- A graph can model a network
- Each node corresponds to a packet switch
- Each link corresponds to a direct connection
104Default Routes
- A graph representing a large WAN may contain many
duplicate entries - Default route or Default routing table
- A long list of entries having same next-hop value
is replaced by a single entry - Only one default entry is allowed in any routing
table - A default entry is present only if more than one
destination has the same next-hop value
105Routing Table Computation
- Static routing
- A program computes and installs routes when a
packet switch boots, the routes do not change - Dynamic routing
- A program builds an initial routing table and
then alters the table as condition changes - Static routing is simple and has low overhead
- Most networks use dynamic routing because
- Handles problems automatically
- Modifies routes to accommodate failures
106Shortest Path Computation
- Dijkstras algorithm
- Finds the distance along a shortest path from a
single source node to each of the other nodes in
the graph - A next-hop routing table is constructed during
the computation of shortest path - Uses weights on edges as a measure of distance
- A path with fewest number of edges may not be the
path with least weight
The shortest path between 4 5 is shown darkened
107Distributed Route Computation
- Each packet switch computes its routing table
locally - Informs the network of the result
- Sends routing information to neighbors
periodically - Each packet switch learns the shortest path to
all destinations - Produces the same next-hop routing table as
Dijkstras algorithm - Allows the network to adapt to a failure
108Distance Vector Routing
- Distance-vector algorithm uses distributed route
computation - Each link in network is assigned a weight
- Distance to a destination is defined to be the
sum of weights along the paths - A packet switch periodically updates the network
- Each message contains pairs of (destination,distan
ce)
109Link-State Routing (SPF)
- Also called shortest path first or SPF routing
- Packet switches sends messages with status of the
link - Message broadcast to all switches
- Each switch collects information and builds the
graph of the network - Switches use Dijkstraa algorithm to produce
routing table - SPF algorithm can adapt to hardware failures
110WAN Technologies
- ARPANET
- One of the first packet switched WANs
- Fast when invented, slow by current standards
- X.25
- Developed an early standard for WAN technology
- More popular in Europe
- Frame relay
- Accepts and delivers blocks of data
- Must operate at high data rates
- Switched Multi-megabit Data Service( SMDS)
- Offered by long-distance carriers
- Operates at speed faster than frame relay
111Asynchronous Transfer Mode (ATM)
- Provides voice, video and data services across a
wide area - Has high data rates, low delay and low jitter
(low variance in delay) - Data divided into fixed sized packets called
cells - Each ATM cell has 53 octets
- 5 for header information and 48 for data
- A Constant Bit Rate (CBR) is specified for voice
or video - Uses switches as primary building blocks
- Uses optical fiber as interconnection media
112Network Ownership
- Private networks
- Owned and used by a single company or an
individual - Public networks
- Owned by common carriers such as telephone
networks - Anyone can subscribe to the service and connect a
computer - LAN technology is most often used for public
networks - Almost all public networks are WANs
- The chief advantage with private networks is
complete control - Public networks are flexible and able to use
state-of-the-art networking without maintaining
technical expertise
113Virtual Private Networks (VPN)
- Combines advantages of both private and public
networks - Allows a company with multiple sites to have
private network - Uses a public network as a carrier
- VPN technology restricts traffic only between the
companys sites - A special hardware and software system is placed
between companys and public network - VPN encrypts each packet before transmission
- Network manager must also configure routing
114Service Paradigm
- Connection-oriented service
- Operates analogous to a telephone system
- Requires a pair of computers to establish a
connection before sending data - Either computer can choose to terminate the
connection - Connection-less service
- Operates analogous to a postal mail system
- Computers do not need to establish a connection
before they can communicate - Accepts and delivers individual frames that each
specify a destination - Less initial overhead
115Connection Types
- Permanent connection
- Dedicated wires between a pair of computers
- Is persistent and always available
- Does not require maintenance
- Always ready to accept data
- Switched connection
- Must establish a connection to communicate
- Each computer maintains physical connection to
network - Is flexible and general
- Permanent connections survive
- either a computer or
- a network reboot
Switched Connections
116Examples of Service Paradigms
117Connection Identifiers
- Connection oriented service uses abbreviated
addresses - A small integer used to communicate after a
connection is established - Ex ATM network uses 28-bit connection
identifiers - The computer places the identifier in each
outgoing cell - ATM divides connection identifiers into two parts
- 12-bit virtual path
- identifier ( VPI )
- 16-bit virtual circuit
- identifier ( VCI)
ID - Identifier
118Network Performance Characteristics
- Delay
- Specifies how long it takes for a bit of data to
travel across the network ( in seconds) - Propagation delay
- Time a signal requires to travel across a wire or
optical fiber - Switching delay
- Delay introduced by electronic devices in network
- Access delay
- Delays caused when waiting to access a shared
media - Queuing delay
- Occurs in packet switched WAN because it enqueues
packets
119Network Performance Characteristics (Cont.)
- Throughput
- Measure of the rate at which data can be sent
through a network - Specified in bits per second, bps
- Throughput is measure of capacity, not speed
- Throughput and delay are related by
- D D0 / ( 1 U)
- D0 idle network delay
- U current utilization between 0 and 1
- D Effective delay
- Volume of data present on the network
- Product of delay and throughput ( T D )
-
120Protocols
- Protocol
- A set of rules that specify the format of
messages and the appropriate action required for
each message - Protocol software
- The software that implements such rules
- Application programs do not interact with network
hardware - Communication software is divided into multiple
protocols - Protocols are designed and developed in complete,
cooperative sets called suites or families
121Protocol Design
- Layering model
- Describes one way a communication problem can be
divided into sub-pieces called layers - ISO defined a 7-layer model
122The Seven Layers
- Layer 1 Physical
- Corresponds to basic network hardware
- Ex RS 232
- Layer 2 Data Link
- Specifies how to organize data into frames and
transmit over a network - Ex Frame format and CRC
- Layer 3 Network
- Specifies how addresses are assigned and how
packets are forwarded - Layer 4 Transport
- Specifies how to handle details of reliable
transfer
123The Seven Layers (Cont.)
- Layer 5 Session
- Specifies how to establish a communication
session with a remote system - Ex Security details
- Layer 6 Presentation
- Specifies how to represent data
- Needed to translate from the representation on
one computer to another - Layer 7 Application
- Specifies how one particular application uses the
network - Ex specifications for an application that
transfers files
124Stacks Layered Software
- When protocol software sends or receives data,
each module only communicates with the next
highest and lowest level - Incoming and outgoing data passes through each
layer
125Stacks Layered Software (Cont.)
- Vendors use the word stack to refer to protocol
software - Software in the given layer on the sending
computer adds information to outgoing data - Software in the same layer on receiving computer
uses the additional information to process
incoming data
Stacks are incompatible
126Multiple, Nested Headers
- Each layer places additional information in a
header before sending data to a lower layer - The header corresponding to the lowest-level
protocol occurs first
127Scientific Basis for Layering
- Layering principle
- Layer N software on the destination computer must
receive the exact message sent by layer N
software on the sending computer - Whatever transformation a protocol applies before
sending a frame must be completely reversed when
the frame is received
128Techniques Protocols Use
- Sequencing for Out-of-order Delivery
- Connectionless networks often deliver packets out
of order - To handle this transfer protocol use sequencing
- Each packet has a sequence number
- Sequencing to Eliminate Duplicate Packets
- Malfunctioning hardware causes packet duplication
- Ex a transceiver using CSMA/CD
- Sequencing solves the problem of duplication
129Techniques Protocols Use (Cont.)
- Retransmit ting Lost Packets
- Protocols use positive acknowledgement with
retransmission - Protocol software uses a timer
- Protocols bound the maximum number of
retransmissions - Avoiding Replay Caused by Excessive Delay
- Replay means that an old, delayed packet affects
later communication - A correct packet may be discarded as a duplicate
- Protocols mark each session with a unique ID
130Techniques Protocols Use (Cont.)
- Flow Control to Prevent Data Overrun
- Data overrun A computer sends data faster than
the destination can absorb - Flow control techniques
4 packet Window
Stop-and-go Flow control
Sliding Window
131Techniques Protocols Use (Cont.)
- Mechanisms to Avoid Net Congestion
- Congestion More packets arrive than can be send
- The queue grows and the effective delay increases
- Congestion collapse
- Persistent congestion causes the entire network
to become unusable - Protocols avoid congestion collapse by
- Arranging for packet switches to inform senders
when congestion occurs - Use packet loss as as estimate of congestion
Congestion prone network
Packet Switches