Transcript Document

Data Link Layer Issues
Dealing with Different Types of Networks
Types of Networks

Network hardware can be categorized into:

Circuit-switched (e.g. telephone)

Prior to communication, the hardware establishes a dedicated
end-to-end connection



Since there is a dedicated connection, a continuous stream of bytes
can be sent
Frequency or time-division multiplexing can be used to share
links in such a network
Packet-switched (e.g. Ethernet, ATM)


Data is divided into packets of limited size, and each is
forwarded through the network to the destination
This can be done by routers or switches
Types of Networks

Disadvantages

Circuit-switched



A dedicated connection that has no transmission means wasted
bandwidth
A connection is time consuming if short, infrequent, or sporadic
communication is to occur
Packet-switched


Forwarding each packet means that each router must decide the
next hop for every packet (even for the same destination)
Routers are typically network slowdowns due to the amount of
processing, as well as input/output buffering
Types of Networks

Circuit-switching is used in a telephone conversation

A connection to the receiver is established by the sender (the caller)



The telephone company reserves a certain bandwidth (64 Kbps for voice
communication) for this call
If the bandwidth is not used by the callers, it is wasted
Packet-switching is similar to the postal service

Each message (envelope) is addressed to the recipient individually,
and the postal service delivers each message to the recipient


The postal service may deliver these envelopes through different cities
and methods of transport (airplane, truck, …)
It can be said that these messages can be delivered using different routes
Circuit-Switching
A
TalkB
Call:
Telephone Company
Switching System
Talk
Disconnect
B
Circuit-Switching
A
Telephone Company
Switching System
B
Packet-Switching
A
Quebec, QC
Buffalo, NY
Montreal, QC
Windsor, ON
Toronto, ON
Niagara Falls, ON
London, ON
Ottawa, ON
Kitchener, ON
Postal Network
B
Types of Packet-Switching

Virtual circuit-switching



A ‘virtual circuit’ is created between source and
destination
This VC is used for all subsequent sending of
packets
Datagram

Each packet is routed individually
Virtual Circuit Packet-Switching
Advantages
 After the first message, routing is faster




Because a connection is created, the connection identifier can
be used (alone) to address packets


A route must only be determined once, for the first message
Once the route has been determined, the path used by the router is
reused for all messages
As a result, routing tables are much smaller (and can be searched
more quickly)
Typically, such as with ATM cells, this can reduce the size of a
cell/packet’s header
Messages do not arrive out of order

As a result, receivers do not need to reorder the cells
Virtual Circuit Packet-Switching
Disadvantages
 Connections take some time to create


Infrequent messaging is not suitable for connection-based
messaging



Routers/switches must intercommunicate in order to create the
connection
The connection may be lost after a timeout, and will have to be
recreated again and again
The time delay for creating the connection may outweigh the speed
benefits of using connection-based transport
Routing tables will be dynamic, and routing algorithms are
more complex
Datagram Packet-Switching
Advantages
 Connections need not be created
 Infrequent messaging is perfect for connectionless messaging


Connectionless messaging can be resumed after any amount of delay,
any number of times, without any delays due to the resumption of
communication
Routing each message separately allows for load balancing

Some messages may be sent through one route, but when that route
becomes saturated, messages may then be sent through a different
route in order to achieve the most optimal communication possible
Datagram Packet-Switching
Disadvantages
 Each message takes a certain amount of time to transmit
(including transmission, routing, reception, etc.)

Nodes communicating large amounts of information in a short time
will:



Use a lot of bandwidth for things such as header information
Waste a lot of time routing messages to the same destination
Messages may arrive out of order

Messages must be reordered by the recipient
Multiple Access Strategies
Schemes for Sharing a Communication
Medium
Multiple Access

Most networks are shared medium


This means that a single medium (e.g. radio frequency) is
shared by all of a network’s hosts
We need a scheme to allow the hosts to share the
medium, without collisions


Collisions occur when two (or more) messages are
transmitted at the same time
The result is constructive and destructive interference in
the carrier wave

This causes the messages to be combined and scrambled
Contention


In contention networks, any node that has a
packet to send, merely sends the packet
It is clear that this type of network frequently
experiences collisions


The more nodes trying to communicate, the
higher the chance of collisions
Thus, contention networks are severely limited in
the number of hosts possible
Contention
Transmit
Contention
Contention
Transmit
Contention
Contention: Collisions
Transmit
Transmit
Contention: Collisions
Scrambled
Signal
Contention

No collision avoidance is present



Messages are just sent
When collisions occur, the messages are simply
resent after some random (or pseudo-random)
amount of time
Collisions can occur anytime
Carrier Sensing
Test the medium
for a signal
Carrier Sensing
Test the medium for a signal: Available
Transmit
Carrier Sensing
Test the medium
for a signal
Carrier Sensing
Test the medium
for a signal: In use
Carrier Sensing
Transmission
Complete
Carrier Sensing
Test the medium
for a signal
Carrier Sensing
Test the medium
for a signal: Available
Carrier Sensing: Collisions
Test the medium
for a signal: Available
Test the medium
for a signal: Available
Carrier Sensing: Collisions
Transmit
Transmit
Carrier Sensing: Collisions
Scrambled data
Carrier Sensing: Collisions
Detect
Transmit
collision
Detect
Transmit
collision
Carrier Sensing (CSMA) 1

To reduce the number of collisions, the medium is
tested for a signal before each transmission



Collisions can still occur (although less often)


If a signal exists, the node waits
Signal testing can be anything from detection of an
electrical signal, to testing for photons
1
If a node tests for a signal before a transmission from
another node, and transmits after, a collision occurs
CSMA/CA is short for:

Carrier Sense Multiple Access with Collision Avoidance
Carrier Sensing Hardware
Transmitter
When a signal is
detected,
transmissions are
blocked by the
signal detector
Receiver
Signal
Detector
If the message is
broadcast or the
address is this
station’s address,
the message is
forwarded to the
receiver
CSMA/CA

CSMA/CA networks (such as wireless 802.11g) also
use carrier sensing and collision detect


However, detecting collisions in wireless networks is
significantly more complicated
Also, after detecting carrier and determining there is
no signal, a CSMA/CA network transmits a ‘Do not
broadcast’ message

If this message is sent without a collision, the host can
assume it is safe to transmit
Carrier Sensing Networks
Advantages
 No tokens


Simple hardware
No need for token transmission
Disadvantages
 Collisions


Wasted bandwidth for re-transmits
Require complicated re-collision avoidance schemes
Token Passing
Transmit
T
Token Passing
Transfer
Token
T
Token Passing
T
Token Passing
Transmit
T
Token Passing

A small packet (the token) is passed from
node to node


When a node has the token, it has sole use of the
network medium
There are no collisions


The nodes must have the token in order to
transmit
The network hardware ensures that there is only
one token at any given time
Token-Based Networks
Advantages
 No collisions, so no bandwidth is wasted by
collisions and re-transmits
 No need for re-collision avoidance schemes
Disadvantages
 Token transmission uses bandwidth
 More complicated hardware

Hardware must be built to use tokens, dynamically
determine token sequence, etc.
Local Area Networks




Networks which span a small geographic area
They typically represent high bandwidth,
short delays, few errors
They commonly support features such as
broadcasting, multicasting
They are typically limited to hundreds of
network nodes (maximum)
Typical Local Area Networks

A collection of computers in the same room


e.g. The basement of the computer centre
All computers within an office building

e.g. The computers in the offices of the professors
and staff in Lambton tower
Local Area Network
Topologies
Structures of LANs
Token Bus Networks

The token is passed in a specific sequence




Nodes must know the address if the next node in the
sequence
The token sequence is not necessarily in the same order as
the physical order of nodes on the communication
medium
When a node has completed transmission, it
forwards the token, addressed to the next node in the
token sequence
The token sequence forms a ‘logical ring’
Common Token Bus Networks

IEEE 802.4 networks

Nodes are share a communication medium
similar to that of Ethernet (IEEE 802.3)

Coaxial cable connection
Token Bus Operation
Transmit
A
C
B
Token sequence: C,A,D,B
D
Token Bus Operation
Transmit
Token
A
C
B
Token sequence: C,A,D,B
D
Token Bus Operation
Receive
Token
A
C
B
Token sequence: C,A,D,B
D
Token Bus Operation
Transmit
A
C
B
Token sequence: C,A,D,B
D
Token Bus Operation
Transmit
Token
A
C
B
Token sequence: C,A,D,B
D
Token Bus Operation
A
C
B
D
Receive
Token
Token sequence: C,A,D,B
Token Bus Operation
A
C
B
D
Transmit
Token sequence: C,A,D,B
Token Ring Networks


The token is passed to each node, in the
physical order on the network
The physical medium must be a closed loop to
meet this network category

So the token can keep going around the network
Common Token Ring Networks

IEEE 802.5 networks


FDDI networks (fibre distributed data interface)


Nodes are share a coaxial communication medium similar
to that of Ethernet (IEEE 802.3)
Nodes use 2 fibre optic rings as the communication
medium
CDDI networks (copper dist. data interface)


Based on FDDI technology, but uses copper wiring
similar to 802.4
However, CDDI uses 2 rings like FDDI
Token Ring Operation
A
Transmit
D
B
C
Token Ring Operation
A
Transmit
Token
D
B
C
Token Ring Operation
A
D
B
C
Receive
Token
Token Ring Operation
A
D
B
C
Transmit
Token Ring Operation
A
D
B
C
Transmit
Token
Token Ring Operation
A
D
B
Receive
Token
C
Token Ring Operation
A
D
B
C
Transmit
Bus and Ring Networks
Advantages
 Less wiring is necessary
Disadvantages
 Node failure can mean partial (or complete)
LAN failure

This can mean locating network problems is also
more difficult
Star Topology




Star networks send all messages through a
central hub
Each node on the network is wired separately
to the hub
Star networks are not a shared bus technology,
but a private bus technology
However, nodes still share the hub
Common Star Networks

Twisted pair Ethernet (logical star):


All nodes connect to a central hub (an Ethernet
hub) via Cat5 cables
The hub forwards messages to all wires, and the
destination node keeps the message


Other nodes ignore the message
An Ethernet switch (similar to an ATM switch)
forwards only in the one correct direction (or not,
if appropriate)
Star Network Operation
Transmit
A
B
Hub
C
D
Star Network Operation
A
B
Hub
C
D
Receive
Star Network Operation
Transmit
A
B
Hub
C
D
Star Network Operation
A
B
Hub
C
Receive
D
Twisted Pair Ethernet

Physically, all Ethernet types are bus networks


However, the actual layout of the cables in
twisted pair Ethernet forms a star topology
Twisted pair is called a logical star topology,
while still a physical bus topology
Twisted Pair Ethernet as a Bus
Short Shared Bus
A
B
C
D
G
H
Hub
E
F
Long Private Lines
Traditional Ethernet as a Bus
Long Shared Bus
A
B
C
D
E
F
G
H
Short Private Lines
Star Topology
Advantages
 Simple installation and wiring
 Node failures do not affect the rest of the system
Disadvantages
 All traffic passes through same hub, so network bandwidth is
limited by hub speed




This can be reduced with buffers inside hubs which store messages
that come in when the hub is busy
Hub failure = LAN failure
More wiring
Duplication of messages
LAN Service Models


In general, most LANs implement (in some sense)
the OSI reference model
The IEEE committee on LAN technology (IEEE
802) chose to subdivide the Data Link Layer into 2
sub-layers:
1.
MAC (Medium Access Control): Deals with issues
specific to each type of LAN

2.
Such as token passing, collision detection, error detection, etc.
LLC (Logical Link Control): Deals with issues common
to all LAN types

Such as data transmission, etc.
Data Link Addressing

The data link layer is represents the network


Addressing, then, is specific to the network hardware


e.g. Ethernet
MAC addresses are typically used for this purpose
These addresses are not used in routing



They are only used on a single network
Thus, they are used for hop to hop delivery
End-to-end delivery is the domain of the Network layer
MAC Addresses

Officially the IEEE 802 committee standardized
addresses to be 16bit, 48bit, and even 60bit

48bit addresses (in use by most LANs covered by the 802
committee) allow for globally unique identifiers (GUIDs)
to be assigned to each network card by the manufacturer



As a result, each NIC can be uniquely identified on any network
These are called MAC addresses, due to the Data
Link sub-layer that deals with them
e.g. 8D-F0-A6-75-9C-13
Data Link Flow Control


Flow control is limiting the packet rate so that
both the source or destination can keep up
At the data link layer, source and destination
are on the same LAN

Thus, limiting the packet rate is relatively easy
Data Link Reliability

Reliability:

Best effort: The network takes no steps to ensure packets
arrive


Reliable: The network uses acknowledgements to ensure
packets arrive



The majority of packets should be received without problems
When packets are lost (for whatever reason), they are handled
appropriately
Error handling: Corrupt packets should be re-sent
Reliability at the Data Link layer is usually
unnecessary, since the Transport layer will typically
be able to do it more efficiently
Error Control


Error control is achieved using one of the
following methods:
Checksum: An n-bit sum is taken of the
binary stream



In other words, a checksum counts the ones
What if one 0 became a 1 and a 1 became a 0??
Cyclical redundancy check:

Should generate different CRC values, despite the
same number of 0s and 1s
Ethernet
An Early Incarnation of LANs
What Started It All
Robert Metcalfe (from Xerox PARC)
Ethernet History


In 1973, Xerox PARC developed a packet-switched
LAN, called Ethernet
In 1978, IEEE created a standard (802.3) based on
the research of Xerox, Intel, and DEC



IEEE: Institute of Electrical and Electronics Engineers
802.3 Ethernet uses a coaxial cable to connect nodes
(called 10Base5 or ThickNet)
Since then, several new forms of Ethernet have
evolved
ThickNet (10Base5)
Outer Insulating Jacket
Inner
Insulating
Layer
Braided
Metal
Shield
(Ground)
Transmission
Wire
½ Inch Diameter
10Base5
5 => 0.5”
ThickNet (10Base5)
10Base5
10 => 10 Mbps
•Each network node uses
a transceiver
•A transceiver ‘taps’ into
the wire through holes
•Maximum throughput is
10 million bits per second
(10 Mbps)
Transceiver
ThinNet (10Base2)




Create as an inexpensive alternative to ThickNet (or
10Base2)
Called thin-wire Ethernet, because it uses a thin
cable with less shielding
Less shielding means more interference, so cable
placement is important
10Base2 does not use transceivers, which are
expensive, which further reduces cost
ThinNet (10Base2)
Node A
Node B
10Base2
10 => 10 Mbps
Node C
10Base2
2 => 0.2”
Node D
•The signal passes through each node
•The network interface card (NIC) retransmits
the signal, so transceivers are not required
•Maximum throughput is 10 million bits per
second (10 Mbps)
Twisted Pair Ethernet (10BaseT)




Uses 4 pairs of twisted wires inside an
unshielded cable
The twisting of the wires reduces interference
The absence of shielding makes the cable
flexible and inexpensive
The cable is capable of 10Mbps
Twisted Pair Ethernet



Connectors on twisted pair Ethernet (RJ45) look
similar to telephone wire connectors (RJ11)
This kind of Ethernet uses unshielded twisted pair
(UTP)
UTP cable has various categories:



Cat3: Can only be used for 10BaseT
Cat5: Can be used for 10BaseT, 100BaseT
Cat5e, Cat6: Can be used for up to 1000BaseT
ThinNet Ethernet
011100110
011100110
Twisted Pair Ethernet
accept
message
011100110
ignore
ignore
ignore
10 Mbps Ethernet Overview


10Base2 and 10Base5 both used coaxial cable
which joined each node in a line
10BaseT uses UTP cabling, where each node
is directly connected with the hub

The hub receives messages and forwards them to
all nodes

The one that is connected to the recipient node
Fast Ethernet

Using the same Cat5 cabling used for 10BaseT, an
Ethernet-based LAN that operates at 100 Mbps
(100BaseT) is possible


Standard: IEEE 802.3u
While using the same cable, network hubs and
network interface cards (NICs) must be upgraded to
transmit messages at 100 Mbps
Fast Ethernet


While very few computers can handle 100 Mbps
throughput (bus speeds of computers are often
slower than this), multiple computers can share this
bandwidth
10/100 Ethernet (or 10/100 switched Ethernet)
allows you to use the same NICs and hubs for both
10BaseT and 100BaseT


If a NIC and hub can both handle 100BaseT, that speed is
used, otherwise 10BaseT is used
10/100 Ethernet allows you to slowly upgrade your
network with minimal downtime
Gigabit Ethernet

Gigabit Ethernet allows for 1000 Mbps throughput

Gigabit Ethernet (Gig-E) can use Cat5 cabling
(1000BaseT) or shielded Cat5E cabling (1000BaseTX)



Standard: IEEE 802.3ab
Gig-E pushes the limits of the speed capable with Cat5
cabling, due to interference with the electrical signal,
Cat5E cabling results in better performance
Gigabit Ethernet is so fast, that it is sometimes used as
a backbone for a Wide Area Network (WAN) instead of
more expensive optical networks

e.g. One of the backbones of the network here at the U
Ethernet Future

Another form of Gigabit Ethernet which uses fibre
optic cabling has been proposed (802.3z)



Using multimode (multiple channel – 1000BaseSX), or
single mode (1000BaseLH, 1000BaseZX)
Research groups are in the process of developing
10 Gigabit Ethernet (802.3ae)
This research is managed by the 10 Gigabit
Ethernet Alliance
http://www.10gea.org
LAN Service Models
LLC (Logical Link Control), for LANs, can
be one of two types:
Type 1: A straight datagram scheme



The packet is delivered using best-effort service
No acknowledgements are used to ensure packet
arrival
Type 2: A reliable scheme


Packets are numbered
Packets are acknowledged as they are received
IEEE 802 Committees

Five 802 committees were developed to research
various technologies associated with LANs:

802.1: Issues common to all LANs


802.2: Issues related to the LLC sub-layer



e.g. reliability schemes, packet transmission
802.3: Issues related to CSMA/CD category LANs


e.g. addressing, management, bridges
e.g. Ethernet
802.4: Issues related to token bus category LANs
802.5: Issues related to token ring category LANs
LAN Addresses

The 48 bit addresses (often called MAC
addresses) are the ones used by Ethernet
LANs


e.g. 02-60-8C-08-E1-0C
All Ethernet cards contain a globally unique
MAC address
Ethernet Overview

Ethernet is not a reliable service



Most Ethernet networks use broadcasting to achieve
messaging


There are no acknowledgements for packet receipt
Ethernet uses best-effort delivery
Each message is received by each node
Ethernet is one network in a category of networks
known as shared bus networks

Each node shares a single communication medium
Ethernet Overview

Ethernet is a carrier-sensing network

Carrier-sensing networks use distributed access
control methods




Each station determines whether it can access the
communication medium
Each station senses whether or not the
transmission medium (wire) is charged
If not, an attempt at transmission is made
If so, the node will wait and sense again
Ethernet Overview

Sometimes, more than one station will attempt to
transmit at roughly the same time


This is called a collision
Due to the finite speed of electrons traversing a wire


Or due to the finite speed of photons moving through
glass


70% of the speed of light
The speed of light
The two (or more) messages collide or interfere with one
another, creating scrambled data packets
Collision Detection in Ethernet

When scrambled messages are read by the
transmitting stations, it is determined to be a
collision


Both (or all) of the stations involved will detect
the collision
This type of network is known as CSMA/CD


Carrier-sensing, multiple access with collision
detection
Each station must retransmit their packets
Collision Avoidance in Ethernet


After a collision occurs, if both stations tried
to transmit after the same period of time,
another collision would occur
To combat this, Ethernet uses a binary
exponential back-off policy

Each subsequent collision would cause the station
to wait double the amount of time before
reattempting transmission
Ethernet Packets (Frames)

Size: 64 octets – 1518 octets


An octet is another term for an 8-bit byte
The frame contains more than just data



The source and destination addresses
An identifier, signifying that the frame is in fact
an Ethernet frame
A Cyclical Redundancy Check (CRC) to ensure
data integrity upon arrival
Ethernet Frames
8 octets
Preamble
6 octets
Dest Address
6 octets
Source Address
2 octets
Frame Type
46-1500
Data
4 octets
CRC
Sequence of 01010101 used to
synchronize the receiving station
The MAC address of the
destination node
The MAC address of the sender
node
The identifier used to identify the
frame as an Ethernet frame
The data to be sent to the
destination
A cyclical redundancy check (CRC)
used to determine if data has
been corrupted
Ethernet Distance Limitations

Coaxial Ethernet cables have a maximum length


Due to signal deterioration
This length could be extended using repeaters


Machines that read signals through a port and recreate
them (at full strength) out another port
The use of more than 2 repeaters between any 2 stations
would interfere with times used in CSMA/CD schemes

As a result, a maximum of 2 repeaters can be placed between any
2 nodes
Ethernet Distance Limitations

Ethernet LAN sizes could also be increased by using
Bridges to connect separate LANs into a single LAN


Bridges filter out erroneous frames, as well as line noise
Some bridges (adaptive bridges) are even intelligent
enough to know when a frame must be forwarded or not

e.g. If the destination node is not on the other side of a Bridge,
the frame need not be forwarded
FDDI
Fiber Distributed Data Interconnect
FDDI

Use optical fibre cabling as a shared communication
medium




Optical fibre cables are made of glass
Because they are so thin, they are fairly flexible
Capable of 100 Mbps
Light is used to transmit data

Light is not susceptible to electrical interference



Optical cabling can span longer distances
Optical cabling does not need to be shielded near devices which
generate electromagnetic interference
Light waves (photons) travel faster than electrons
FDDI

Is a token-ring category network

A token is passed from station to station



When a station receives the token, it may transmit data
If a station has no data, it allows the token to pass to the next
station
FDDI uses 2 rings of cabling, moving in opposite
directions


The second ring is used to allow twice the flow of data
The purpose of the second ring is to allow data to reach its
destination, even when one station has failed (and cannot
forward messages)
FDDI Ring Technology
FDDI With Node Failure
FDDI Token Passing
S:12
D:07
T 12
1
S:12
D:07
S:12
D:07
2
S:12
D:07
S:12
D:07
3
S:12
D:07
S:12
D:07
4
S:12
D:07
S:12
D:07
5
S:12
D:07
S:12
D:07
11
10
9
8
S:12
D:07
6
7
S:12
D:07
S:12
D:07
FDDI Token Passing
T
T
1
2
3
4
12
5
11
6
10
9
8
7
FDDI Frames
octets: 2+
1
1
2 or 6
2 or 6
0-30
0-4500
4
0.5
1.5+
Preamble
Start Delimiter
Frame Control
Dest Address
Source Address
Routing Info
Data
FCS
End Delimiter
Frame Status
Data Used to Synchronize Stations
Indicates Start of Frame
Identifies the Type of Frame
Address of the Destination Node
Address of the Source Node
Routing Information
Frame Data
Frame Check Sequence
Indicates End of Frame
Status of Frame
Wireless Networks
Radio-Based LANs
Wireless LANs


Contrary to one’s initial guess, wireless LANs
are very similar to ‘wired’ LANs
Wireless LANs are a shared media network,
just like Ethernet

However, in a wireless LAN, the shared medium
is not the air, but something called a base station
or wireless access point
Wireless LANs (WLANs)

The wireless access point, which is similar to a hub,
is the shared medium


Despite the fact that radio waves using the same
frequency will cause mutual interference, the air is not
generally considered a shared medium
Technically speaking, twisted pair Ethernet is similar
to WLANs


The cables themselves are just point-to-point connectors
and are not shared
The hub/switch, however, is shared
Wireless LANs (WLAN)

Wireless Access Point (WAP): A base station that coordinates
transmission between one or more wireless hosts




Analogous to a cell tower in a mobile phone network
Wireless hosts must be a certain distance away from a WAP to
participate on a WLAN
The communicable area of all of the WAPs in a WLAN, define the
coverage area for the WLAN
Some WLANs do without a WAP, but pass messages directly
to one another

These are typically small (2-3 hosts) networks, and are called ad hoc
networks
802.11 Operation

802.11 networks (such as 802.11g) use CSMA/CA multiple
access scheme


Hosts try to detect carrier before sending (CS)
This is not adequate, since there could be hidden hosts

These are hosts out of range of this host, but in range of the same base
station:
802.11 Operation

To avoid collisions with hidden hosts:


The host will send a ‘request to send’ (RTS) frame before
transmitting
The base station will respond with a ‘clear to send’ (CTS)
frame if the channel is clear



Once a base station sends a CTS, it will reject any further RTS
requests until the data is received by the host who sent the first
RTS
This is called collision avoidance (CA)
Frames are acknowledged at the data link layer in
802.11 networks
802.11 Frame Format
Frame Control (2 octets)
Source Address (6)
Destination Address (6)
Receiving Station Address (6)
Sequence Control (2)
Transmitting Station Address (6)
Data (0-2312)
Frame Check Sequence (2)
Flags
MAC Address of sending host
MAC Address of receiving host
MAC Address of sender base station
Fragment number, sequence number
MAC Address of receiver base station
Frame data
CRC for frame header and data
802.11 Frame Header: Frame Control
Protocol Version (2 bits)
Type (2)
Subtype (4)
To AP (1)
From AP (1)
More Fragments (1)
Retry (1)
Power Management (1)
More Data (1)
WEP (1)
Order (1)
Flags
Management, control or data frame
Type of management or control frame
Sent to an access point?
Sent by an access point?
Are there more fragments from this frame?
Is this a retransmission of a previous frame?
Power state of sender after transmission
Is there more data to come?
Has WEP encryption been applied to frame?
Are the packets strictly ordered?
Wireless Access Points
WAP2
WAP1
WAP3
Ad Hoc Networks

In ad hoc networks, stations directly transmit
to one another


Hosts are responsible for routing, addressing,
name translation, security, etc.
Two ad hoc networks using the same
frequency, within range of one another will
cause conflicts

Thus, different frequencies should be used
Handoffs in WAPs

For WLANs with WAPs, roaming hosts must be
considered


If a host moves into the range of another WAP, then out of
range of their current WAP, a handoff takes place
A handoff is when one WAP gives the responsibility for a
particular host to one of its neighbouring WAPs

The two WAPs must communicate for this to happen, and thus
neighbouring WAPs must be within each other’s transmission
range
Wireless LAN Standards

Some of the main standardized WLANs:


802.11a, 802.11g: 54Mbps, comparable with 100BaseT
Ethernet, under 100M range
802.11b: 11Mbps, comparable to 10BaseT Ethernet,
under 100M range


These technologies are intended for LANs within the same small
to medium-sized building
BlueTooth/802.15: 721 kbps, under 10M range

This technology is intended for communicate within one room or
vehicle