Computer Communications - FTP Directory Listing

curvyrawrNetworking and Communications

Oct 23, 2013 (3 years and 7 months ago)

67 views


By José M. Matos.


Do not use without permission.


Computer Communications



The OSI reference model

................................
................................
................................
....

2

The Application layer
................................
................................
................................
..........

4

The Transport layer

................................
................................
................................
.............

5

The Network layer

................................
................................
................................
..............

6

The Data Link layer

................................
................................
................................
............

7

Medium Access Control (MAC) layer
................................
................................
................

7

Physical layer

................................
................................
................................
......................

8

Circuit switching

................................
................................
................................
.................

9

Packet switching

................................
................................
................................
...............

10

Routing
................................
................................
................................
..............................

11

X.25
................................
................................
................................
................................
...

12

Local Area

Network (LAN)

................................
................................
..............................

13

Medium Access Control (MAC)
................................
................................
.......................

14

Frequency Division Multiplexing (FDM)
................................
................................
.........

15

Time Division Multiplexing (TDM)

................................
................................
.................

16

ALOHA
................................
................................
................................
.............................

17

Carrier Sense Multiple Access (CSMA)

................................
................................
...........

18

IEEE Standard 802.3: Ethernet

................................
................................
.........................

19

IEEE Standard 802.4: Token bus

................................
................................
......................

20

IEEE Standard
802.11: Wireless LANs

................................
................................
............

21

Asynchronous Transfer Mode (ATM)

................................
................................
..............

23

Internet Protocol (IP)

................................
................................
................................
........

25


By José M. Matos.


Do not use without permission.


The OSI
reference model


The International Standards Organization (ISO) has developed a model to describe the
communication process between two open systems. An open system is any computer that
is available to communicate with other computer. The name of the model

is Open System
Interconnection Reference Model, OSI for short. The model is divided in layers and they
“sit” one over the other in the following order:




Application protocol



Application layer






Application layer





Presentation protocol



Pr
esentation layer






Presentation layer





Session protocol



Session layer






Session layer





Transport protocol



Transport layer






Transport layer




Network layer host
-
router protocol


Network layer



Network layer



Network

layer




Data link layer host
-
router protocol


Data link layer



Data link layer



Data link layer




Physical layer host
-
router protocol


Physical layer



Physical layer



Physical layer

Host A



Router



Host B


They idea of having different

layers is to separate the tasks needed to communicate in a
network. Each layer communicates with its analogous layer in the destination system
without a care of what’s going on the rest of the layers; the operation of the other layers is
transparent to a
given layer. For all intents and purposes, every layer is talking
directly
to
the equivalent layer in the computer on the other end using the corresponding protocol. A
protocol is a set of prearranged rules for a given layer.



By José M. Matos.


Do not use without permission.


One advantage of this modular

approach is that a layer can be modified or overhauled
completely transparently to the rest of the layers as long as the interface with the
contiguous layers is not changed.


By José M. Matos.


Do not use without permission.


The Application layer


The application layers sits at the top of the OSI referen
ce model and I see it as the
interface between the user and the rest of the network. A better understanding of this
layer can be attained if we consider that some of the applications that belong to the
Application layer protocol are:




File Transfer Protoco
l (FTP)



HyperText Transfer Protocol (HTTP)



Simple Mail Transfer Protocol (SMTP)



Telnet


Most probably, this layer will append a header to the data it will send to the Application
layer in the destination computer and will pass the whole source data to the
Presentation
layer.


By José M. Matos.


Do not use without permission.


The Transport layer


Some of the tasks of the Transport layer are:




To set
-
up and tear down network connections between computers in a network.
At the beginning of a transmission a connection between the computers that will
communicat
e might not exist and is the job of the Transport layer to set
-
up that
connection.



To control the flow of datagrams so the receiving end is not overflown by the
packets sent.


The Transport layer receives a data unit from the Presentation layer, most proba
bly
appends a header to it, ands sends it to the Network layer.


The Transport layer does not determine the route the information will follow through the
network. That is the work of the Network layer.


By José M. Matos.


Do not use without permission.


The Network layer


Once a connection is set
-
up by the

Transport layer, the Network layer determines the
route the
packets

will follow through the network. A
packet

is the data unit in which the
data received from the Transport layer is divided in the Network layer.


The Network layer receives a data unit fro
m the Transport layer, appends a header to it,
ands sends it to the Data link layer.


By José M. Matos.


Do not use without permission.


The Data Link layer


The job of the Data Link layer is to provide to the Network layer, and as a result, to all
the layers above the Network layer, and error
-
free channel
. The thing is that the layer
below the Data Link layer, the Physical layer, although tries to be error
-
free, it is not
guaranteed to be error
-
free. Frame acknowledgement is used as the method to ensure
error
-
free communication. The Data Link layer also re
gulates the flow of frames so that
fast senders do not choke slow receivers.


The packets received from the Network layer are divided into
frames

in the Data Link
layer.


Medium Access Control (MAC) layer


The responsibilities of the Data link layer compli
cate when the transmission medium is
shared by more than one computer. There must be a way of regulating the traffic in this
shared medium. For this case exits a sublayer belonging to the Data Link layer called the
Medium Access Control (MAC) sublayer.



By José M. Matos.


Do not use without permission.


P
hysical layer


The Physical layer is occupied in sending binary data through a physical medium.
Examples of medium are coaxial cable, optical fiber, and air in the case of wireless
communications. The physical layer may be seen as a blind agent that receiv
es a group of
bytes and send them through the physical medium maximizing the probability that the
received message is exactly the same to the sent message.


By José M. Matos.


Do not use without permission.


Circuit switching


In
circuit switching
, a dedicated path is set
-
up for communication. An example o
f circuit
switching is the Public Switch Telephone Network (PSTN).


In
circuit switching
, a physical path is set
-
up and it is kept until the end of the
communication. You may have
space division switching

or
time division switching
. In
space division switc
hing
, multiple switches are used and multiple dedicated links might
be created. In
time division switching
, different computers share that same
communication link by taking turns in time.


A disadvantage of circuit switching is that a dedicated circuit’s b
andwidth might be
underused at times.


By José M. Matos.


Do not use without permission.


Packet switching


In
packet switching
each packet sent runs independently throughout the network. The
packets are called datagrams. An example will be the Internet.


An advantage of packet switching is that since ther
e in no physical link dedicated to any
given transmission for a long period of time, the bandwidth of the network is used more
efficiently.


A
virtual circuit

is the route a datagram will follow during transmission. The difference to
circuit switching is t
hat it is valid while there is datagrams to send. As soon as the last
datagram is sent, the virtual circuit is teared down.


By José M. Matos.


Do not use without permission.


Routing


Routing algorithms exist to come up with the most efficient route to send datagrams or
set
-
up virtual circuits.


There are

several types of routing:



Distributed: each node calculates the routing



Centralized: a central node calculates the routing



Source: the source node decides the routing


Routing strategies:



Fixed routing. Calculated before the transmission and remains const
ant during
transmission. Algorithms used: Dijkstra and Bellman
-
Ford.



Flooding. Every hop sends the packet received to every other hop. This, in
practice, floods the network with packets. The most efficient route is always used
because every route is tried.

Military applications might benefit from this flooding
strategy because a battlefield could be very dynamic where hosts can be blown to
bits and new routes need to be found.



Random routing. Similar to flooding with the difference that a node sends a
recei
ved packet to just one node that is chosen at random.



Adaptive routing. The route changes with time.


By José M. Matos.


Do not use without permission.


X.25


It is an old standard used in public networks mostly outside of USA. It uses a packet size
of up to 128 bytes and reach speeds of 64 kbps. X.25 uses

the technique of virtual
circuits.


There are two types of virtual circuits (VC): Switched VC and Permanent VC. In
switched VC, a packet is sent asking the receiving computer to set
-
up a link and all
subsequent packets follow the same VC. In permanent VC,

the carrier and costumer
decide beforehand VC that will be used.


There are several packet types: call setup, flow control, diagnostic, reset & start, and
interrupts.


By José M. Matos.


Do not use without permission.


Local Area Network (LAN)


A relatively small network of computers that comprises a buil
ding or a campus and not
bigger that a few kilometers. The advantage of a LAN is that computers connected to it
can share resources and communicate with each other. No routers are needed.


Topologies used in LAN’s:




In a
bus

topology, the computers are all

directly connected to the same central
medium (coaxial cable, optical fiber, etc.) called a bus.



In a
ring

topology, every computer is connected to two contiguous computers.



In a
star

topology, each computer is connected to a central element called a hub
with two links: an uplink and a downlink.


By José M. Matos.


Do not use without permission.


Medium Access Control (MAC)


See
Medium Access Control (MAC) layer

in
The Data Link layer
.


By José M. Matos.


Do not use without permission.



Frequency Division Multiplexing (FDM)


F
DM is a static allocation scheme that belongs to the MAC sub
-
layer. The total
bandwidth is allocated equally among the users. It is a simple technique but has the
drawback that bandwidth is wasted if some users use the channel more that others.


By José M. Matos.


Do not use without permission.


Time Divis
ion Multiplexing (TDM)


TDM is another static allocation scheme that belongs to the MAC sub
-
layer. This time
the medium is shared by the users one at a time. It is also a simple technique but again is
not efficient if some users have much more packets to s
end that others.


By José M. Matos.


Do not use without permission.


ALOHA


ALOHA is a dynamic allocation scheme that belongs to the MAC sub
-
layer. It was
invented at the University of Hawaii for a radio broadcast system. The scheme is simple:



When a station is ready to send a frame it just sends it witho
ut knowing if the
medium is occupied with another transmission or not. (Remember that we use the
term frame here because the MAC sub
-
layer belongs to the Data Link layer that
takes the packets it receives from the Network layer and arrange them into
frames
.)



Of course, it will happen that sometimes two users or more will try to transmit at
the same time and the data will be garbled in the shared medium. This is called a
collision
.



The station will “listen” the medium
after

it transmits and determine if a c
ollision
has occurred. “Listening” is done by sampling the channel and comparing the
power or pulse width of the signal sampled to the signal sent.



If a collision does occur, the station will wait a random amount of time and will
retransmit.



For maximum ef
ficiency, the frames sent should be of the same size.


There are two types of ALOHA schemes: Pure ALOHA and Slotted ALOHA.


In pure ALOHA, the stations transmit their frames whenever they are ready. The
efficiency attained is 18.4% if same
-
size frames are
used. (Efficiency means that if the
network has a theoretical throughput S, the real throughput will be reduced to 0.184S.)


In slotted ALOHA, time is divided in time slots. Stations still transmit when they have a
frame to transmit with the constraint tha
t they wait for the beginning of a time slot. The
efficiency attained now is 36.8% if same
-
size frames are used.


By José M. Matos.


Do not use without permission.


Carrier Sense Multiple Access (CSMA)


This is another family of dynamic allocation schemes that belong to the MAC sub
-
layer.
It is different t
o ALOHA in that stations sense the channel
before

sending a frame. In
comparison, ALOHA stations just send their frames without listening the medium before
for ongoing transmissions. There are different types of CSMA schemes and I compare
them below.


CSMA

If channel idle

If channel busy

Efficiency

1
-
persistent CSMA

Station sends frame
immediately.

Station continues
sensing the channel until
it clears out and
transmits immediately.

~55%

nonpersistent CSMA

Station sends frame
immediately.

Station does not
sense
the channel
continuously. Instead, it
waits a random amount
of time before listening
to the channel again and
the process is repeated.

~90%

p
-
persistent CSMA

(Applies to slotted
systems)

Station does not send
immediately. If a station
sees an idle s
lot, it will
transmit with probability
p. If it sent, good; if it
did not send, it waits
until the next time slot
and repeats the process.

It waits until the next
time slot and repeats the
procedure.

Depends on p. Close to
100% efficiency can be
attained i
f p is small
(e.g. 0.01). However,
the delay experimented
by the users will be big.


CSMA with Collision Detection (CSMA/CD) is another scheme where the stations
transmitting sense the channel to detect collisions. If they detect a collision, they stop
tr
ansmitting immediately.


By José M. Matos.


Do not use without permission.


IEEE Standard 802.3: Ethernet


The IEEE Standard 802.3 is a standard for LANs that evolved from the Ethernet standard
developed in part by Xerox. It is an 1
-
persistent CSMA/CD scheme discussed in
Carrier
Sense Multiple Access (CSMA)
.


Cabling


The cabling in 802.3 is referred as 10Base”x” where “10” refers the capacity of the LAN
of 10 Mbps, “Base” refers to the baseband modulation. Refer to the table below to know
what the “x” stands for
.


10Base5

Refers to a coaxial cable of a maximum length of 500 meters

10Base2

Refers to a coaxial cable of a maximum length of 200 meters

10Base
-
T

Refers to a twisted pair

10Base
-
F

Refers to fiber optics



By José M. Matos.


Do not use without permission.


IEEE Standard 802.4: Token bus


The IEEE Stan
dard 802.4 is a standard for LANs where the stations are connected to a
central coaxial cable called a bus. A special frame called a token will travel through the
bus starting at the station with the highest address down to the station with the lowest
addr
ess and back to the first station again. Each station knows it position in this order and
the station before and after itself.


The token travels through the bus and when a computer is ready to send frames and
receives the token, it does not send the token

but sends its frames. When the station has
finished transmission, it will send the token into the LAN.


Collisions do not occur in this scheme since only one station at a time will have the token
and transmit. The token bus is inefficient in low traffic a
nd more efficient in high traffic.


By José M. Matos.


Do not use without permission.


IEEE Standard 802.11: Wireless LANs


A WLAN is similar to a wired network the fundamental difference being that the access
medium is air in a WLAN instead of copper wire, coaxial cable or fiber optic. The
advantages of a

WLAN are several:


1.

The cost of installation is minimal. There are no wires to install from access point to
the stations. A network might be needed in an existing building where the installation
of wires might be difficult or impossible; a WLAN will avoid
this problem.

2.

A station is may be easily added to the existing network just by installing a card to the
new station.

3.

Mobility of the users.

4.

Ad Hoc Networking is the ability of setting up a peer
-
to
-
peer network without the
need of wires and a centralized se
rver. An Ad Hoc network may be used in a
battlefield or in a natural disaster area where a rapid deployment of a network is
needed and there is no infrastructure to support it.


An experimental WLANs has been deployed in the Carnegie Mellon University [
i
].

The
authors describe the infrastructure used and the lessons learned. Another example of an
experimental implementation is described in [
ii
]. The IEEE 802.11 committee has
developed the standards for the WLANs.


The challenges of a WLAN are:


1.

New MAC proto
cols are needed for the new medium: air. Power consumption of the
units is also a concern.

2.

Battery life of stations in an Ad Hoc network, for example. In natural disaster area
power may not be available and the communication will depend on the batteries of

the stations.

3.

Routing

4.

Hidden node problem: a station is in the range of the destination node but out of the
range of the source. See picture below.




B





A


C


E



D





Letters A trough E represent the physical positions of five nodes and the cir
cles
represent the range of nodes A and E. Suppose that node A is transmitting a packet to
node C and node E wants to begin a transmission to node C. Since node A is out of its

By José M. Matos.


Do not use without permission.


range, node E is not aware of the transmission to C and will interfere with the

ongoing communication if it begins to transmit to node C. Node A is “hidden” from
node E.

5.

Exposed node problem: a node is in the range of the source node but out of the range
of the destination node. The exposed node will not transmit because does not kno
w
with what node is the communication of the source.


MACs proposed


Gumalla and Limb in [
iii
] propose a MAC protocol that they call Wireless Collision
Detect (WCD) because it is based in collision detection techniques. In summary, in
WCD, the channel is div
ided in a data channel and a feedback channel. The nodes sample
the feedback channel when they want to transmit. When they receive a packet, they
acknowledge thorough the feedback channel. The authors include simulation results.


Another MAC protocol propo
sed [
iv
] is one where a Spreading Code Control Server
(SPCCS) acts like the referee between the stations that want to access the medium. A
station that wants to transmit makes a request to transmit to the SPCCS and the SPCCS
responds with a permission or a
denial. When the station finishes transmission, it must
tell the SPCCS. Another function of the SPCCS is synchronization.


Power Consumption


In a seminar offered by Dr. Raghavendra from USC he presented a scheme where
stations turn themselves on and off d
epending on the traffic in order to save power.


Routing


Johnson and Maltz suggest in [
v
] a routing protocol called dynamic source routing. It uses
the concept of source routing borrowed from wired LANs. In source routing, the sender
determines the route
to the receiver and includes that route in the header of the packet;
the stations in the route just retransmit the packet to the next hop listed in the header. The
idea proposed by in the paper is that each station in the ad hoc network maintains a route
c
ache with the routes it has learned. If it originates a packet to be sent, it uses the routes
in memory; if the destination in not in its routes, it uses a route discovery protocol. In a
route discovery protocol, a route request packet is sent which propag
ates to the intended
destination or to a station that can reply with a route to the destination.


By José M. Matos.


Do not use without permission.


Asynchronous Transfer Mode (ATM)


ATM is a packet switching network similar to X.25. ATM packets are called cells and
have a fixed length of 53 bytes: 5 bytes

for the header and 48 bytes for the user data.
Having a fixed length cells provides an advantage to the network since it reduces
processing in the stations.


ATM is connection oriented in the sense that it does not require acknowledge and
guarantees that
the cells will arrive in the order they were sent.


Virtual circuits are used to route the cells.
Virtual paths

(VP) are collections of virtual
circuits. Some of the advantages of using VPs are:


1.

The management of the VCs is easier because managing a VP yo
u are able to
manage a bunch of VCs.

2.

The routing tables are smaller because they are based just in the VP and not in the
VCs.

3.

The routing tables are also local.

4.

Because the routing decision is based on the VPs, the routing decision for all the
VCs within a

VP is the same.


ATM header


There are two types of interfaces in an ATM network: User
-
Network Interface (UNI) and
Network
-
Network Interface (NNI). UNI defines the interface between a user and the
ATM network. NNI defines the interface between routers, ca
lled switches, in the
network.


The header for a UNI cell looks like this


GFC (4 bits)

VPI (8 bits)

VCI (16 bits)

PTI (3 bits)

CLP (1 bit)

HEC (8 bits)


The header for a NNI cell looks like this


VPI (12 bits)

VCI (16 bits)

PTI (3 bits)

CLP (1 bit)

HEC (
8 bits)


GFC: Generic Flow Control (Not used)



PTI: Payload Type

VPI: Virtual Path ID






CLP: Cell Loss Priority

VCI: Virtual Channel ID





HEC: Header Error Check


A VPI is a number that identifies a VP. The VPI is included in the header of each cell

so
each ATM network may direct correctly each cell. The switch assigns a VPI to each VP
used so when an ATM cell arrives, based on the VPI, it routes the cell through the
appropriate outgoing VP. Theoretically, the network could set
-
up up to 2
8

= 256 (UNI
) or
2
12

= 4096 (NNI) VPs.



By José M. Matos.


Do not use without permission.


Similarly, the VCI identifies a VC. Theoretically, the network can set
-
up up to 2
16

=
65536 VCs.


The PTI can convey information about network congestion and determines if a cell is
user data or if it is used for network OAM: op
eration, administration, and maintenance.


The CLP is set to 0 if the cell is high priority and should not be discarded by the network
and 1 if the cell is low priority and can be discarded.


The HEC is used to detect errors in the header, not in the paylo
ad. The code used can
detect single
-
bit errors and 90% of multibit errors.


Service Categories


Category

Characteristic

Application

Constant Bit Rate (CBR)

Its intended to work like a
dedicated line.


real time Variable Bit Rate(rt
-
VBR)


Videoconferencin
g

non real time Variable Bit Rate(nrt
-
VBR)



Available Bit Rate (ABR)


Web browsing

Unspecified Bit Rate (UBR)




Quality of Service (QoS)


The carrier and a costumer negotiate QoS parameters beforehand. Some QoS parameters
are:




Cell rate (Throughput)



Maximum cell delay



Cell delay variation (jitter)



Cell loss probability


By José M. Matos.


Do not use without permission.


Internet Protocol (IP)


IP v. 4

bits

0 4


8


16


19


31

Version

IHL

Type of Service

Total Length

Identification

Flags

Fragment Offset

Time to Live

Protocol

Header Checksum

Source Address

Destination Address

Options + Padding



Data filed (65,535 bytes max)





For purpose of showing the IP v.4 header
, each row above has 32 bits or 4 bytes. The first
six rows or 24 bytes are the IP header. The first 5 rows or 20 bytes are mandatory; row 6
is optional and is called the header overhead.


Field

IPv4

Version

Indicates IP version number

IHL

Since the leng
th of the header is variable, this field indicates the length in 32
-
bit
words.

Type of service

Specifies reliability, precedence, delay, and throughput. Tries to provide some QoS.

Total length

Total datagram length in bytes.

Identification

Together with

the source and destination address and the protocol that generated the
data, tries to identify uniquely a packet.

Flags

More flag: used in fragmentation

Don’t fragment flag: a packet that will not be fragmented

cragment offset

fndicated where in the ori
ginal data the packet belongsK

Time to live

eow longI in secondsI the packet will remain in the internetK

mrotocol

The protocol that should receive the payload at the receiverK

eeader checksum

brror detecting code applied to the header

pource address

l
rigination
J
system address

aestination address

bnd
J
system address


IP addresses are 32 bits long and for convenience are represented in
dotted decimal
format
. In the decimal format, the IP address is taken 8 bits at a time and represented
with decimal num
bers separated by dots. In decimal format, theoretically the lowest IP
address would be 0.0.0.0 and the highest would be 255.255.255.255.


There are five different classes of IP v.4 addresses: A, B, C, D, and E.


By José M. Matos.


Do not use without permission.


Class A (0.0.0.0 through 127.255.255.255)

0

Network ID (7 bits)

Host ID (24 bits)


Class B (128.0.0.0 through 191.255.255.255)

1

0

Network ID (14 bits)

Host ID(16 bits)


Class C (192.0.0.0 through 223.255.255.255)

1

1

0

Network ID (21 bits)

Host ID (8 bits)


Class D (224.0.0.0 through 239.255.25
5.255)

1

1

1

0

Multicast


Class E (240.0.0.0 through 247.255.255.255)

1

1

1

1

0

Future use


For example, the University of Florida has been assigned a Class B address: 128.227.x.y.
Accordingly, The ECE department at UF was assigned the address 128.227.22
0.x.


Subnet Addressing


Class A and B addresses have too many bits reserved for hosts. In subneting, some of the
Host ID bits are taken and used to identify networks.


How a source finds the address of a destination


The source sends a message to a Domain

Name System (DSN) server with the name of
the destination, i.e.
www.ufl.edu
. The DSN server looks up the name and returns the IP
address to the source.


Internet Control Message Protocol (ICMP)


When something is wrong
with the network and the datagrams do not reach their
destination, router has the option of sending ICMP messages to notify others routers of
problems or to try to figure out what’s wrong. Some of the messages are:


Message

Meaning

Destination Unreachable

Packet could not be delivered

Time exceeded

Time to live field reached 0

Source quench

Reduce data rate at source

Redirect

Advise host of a better route

Echo/Echo reply

Test communication link

Time stamp/Time stamp reply

Test delay over the communica
tion link



IP v. 6 (IPng)



By José M. Matos.


Do not use without permission.


The 32
-
bit address format in IP v.4 is only one of the limitations that encourage the
creation of IPng (IP next generation) which has addresses of 128 bits.


bits

0 4


12



16


24


31

Version

Traffic Class

Flow Label

Payload Length

Next Header

Hop Limit


Source Address (128 bits)




Destination Address (128 bits)




For purpose of showing the IP v
.4 header, each row above has 32 bits or 4 bytes. All ten
rows shown are mandatory making the IPng header at least 40 bytes long. However,
Extension Headers

might be added.


Field

IPng

Version

It provides a field fo
r IP version. The version is 6.

Traffic Class

Experimental. May distinguish between packets in terms of QoS.

Flow label

Flow Label

may also provide some QoS. It emulates some type of virtual
circuit operation.

Pay Load Length

Length of
extension headers

plus payload.

Next header

Points to the next header following the current one or the protocol that should
receive the payload at the receiver.

Hop limit

How long, in hop counts, the packet will remain in the inte
rnet.

Source address

Origination
-
system address

Destination address

End
-
system address


For convenience, an IPng address is written in hexadecimal notation. Every four bits of
the address are converted to hex ending up with 32 characters. Those 32 chara
cters are
grouped in eight groups of four each one separated by colons (:). For example:


4000:0000:0000:0000:BA5F:039A:000A:2176


Leading zeros are omitted:


4000:0:0:0:BA5F:39A:A:2176


Consecutive zeros are also omitted:


4000:: BA5F:39A:A:2176


IP v.4 a
ddresses are written as ::128.127.x.y, for example.



By José M. Matos.


Do not use without permission.


Extension Headers


Hop
-
by
-
Hop Options header
: is used for information that every router along the path
must see. In addition, can be used to specify jumbograms: datagrams with a payload
greater than 64Kb
ytes. Supports datagrams up to 4 billions bytes.


Routing header
: contains the addresses of intermediate nodes that should be visited
during transmission.


Fragment header
: used when the datagram has been fragmented for transmission.
Fragmentation of a dat
agram can be done only by the source. If a node along the way
cannot handle a datagram due to its size, the node discards the datagram and sends an
ICPM message to the source.


Multicasting

is the act of sending a packet from a single source to multiple de
stinations.




i

A. Hills and D.B. Johnson, “A Wireless Data Network Infrastructure at Carnegie Mellon University,”
IEEE Personal Communications,
Feb. 1996, pp. 56
-
63.

ii

S. Miura and Y. Suzuk
i, “Personalized High Speed Wireless LAN System: JPLAN”,
IEEE Vehicular
Technology Conference,
v. 5, 1999, pp. 2682
-
2685.

iii

A. C. Gumalla and J. O. Limb, “Design of an Access Mechanism for a High Speed Distributed Wireless
LAN,”
IEEE Journal on Areas in Co
mmunications
, Sept. 2000, pp 1740
-
1750.

iv

G.S. Kuo and P.C. Ko, “A Collision
-
Free Medium Access Control Protocol for Flow
-
Oriented Ad Hoc
Wireless LAN,”
IEEE Vehicular Technology Conference,

v.1, 1999, pp. 325
-
331.

v

D. B. Johnson and D.A. Maltz, “Protocol
s for Adaptive Wireless and Mobile Networking,”
IEEE
Personal Communications,
Feb. 1996, pp. 34
-
42.