Computer Networks

and IP Technology

Shen Jinlong

Nanjing University of Posts and Telecommunications

September 2006

Contents
1

INTRODUCTION
1.1 The Definition about Computer Network
1.2 Network Hardware and his classification
1.3 Development of the telecoms networks in China

NETWORK ARCHITECTURE
2.1 The OSI Reference Model
2.2 Data Transmission in the OSI Model
2.3 The TCP/IP Reference Model

PUBLIC DATA NETWORKS TECHNOLGY

3.1 X.25 Packet Switching Public Data Networks
3.2 Frame Relay
3.3 Broadband ISDN and ATM

LAN&MAN
4.1 LAN Architecture
4.2 IEEE 802.3/ Ethernet
4.3 Wireless LAN

INTERNET AND INTRANET

5..1. The IP Protocol
5.2 IP Addresses and Subnet
5.3 Internet Control Protocols

BROADBAND AND IP QOS

6.1
6.2
6.3
6.4

The Transport Service

The Internet Transport Protocols (TCP and UDP )
IP QoS technology
10 Hottest Technologies

REFERENCES

INTRODUCTION

Each of the past three centuries has been dominated by a single technology. The 18th Century
was the time of the great mechanical systems accompanying the Industrial Revolution. The 19th
Century was the age of the steam engine. During the 20th Century, the key technology has been
information gathering, processing, and distribution. Among other developments, we have seen the
installation of worldwide telephone networks, the invention of radio and television, the birth and
unprecedented growth of the computer industry, and the launching of communication satellites.
Due to rapid technological progress, these areas are rapidly converging, and the differences
between collecting, transporting, storing, and processing information are quickly disappearing.
Organizations with hundreds of offices spread over a wide geographical area routinely expect to
be able to examine the current status of even their most remote outpost at the push of a button. As
our ability to gather, process, and distribute information grows, the demand for even more
sophisticated information processing grows even faster.[1] In the new century we are entering the
era of Information Technology with would.
1.1 The Definition about Computer Network
Although the computer industry is young compared to other industries (e.g., automobiles and
air transportation), computers have made spectacular progress in a short time. During the first two
decades of their existence, computer systems were highly centralized, usually within a single large
room. A medium-size company or university might have had one or two computers, while large
institutions had at most a few dozen. It is called as the computer system based on Master/Slave
model as shown in Fig. 1-1.
(a) point-to-point comm..
point to point link
(b) point-to-multipoint comm.

multipoint link
Terminal

FEP

Host

(c) multiplexer

Multiplexer

Multiplexer

(d) concerntrator

(e) dial-up
PSTN
Modem

Modem

Fig.1-1 Master/Slave computer system

1

The merging of computers and communications has had a profound influence on the way
computer systems are organized. The concept of the "computer center" as a room with a large
computer to which users bring their work for processing is now totally obsolete. The old model of
a single computer serving all of the organization's computational needs has been replaced by one
in which a large number of separate but interconnected computers do the job. These systems are
called computer networks, as shown in Fig.1-2.
Router

LAN

Resource Subnet

Comm. Subnet

Fig.1-2 computer networks

The term "computer network" is to mean an interconnected collection of autonomous
computers. Two computers are said to be interconnected if they are able to exchange information.
The connection need not be via a copper wire; fiber optics, microwaves, and communication
satellites can also be used. By requiring the computers to be autonomous, we wish to exclude from
our definition systems in which there is a clear master/slave relation. If one computer can forcibly
start, stop, or control another one, the computers are not autonomous. A system with one control
unit and many slaves is not a network; nor is a large computer with remote printers and terminals.
There is considerable confusion in the literature between a computer network and a distributed
system. The key distinction is that in a distributed system, the existence of multiple autonomous
computers is transparent (i.e., not visible) to the user. He or she can type a command to run a
program, and it runs. It is up to the operating system to select the best processor, find and transport
all the input files to that processor, and put the results in the appropriate place.
In other words, the user of a distributed system is not aware that there are multiple processors;
it looks like a virtual uniprocessor. Allocation of jobs to processors and files to disks, movement of
files between where they are stored and where they are needed, and all other system functions
must be automatic.
With a network, users must explicitly log onto one machine, explicitly submit jobs remotely,
explicitly move files around and generally handle all the network management personally. With a
distributed system, nothing has to be done explicitly; it is all automatically done by the system
without the users' knowledge.
In effect, a distributed system is a software system built on top of a network. The software
gives it a high degree of cohesiveness and transparency. Thus the distinction between a network
and a distributed system lies with the software (especially the operating system), rather than with
the hardware.
Nevertheless, there is considerable overlap between the two subjects. For example, both
distributed systems and computer networks need to move files around. The difference lies in who
invokes the movement, the system or the user.
2

1.2. Network Hardware and his classification

It is now time to focus on the technical issues involved in network design. There is no generally
accepted taxonomy into which all computer networks fit, but two dimensions stand out as
important: transmission technology and scale. We will now examine each of these in turn.
Broadly speaking, there are two types of transmission technology: Broadcast networks, Pointto-point networks.
(1) Broadcast networks have a single communication channel that is shared by all the
machines on the network. Short messages, called packets in certain contexts, sent by any machine
are received by all the others. An address field within the packet specifies for whom it is intended.
Upon receiving a packet, a machine checks the address field. If the packet is intended for itself, it
processes the packet; if the packet is intended for some other machine, it is just ignored.
Broadcast systems generally also allow the possibility of addressing a packet to all destinations
by using a special code in the address field. When a packet with this code is transmitted, it is
received and processed by every machine on the network. This mode of operation is called
broadcasting. Some broadcast systems also support transmission to a subset of the machines,
something known as multicasting. One possible scheme is to reserve one bit to indicate
multicasting. The remaining (n 1) address bits can hold a group number. Each machine can
"subscribe" to any or all of the groups. When a packet is sent to a certain group, it is delivered to
ail machines subscribing to that group.
(2) Point-to-point networks consist of many connections between individual pairs of
machines . To go from the source to the destination, a packet on this type of network may have to
first visit one or more intermediate machines. Often multiple mutes, of different lengths are
possible, so routing algorithms play an important role in point-to-point networks. As a general
rule (although there are many exceptions), smaller, geographicaily localized networks tend to use
broadcasting, whereas larger networks usually are point-to-point.
An alternative criterion for classifying networks is their scale. A classification of multiple
processor systems arranged by their physical size is given in Fig. 1-3.

Personal Operating Space

WAN
WAN-MAN

MAN

PAN
MAN-LAN
LAN-PAN

Pico-Cell

~50km

Interprocessor

~2km

Processors located

0km

~10m

Example

distance

in same

0.1m

Circuit board

1m

System

Multicomputer

10m

Room

PAN, Personal Area Network

100m

Building

1km

Campus

10km

City

100km

Country

1000km

Continent

10000km

Planet

Data flow machine

LAN, Local Area Network

MAN, Metropolitan Area Network
WAN, Wide Area Network
The internet

Fig. 1-3. Classification of interconnected processors or computers by scale.

At the top are data flow machines, highly parallel-computers with many functional units all
working on the same program. Next come the multi-computers, systems that communicate by
sending messages over very short, very fast buses. Beyond the multi-computers are the tree
networks, computers that communicate by exchanging messages over longer cables. These can be
divided into local, metropolitan, and wide area networks. Finally, the connection of two or more
networks is called an internetwork. The worldwide Internet is a well-known example of an
internetwork. Distance is important as a classification metric because different techniques are used
at different scales. In this book we will be concerned with only the true networks and their
interconnection. Below we give a brief introduction to the subject of network hardware.
1.2.1. Local Area Networks
Local area networks, generally called LANs, are privately-owned networks within a single
building or campus of up to a few kilometers in size. They are widely, used to connect personal
computers and workstations in company offices and factories to share resources (e.g., printers)
and exchange information. LANs are distinguished from other kinds of networks by three
characteristics: (1) their size, (2) their transmission technology, and (3) their topology.
LANs are restricted in size, which means that the worst-case transmission time is bounded and
known in advance. Knowing this bound makes it possible to use certain kinds of designs that
would not otherwise be possible. It also simplifies network management.
LANs often use a transmission technology consisting of a single cable to which all the
machines are attached, like the telephone company party lines once used in rural areas. Traditional
LANs run at speeds of 10 to 100 Mbps, have low delay, and make very few errors(10-11). Newer
LANs may operate at higher speeds, up to and up to 1Gbps or 10Gbps.

Bus

workstantion

Tower System

Server
(a) Bus

(b) Ring

Fig. 1-4. Two broadcast networks. (a) Bus, (b) Ring.

Various topologies are possible for broadcast LANs. Figure 1-4 shows two of them. In a bus
(i.e., a linear cable) network, at any instant one machine is the master and is allowed to transmit.
All other machines are required to refrain from sending. An arbitration mechanism is needed to
resolve conflicts when two or more machines want to transmit simultaneously. The arbitration
mechanism may be centralized or distributed. IEEE 802.3, popularly called Ethernet, for
example, is a bus-based broadcast network with decentralized control operating at 10 or 100
Mbps. Computers on an Ethernet can transmit whenever they want to; if two or more packets
collide, each computer just waits a random time and tries again later.
A second type of broadcast system is the ring. In a ring, each bit propagates around on its own,
not waiting for the rest of the packet to which it belongs. Typically, each bit circumnavigates the
entire ring in the time it takes to transmit a few bits, often before the complete packet has even
been transmitted. Like all other broadcast systems, some rule is needed for arbitrating
simultaneous accesses to the ring. Various methods are in use and will be discussed later in this
book. IEEE 802.5 (the IBM token ring), is a popular ring-based LAN operating at 4 and 16
Mbps(now up to 100 Mbps).
Broadcast networks can be further divided into static and dynamic, depending on how the
channel is allocated. A typical static allocation would be to divide up time into discrete intervals
and run a round robin algorithm, allowing each machine to broadcast only when its time slot
comes up. Static allocation wastes channel capacity when a machine has nothing to say during its
allocated slot, so most systems attempt to allocate the channel dynamically (i.e., on demand).
Dynamic allocation methods for a common channel are either centralized or decentralized. In
the centralized ,channel allocation method, there is a single entity, for example a bus arbitration
unit, which determines who goes next. It might do this by accepting requests and making a
decision according to some internal algorithm. In the decentralized channel allocation method,
there is no central entity; each machine must decide for itself whether or not to transmit. You
might think that this always leads to chaos, but it does not. Later we will study many algorithms
designed to bring order out of the potential chaos.
The other kind of LAN is built using point-to-point lines. Individual lines connect a specific
machine with another specific machine. Such a LAN is really a miniature wide area network. We
will look at these later.
1.2.2. Metropolitan Area Networks
A metropolitan area network, or MAN, is basically a bigger version of a LAN and normally
uses similar technology. It might cover a group Of nearby corporate offices or a city and might be
either private or public. A MAN can support both data and voice, add might even be related to the
local cable television network. A MAN just has one or two cables and does not contain switching
elements, which shunt packets over one of several potential output lines. Not having to switch
simplifies the design.
The main reason for even distinguishing MANs as a special category is that a standard has been
adopted for them, and this standard is now being implemented. It is called DQDB (Distributed
Queue Dual Bus) or for people who prefer numbers to letters, 802.6 (the number of the IEEE
5

standard that defines it). DQDB consists of two unidirectional buses (cables) to which all the
computers are connected, as shown in Fig. 1-5. Each bus has a bead-end, a device that initiates
transmission activity. Traffic that is destined for a computer to the right of the sender uses the
upper bus. Traffic to the left uses the lower one.

Bus A

RQ

B
End

Source
WS1

WS2

WS3

WS n
Source

End
B

RQ

Bus B

Fig, 1-5. Architecture of the DQDB metropolitan area network.

A key aspect of a MAN is that there is a broadcast medium (for 802.6, two cables) to which
all the computers are attached. This greatly simplifies the design compared to other kinds of
networks.
1.2.3. Wide Area Networks
A wide area network, or WAN, spans a large geographical area, often a country or continent. If
contains a collection of machines intended for running user (i.e., application) programs. We will
follow traditional usage and call these machines hosts. The term end system is sometimes also
used in the literature. The hosts are connected by a communication subnet, or just subnet for
short. The job of the subnet is to carry messages from host to host, just as the telephone system
carries words from speaker to listener. By separating the pure communication aspects of the
network (the subnet) from the application aspects (the hosts), the complete network design is
greatly simplified.
In most wide area networks, the subnet consists of two distinct components: transmission lines
and switching elements. Transmission lines (also called circuits, channels, or trunks) move bits
between machines.
The switching elements are specialized computers used to connect two or more transmission
lines. When data arrive on an incoming line, the switching element must choose an outgoing line
to forward them on. Unfortunately, there is no standard terminology used to name these
computers. They are variously called packet switching nodes, intermediate systems, and data
switching exchanges, among other things. As a generic term for the switching computers, we will
use the word router, but the reader should be aware that no consensus on terminology exists here.
In this model, each host is generally connected to a LAN on which a router is present, although in
some cases a host can be connected directly to a router. The collection of communication tines and
routers (but not the hosts) form the subnet.
In most WANs, the network contains numerous cables or telephone lines, each one
connecting a pair of routers. If two routers that do not share a cable nevertheless wish to
communicate, they must do this indirectly, via other routers. When a packet is sent from one router
6

to another via one or more intermediate routers, the packet is received at each intermediate router
in its entirety, stored there until the required output line is free, and then forwarded. A subnet using
this principle is called a point-to-point, store-and-forward, or packet switched subnet. Nearly all
wide area networks (except those using satellites) have store and-forward subnets. When the
packets are small and all the same size, they are often called cells.
When a point-to-point subnet is used, an important design issue is what the router
interconnection topology should look like. Figure 1-6 shows several possible topologies. Local
networks that were designed as such usually have a symmetric topology. In contrast, wide area
networks typically have irregular topologies.
Backbone layer

Distributed layer

Access layer
Fig.1-6 Topology for WAN

A second possibility for a WAN is a satellite or ground radio system. Each router has an
antenna through which it can send and receive. All routers can hear the output from the satellite,
and in some cases they can also hear the upward transmissions of their fellow routers to the
satellite as well. Sometimes the routers are connected to a substantial point-to-point subnet, with
only some of them having a satellite antenna. Satellite networks are inherently broadcast and are
most useful when the broadcast property is important.
1.2.4. Wireless Networks
Mobile computers, such as notebook computers and personal digital assistants (PDAs), are the
fastest-growing segment of the computer industry. Many of the owners of these computers have
desktop machines on LANs and WANs back at the office and want to be connected to their home
base even when away from home or en route. Since having a wired connection is impossible in
cars and air-planes, there is a lot of interest in wireless networks.
Although wireless networking and mobile computing are often related, they are not identical.
Portable computers are sometimes wired. For example, if a traveler plugs a portable computer into
the telephone jack in a hotel, we have mobility without a wireless network. Another example is
someone carrying a portable computer along as he inspects a train for technical problems. Here a
long cord can trail along behind (vacuum cleaner model).
On the other hand, some wireless computers are not portable. An important example here is a
company that owns an older building that does not have network cabling installed and wants to
connect its computers. Installing a wireless LAN may require little more than buying a small box
with some electronics and setting up some antennas. This solution may be cheaper than wiring the
building.
Although wireless LANs are easy to install, they also have some disadvantages. The error
7

rates are often much higher, too, and the transmissions from different computers can interfere with
one another.Typically they have a capacity of 11 Mbps (IEEE 802.11b), which is much slower
than wired LANs.
Wireless networks come in many forms. Some universities are already installing antennas all
over campus to allow students to sit under the trees and consult the library's card catalog. Here the
computers communicate directly with the wireless LAN in digital form. Another possibility is
using a cellular (i.e., portable) telephone with a traditional analog modem. Direct digital cellular
service, such as CDPD (Cellular Digital Packet Data), GPRS(General Packet Radio Service),
SMS(Short Message System) and MMS(Media Message System) is becoming available in many
cities.
While many people believe that wireless portable computers are the wave of the future, at least
one dissenting voice has been heard. Bob Metcalfe, the inventor of Ethernet, has written: "Mobile
wireless computers are like mobile pipeless bathrooms--portapotties. They will be common on
vehicles, and at construction sites, and rock concerts. My advice is to wire up your home and stay
there" (Metcalfe, 1995). Will most people follow Metcalfe's advice? Time will tell.
1.2.5. lnternetworks
Many networks exist in the world, often with different hardware and software. People
connected to one network often want to communicate with people attached to a different one. This
desire requires connecting together different, and frequently incompatible networks, sometimes by
using machines called gateways to make the connection and provide the necessary translation,
both in terms of hardware and software. A collection of interconnected networks is called an
internetwork or just internet.
A common form of internet is a collection of LANs connected by a WAN. In fact, if we were
to replace the label "subnet" in Fig. 1-2 by "WAN," nothing else in the figure would have to
change. The only real distinction between a subnet and a WAN in this case is whether or not hosts
are present. If the system within the closed dot curve contains only routers, it is a subnet. If it
contains both routers and hosts with their own users, it is a WAN.
To avoid confusion, please note that the word "internet" will always be used in a generic sense.
In contrast, the Internet (note uppercase I) means a specific worldwide internet that is widely used
to connect universities, government offices, companies, and of late, private individuals. We will
have much to say about both internets and the Internet later in this material.
Subnets, networks, and internetworks are often confused. Subnet makes the most sense in the
context of a wide area network, where it refers to the collection of routers and communication
lines owned by the network operator, for example, companies like AOL(America Online) and
CompuServe. As an analogy, the telephone system consists of telephone switching offices
connected to each other by high-speed lines, and to houses and businesses by low-speed lines.
These lines and equipment, owned and managed by the telephone company, form the subnet of the
telephone system. The telephones themselves (the hosts in this analogy) are not part of the subnet.
The combination of a subnet and its hosts forms a network. In the case of a LAN, the cable and the
hosts form the network. There really is no subnet.
An internetwork is formed when distinct networks are connected together. In our view,
connecting a LAN and a WAN or connecting two LANs forms an internetwork, but there is little
agreement in the industry over terminology in this area.
8

1.3 Development of the telecoms networks in China[2]

Since the inception of the open-door policy, China's telecoms network has made astounding
progress as it has pursued a policy of "reform and development''. It has experienced growth at a
pace that is unprecedented in the country and unique in the world, and its overall standard has
been raised to a historical high, and making a vital contribution to the national economy and social
advancement.
After many years of development, the overall capacity of China's telecoms network has been
significantly enhanced. The long-time problem of the under-supply of phone services has been
solved. A state-of-the-art, reliable and diversified telephone network has gradually been put in
place, thereby promoting the development of the country's IT infrastructure, as shown in Fig. 1-7 .
Information Home
Electrical Products

Satellite / Mobile
CPNoC

TV

Access
Access Net
Net

N-ISDN
N-ISDN // PSTN
PSTN

CCSS
No.7

X.25 PSPDN

Internet

DDN
FRN

Telephone

IN
ATM
ATM

Gateway
Gateway
SDH
Enterprise Net

Fibers
TMN

Fig. 1-7 Backbone of IT infrastructure in China

Public data and multimedia networks as well as the Internet are accessible nationwide.
Network platform resources have expanded substantially, as port capacity for digital data networks
and broadband access as well as the total bandwidth for international entry/exit to the Internet
have been significantly enhanced. The constraint of bandwidth "bottleneck" has been effectively
relieved. The telecoms sector provides channels for information transfer for government
departments, industries and business enterprises. A variety of services, such as leased-line access,
VPN, web-hosting, system integration and network configuration, are provided to facilitate the
development of e-government, e-business, remote learning, remote medical care, and business
information management. These developments have played a pivotal role in reengineering
traditional industries, and in promoting the use of information in the society. Telecoms companies
have cooperated with relevant government departments to implement the three online projects government online, enterprise online and home online - by leveraging the existing resources in
joint efforts to expedite the construction of IT infrastructure.
In the areas of finance, custom, taxation and foreign trade, the telecoms sector has
9

collaborated with relevant government authorities to carry out a number of online projects with
significant results. Telecoms companies provide preferential, quality services and network
platform support to national financial information networks, electronic enforcement systems at
ports and airports, and the information system of the tax authorities. Currently, the People's Bank
of China and the General Administration for Taxation are leasing 10,000 lines from the telecoms
companies for various services. Industries and government departments configured more than 187
application systems using public telecoms networks. In pursuing the national strategy of IT-driven
industrialization and in advancing the construction of IT infrastructure, the telecoms sector has
made a vital contribution, proving itself to be irreplaceable.
In the past 20-odd years, China's telecom country's IT infrastructure services and
communications capacity has increased enormously. Today, public telephone networks cover the
entire country, giving access to the whole world. By the end of June 2002, the total length of the
fiber optic cable backbone of the system was 1.97 million km, and the length of long-distance
fiber optic cable was 340,000 km. A number of terrestrial and undersea cables connecting China
and Japan, Asia and Europe, and China and the Americas had been built. Long-distance circuits
totaled 4.82 million lines at the end of June 2002, 145 times the total number in 1978. Local
switches had a capacity of 210 million lines at the end of June 2002, 50 times the capacity in
1978.
Networks for new services had been built from scratch. Switches for mobile communications
had the capacity to serve 250 million subscribers at the end of June 2002. Over 1 million ports
had been built for multimedia data networks and broadband networks. Internet access bandwidth
reached 7.6G. Seventy-one countries/territories had built circuits directly linked to China. Longdistance services had been opened with over 200 countries/territories (including transfer services),
and roaming services had been established with 90 countries/territories.
The improvement in service quality is evident. Common problems, as reflected by
subscribers, are gradually being dealt with. At present, the waiting time for installation of fixedline service has been shortened to 10.5 days across the country (success rate: 98.5%). The average
time required to restore disrupted services is 14.4 hours (success rate: 97.9%). Services for mobile
phone, data communications, leased-line users, large usage customers and corporate
clients have also been improved.

10

NETWORK ARCHITECTURE

Now that we have discussed layered networks in the abstract, it is time to look at some
examples. In the next two sections we will discuss two important network architectures, the OSI
reference model and the TCP/IP reference model.
2.1. The OSI Reference Model
The OSI model is shown in Fig. 2-1 (minus the physical medium). This model is based on a
proposal developed by the International Standards Organization (ISO) as a first step toward
international standardization of the protocols used in the various layers (Day and Zimmermann,
1983). The model is called the ISO OSI (Open Systems Interconnection) Reference Model
because it deals with connecting open systems--that is, systems that are open for communication
with other systems. We will usually just call it the OSI model for short.
Below we will discuss each layer of the model in turn, starting at the bottom layer. Note that
the OSI model itself is not a network architecture because it does not specify the exact services
and protocols to be used in each layer. It just tells what each layer should do. However, ISO has
also produced standards for all the layers, although these are not part of the reference model itself.
Each one has been published as a separate international standard.

A Open Real System

A real System
APA
LSM

B Open Real System

A Open System

OSI environment
7
6

7
6

5
Relay Open System

4
Local System

B Real System

B Open System

APB
LSM

Host A

Host B
Comm. Subnet

AP _Application Process

LSM _Local System Management

_Implementation Module

Fig.2-1. The OSI reference model.

11

2.1.1 The Physical Layer

The physical layer is concerned with transmitting raw bits over a communication channel.
The design issues have to do with making sure that when one side sends a 1 bit, it is received by
the other side as a 1 bit, not as a 0 bit. Typical questions here are how many volts should be used
to represent a 1 and how many for a 0, how many microseconds a bit lasts, whether transmission
may proceed simultaneously in both directions, how the initial connection is established and how
it is tom down when both sides are finished, and how many pips the network connector has and
what each pin is used for. The design issues here largely deal with mechanical, electrical, and
procedural interfaces, and the physical transmission medium, which lies below the physical layer.
2.1.2 The Data Link Layer
The main task of the data link layer is to take a raw transmission facility and transform it into
a line that appears free of undetected transmission errors to the network layer. It accomplishes this
task by having the sender break the input data up into data frames (typically a few hundred or a
few thousand bytes), transmit the frames sequentially, and process the acknowledgement frames
sent back by the receiver. Since the physical layer merely accepts and transmits a stream of bits
without any regard to meaning or structure, it is up to the data link layer to create and recognize
frame boundaries. This can be accomplished by attaching special bit patterns to the beginning and
end of the frame. If these bit patterns can accidentally occur in the data, special care must be take
n to make sure these patterns are not incorrectly interpreted as frame delimiters.
A noise burst on the line can destroy a frame completely. In this case, the data link layer
software on the source machine can retransmit the frame. However, multiple transmissions of the
same frame introduce the possibility of duplicate frames. A duplicate frame could be sent if the
acknowledgement frame from the receiver back to the sender were lost. It is up to this layer to
solve the problems caused by damaged, lost, and duplicate frames. The data link layer may offer
several different service classes to the network layer, each of a different quality and with a
different price.
Another issue that arises in the data link layer (and most of the higher layers as well) is how
to keep a fast transmitter from drowning a slow receiver in data. Some traffic regulation
mechanism must be employed to let the transmitter know how much buffer space the receiver has
at the moment. Frequently, this flow regulation and the error handling are integrated.
If the line can be used to transmit data in both directions, this introduces a new complication
that the data link layer software must deal with. The problem is that the acknowledgement frames
for A to B traffic compete for the use of the line with data frames for the B to A traffic. A clever
solution (piggybacking) has been devised; we will discuss it in detail later.
Broadcast networks have an additional issue in the data link layer: how to control access to
the shared channel. A special sublayer of the data link layer, the medium access sublayer, deals
with this problem.

12

2.1.3 The Network Layer

The network layer is concerned with controlling the operation of the subnet A key design issue
is determining how packets are routed from source to destination. Routes can be based on static
tables that are "wired into" the network and rarely changed. They can also be determined at the
start of each conversation, for example a terminal session. Finally, they can be highly dynamic,
being determined anew for each packet, to reflect the current network load.
If too many packets are present in the subnet at the same time, they will get in each other's way,
forming bottlenecks. The control of such congestion also belongs to the network layer.
Since the operators of the subnet may well expect remuneration for their efforts, there is often
some accounting function built into the network layer. At the very least, the software must count
how many packets or characters or bits are sent by each customer, to produce billing information.
When a packet crosses a national border, with different rates on each side, the accounting can
become complicated.
When a packet has to travel from one network to another to get to its destination, many
problems can arise. The addressing used by the second network may be .different from the first
one. The second one may not accept the packet at all because it is too large. The protocols may
differ, and so on. It is up to the network layer to overcome all these problems to allow
heterogeneous networks to be interconnected.
In broadcast networks, the routing problem is simple, so the network layer is often thin or
even nonexistent.
2.1.1 The Transport Layer
The basic function of the transport layer is to accept data from the session layer, split it up into
smaller units if need be, pass these to the network layer, and ensure that the pieces all arrive
correctly at the other end. Furthermore, all this must be done efficiently, and in a way that isolates
the upper layers from the inevitable changes in the hardware technology.
Under normal conditions, the transport layer creates a distinct network connection for each
transport connection required by the session layer. If the transport connection requires a high
throughput, however, the transport layer might create multiple network connections, dividing the
data among the network connections to improve throughput. On the other hand, if creating or
maintaining a network connection is expensive, the transport layer might multiplex several
transport connections onto the same network connection to reduce the cost. In all cases, the
transport layer is required to make the multiplexing transparent to the session layer.
The transport layer also determines what type of service to provide the session layer, and
ultimately, the users of the network. The most popular type of transport connection is an error-free
point-to-point channel that delivers messages or bytes in the order in which they were sent.
However, other possible kinds of transport service are transport of isolated messages with no
guarantee about the order of delivery, and broadcasting of messages to multiple destinations. The
type of service is determined when the connection is established.
The transport layer is a true end-to-end layer, from source to destination. In other words, a
program on the source machine carries on a conversation with a similar program on the
destination machine, using the message headers and control messages. In the lower layers, the
protocols are between each machine and its immediate neighbors, and not by the ultimate source
and destination machines, which may be separated by many routers. The difference between layers
13

1 through 3, which are chained, and layers 4 through 7, which are end-to-end, is illustrated in Fig.
2-1.
Many hosts are multiprogrammed, which implies that multiple connections will be entering
and leaving each host. There needs to be some way to tell which message belongs to which
connection. The transport header is one place this information can be put.
In addition to multiplexing several message streams onto one channel, the transport layer
must take care of establishing and deleting connections across the network. This requires some
kind of naming mechanism, so that a process on one machine has a way of describing with whom
it wishes to converse. There must also be a mechanism to regulate the flow of information, so that
a fast host cannot overrun a slow one. Such a mechanism is called flow control and plays a key
role in the transport layer (also in other layers). Flow control between hosts is distinct from flow
control between routers, although we will later see that similar principles apply to both.
2.1.5 The Session Layer
The session layer allows users on different machines to establish sessions between them. A
session allows ordinary data transport, as does the transport layer, but it also provides enhanced
services useful in some applications. A session might be used to allow a user to log into a remote
timesharing system or to transfer a file between two machines.
One of the services of the session layer is to manage dialogue control. Sessions can allow traffic
to go in both directions at the same time, or in only one direction at a time. If traffic can only go
one way at a time (analogous to a single railroad track), the session layer can help keep track of
whose turn it is.
A related session service is token management. For some protocols, it is essential that both sides
do not attempt the same operation at the same time. To manage these activities, the session layer
provides tokens that can be exchanged. Only the side holding the token may perform the critical
operation.
Another session service is synchronization. Consider the problems that might occur when trying
to do a 2-hour file transfer between two machines with a l-hour mean time between crashes. After
each transfer was aborted, the whole transfer would have to start over again and would probably
fail again the next time as well. To eliminate this problem', the session layer provides a way to
insert checkpoints into the data stream, so that after a crash, only the data transferred
after the last checkpoint have to be repeated.
2.1.6 The Presentation Layer
The presentation layer performs certain functions that are requested sufficiently often to
warrant finding a general solution for them, rather than letting each user solve the problems. In
particular, unlike all the lower layers, which are just interested in moving bits reliably from here to
there, the presentation layer is concerned with the syntax and semantics of the information
transmitted.
A typical example of a presentation service is encoding data in a standard agreed upon way.
Most user programs do not exchange random binary bit strings. They exchange things such as
people's names, dates, amounts of money, and invoices. These items are represented as character
strings, integers, floating-point numbers, and data structures composed of several simpler items.
Different computers have different codes for representing character strings (e.g., ASCII and
14

Unicode), integers (e.g., one's complement and two's complement), and so on. In order to make it
possible for computers with different representations to communicate, the data structures to be
exchanged can be defined in an abstract way, along with a standard encoding to be used "on the
wire." The presentation layer manages these abstract data structures and converts from the
representation used inside the computer to the network standard representation and back.
2.1.7 The Application Layer
The application layer contains a variety of protocols that are commonly needed. For example,
there are hundreds of incompatible terminal types in the world. Consider the plight of a full screen
editor, that is supposed to work over a network with many different terminal types, each with
different screen layouts, escape sequences for inserting and deleting text, moving the cursor, etc.
One way to solve this problem is to define an abstract network virtual terminal that editors and
other programs can be written to deal with. To handle each terminal type, a piece of software must
be written to map the functions of the network virtual terminal onto the real terminal. For
example, when the editor moves the virtual terminal's cursor to the upper left-hand corner of the
screen, this software must issue the proper command sequence to the real terminal to get its cursor
there too. All the virtual terminal software is in the application layer.
Another application layer function is file transfer. Different file systems have different file
naming conventions, different ways of representing text lines, and so on. Transferring a file
between two different systems requires handling these and other incompatibilities. This work, too,
belongs to the application layer, as do electronic mail, remote job entry, directory lookup, and
various other general purpose and special-purpose facilities.
2.2 Data Transmission in the OSI Model
Figure 2-2 shows an example of how data can be transmitted using the OSI model The sending
process has some data it wants to Send to the receiving process. It gives the data to the application
layer, which then attaches the application header, AH (which may be null), to the front of it and
AP A
7
6

4
3
2
1

7
6

PPDU

5
End-to-end

AP B

AP dat
a
APDU

SPDU

TPDU
NPDU

packet

LPDU

frame
Bit stream

3
2
1

media
Fig. 2-2. An example of how the OSI model is used. Some of the headers may be null

15

gives the resulting item to the presentation layer.

The presentation layer may transform this item in various ways and possibly add a header to the
front, giving the result to the session layer. It is important to realize that the presentation layer is
not aware of which portion of the data given to it by the application layer is AH, if any, and which
is true user data.
This process is repeated until the data reach the physical layer, where they are actually
transmitted to the receiving machine. On that machine the various headers are stripped off one by
one as the message propagates up the layers until it finally arrives at the receiving process.
The key idea throughout is that although actual data transmission is vertical in Fig. 2-2, each
layer is programmed as though it were horizontal. When the sending transport layer, for example,
gets a message from the session layer, it attaches a transport header and sends it to the receiving
transport layer, From its point of view, the fact that it must actually hand the message to the
network layer on its own machine is an unimportant technicality. As an analogy, when a Tagalogspeaking diplomat is addressing the United Nations, he thinks of himself as addressing the other
assembled diplomats. That, in fact, he is really only speaking to his translator is seen as a technical
detail.
2.3 The TCP/IP Reference Model
Let us now turn from the OSI reference model to the reference model used in the grandparent
of all computer networks, the ARPANET, and its successor, the worldwide Internet. As well
known, the ARPANET was a research network sponsored by the DoD (U.S. Department of
Defense). It eventually connected hundreds of universities and government installations using
leased telephone lines. When satellite and radio networks were added later, the existing protocols
had trouble interworking with them, so a new reference architecture was needed. Thus the ability
to connect multiple networks together in a seamless way was one of the major design goals from
the very beginning. This architecture later became known as the TCP/IP Reference Model, after its
two primary protocols. It was first defined in (Ced and Kahn, 1974). A later perspective is given in
(Leiner et al., 1985). The design philosophy behind the model is discussed in (Clark, 1988).
Given the DoD's worry that some of its precious hosts, routers, and internetwork gateways
might get blown to pieces at a moment's notice, another major goal was that the network be able to
survive loss of subnet hardware, with existing conversations not being broken off. In other words,
DoD wanted connections to remain intact as long as the source and destination machines were
functioning, even if some of the machines or transmission lines in between were suddenly put out
of operation. Furthermore, a flexible architecture was needed, since applications with divergent
requirements were envisioned, ranging from transferring files to real-time speech transmission.
2.3.1 The Internet Layer
All these requirements led to the choice of a packet-switching network based on a
connectionless internetwork layer. This layer, called the internet layer, is the linchpin that holds
the whole architecture together. Its job is to permit hosts to inject packets into any network and
have them travel independently to the destination (potentially on a different network). They may
even arrive in a different order than they were sent, in which case it is the job of higher layers to
rearrange them, if in-order delivery is desired.
The analogy here is with the (snail) mail system. A person can drop a sequence of international
16

letters into a mall box in one country, and with a little luck, most of them will be delivered to the
correct address in the destination country. Probably the letters will travel through one or more
international mail gateways along the way, but this is transparent to the users. Furthermore, that
each country (i.e., each network) has its own stamps, preferred envelope sizes, and delivery rules
is bidden from the users.
The internet layer defines an official packet format and protocol called IP (Internet Protocol).
The job of the internet layer is to deliver IP packets where they are supposed to go. Packet routing
is clearly the major issue here, as is avoiding congestion. For these reasons, it is reasonable to say
that the TCP/IP internet layer is very similar in functionality to the OSI network layer. Figure 2-3
shows this correspondence.
Internet / Intranet

OSI_RM
57
4

Application
Transport

Telnet

FTP

TCP

SMTP

DNS

UDP

Others
NVP

ICMP
IP

Rou
Rou
ter
ter

1
NIL
Hardware

IP

Brid
Brid
ger
ger
LAN
Hu
Hu
bb

ARP

RARP

Others
WAN

Figure 2-3 Protocols and networks in the TCP/IP stacks

2.3.2 The Transport Layer

The layer above the internet layer in the TCP/lP model is now usually called the transport layer.
It is designed to allow peer entities on the source and destination hosts to carry on a conversation,
the same as in the OSI transport layer.
Two end-to-end protocols have been defined here. The first one, TCP (Transmission Control
Protocol) is a reliable connection-oriented protocol that allows a byte stream originating on one
machine to be delivered without error on any Other machine in the internet. It fragments the
incoming byte stream into discrete messages and passes each one onto the internet layer. At the
destination, receiving TCP process reassembles the received messages into the output stream. TCP
also handles flow control to make sure a fast sender cannot swamp a slow receiver with more
messages than it can handle.
The second protocol in this layer, UDP (User Datagram Protocol), is an unreliable,
connectionless protocol for applications that do not want TCP's sequencing or flow control and
wish to provide their own. It is also widely used for one-shot, client-server type request-reply
queries and applications in which prompt delivery is more important than accurate delivery, such
as transmitting speech or video. The relation of IP, TCP, and UDP is shown in Fig. 2-3. Since the
model was developed; IP has been implemented on many other networks.

17

2.3.3 The Application Layer

The TCP/IP model does not have session or presentation layers. No need for them was
perceived, so they were not included. Experience with the OSI model has proven this view
correct: they are of little use to most applications.
On top of the transport layer is the application layer. It contains all the higher-level protocols.
The early ones included virtual terminal (TELNET), file transfer (FTP), and electronic mail
(SMTP), as shown in Fig. 2-3. The virtual terminal protocol allows a user on one machine to log
into a distant machine and work there. The file transfer protocol provides a way to move data
efficiently from one machine to another. Electronic mall was originally just a kind of file transfer,
but later a specialized protocol was developed for it. Many other protocols have been added to
these over the years, such as the Domain Name Service (DNS) for mapping host names onto their
network addresses, NNTP, the protocol used for moving news articles around, and HTTP, the
protocol used for fetching pages on the World Wide Web, and many others.
2.3.4 The Host-to-Network Layer
Below the internet layer is a great void. The TCP/IP reference model does not really say
much about what happens here, except to point out that the host has to connect to the network
using some protocol so it can send IP packets over it. This protocol is not defined and varies from
host to host and network to network.

18

PUBLIC DATA NETWORKS TECHNOLGY

3.1 X.25 Packet Switching Public Data Networks

Many older public data networks in the would follow a standard called X.25 It was developed
during the 1970s by ITU-T(old name CCITT ) to provide an interface protocol including three
levels between public packet-switched networks (DCE, Data Communication Equipment) and
their customers(DTE, Data Terminal Equipment) ,as shown in FIG. 3-1.
DCE
P_DTE

VC1

P_DTE

PSPDN
DCE
X.25

High
level

X.25

VC 2
DCE

P_DTE
Logical Channel
TPDU

3
2
1

Packet
Packet
Fram
Fram
LAPB
e
e
Physica
Physica
l
l
DTE

H
F AC

D
I

FCSF

Bit stream

DCE

Figure 3-1

X.25 Interface Protocol

The physical layer protocol, called X.21, specifies the physical, electrical, and procedural
interface between the host and the network. Very few public networks actually support this
standard, because it requires digital, rather than analog signaling on the telephone lines. As an
interim measure, an analog interface similar to the familiar RS-232 standard was defined.
The data link layer (Frame Level) standard has a number of (slightly incompatible) variations.
They all are designed to deal with transmission errors on the telephone line between the user's
equipment (host or terminal) and the public network (router). The data link layer standard is
referred to as LAPB ( Link Access Protocol for Balance ), which is a subset of HDLC ( High
Data Link Control ).
The network layer (Packet Level) protocol deals with addressing, flow control, delivery
confirmation, interrupts, and related issues. Basically, it allows the user to establish virtual circuits
and then send packets of up to 128 bytes on them. These packets are delivered reliably and in
order. Most X.25 networks work at speeds up to 64 kbps, which makes them obsolete for many
purposes. Nevertheless, they are still widespread, so readers should be aware of their existence.
19

X.25 is connection-oriented and supports both switched virtual circuits (SVC) and
permanent virtual circuits(PVC). A switched virtual circuit is created when one computer sends a
packet to the network asking to make a call to a remote computer. DTE / DCE address number is
assigned in term of X.121. Once established, packets can be sent over the connection, always
arriving in order. X.25 provides flow control, to make sure a fast sender cannot swamp a slow or
busy receiver. It is used to distinguish which the virtual circuit to use according to the
LCGN+LCN (Logical Channel Number).
A permanent virtual circuit is used the same way as a switched one, but it is set up in advance
by agreement between the customer and the carrier. It is always present, and no call setup is
required to use it. It is analogous to a leased line.
Because the world is still full of terminals that do not speak X.25, another set of standards
was defined that describes how an ordinary (non-intelligent) terminal communicates with an X.25
public network. In effect, the user or network operator installs a "black box" to which these
terminals can connect. The black box is called a PAD (Packet Assembler Disassembler), and its
function is described in a document known as X.3. A standard protocol has been defined between
the terminal and the PAD, called X.28; another standard protocol exists between the PAD and the
network, called X.29. Together, these three recommendations are often called triple X.
3.2 Frame Relay
Frame relay is a service for people who want an absolute bare-bones connection-oriented
way to move bits from A to B at reasonable speed and low cost (Smith, 1993). Its existence is due
to changes in technology over the past two decades. Twenty years ago, communication using
telephone lines was slow, analog, and unreliable, and computers were slow and expensive. As a
result, complex protocols were required to mask errors, and the users' computers were too
expensive to have them do this work.
The situation has changed radically. Leased telephone lines are now fast, digital, and reliable,
and computers are fast and inexpensive. This suggests the use of simple protocols, with most of
the work being done by the users' computers, rather than by the network. It is this environment
that frame relay addresses.
The architecture of Frame relay is shown in Figure 3-2 . Frame relay can best be thought of
as a virtual leased line. The customer leases a permanent virtual circuit between two points and
can then send frames (i.e., packets) of up to 1600 bytes between them. It is also possible to lease
permanent virtual circuits between a given site and multiple other sites, so each frame carries a 10bit number (DLCI, Data Link Connection Identification) telling which virtual circuit to use.
OSI

C-plane

U-plane

Q.931
orI.451

Option for
users

Q.922
orI.441

U-plane

C-plane
Q.931
or I.451

Q.922
core

Q.922
core

Q.922
or I.441

LAPF
I.430orI.431

I.430 I.431

Figure 3-2 The architecture of Frame relay

20

The difference between an actual leased line and a virtual leased line is that with an actual
one, the user can send traffic all day long at the maximum speed. With a virtual one, data bursts
may be sent at full speed, but the long-term average usage must be below a predetermined level.
In return, the carrier charges much less for a virtual line than a physical one.
In addition to competing with leased lines, frame relay also competes with X.25 permanent
virtual circuits, except that it operates at higher speeds, usually 1.544 Mbps(T 1) or 2.048
Mbps(E1), and provides fewer features.
Frame relay provides a minimal service, primarily a way to determine the start and end of each
frame, and detection of transmission errors. If a bad frame is received, the frame relay service
simply discards it. It is up to the user to discover that a frame is missing and take the necessary
action to recover. Unlike X.25, frame relay does not provide acknowledgements or normal flow
control. It does have a bit in the header, however, which one end of a connection ~:an set to
indicate to the other end that problems exist. The use of this bit is up to the users.
3.3 Broadband ISDN and ATM
Even if the above services become popular, the telephone companies are still faced with a far
more fundamental problem: multiple networks. POTS (Plain Old Telephone Service) and Telex
use the old circuit-switched network. Each of the new data services such as SMDS and frame
relay uses its own packet-switching network. DQDB is different from these, and the internal
telephone company call management network (SSN 7) is yet another network. Maintaining all
these separate networks is a major headache, and there is another network, cable television, that
the telephone companies do not control and would like to.
The perceived solution is to invent a single new network for the future that will replace the
entire telephone system and all the specialized networks with a single integrated network for all
kinds of information transfer. This new network will have a huge data rate compared to all existing
networks and services and will make it possible to offer a large variety of new services. This is not
a small project, and it is certainly not going to happen overnight, but it is now under way.
The new wide area service is called B-ISDN (Broadband Integrated Services Digital
Network). It will offer video on demand (VOD), live television from many sources, full motion
multimedia electronic mail, CD-quality music, LAN interconnection, high-speed data transport for
science and industry and many other services that have not yet even been thought of, all over the
telephone line.
The underlying technology that makes B-ISDN possible is called ATM (Asynchronous Transfer
Mode) because it is not synchronous (tied to a master lock), as most long distance telephone lines
are. Note that the acronym ATM here has nothing to do with the Automated Teller Machines many
banks provide (although an ATM machine can use an ATM network to talk to its bank).
The basic idea behind ATM is to transmit all information in small, fixed-size packets called
cells. The cells are 53 bytes long, of which 5 bytes are header and 48 bytes are payload, the ATM
cell format is shown in Fig. 3-3. ATM is both a technology (hidden from the users) and potentially
a service (visible to the users). Sometimes the service is called cell relay, as an analogy to frame
relay.
The use of a cell-switching technology is a gigantic break with the 100-year old tradition of
circuit switching (establishing a copper path) within the telephone system. There are a variety of
reasons why cell switching was chosen, among them are the following. First, cell switching is
21

highly flexible and can handle both constant rate traffic (audio, video) and variable rate traffic
Byte
1
2
3
1
4
5

GFC/VPI

VPI

VPI

VCI
VCI

VCI

GFC general flow control

PT

CLP

HEC

VPI virtual path identification

VCI virtual channel identification
PT payload type

PAYLOAD

CLP cell loss priority

HEC header error control

Fig. 3-2 ATM cell format

(data) easily. Second, at the very high speeds envisioned (gigabits per second are within reach),
digital switching of cells is easier than using traditional multiplexing techniques, especially using
fiber optics. Third, for television distribution, broadcasting is essential; cell switching can provide
this and circuit switching cannot.
ATM networks are connection-oriented. Making a call requires first sending a message to set
up the connection. After that, subsequent cells all follow the same path to the destination. Cell
delivery is not guaranteed, but their order is. If cells 1 and 2 are sent in that order, then if both
arrive, they will arrive in that order, never first 2 then 1.
ATM networks are organized like traditional WANs, with lines and switches (routers). The
intended speeds for ATM networks are 155 .520Mbps and 622.080 Mbps, with the possibility of
gigabit speeds later. The 155-Mbps speed was chosen because this is about what is needed to
transmit high definition television. The exact choice of 155.52 Mbps was made for compatibility
with AT&T's SONET transmission system. The 622 Mbps speed was chosen so four 155-Mbps
channels could be sent over it. By now it should be clear why some of the gigabit testbeds
operated at 622 Mbps: they used ATM.
When ATM was proposed, virtually all the discussion (i.e., the hype) was about video on
demand to every home and replacing the telephone system, as described above. Since then, other
developments have become important. Many organizations have run out of bandwidth on their
campus or building-wide LANs and are being forced to go to some kind of switched system that
has more bandwidth than does a single LAN. Also, in client-server computing, some applications
need the ability to talk to certain servers at high speed. ATM is certainly a major candidate for
both of these applications. Nevertheless, it is a bit of a let down to go from a goal of trying to
replace the entire low-speed analog telephone system with a high-speed digital one to a goal of
trying connect all the Ethernets on campus.
It is also worth pointing out that different organizations involved in ATM have different
(financial) interests. The long-distance telephone carriers and PTTs are mostly interested in using
ATM to upgrade the telephone system and compete with the cable TV companies in electronic
video distribution. All these competing interests do not make the ongoing standardization process
any easier, faster, or more coherent. Also, politics and power within the organization standardizing
ATM (The ATM Forum) have considerable influence on where ATM is going.
22

3.3.1 The B-ISDN ATM Reference Model

Let us now turn back to the technology of ATM, especially as used in the telephone system.
Broadband ISDN using ATM has its own reference model, different from the OSI model and also
different from the TCP/IP model. This model is shown in Fig. 3-4. It consists of three layers, the
physical, ATM, and ATM adaptation layers, plus whatever the users want to put on top of that.
Management-plane
Plane Management
Control -plane
High level

User-plane
High level

ATM Adapt Layer I . 362/363

Layer Management

ATM Layer I . 361

Physical Layer

Fig. 3-4 B_ISDN ATM reference model

The physical layer deals with the physical medium: voltages, bit timing, and various other
issues. ATM does not prescribe a particular set of rules, but instead says that ATM cells may be
sent on a wire or fiber by themselves, but they may also be packaged inside the payload of other
carrier systems. In other words, ATM has been designed to be independent of the transmission
medium.
The ATM layer deals with cells and cell transport. It defines the layout of a cell and tells what
the header fields mean. It also deals with establishment and release of virtual circuits. Congestion
control is also located here.
Because most applications do not want to work directly with cells (although some may), a
layer above the ATM layer has been defined that allows users to send packets larger than a cell.
The ATM interface segments these packets, transmits the cells individually, and reassembles them
at the other end. This layer is the AAL (ATM Adaptation Layer).
Unlike the earlier two-dimensional reference models, the ATM model is defined as being threedimensional, as shown in Fig. 3-4. The user plane deals with data transport, flow control, error
correction, and other user functions. In contrast, the control plane is concerned with connection
management. The layer and plane management functions relate to resource management and
interlayer coordination.
The physical and AAL layers are each divided into two sublayers, one at the bottom that does
the work and a convergence sublayer on top that provides the proper interface to the layer above
it.
The PMD (Physical Medium Dependent) sublayer interfaces to the actual cable. It moves the
bits on and off and handles the bit timing. For different carders and cables, this layer will be
different.
23

The other sublayer of the physical layer is the TC (Transmission Convergence) sublayer. When
cells are transmitted, the TC layer sends them as a string of bits to the PMD layer. Doing this is
easy. At the other end, the TC sublayer gets a pure incoming bit stream from the PMD sublayer. Its
job is to convert this bit stream into a cell stream for the ATM layer. It handles all the issues
related to telling where cells begin and end in the bit stream. In the ATM model, this functionality
is in the physical layer. In the OSI model and in pretty much all other networks, the job of
framing, that is, turning a raw bit stream into a sequence of frames or cells, is the data link layer's
task. For that reason we will discuss it in this book along with the data link layer, not with the
physical layer.
As we mentioned earlier, the ATM layer manages cells, including their generation and
transport. Most of the interesting aspects of ATM are located here. It is a mixture of the OSI data
link and network layers, but it is not split into sublayers.
The AAL layer is split into a SAR (Segmentation And Reassembly) sublayer and a CS
(Convergence Sublayer). The lower sublayer breaks packets up into cells on the transmission side
and puts them back together again at the destination. The upper sublayer makes it possible to have
ATM systems offer different kinds of services to different applications (e.g., file transfer and video
on demand have different requirements concerning error handling, timing, etc.).

24

LAN AND MAN

In this part, we examine local area networks (LANs) and metropolitan area networks
(MANs). These networks share the characteristic of being packet broadcasting networks. With a
broadcast communications network, each station is attached to a transmission medium shared by
other stations. In its simplest form, a transmission from any one station is broadcast to and
received by all other stations. As with packet-switched networks, transmission on a packet
broadcasting network is in the form of packets.
The useful definitions of LANs and MANs taken from one of the IEEE 802 standards
documents are described as following.
The LANs described herein are distinguished from other types of data networks in that they
are optimized for a moderate size geographic area such as a single office building, a warehouse, or
a campus. The IEEE 802 LAN is a shared medium peer-to-peer communications network that
broadcasts information for all stations to receive. As a consequence, it does not inherently provide
privacy. The LAN enables stations to communicate directly using a common physical medium on
a point-to-point basis without any intermediate switching node being required. There is always
need for an access sublayer in order to arbitrate the access to the shared medium. The network is
generally owned, used, and operated by a single organization. This is in contrast to Wide Area
Networks (WANs) that interconnect communication facilities in different parts of a country or are
used as a public utility. These LANs are also different from networks that are optimized for the
interconnection of devices on a desk top or components within a single piece of equipment.
A MAN is optimized for a larger geographical area than a LAN, ranging from several blocks
of buildings to entire cities. As with local networks, MANs can also depend on communications
channels of moderate-to-high data rates. Error rates and delay may be slightly higher than might
be obtained on a LAN. A MAN might be owned and operated by a single organization, but usually
will be used by many individuals and organizations. MANs might also be owned and operated as
public utilities. They will often provide means for internetworking of local networks. Although not
a requirement for all LANs, the capability to perform local networking of integrated voice and
data (IVD) devices is considered an optional function for a LAN. Likewise, such capabilities in a
network covering a metropolitan area are optional functions of a MAN.
The key technology ingredients that determine the nature of a LAN or MAN are
Topology
Transmission medium
Medium access control technique
4.1 LAN Architecture
The architecture of a LAN is best described in terms of a layering of protocols that
organize the basic functions of a LAN. This section opens with a description of the standardized
protocol architecture for LANs, which encompasses physical, medium access control, and logical
link control layers. Each of these layers is then examined in turn.
25

4.1.1 Protocol Architecture

Protocols defined specifically for LAN and MAN transmission address issues relating to the
transmission of blocks of data over the network. In OSI terms, higher layer protocols (layer 3 or 4
and above) are independent of network architecture and are applicable to LANs, MANs, and
WANs. Thus, a discussion of LAN protocols is concerned principally with lower layers of the OSI
model.
Figure 4-1 relates the LAN protocols to the OSI architecture (first introduced in Figure 2-1).
This architecture was developed by the IEEE 802 committee and has been adopted by all
organizations working on the specification of LAN standards. It is generally referred to as the
IEEE 802 reference model.
OSI-RM

Application

Presentation

Session

Transport

Network

internetwork

Data Link

LLC

MSAP

Physical

MAC

PSAP

IEEE LAN&MAN reference model

high level protocol

Physical

NSAP
LSAP
MSAP
PSAP
Physical

Fig. 4-1 IEEE802 LAN&MAN reference model (protocol layers compared to OSI model)
For the sake of brevity, the material often uses LAN when referring to LAN and MAN
concerns. The context should clarify when only LAN or both LAN and MAN is meant. Working
from the bottom up, the lowest layer of the IEEE 802 reference model corresponds to the physical
layer of the OSI model, and includes such functions as
Encoding/decoding of signals
Preamble generation/removal (for synchronization)
Bit transmission/reception
In addition, the physical layer of the 802 model includes a specification of the transmission
medium and the topology. Generally, this is considered below the lowest layer of the OSI model.
However, the choice of transmission medium and topology is critical in LAN design, and so a
specification of the medium is included.
Above the physical layer are the functions associated with providing service to LAN users.
These include
On transmission, assemble data into a frame with address and error-detection fields.
On reception, disassemble frame, perform address recognition and error detection.
Govern access to the LAN transmission medium.
Provide an interface to higher layers and perform flow and error control.
26

These are functions typically associated with OSI layer 2. The set of functions in the last
bulleted item are grouped into a logical link control (LLC) layer. The functions in the first three
bullet items are treated as a separate layer, called medium access control (MAC). The separation is
done for the following reasons:
The logic required to manage access to a shared-access medium is not found in
traditional layer-2 data link control.
For the same LLC, several MAC options may be provided.
The standards that have been issued are illustrated in Figure 4-2. Most of the standards were
developed by a committee known as IEEE 802, sponsored by the Institute for Electrical and
Electronics Engineers. All of these standards have subsequently been adopted as international
standards by the International Organization for Standardization (ISO).
ISO/OSI_RM

IEEE 802 standard

802.10 security

802.1:Architecture, Addressing, Inter-network, Management

5
4

802.2 LLC, Logical Link Control

3
2

MAC

802.3 802.4 802.5

CSMA/ Token Token
Bus
CD
Ring

802.6

802.9

802.11 802.12

MAN

IntLAN

100VG
W-LAN -Any

Phusical

Media
802.7 Broadband-LAN

802.8 fiber transmission

Figure 4-2 IEEE 802 standard

4.1.2 LAN Topologies
For the physical layer, we confine our discussion for now to an introduction of the basic LAN
topologies. The common topologies for LANs are bus, tree, ring, and star (Figure 4-3). The bus is
a special case of the tree, with only one trunk and no branches; we shall use the term bus/tree
when the distinction is unimportant.
1. Bus topology
Both bus and tree topologies are characterized by the use of a multipoint medium. For the
bus, all stations attach, through appropriate hardware interfacing known as a tap, directly to a
linear transmission medium, or bus. Full-duplex operation between the station and the tap allows
data to be transmitted onto the bus and received from the bus. A transmission from any station
propagates the length of the medium in both directions and can be received by all other stations.
At each end of the bus is a terminator, which absorbs any signal, removing it from the bus.
2. Tree topology
The tree topology is a generalization of the bus topology. The transmission medium is a
27

branching cable with no closed loops. The tree layout begins at a point known as the headend,
where one or more cables start, and each of these may have branches. The branches in turn may
have additional branches to allow quite complex layouts. Again, a transmission from any station
propagates throughout the medium and can be received by all other stations.
3. Ring topology
In the ring topology, the network consists of a set of repeaters joined by point-to-point links in a
closed loop. The repeater is a comparatively simple device, capable of receiving data on one link
and transmitting them, bit by bit, on the other link as fast as they are received, with no buffering at
the repeater. The 'links are unidirectional; that is, data are transmitted in one direction only and all
are oriented in the same way. Thus, data circulate around the ring in one direction (clockwise or
counterclockwise). Each station attaches to the network at a repeater and can transmit data onto
the network through that repeater.
Tree

Bus

Star
Ring

Figure 4-3 LAN&MAN topologies

4. Star Topology
In the star LAN topology, each station is directly connected to a common central node.
Typically, each station attaches to a central node, referred to as the star coupler, via two point-topoint links, one for transmission in each direction.
In general, there are two alternatives for the operation of the central node. One approach is
for the central node to operate in a broadcast fashion. A transmission of a frame from one station
to the node is retransmitted on all of the outgoing links. In this case, although the arrangement is
physically a star, it is logically a bus; a transmission from any station is received by all other
stations, and only one station at a time may successfully transmit.
Another approach is for the central node to act as a frame switching device. An incoming
frame is buffered in the node and then retransmitted on an outgoing link to the destination station.
4.2 IEEE 802.3/ Ethernet
28

The most commonly used medium access control technique for bus/tree and star
topologies is carrier-sense multiple access with collision detection (CSMA/CD). The original
baseband version of this technique was developed by Xerox as part of the Ethernet LAN. The
original broadband version was developed by MITRE as part of its MITREnet LAN. All of this
work formed the basis for the IEEE 802.3 standard.
4.2.1 IEEE 802.3 Medium Access Control
In this section, we will focus on the IEEE 802.3 / Ethernet Medium Access Control schemes
-- CSMA/CD (Carrier Sense Multiple Access / Collision Detection).
The operation rules for CSMA/CD are as following:
1. If the medium is idle, transmit; otherwise, go to step 2.
2. If the medium is busy, continue to listen until the channel is idle, then transmit
immediately.
3. If a collision is detected during transmission, transmit a brief jamming signal to assure that
all stations know that there has been a collision and then cease transmission.
4. After transmitting the jamming signal, wait a random amount of time, then attempt to
transmit again. (Repeat from step 1.)
Figure 4-4 illustrates the technique for a baseband bus. At time to, station A begins
transmitting a packet addressed to B. At t y, station B is ready to transmit. B senses a transmission
and it is empty. So B, however, begins its own transmission. When A's transmission reaches on the
way, at t, B detects the collision and ceases transmission. The effect of the collision propagates
back to A, where it is detected some time later, at t2, at which time A ceases transmission .
A

1 km

B
Y

B finds the collision

A finds the collision

tj

Legend:
Y random variable
tj jam time

tj
Fig. 4-4 CSMA/CD operation and collision detection
4.2.2 IEEE 802.3 / Ethernet Frame Format
Figure 4-5 depicts the frame format for the 802.3 / Ethernet protocol, it consists of the
following fields:
Pr: Preamble (7byte)
SD: Start frame Delimiter (1byte)
DA: Destination Address (6byte)
SA: Source Address (6byte)
L: Length of the LLC data field (2byte) for 802.3
T: Type
for Ethernet
29

I: data unit supplied by LLC

FCS: Frame Check Sequence (4byte)
DSAP: Destination Service Access Point
SSAP: Source Service Access Point
C: Control field

High PDU
Byte 1 1 1-2
DSAP SSAP C

LLC PDU

Data

MAC Frame

T
1500

Pr

802.3

SD DA SA L

PAD

FCS

Figure 4-5 the frame format for the 802.3 / Ethernet protocol
4.2.3 IEEE 802.3 / Ethernet Series Specification
IEEE 802.3 / Ethernet 10 Mbps specifications are shown in Table 4-1.
Table 4-1 IEEE 802.3 / Ethernet 10 Mbps specifications
10Base5

10Base2

10BaseT

10Broad36

10Base-F

Transmission

Coaxial Cable

Coaxial Cable

Unshielded

Coaxial Cable

850-nm Optical

medium

(50 ohm)

(50 ohm)

Twisted Pair

(75 ohm)

Fiber Pair

Broadband

Manchester

(DPSK)

On-off

1800

500

Signaling

Baseband (Manchester)

technique
Maximum

500

185

100

30

100

segment
length(m)
Nodes per

33

segment

Fast Ethernet refers to a set of specifications developed by the IEEE 802.3 committee to
provide a low-cost, Ethernet-compatible LAN operating at 100 Mbps. The blanket designation for
these standards is 100BASE-T. The committee defined a number of alternatives to be used with
different transmission media.
Figure 4-6 shows the terminology used in labeling the specifications and indicates the media
used. All of the 100BASE-T options use the IEEE 802.3 MAC protocol and frame format.
100BASE-X refers to a set of options that use the physical medium specifications originally
defined for Fiber Distributed Data Interface (FDDI). All of the 100BASE-X schemes use two
physical links between nodes: one for transmission and one for reception. 100BASE-TX makes
use of shielded twisted pair (STP) or high-quality (Category 5) unshielded twisted pair (UTP).

30

100BASE-FX uses optical fiber.

In many buildings, each of the 100BASE-X options requires the installation of new cable.
For such cases, 100BASE-T4 defines a lower-cost alternative that can use Category-3, voice grade
UTP in addition to the higher-quality Category 5 UTP.4 To achieve the 100-Mbps data rate over
lower-quality cable, 100BASE-T4 dictates the use of four twisted pair lines between nodes, with
the data transmission making use of three pairs in one direction at a time.
For all of the 100BASE-T options, the topology is similar to that of 10BASE-T, namely a
star-wire topology.
For all of the transmission media specified under 100BASE-X, a unidirectional data rate of
I00 Mbps is achieved by transmitting over a single link (single twisted pair, single optical fiber).
For all of these media, an efficient and effective signal
IEEE 802.3 (100mbps)

100Base-X

100Base-TX

2 Category5 UTP 2 STP

100Base-FX

2 Optical fiber

100Base-T4
4 Category3 or
Category5 UTP

Fig.4-6 IEEE 802.3 100Base options

4.2.4 Gigabit Ethernet
Gigabit Ethernet builds on top of the Ethernet protocol, but increases speed tenfold over Fast
Ethernet to 1000 Mbps, or 1 gigabit per second (Gbps). Gigabit Ethernet allows Ethernet to scale
from 10/100 Mbps at the desktop to 100 Mbps up the riser to 1000 Mbps in the data center.
By leveraging the current Ethernet standard as well as the installed base of Ethernet and Fast
Ethernet switches and routers, network managers do not need to retrain and relearn a new
technology in order to provide support for Gigabit Ethernet.
In order to accelerate speeds from 100 Mbps Fast Ethernet up to 1 Gbps, several changes
need to be made to the physical interface. It has been decided that Gigabit Ethernet will look
identical to Ethernet from the data link layer upward. The challenges involved in accelerating to
1 Gbps have been resolved by merging two technologies together: IEEE 802.3 Ethernet and
ANSI X3T11 Fiber Channel. Figure 4-7 shows how key components from each technology have
been leveraged to form Gigabit Ethernet.

31

Fig. 4-7 Gigabit Ethernet Protocol Stack

Leveraging these two technologies means that the standard can take advantage of the existing
high-speed physical interface technology of Fibre Channel while maintaining the IEEE 802.3
Ethernet frame format, backward compatibility for installed media, and use of full- or half-duplex
carrier sense multiple access collision detect (CSMA/CD). This scenario helps minimize the
technology complexity, resulting in a stable technology that can be quickly developed. The actual
model of Gigabit Ethernet is shown in Figure 2. Each of the layers will be discussed in detail.

Fig.4-8 Architectural Model of IEEE 802.3z Gigabit Ethernet

(source: IEEE Media Access Control parameters, physical layers, repeater and management
parameters for 1000-Mbps operation)
4.2.5 Migration to Gigabit Ethernet Related Standards
The following sections briefly summarize four related IEEE standards.
1. IEEE 802.IP
Quality of service has become increasingly important to network managers. In June 1998, the
32

IEEE 802.IP committee standardized a means of individual end station requesting a particular QoS
of the network and the network being able to respond accordingly. This standard also specifies
multicast group management.
A new protocol is defined in 802.IP, generic attribute registration protocol (GARP). GARP is
a generic protocol that will be used by specific GARP applications; for example, GARP multicast
registration protocol (GMRP), and GARP VLAN registration protocol (GVRP). GMRP is defined
in 802.IP; GMRP provides registration services for multicast MAC address groups.
2. IEEE 802.1Q
The introduction of virtual LANs (VLANs) into switched internetworks has created
significant advantages to networking vendors because they can offer value-added features to their
products such as VLAN trunking, reduction in spanning-tree recalculations effects, and broadcast
control. However, with the exception of ATM LAN emulation, there is no industry standard means
of creating VLANs.
The 802.1Q committee has worked to create standards-based VLANs. This standard is based
on a frame-tagging mechanism that will work over Ethernet, Fast Ethernet, Token Ring, and
FDDI. The standard will allow a means of VLAN tagging over switches and routers and will allow
vendor VLAN interoperability. GVRP has been introduced in 802.1Q; this protocol provides
registration services for VLAN membership.
3. IEEE 802.3x
The IEEE 802.3x committee standardized a method of flow control for full-duplex Ethernet.
This mechanism is set up between the two stations on the point-to-point link. If the receiving
station at the end becomes congested, it can send back a frame called a "pause frame" to the
source at the opposite end of the connection, instructing that station to stop sending packets for a
specific period of time. The sending station waits the requested time before sending more data.
The receiving station can also send a frame back to the source with a time-to-wait of zero,
instructing the source to begin sending data again. (See Figure 4-9)

Fig. 4-9 Operation of IEEE 802.3x Flow Control

This flow-control mechanism was developed to match the sending and receiving device
throughput. For example, a server can transmit to a client at a rate of 3000 pps. The client,
however, may not be able to accept packets at that rate because of CPU interrupts, excessive
network broadcasts, or multitasking within the system. In this example, the client sends out a
pause frame and requests that the server delay transmission for a certain period of time. This
mechanism, though separate from the IEEE 802.3z work, complements Gigabit Ethernet by
allowing gigabit devices to participate in this flow-control mechanism.
4. IEEE 802.3ab
33

The IEEE 802.3ab committee specified Gigabit Ethernet transmission over Category 5
copper cable (1000BASE-T). For more information, see "1000BASE-T: Delivering Gigabit
Intelligence on Copper Infrastructure."
http://www.cisco.com/warp/public/cc/techno/lnty/etty/ggetty/tech/1000b_sd.htm
Gigabit Ethernet is a viable technology that allows Ethernet to scale from 10/100 Mbps at the
desktop to 100 Mbps up the riser to 1000 Mbps in the data center. By leveraging the current
Ethernet standard as well as the installed base of Ethernet and Fast Ethernet switches and routers,
network managers do not need to retrain and relearn a new technology in order to provide support
for Gigabit Ethernet.
4.3 Wireless LAN
A set of wireless LAN standards has been developed by the IEEE 802.11 committee. The
terminology and some of the specific features of 802.11 are unique to this standard and are not
reflected in all commercial products. However, it is useful to be familiar with the standard as its
features are representative of required wireless LAN capabilities.
Figure 4-10 indicates the model developed by the 802.11 working group. The smallest
building block of a wireless LAN is a basic service set (BSS), which consists of some number of
stations executing the same MAC protocol and competing for access to the same shared medium.
A basic service set may be isolated, or it may connect to a backbone distribution system through

ESS

Extended Service Set

PC

Server
PC
AP
PC

AP
BSS
Fig. 4-10

BSS

Distribution System

IEEE802.11 WLAN topologies

an access point. The access point functions as a bridge. The MAC protocol may be fully
distributed or controlled by a central coordination function housed in the access point. The basic
service set generally corresponds to what is referred to as a cell in the literature.
An extended service set (ESS) consists of two or more basic service sets interconnected by a
distribution system. Typically, the distribution system is a wired backbone LAN. The extended
service set appears as a single logical LAN to the logical link control (LLC) level.
The standard defines three types of stations, based on mobility:
No-transition. A station of this type is either stationary or moves only within the direct
communication range of the communicating stations of a single BSS.
BSS-transition. This is defined as a station movement from one BSS to another BSS
within the same ESS. In this case, delivery of data to the station requires that the
34

addressing capability be able to recognize the new location of the station.

ESS-transition. This is defined as a station movement from a BSS in one ESS to a BSS
within another ESS. This case is supported only in the sense that the Station can move.
Maintenance of upper-layer connections supported by 802.11 cannot be guaranteed. In
fact, disruption of service is likely to occur.

4.3.1 Physical Medium Specification

Three physical media are defined in the current 802.11 standard:
Infrared at 1 Mbps and 2 Mbps operating at a wavelength between 850 and 950 nm.
Direct-sequence spread spectrum operating in the 2.4-GHz ISM band. Up to 7 channels,
each with a data rate of 1 Mbps or 2 Mbps, can be used.
Frequency-hopping spread spectrum operating in the 2.4-GHz ISM hand. The details of
this option are for further study:
4.3.2 Medium Access Control
The 802.11 working group considered two types of proposals for a MAC algorithm:
distributed-access protocols which, like CSMA/CD, distributed the decision to transmit over
all the nodes using a carrier-sense mechanism; and centralized access protocols, which involve
regulation of transmission by a centralized decision maker.
A distributed access protocol makes sense of an ad hoc network of peer workstations and
may also be attractive in other wireless LAN configurations that consist primarily of bursty traffic.
A centralized access protocol is natural for configurations in which a number of wireless stations
are interconnected with each other and with some sort of base station that attaches to a backbone
wired LAN; it is especially useful if some of the data is time-sensitive or high priority.
The end result of the 802.11 is a MAC algorithm called DFWMAC (distributed foundation
wireless MAC) that provides a distributed access-control mechanism with an optional centralized
control built on top of that. Figure 4-11 illustrates the architecture. The lower sublayer of the MAC
layer is the distributed coordination function (DCF). DCF uses a contention algorithm to provide
access to all traffic. Ordinary asynchronous traffic directly uses DCF. The point coordination
function (PCF) is a centralized MAC algorithm used to provide contention-free service. PCF is
built on top of DCF and exploits features of DCF to assure access for its users. Let us consider
these two sublayers in turn.
Contention-free
service

Contention Service

PCF
DCF
Fig. 4-11 IEEE802.11 protocol architecture
1. Distributed Coordination Function
The DCF sublayer makes use of a simple CSMA algorithm. If a station has a MAC frame to
35

transmit, it listens to the medium. If the medium is idle, the station may transmit; otherwise, the
station must wait until the current transmission is complete before transmitting. The DCF does not
include a collision-detection function (i.e., CSMA/CD) because collision detection is not practical
on a wireless network. The dynamic range of the signals on the medium is very large, so that a
transmitting station cannot effectively distinguish incoming weak signals from noise and the
effects of its own transmission.
To ensure the smooth and fair functioning of this algorithm, DCF includes a set of delays that
amounts to a priority scheme. Let us start by considering a single delay known as an inter-frame
space (IFS). In fact, there are three different IFS values, but the algorithm is best explained by
initially ignoring this detail. Using an IFS, the rules for CSMA access are as follows:
(1) A station with a frame to transmit senses the medium. If the medium is idle, the station waits
to see if the medium remains idle for a time equal to IFS, and, if this is so, the station may
immediately transmit.
(2) If the medium is busy (either because the station initially finds the medium busy or because
the medium becomes busy during the IFS idle time), the sta tion defers transmission and continues
to monitor the medium until the cur rent transmission is over.
(3) Once the current transmission is over, the station delays another IFS. If the medium remains
idle for this period, then the station backs off using a binary exponential backoff scheme and again
senses the medium. If the medium is still idle, the station may transmit.
As with Ethernet, the binary exponential backoff provides a means of handling a heavy load.
If a station attempts to transmit and finds the medium busy, it backs off a certain amount and tries
again. Repeated failed attempts to transmit result in longer and longer backoff times.
The above scheme is refined for DCF to provide priority-based access by the simple
expedient of using three values for IFS:
SIFS (short IFS). The shortest IFS, used for all immediate response actions, as
explained below.
PIFS (point coordination function IFS). A mid-length IFS, used by the centralized
controller in the PCF scheme when issuing polls.
DIFS (distributed coordination function IFS), The longest IFS, used as a minimum
delay for asynchronous frames contending for access.
Figure 4-12 illustrates the use of these time values. Consider first the SIFS. Any station using
SIFS to determine transmission opportunity has, in effect, the highest priority, because it will
always gain access in preference to a station waiting an amount of time equal to PIFS or DIFS.

36

DATA

RTS

Source
DIFS
Destination

CTS

ACK
DIFS

others

NAV RTS

NAV CTS

Delay access

Fig.4-12 IEEE 802.11 MAC timing

The SIFS is used in the following circumstances:
Acknowledgment (ACK). When a station receives a frame addressed only to
itself
(not multicast or broadcast) it responds with an ACK frame after waiting only for an
SIFS gap; this has two desirable effects. First, because collision detection is not used,
the likelihood of collisions is greater than with CSMA/CD, and the MAC-level ACK
provides for efficient collision recovery. Second, the SIFS can be used to provide
efficient delivery of an LLC protocol data unit (PDU) that requires multiple MAC
frames. In this case, the following scenario occurs. A station with a multiframe LLC
PDU to transmit sends out the MAC frames one at a time. Each frame is acknowledged
by the recipient after SIFS. When the source receives an ACK, it immediately (after
SIFS) sends the next frame in the sequence. The result is that once a station has
contended for the channel, it will maintain control of the channel until it has sent all of
the fragments of an LLC PDU.
Clear to Send (CTS). A station can ensure that its data frame will get through by first
issuing a small Request to Send (RTS) frame. The station to which this frame is
addressed should immediately respond with a CTS frame if it is ready to receive. All
other stations receive the RTS and defer using the medium until they see a
corresponding CTS, or until a timeout occurs.
Poll response. This is explained in the discussion of PCF, below. The next longest IFS
interval is the PIFS; this is used by the centralized controller in issuing polls and takes
precedence over normal-contention traffic. However, those frames transmitted using
SIFS have precedence over a PCF poll. Finally, the DIFS interval is used for all ordinary
asynchronous traffic.
2. Point Coordination Function
PCF is an alternative access method implemented on top of the DCF. The operation consists
of polling with the centralized polling master (point coordinator). The point coordinator makes use
of PIFS when issuing polls. Because PIFS is smaller than DIFS, the point coordinator can seize
the medium and lock out all asynchronous traffic while it issues-polls and receives responses.
As an extreme, consider the following possible scenario. A wireless network is configured so
that a number of stations with time-sensitive traffic are controlled by the point coordinator while
37

remaining traffic, using CSMA, contends for access. The point coordinator could issue polls in a
round-robin fashion to all stations configured for polling. When a poll is issued, the polled station
may respond using SIFS. If the point coordinator receives a response, it issues another poll using
PIFS. If no response is received during the expected turnaround time, the coordinator issues a poll.
If the discipline of the preceding paragraph were implemented, the point coordinator would
lock out all asynchronous traffic by repeatedly issuing polls. To prevent this situation, an interval
known as the super-frame is defined. During the first part of this interval, the point coordinator
issues polls in a round-robin fashion to all stations configured for polling. The point coordinator
then idles for the remainder of the super-frame, allowing a contention period for asynchronous
access.
Figure 4-12 illustrates the use of the super-frame. At the beginning of a super-frame, the point
coordinator may optionally seize control and issue polls for a give period of time. This interval
varies because of the variable frame size issued by responding stations. The remainder of the
super-frame is available for contention-based access. At the end of the super-frame interval, the
point coordinator contends for access to the medium using PIPS. If the medium is idle, the point
coordinator gains immediate access, and a full super-frame period follows. However, the medium
may be busy at the end of a super-frame. In this case, the point coordinator must wait until the
medium is idle to gain access; this results in a foreshortened super-frame period for the next cycle.

INTERNET AND INTRANET

The Internet can be viewed as a collection of subnetworks or Autonomous Systems(ASes)

that are connected together. There are no real structure, but several major backbones exist. These
are constructed high-bandwidth and fast routers. Attached to the backbone are regional (midlevel)
networks , attached to the these regional networks are the LANas at many enterprises, companies
and universities, and Internet Service Providers, A sketch of this quasihierachical organization is
given in Fig. 5-1.
AS1
AS2
leased trunk
Backbone layer

Distributed layer

Access layer

38

LAN

Fig.5-1 The Internet is an interconnected collection of many Networks

The glue that holds the Internet together is the Network layer protocol, IP ( Internet Protocol )
. It was designed from the beginning with the internetworking and intercommunication in mind.
Its job is to provide a best-efforts way to transport datagrams from source to destination, without
regard to whether or not these machines are on the same network, or whether or not there are other
networks in between them.
Communication in the Internet works as follows. The transport layer takes data streams and
breaks them up into datagrams. In theory, datagrams can be up to 64 Kbytes each, but in practice
they are usually around 1500 bytes. Each datagram is transmitted through the Internet, possibly
being fragmented into smaller units as it goes. When all the pieces finally get to the destination
machine, they are reassembled by the network layer into the original datagram. This datagram is
then handed to the transport layer, which inserts it into the receiving process' input stream.
5..1. The IP Protocol
An appropriate place to start our study of the network layer in the Internet is the format of the IP
datagrams themselves. An IP datagram consists of a header part and a text part. The header has a
20-byte fixed part and a variable length optional part. The header format is shown in Fig. 5-2. It is
transmitted in big endian order: from left to right, with the high-order bit of the Version field going
first. (The SPARC is big endian; the Pentium is little endian.) On little endian machines, software
conversion is required on both transmission and reception.
Version field
The Version field keeps track of which version of the protocol the datagram belongs to. By
including the version in each datagram, it becomes possible to have the transition between
versions take months, or even years, with some machines running the old version and others
running the new one.
header length
Since the header length is not constant, a field in the header, IHL, is provided to tell how
long the header is, in 32-bit words. The minimum value is 5, which applies when no options are
present. The maximum value of this 4-bit field is 15, which limits the header to 60 bytes, and thus
the options field to 40 bytes. For some options, such as one that records the route a packet has
taken, 40 bytes is far too small, making the option useless.

39

3 4

Version

7 8 15 16
IHL

31 bits
ToS

Total length

Identification
Time To Live

DF MF
Protocol

Fragment offset
Header checksum

Source address
Destination address
Option

Pad
Data

Fig. 5-2. The IP (Internet Protocol) header

Type of service field
The Type of service field allows the host to tell the subnet what kind of service it wants.
Various combinations of reliability and speed are possible. For digitized voice, fast delivery beats
accurate delivery. For file transfer, error-free transmission is more important than fast
transmission.
The field itself contains (from left to right), a three-bit Precedence field, three
flags, D, T, and R, and 2 unused bits. The Precedence field is a priority, from 0 (normal) to 7
(network control packet). The three flag bits allow the host to specify what it cares most about
from the set {Delay, Throughput, Reliability}. In theory, these fields allow routers to make choices
between, for example, a satellite link with high throughput and high delay or a leased line with
low throughput and low delay. In practice, current routers ignore the Type of Service field
altogether.
Total length
The Total length includes everything in the datagram--both header and data. The maximum
length is 65,535 bytes. At present, this upper limit is tolerable, but with future gigabit networks
larger datagrams will be needed.
Identification field
The Identification field is needed to allow the destination host to determine which datagram
a newly arrived fragment belongs to. All the fragments of a datagram contain the same
Identification value.
DF /MF- Fragment offset
DF stands for Don't Fragment. It is an order to the routers not to fragment the datagram
because the destination is incapable of putting the pieces back together again. For example, when
a computer boots, its ROM might ask for a memory image to be sent to it as a single datagram. By
marking the datagram with the DF bit, the sender knows it will arrive in one piece, even if this
means that the datagram must avoid a small packet network on the best path and take a suboptimal
route. All machines are required to accept fragments of 576 bytes or less.
MF stands for More Fragments. All fragments except the last one have this bit set. It is
needed to know when all fragments of a datagram have arrived.

40

The Fragment offset tells where in the current datagram this fragment belongs. All fragments
except the last one in a datagram must be a multiple of 8 bytes, the elementary fragment unit.
Since 13 bits are provided, there is a maximum of 8192 fragments per datagram, giving a
maximum datagram length of 65,536 bytes, one more than the Total length field.
Time to live field
The Time to live field is a counter used to limit packet lifetimes. It is supposed to count time
in seconds, allowing a maximum lifetime of 255 sec. It must be decremented on each hop and is
supposed to be decremented multiple times when queued for a long time in a router. In practice, it
just counts hops. When it hits zero, the packet is discarded and a warning packet is sent back to the
source host. This feature prevents datagrams for wandering around forever, something that
otherwise might happen if the routing tables ever become corrupted.
Protocol field
When the network layer has assembled a complete datagram, it needs to know what to do
with it. The Protocol field tells it which transport process to give it to. TCP is one possibility, but
so are UDP and some others. The numbering of protocols is global across the entire Internet and is
defined in RFC 1700.
Header checksum
The Header checksum verifies the header only. Such a checksum is useful for detecting
errors generated by bad memory words inside a router. The algorithm is to add up all the 16-bit
halfwords as they arrive, using one's complement arithmetic and then take the one's complement
of the result. For purposes of this algorithm, the Header checksum is assumed to be zero upon
arrival. This algorithm is more robust than using a normal add. Note that the Header checksum
must be recomputed at each hop, because at least one field always changes (the Time to live field),
but tricks can be used to speed up the computation.
The Source address and Destination address indicate the network number and host number.
We will discuss Internet addresses in the next section.
The Options field was designed to provide an escape to allow subsequent versions of the protocol
to include information not present in the original design, to permit experimenters to try out new
ideas, and to avoid allocating header bits to information that is rarely needed. The options are
variable length. Each begins with a 1-byte code identifying the option. Some options are followed
by a 1-byte option length field, and then one or more data bytes. The Options field is padded out to
a multiple of four bytes.
The Security option tells how secret the information is. In theory, a military router might use
this field to specify not to route through certain countries the military considers to be "bad guys."
In practice, all routers ignore it, so its only practical function is to help spies find the good stuff
more easily.
The Strict source routing option gives the complete path from source to destination as a
sequence of IP addresses. The datagram is required to follow that exact route. It is most useful for
system managers to send emergency packets when the routing tables are corrupted, or for making
timing measurements.
The Loose source routing option requires the packet to traverse the list of routers specified,
and in the order specified, but it is allowed to pass through other routers on the way. Normally, this
option would only provide a few routers, to force a particular path. For example, to force a packet
from London to Sydney to go west instead of east, this option might specify routers in New York,
41

Los Angeles, and Honolulu. This option is most useful when political or economic
considerations dictate passing through or avoiding certain countries.
The Record route option tells the routers along the path to append their IP address to the
option field. This allows system managers to track down bugs in the routing algorithms ("Why are
packets from Houston to Dallas all visiting Tokyo first?") When the ARPANET was first set up, no
packet ever passed through more than nine routers, so 40 bytes of option was ample. As mentioned
above, now it is too small.
The Timestamp option is like the Record route option, except that in addition to recording
its 32-bit IP address, each router also records a 32-bit time- stamp. This option, too, is mostly for
debugging routing algorithms.
5.2 IP Addresses and Subnet
5.2.1 IP Addresses
Every host and router on the Internet has an IP address, which encodes its network number
and host number. The combination is unique: no two machines have the same IP address. All IP
addresses are 32 bits long and are used in the Source address and Destination address fields of IP
packets. The formats used for IP address are shown in Fig. 5-3. Those machines connected to
multiple networks have a different IP address on each network.
Byte 1
0

available left N-ID

N-ID

H-ID

2
10

1110

11110

1126

2
N-ID

H-ID
3

110

128191

192223

224239

N-ID

H-ID

multicast

Reserved for future use

Fig. 5-3 IP address firmats

The class A, B, C, and D formats allow for up to 126 networks with 16 million hosts each, ,
16,382 networks with up to 64K hosts, 2 million networks, (e.g., LANs), with up to 254 hosts
each, and multicast, in which a datagram is directed to multiple hosts Addresses beginning with
11110 are reserved for future use. Tens of thousands of networks are now connected to the
Internet, and the number doubles every year. Network numbers are assigned by the NIC (Network
Information Center) to avoid conflicts.
Network addresses, which are 32-bit numbers, are usually, written in dotted decimal notation.
In this format, each of the 4 bytes is written in decimal, from 0 to 255. For example, the
hexadecimal address C0290614 is written as 192.41.6.20. The lowest IP address is 0.0.0.0 and the
highest is 255.255.255.255.
The values 0 and -1 have special meanings. The value 0 means this network or this host. The
value of-1 is Used as a broadcast address to mean all hosts on the indicated network.
42

The IP address 0.0.0.0 is used by hosts when they are being booted but is not used afterward. IP
addresses with 0 as network number refer to the current network. These addresses allow machines
to refer to their own network without knowing its number (but they have to know its class to know
how many Os to include). The address consisting of all ls allows broadcasting on the local
network, typically a LAN. The addresses with a proper network number and all 1 s in
the host field allow machines to send broadcast packets to distant LANs anywhere in the Internet.
Finally, all addresses of the form 127.xx.yy.zz are reserved for loopback testing. Packets sent to
that address are not put out onto the wire; they are processed locally and treated as incoming
packets. This allows packets to be sent to the local network without the sender knowing its
number. This feature is also used for debugging network software.
5.2.2. Subnets
As we have seen, all the hosts in a network must have the same network number. This property
of IP addressing can cause problems as networks grow. For example, consider a company that
starts out with one class C LAN on the Internet. As time goes on, it might acquire more than 254
machines, and thus need a second class C address. Alternatively, it might acquire a second LAN of
a different type and want a separate IP address for it (the LANs could be bridged to form a single
IP network, but bridges have their own problems). Eventually, it might end up with many LANs,
each with its own router and each with its own class C network number.
As the number of distinct local networks grows, managing them can become a serious
headache. Every time a new network is installed the system administrator has to contact NIC to
get a new network number. Then this number must be announced worldwide. Furthermore,
moving a machine from one LAN to another requires it to change its IP address, which in turn
may mean modifying its configuration files and also announcing the new IP address to the world.
If some other machine is given the newly-released IP address, that machine will get email and
other data intended for the original machine until the address has propagated all over the world.
The solution to these problems is to allow a network to be split into several parts for internal Use
but still act like a single network to the outside world. In the Internet literature, these parts are
called subnets.
In this section, the new definition will be the one used. If our growing company started up
with a class B address instead of a class C address, it could start out just numbering the hosts from
1 to 254. When the second LAN arrived, it could decide, for example, to split the 16-bit host
number into a 6-bit subnet number and a 10-bit host number, as shown in Fig. 5-4. This split
allows 62 LANs (0 and -1 are reserved), each with up to 1022 hosts.
Outside the network, the subnetting is not visible, so allocating a new subnet does not require
contacting NIC or changing any external databases. In this example, the first subnet might use IP
addresses starting at 130.50.4.1, the second subnet might start at 130.50.8.1, and so on.
To see how subnets work, it is necessary to explain how IP packets are processed at a router.
Each router has a table listing some number of (network, 0) IP addresses and some number of
(this-network, host) IP addresses. The first kind tells how to get to distant networks. The second
kind tells how to get to local hosts. Associated with each table is the network interface to use to
reach the destination, and certain other information.

43

internetwork

Network

Local

Subnet

Host

Subnet mask
all 1
Subnet
mask

10

Network

11111111 11111111

all 0
Subnet

Host

111111 0000000000

Fig.5-4 One of the way to subnet a class B network

When an IP packet arrives, its destination address is looked up in the routing table. If the
packet is for a distant network, it is forwarded to the next router on the interface given in the table.
If it is a local host (e.g., on the router's LAN), it is sent directly to the destination. If the network is
not present, the packet is forwarded to a default router with more extensive tables. This algorithm
means that each router only has to keep track of other networks and local hosts, not (network,
host) pairs, greatly reducing the size of the routing table.
When subnetting is introduced, the routing tables are changed, adding entries of the form
(this-network, subnet, 0) and (this-network, this-subnet, host). Thus a router on subnet k knows
how to get to all the other subnets and also how to get to all the hosts on subnet k. It does not have
to know the details about hosts on other subnets. In fact, all that needs to be changed is to have
each router do a Boolean AND with the network's subnet mask to get rid of the host number and
look up the resulting address in its tables (after determining which network class it is). For
example, a packet addressed to 130.50.15.6 and arriving at a router on subnet 5 is ANDed with the
subnet mask of Fig. 5-5 to give the address 130.50.12.0. This address is looked up in the routing
tables to find out how to get to hosts on subnet 3. The router on subnet 5 is thus spared the work of
keeping track of the data link addresses of hosts other than those on subnet 5. Subnetting thus
reduces router table space by creating a three-level hierarchy.
5.3 Internet Control Protocols
In addition to IP, which is used for data transfer, the Internet has several control protocols
used in the network layer, including ICMP, ARP, RARP and so on.
5.3.1 Internet Control Message Protocol
The operation of the Internet is monitored closely by the routers. When something unexpected
occurs, the event is reported by the ICMP (Internet Control Message Protocol), which is also used
to test the Internet. About a dozen types 0f ICMP messages are defined. Each ICMP message type
is encapsulated in an IP packet.
The DESTINATION UNREACHABLE message is used when the subnet or a router
cannot locate the destination, or a packet with the DF bit cannot be delivered because a smallpacket network stands in the way.
The TIME EXCEEDED message is sent when a packet is dropped due to its counter
reaching zero. This event is a symptom that packets are looping, that there is enormous
congestion, or that the timer values are being set too low.
44

The PARAMETER PROBLEM message indicates that an illegal value has been detected in
a header field. This problem indicates a bug in the sending host's IP software, or possibly in the
software of a router transited.
The SOURCE QUENCH message was formerly used to throttle hosts that were sending too
many packets. When a host received this message, it was expected to slow down. It is rarely used
any more because when congestion occurs, these packets tend to add more fuel to the fire.
Congestion control in the Internet is now done largely in the transport layer.
The REDIRECT message is used when a router notices that a packet seems to be routed
wrong. It is used by the router to tell the sending host about the probable error.
The ECHO REQUEST and ECHO REPLY messages are used to see if a given destination
is reachable and alive. Upon receiving the ECHO message, the destination is expected to send an
ECHO REPLY message back.
The TIMESTAMP REQUEST and TIMESTAMP REPLY messages are similar, except
that the arrival time of the message and the departure time of the reply are recorded in the reply.
This facility is used to measure network performance.
In addition to these messages, there are four others that deal with Internet addressing, to allow
hosts to discover their network numbers and to handle the case of multiple LANs sharing a single
IP address. ICMP is defined in RFC 792.
5.3.2 The Address Resolution Protocol
Although every machine on the Internet has one (or more) LP addresses, these cannot actually
be used for sending packets because the data link layer hardware does not understand Internet
addresses. Nowadays, most hosts are attached to a LAN by an interface board that only
understands LAN addresses. For example, every Ethernet board ever manufactured comes
equipped with a 48-bit Ethernet MAC address. Manufacturers of Ethernet boards request a block
of addresses from a central authority to ensure that no two boards have the same address (to avoid
conflicts should the two boards ever appear on the same LAN). The boards send and receive
frames based on 48-bit Ethernet addresses. They know nothing at all about 32-bit IP addresses.
The question now arises: How do IP addresses get mapped onto data link layer addresses,
such as Ethernet? To explain how this works, let us use the example of Fig. 5-5, in which a small
university with several class C networks is illustrated. Here we have two Ethernets, one in the
Computer Science and Technology department, with IP address 202.119.65.0 and one in Electrical
Engineering department, with IP address 202.119.63.0. These are connected by a campus FDDI
ring with IP address 202.119.60.0. Each machine on an Ethernet has a unique Ethernet address,
labeled El through E6, and each machine on the FDDI ring has an FDDI address, labeled Fl
through F3.
Let us start out by seeing how a user on host 1 sends a packet to a user on host 2. Let us assume
the sender knows the name of the intended receiver, possibly something like y00416@
cst.njupt.edu.cn. The first step is to find the IP address for host 2, known as cst.njupt.edu.cn. This
lookup is performed by the Domain Name System. For the moment, we will just assume that DNS
returns the IP address for host 2 (202.119.65.5).
The upper layer software on host 1 now builds a packet with 202.119.65.5 in the Destination
address field and gives it to the IP software to transmit. The IP software can look at the address
and see that the destination is on its own network, but it needs a way to find the destination's
45

Ethernet address. One solution is to have a configuration file somewhere in the system that maps
IP addresses onto Ethernet addresses. This solution is certainly possible, but for organizations with
thousands of machines, keeping these files up to date is an error-prone, time-consuming job.
202.119.60.4
202.119.65.1
LAN

202.119.60.7
202.119.63.3
F1 FDDI F3
E4 LAN

E3
Router 1

E1
E2
202.119.65.7 202.119.65.5
CST Ethernet
202.119.65.0

Router 2
F2
202.119.60.0

5
E5

E6
202.119.63.8

EE Ethernet
202.119.63.0

Fig. 5-5. Three interconnected class C networks: two Ethernets and an FDDI ring
A better solution is for host I to output a broadcast packet onto the Ethernet asking: "Who
owns IP address 202.119.65.5?" The broadcast will arrive at every machine on Ethernet
202.119.65.0, and each one will check its IP address. Host 2 alone will respond with its Ethernet
address (E2). In this way host 1 learns that IP address 202.119.65.5 is on the host with Ethernet
address E2. The protocol for asking this question and getting the reply is called ARP (Address
Resolution Protocol). Almost every machine on the Internet runs it. It is defined in RFC 826.
At this point, the IP software on host 1 builds an Ethernet frame addressed to E2, puts the IP
packet (addressed to 202.119.65.5) in the payload field, and dumps it onto the Ethernet. The
Ethernet board of host 2 detects this frame, recognizes it as a frame for itself, scoops it up, and
causes an interrupt. The Ethernet driver extracts the IP packet from the payload and passes it to the
IP software, which sees that it is correctly addressed, and processes it.
Various optimizations are possible to make ARP more efficient. To start with, once a machine
has run ARP, it caches the result in case it needs to contact the same machine shortly. Next time it
will find the mapping in its own cache, thus eliminating the need for a second broadcast. In many
cases host 2 will need to send back a reply, forcing it, too, to run ARP to determine the sender's
Ethernet address. This ARP broadcast can be avoided by having host 1 include its IP to Ethernet
mapping in the ARP packet. When ARP broadcast arrives at host 2, the pair (202.119.65.7, El) is
entered into host 2's ARP cache for future use. In fact, all machines on the Ethernet can enter this
mapping into their ARP caches.
Yet another optimization is to have every machine broadcast its mapping when it boots. This
broadcast is generally done in the form of an ARP looking for its own IP address. There should not
be a response, but a side effect of the broadcast is to make any entry in everyone's ARP cache. If a
response does arrive, two machines have been assigned the same IP address. The new one should
inform the system manager and not boot.
To allow mappings to change, for example, when an Ethernet board breaks and is replaced
with a new one (and thus a new Ethernet address), entries in the ARP cache should time out after a
46

few minutes.
Now let us look at Fig. 5-5 again, only this time host 1 wants to send a packet to host 6
(202.119.63.8). Using ARP will fail because host 4 will not see the broadcast (routers do not
forward Ethernet-level broadcasts). There are two solutions. First, the CST router could be
configured to respond to ARP requests for network 202.119.63.0 (and possibly other local
networks). In this case, host 1 will make an ARP cache entry of (202.119.63.8, E3) and happily
send all traffic for host 4 to the local router. This solution is called proxy APP. The second
solution is to have host 1 immediately see that the destination is on a remote network and just rend
all such traffic to a default Ethernet address that handles all remote traffic, in this case E3. This
solution does not require having the CST router know which remote networks it is serving.
Either way, what happens is that host 1 packs the IP packet into the payload field of an
Ethernet frame addressed to E3. When the CST router gets the Ethernet frame, it removes the IP
packet from the payload field and looks up the IP address in its routing tables. It discovers that
packets for network 202.119.63.0 are supposed to go to router 202.119.60.7. If it does not already
know the FDDI address of 202.119.60.7, it broadcasts an ARP packet onto the ring and learns that
its ring address is F3. It then inserts the packet into the payload field of an FDDI frame addressed
to F3 and puts it on the ring.
At the EE router, the FDDI driver removes the packet from the payload field and gives it to
the IP software, which sees that it needs to send the packet to 202.119.63.8. If this IP address is not
in its ARP cache, it broadcasts an ARP request on the EE Ethernet and learns that the destination
address is E6 so it builds an Ethernet frame addressed to E6, puts the packet in the payload field,
and sends it over the Ethernet. When the Ethernet frame arrives at host 4, the packet is extracted
from the frame and passed to the IP software for processing.
Going from host 1 to a distant network over a WAN works essentially the same way, except
that this time the CST router's tables tell it to use the WAN whose FDDI address is F2.
5.3.3 The Reverse Address Resolution Protocol
ARP solves the problem of finding out which Ethernet address corresponds to given IP
address., Sometimes the reverse problem has to solved: Given an Ethernet address, what is the
corresponding IP address? In particular, this problem occurs when booting a diskless workstation.
Such a machine will normally get the binary image of its operating system from a remote file
server.
5.3.4, The Interior Gateway Routing Protocol: OSPF
As we mentioned earlier, the Internet is made up of a large number of Autonomous Systems.
Each AS is operated by a different organization and can use its own routing algorithm inside. For
example, the internal networks of companies X, Y, and Z would usually be seen as three ASes if
all three were on the Internet. All three may use different routing algorithms internally.
Nevertheless, having standards, even for internal routing, simplifies the implementation at the
boundaries between ASes and allows reuse of code. In this section we will study routing within an
AS. In the next one, we will look at routing between ASes. A routing algorithm within an AS is
called an interior gateway protocol; an algorithm for routing between ASes is called an
EGP(exterior gateway protocol).
The original Internet interior gateway protocol was a distance vector protocol (RIP,Routing
47

Information Protocol) based on the Bellman-Ford algorithm. It worked well in small systems,
but less well as ASes got larger. It also suffered from the count-to-infinity problem and generally
slow convergence, so it was replaced in May 1979 by a link state protocol. In 1988, the
IETF( Internet Engineering Task Force )began Work on a successor. That successor, called
OSPF (Open Shortest Path First) became a standard in 1990. Many router vendors are now
supporting it, and it will become the main interior gateway protocol in the near future. Below we
will give a sketch of how OSPF works. For the complete story, see RFC 1247.
Given the long experience with other routing protocols, the group designing the new protocol
had a long list of requirements that had to be met.
(1) The algorithm had to be published in the open literature, hence the "O" in OSPF. A
proprietary solution owned by one company would not do.
(2) The new protocol had to support a variety of distance metrics, including physical
distance, delay, and so on.
(3) Iit had to be a dynamic algorithm, one that adapted to changes in the topology
automatically and quickly.
(4) New for OSPF, it had to support routing based on type of service. The new protocol had to
be able to route real-time traffic one way and other traffic a different way. The IP protocol has a
Type of Service field, but no existing routing protocol used it.
(5) Related to the above, the new protocol had to do load balancing, splitting the load over
multiple lines. Most previous protocols sent all packets over the best route. The second-best route
was not used at all. In many cases, splitting the load over multiple lines gives better performance.
(6) Support for hierarchical systems was needed. By 1988, the Internet had grown so large
that no router could be expected to know the entire topology. The new routing protocol had to be
designed so that no router would have to.
(7) Some modicum of security was required to prevent fun-loving students from spoofing
routers by sending them false routing information.
(8) Provision was needed for dealing with routers that were connected to the Internet via a
tunnel. Previous protocols did not handle this well.
OSPF supports three kinds of connections and networks:
1. Point-to-point lines between exactly two routers.
2. Multiaccess networks with broadcasting (e.g., most LANs).
3. Multiaccess networks without broadcasting (e.g., most packet switched WANs).
A multiaceess network is one that can have multiple routers on it, each of which can directly
communicate with all the others. All LANs and WANs have this property. Figure 5-6 shows an AS
containing all three kinds of networks. Note that hosts do not generally play a role in OSPF.
OSPF works by abstracting the collection of actual networks, routers, and lines into a directed
graph in which each arc is assigned a cost (distance, delay, etc.). It then computes the shortest path
based on the weights on the arcs. A serial connection between two routers is represented by a pair
of arcs, one in each direction. Their weights may be different. A multiaccess network is
represented by a node for the network itself plus a node for each router. The arcs from the network
node to the routers have weight 0 and are omitted from the graph.
What OSPF fundamentally does is represent the actual network as a graph like this and then
compute the shortest path from every router to every other router.

48

Many of the ASes in the Internet are themselves large and nontrivial to manage. OSPF allows
them to be divided up into numbered areas, where an area is a network or a set of contiguous
networks. Areas do not overlap but need not be exhaustive, that is, some routers may belong to no
area. An area is a generalization of a subnet. Outside an area, its topology and details are not
visible.
WAN 1
LAN 1
C
A

WAN 2
H

WAN 3

LAN 2

Fig. 5-6. An autonomous system

Every AS has a backbone area, called area 0. All areas are connected to the backbone, possibly by
tunnels, so it is possible to go from any area in the AS to any other area in the AS via the
backbone. A tunnel is represented in the graph as an arc and has a cost. Each router that is
connected to two or more areas is part of the backbone; As with other areas, the topology of the
backbone is not visible outside the backbone.
Within an area, each router has the same link state database and runs the same shortest path
algorithm. Its main job is to calculate the shortest path from itself to every other router in the area,
including the router that is connected to the backbone, of which there must be at least one. A
router that connects to two areas needs the databases for both areas and must run the shortest path
algorithm for each one separately.
The way OSPF handles type of service routing is to have multiple graphs, one labeled with
the costs when delay is the metric, one labeled with the costs when throughput is the metric, and
one labeled with the costs when reliability is the metric. Although this triples the computation
needed, it allows separate routes for optimizing delay, throughput, and reliability.
During normal operation, three kinds of routes may be needed: intra-area, inter-area, and
interAS. Intra-area routes are the easiest, since the source router already knows the shortest path to
the destination router. Inter-area routing always proceeds in three steps: go from the source to the
backbone; go across the backbone to the destination area; go to the destination. This algorithm
forces a star configuration on OSPF with the backbone being the hub and the other areas being
spokes. Packets are routed from source to destination "as is." They are not encapsulated or
tunneled, unless going to an area whose only connection to the backbone is a tunnel.
When a router boots, it sends HELLO messages on all of its point-to-point lines and multicasts
them on LANs to the group consisting of all the other routers. On WANs, it needs some
configuration information to know who to contact. From the responses, each router learns who its
neighbors are.
OSPF works by exchanging information between adjacent routers, which is not the same as
between neighboring routers. In particular, it is inefficient to have every router on a LAN talk to
every other router on the LAN. To avoid this situation, one router is elected as the designated
49

router. It is said to be adjacent to all the other routers, and exchanges information with them.
Neighboring routers that are not adjacent do not exchange information with each other. A backup
designated router is always kept up to date to ease the transition should the primary designated
router crash.
During normal operation, each router periodically floods LINK STATE UPDATE messages to
each of its adjacent routers. This message gives its state and provides the costs used in the
topological database. The flooding messages are acknowledged (LINK STATE ACK), to make
them reliable. Each message has a sequence number, so a router can see whether an incoming
LINK STATE UPDATE is older or newer than what it currently has. Routers also send these
messages when a line goes up or down or its cost changes.
DATABASE DESCRIPTION messages give the sequence numbers of all the link state
entries currently held by the sender. By comparing its own values with those of the sender, the
receiver can determine who has the most recent values. These messages are used when a line is
brought up.
Either partner can request link state information from the other one using LINK STATE
REQUEST messages. The net result of this algorithm is that each pair of adjacent routers checks
to see who has the most recent data, and new information is spread throughout the area this way.
All these messages are sent as raw IP packets.
Finally, we can put all the pieces together. Using flooding, each router informs all the other
routers in its area of its neighbors and costs. This information allows each router to construct the
graph for its area(s) and compute the shortest path. The backbone area does this too. In addition,
the backbone routers accept information from the area border routers in order to compute the best
route from each backbone router to every other router. This information is propagated back to the
area border routers, which advertise it within their areas. Using this information, a router about to
send an inter-area packet can select the best exit router to the backbone.
5.3.5 The Exterior Gateway Routing Protocol: BGP
Within a single AS, the recommended routing protocol on the Internet is OSPF (although it is
certainly not the only one in use). Between ASes, a different protocol, BGP (Border Gateway
Protocol), is used. A different protocol is needed between ASes because the goals of an interior
gateway protocol and an exterior gateway protocol are not the same. All an interior gateway
protocol has to do is move packets as efficiently as possible from the source to the destination. It
does not have to worry about politics.
Exterior gateway protocol routers have to worry about politics a great deal. For example, a
corporate AS might want the ability to send packets to any Internet site and receive packets from
any Internet site. However, it might be unwilling to carry transit packets originating in a foreign
AS and ending in a different foreign AS, even if its own AS was on the shortest path between the
two foreign Ases ("That's their problem, not ours"). On the other hand, it might be willing to carry
transit traffic for its neighbors, or even for specific other ASes that paid it for this service.
Telephone companies, for example, might be happy to act as a carrier for their customers, but not
for others. Exterior gateway protocols in general, and BGP in particular, have been designed to
allow many kinds of routing policies to be enforced in the inter-AS traffic.
Typical policies involve political, security, or economic considerations. A few examples of
routing constraints are:
50

1. No transit traffic through certain ASes.

2. Never put Iraq on a route starting at the Pentagon.
3. Do not use the United States to get from British Columbia to Ontario.
4. Only transit Albania if there is no alternative to the destination.
5. Traffic starting or ending at IBM should not transit Microsoft.
Policies are manually configured into each BGP router. They are not part of the protocol
itself.
From the point of view of a BGP router, the world consists of other BGP routers and the lines
connecting them. Two BGP routers are considered connected if they share a common network.
Given BGP's special interest in transit traffic, networks are grouped into one of three categories.
The first category is the stub networks, which have only one connection to the BGP graph. These
cannot be used for transit traffic because there is no one on the other side. Then come the
multiconneeted networks. These could be used for transit traffic, except that they refuse. Finally,
there are the transit networks, such as backbones, which are willing to handle third-party packets,
possibly with some restrictions.
Pairs of BGP routers communicate with each other by establishing TCP connections. Operating
this way provides reliable communication and hides all the details of the network being passed
through.
BGP is fundamentally a distance vector protocol, but quite different from most others such as
RIP. Instead of maintaining just the cost to each destination, each BGP router keeps track of the
exact path used. Similarly, instead of periodically giving each neighbor its estimated cost to each
possible destination, each BGP router tells its neighbors the exact path it is using.
The current definition of BGP is in RFC 1654. Additional useful information can be found in
RFC 1268.
5.3.6 Internet Multicasting
Normal IP communication is between one sender and one receiver. However, for some
applications it is useful for a process to be able to send to a large number of receivers
simultaneously. Examples are updating replicated, distributed databases, transmitting stock quotes
to multiple brokers, and handling digital conference (i.e., multiparty) telephone calls.
IP supports multicasting, using class D addresses. Each class D address identifies a group of
hosts. Twenty-eight bits are available for identifying groups, so over 250 million groups can exist
at the same time. When a process sends a packet to a class D address, a best-efforts attempt is
made to deliver it to all the members of the group addressed, but no guarantees are given. Some
members may not get the packet.
Two kinds of group addresses are supported: permanent addresses and temporary ones. A
permanent group is always there and does not have to be set up. Each permanent group has a
permanent group address. Some examples of permanent group addresses are
224.0.0.1 All systems on a LAN
224.0.0.2 All routers on a LAN
224.0.0.5 All OSPF routers on a LAN
224.0.0.6 All designated OSPF routers on a LAN
Temporary groups must be created before they can be used. A process can ask its host to join
a specific group. It can also ask its host to leave the group. When the last process on a host leaves
51

a group, that group is no longer present on the host. Each host keeps track of which groups its
processes currently belong to.
Multicasting is implemented by special multicast routers, which may or may not be colocated
with the standard routers. About once a minute, each multicast router sends a hardware (i.e., data
link layer) multicast to the hosts on its LAN (address 224.0.0.1) asking them to report back on the
groups their processes currently belong to. Each host sends back responses for all the class D
addresses it is interested in.
These query and response packets use a protocol called IGMP (Internet Group Management
Protocol), which is vaguely analogous to ICMP. It has only two kinds of packets: query and
response, each with a simple fixed format containing some control information in the first word of
the payload field and a class D address in the second word. It is described in RFC 1112.
Multicast routing is done using spanning trees. Each multicast router exchanges information
with its neighbors using a modified distance vector protocol in order for each one to construct a
spanning tree per group covering ail group members. Various optimizations are used to prune the
tree to eliminate routers and networks not interested in particular groups. The protocol makes
heavy use of tunneling to avoid bothering nodes not in a spanning tree.
At present there are many reliable multicast protocols such as MFTP (Multicast File Transfer
Protocol), PGM (Pretty Good Multicast), HRM (Hierarchical Reliable Multicast), LRMP (Lightweight Multicast Reliable Multicast Protocol), RRMP (Randomized Error Recovery Algorithm for
Reliable Multicast).
5.3.7. Mobile IP
Many users of the Internet have portable computers and want to stay connected to the
Internet when they visit a distant Internet site and even on the road in between. Unfortunately, the
IP addressing system makes working far from home easier said than done . In this section we will
examine the problem and the solution.
The real villain is the addressing scheme itself. Every IP address contains three fields: the
class, the network number, and the host number. For example, consider the machine with IP
address 160.80.40.20. The 160.80 gives the class (B) and network number (8272); the 40.20 is the
host number (10260). Routers ail over the world have routing tables telling which line to use to get
to network 160.80. Whenever a packet comes in with a destination IP address of the form
160.80.xxx.yyy, it goes out on that line.
If ail of a sudden, the machine with that address is carted off to some distant site, the packets
for it will continue to be routed to its home LAN (or router). The owner will no longer get email,
and so on. Giving the machine a new IP address corresponding to its new location is unattractive
because large numbers of people, programs, and databases would have to be informed of the
change.
Another approach is to have the routers use complete IP addresses for routing, instead of just
the class and network. However, this strategy would require each router to have millions of table
entries, at astronomical cost to the Internet. When people began demanding the ability to have
mobile hosts, the IETF set up a Working Group to find a solution. The Working Group quickly
formulated a number of goals considered desirable in any solution. The major ones were
1. Each mobile host must be able to use its home IP address anywhere.
2. Software changes to the fixed hosts were not permitted.
52

3. Changes to the router software and tables were not permitted.

4. Most packets for mobile hosts should not make detours on the way.
5. No overhead should be incurred when a mobile host is at home.
To review it briefly, every site that wants to allow its users to roam has to create a home
agent. Every site that wants to allow visitors has to create a foreign agent. When a mobile host
shows up at a foreign site, it contacts the foreign host there and registers. The foreign host then
contacts the user's home agent and gives it a care-of address, normally the foreign agent's own IP
address.
When a packet arrives at the user's home LAN, it comes in at some router attached to the
LAN. The router then tries to locate the host in the usual way, by broadcasting an ARP packet
asking, for example: "What is the Ethernet address of 160.80.40.20?" The home agent responds to
this query by giving its own Ethernet address. The router then sends packets for 160.80.40.20 to
the home agent. It, in turn, tunnels them to the care-of address by encapsulating them in the
payload field of an IP packet addressed to the foreign agent. The foreign agent then decapsulates
and delivers them to the data link address of the mobile host. In addition, the home agent gives the
care-of address to the sender, so future packets can be tunnelled directly to the foreign agent. This
solution meets all the requirements stated above.
One small detail is probably worth mentioning. At the time the mobile host moves, the ~outer
probably has its (soon-to-be-invalid) Ethernet address cached. To replace that Ethernet address
with the home agent's, a trick called gratuitious ARP is used. This is a special, unsolicited
message to the router that causes it to replace a specific cache entry, in this case, that of the mobile
host about to leave. When the mobile host returns later, the same trick is used to update the router's
cache again.
Nothing in the design prevents a mobile host from being its own foreign agent, but that
approach only works if the mobile host (in its capacity as foreign agent) is logically connected to
the Internet at its current site. Also, it must be able to acquire a (temporary) care-of IP address to
use. That IP address must belong to the LAN to which it is currently attached.
The IETF solution for mobile hosts solves a number of other problems not mentioned so far.
For example, how are agents located? The solution is for each agent to periodically broadcast its
address and the type of services it is willing to provide (e.g., home, foreign, or both). When a
mobile host arrives somewhere, it can just listen for these broadcasts, called advertisements.
Alternatively, it can broadcast a packet announcing its arrival and hope that the local foreign agent
responds to it.
Another problem that had to be solved is what to do about impolite mobile hosts that leave
without saying goodbye. The solution is to make registration valid only for a fixed time interval. If
it is not refreshed periodically, it times out, so the foreign host can clear its tables.
Yet another issue is security. When a home agent gets a message asking it to please forward all
of Nora's packets to some IP address, it had better not comply unless it is convinced that Nora is
the source of this request, and not somebody trying to impersonate her. Cryptographic
authentication protocols are used for this purpose.
A final point addressed by the Working Group relates to levels of mobility. Imagine an
airplane with an on-board Ethernet used by the navigation and avionics computers. On this
Ethernet is a standard router that talks to the wired Internet on the ground over a radio link. One
fine day, some clever marketing executive gets the idea to install Ethernet connectors in all the
53

arm rests so passengers with mobile computers can also plug in.
Now we have two levels of mobility: the aircraft's own computers, which are stationary with
respect to the Ethernet, and the passengers' computers, which are mobile with respect to it. In
addition, the on-board router is mobile with respect to routers on the ground. Being mobile with
respect to a system that is itself mobile can be handled using recursive tunneling.
5.3.8 CIDR--Classless Inter Domain Routing
IP has been in heavy use for over a decade. It has worked extremely well, as demonstrated by
the exponential growth of the Internet. Unfortunately, IP is rapidly becoming a victim of its own
popularity: it is running out of addresses. This looming disaster has sparked a great deal of
discussion land controversy within the Internet community about what to do about it. In this
section we will describe both the problem and several proposed solutions.
Back in 1987, a few visionaries predicted that some day the Internet might grow to 100,000
networks. Most experts pooh-poohed this as being decades in the future, if ever. The 100,000th
network was connected in 1996. The problem, simply stated, is that the Internet is rapidly running
out of IP addresses. In principle, over 2 billion addresses exist, but the practice of organizing the
address space by classes, wastes millions of them. In particular, the real villain is the class B
network. For most organizations, a class A network, with 16 million addresses is too big, and a
class C network, with 256 addresses is too small. A class B network, with 65,536, is just right. In
Internet folk{ore, this situation is known as the three bears problem (as in Goldilocks and the
Three Bears).
In reality, a class B address is far too large for most organizations. Studies have shown that
more than half of all class B networks have fewer than 50 hosts. A class C network would have
done the job, but no doubt every organization that asked for a class B address thought that one day
it would outgrow the 8-bit host field. In retrospect, it might have been better to have had class C
networks use 10 bits instead of eight for the host number, allowing 1022 hosts per network. Had
this been the case, most organizations would have probably settled for a class C network, and
there would have been half a million of them (versus only 16,384 class B networks).
However, then another problem would have emerged more quickly: the routing table
explosion. From the point of view of the routers, the IP address space is two-level hierarchy, with
network numbers and host numbers. Routers do not have to know about all the hosts, but they do
have to know about all the networks. If half a million class C networks were in use, every router in
the entire Internet would need a table with half a million entries, one per network, telling which
line to use to get to that network, as well as other information.
The actual physical storage of half a million entry tables is probably doable, although
expensive for critical routers that keep the tables in static RAM on I/O boards. A more serious
problem is that the complexity of various algorithms relating to management of the tables grows
faster than linear. Worse yet, much of the existing router software and firmware was designed at a
time when the Internet had 1000 connected networks and 10,000 networks seemed decades away.
Design choices made then often are far from optimal now.
In addition, various routing algorithms require each router to transmit its tables periodically.
The larger the tables, the more likely some parts will get lost underway, leading to incomplete data
at the other end and possibly routing instabilities.
The routing table problem could have been solved by going to a deeper hierarchy. For
54

example, having each IP address contain a country, state, city, network, and host field might work.
Then each router would only need to know how to get to each country, the states or provinces in
its own country, the cities in its state or province, and the networks in its city. Unfortunately, this
solution would require considerably more than 32 bits for IP addresses and would use addresses
inefficiently (Liechtenstein would have as many bits as the United States).
In short, most solutions solve one problem but create a new one. One solution that is now
being implemented and which will give the Internet a bit of extra breathing room is CIDR
(Classless Inter Domain Routing). The basic idea behind CIDR, which is described in RFC 1519,
is to allocate the remaining class C networks, of which there are almost two million, in variablesized blocks. If a site needs, say, 2000 addresses, it is given a black of 2048 addresses (eight
contiguous class C networks), and not a full class B address. Similarly, a site needing 8000
addresses gets 8192 addresses (32 contiguous class C networks).
In addition to using blocks of contiguous class C networks as units, the allocation rules for the
class C addresses were also changed in RFC 1519. The world was partitioned into four zones, and
each one given a port/on of the class C address space. The allocation was as follows:
Addressee 194.0.0.0 to 195.255.255.255 are for Europe
Addresses 198.0.0.0 to 199.255.255.255 are for North America
Addresses 200.0.0.0 to 201.255.255,255 are for Central and South America
Addresses 202.0.0.0 to 203.255.255.255 are for Asia and the Pacific
In this way, each region was given about 32 million addresses to allocate, with another 320
million class C addresses from 204.0.00 through 223.255.255.255 held in reserve for the future.
The advantage of this allocation is that now any router outside of Europe that gets a packet
addressed to 194.xx.yy.zz or 195.xx.yy.zz can just send it to its standard European gateway. In
effect 32 million addresses have noe been compressed into one routing table entry. Similarly for
the other region.
Of course, once a 194.xx.yy.zz packet gets to Europe, more detailed routing tables are
needed.
One possibility is to have 131,072 entries for networks 194.0.0.xx through
195.255.255.xx, but this is precisely this routing table explosion that we are trying to avoid.
Instead, each routing table entry is extended by giving it a 32-bit mask. When a packet comes in,
its destination address is first extracted. Then (conceptually) the routing table is scanned entry by
entry, masking the destination address and comparing it to the table entry looking for a match. To
make this comparison process clearer, let us consider an example. Suppose that Cambridge
University needs 2048 addresses and is assigned the addresses 194.24.0.0 through 194.24.7.255,
along with mask 255.255.248.0. Next, Oxford University asks for 4096 addresses. Since a block
of 4096 addresses must lie on a 4096-byte boundary, they cannot be given addresses starting at
194.8.0.0. Instead they get 194.24.16.0 through 194.24.31.255 along with mask 255.255.240.0.
Now the University of Edinburgh asks for 1024 addresses and is assigned addresses 194.24.8.0
through 194.24.11.255 and mask 255.255.252.0.
The routing tables all over Europe are now updated with three entries, each one containing a
base address and a mask. These entries (in binary) are:
Address
Mask
11000010 00011000 00000000 00000000 11111111 11111111 11111000 00000000
11000010 00011000 00010000 00000000 11111111 11111111 11110000 00000000
11000010 00011000 00001000 00000000 11111111 11111111 11111100 00000000
55

Now consider what.happens when a packet comes in addressed to 194.24.17.4,

which in binary is
110000100001100000010001 00000100
First it is Boolean ANDed with the Cambridge mask to get
110000 I0 0001 I000 000 I0000 00000000
This value does not match the Cambridge base address, so the original address is
next ANDed with the Oxford mask to get
11000010 00011000 00010000 0000000
This value does match the Oxford mask, so the packet is sent to the Oxford router. In practice, the
router entries are not tried sequentially; indexing nicks are used to speed up the search. Also, it is
possible for two entries to match, in which case the one whose mask has the most 1 bits wins.
Finally, the same idea can be applied to all addresses, not just the new class C addresses, so with
CIDR, the old class A, B, and C networks are no longer used for routing. This is why CIDR is
called classless routing.
5.3.9 IPv6
While CIDR may buy a few more years' time, everyone realizes that the days of IP in its
current form (IPv4) are numbered. In addition to these technical problems, there is another issue
looming in the background. Up until recently, the Internet has been used largely by universities,
high-tech industry, and the government (especially the Dept. of Defense). With the explosion of
interest in the Internet starting in the mid 1990s, it is likely that in the next millenium, it will be
used by a much larger group of people, especially people with different requirements. For one
thing, millions of people with wireless portables may use it to keep in contact with their home
bases. For another, with the impending convergence of the computer, communication, and
entertainment industries, it may not be long before every television set in the world is an Internet
node, producing a billion machines being used for video on demand. Under these circumstances, it
became apparent that IP had to evolve and become more flexible. Seeing these problems on the
horizon, in 1990, IETF started work on a new version of IP, one which would never run out of
addresses, would solve a variety of other problems, and be more flexible and efficient as well. Its
major goals were to
1. Support billions of hosts, even with inefficient address space allocation.
2. Reduce the size of the routing tables.
3. Simplify the protocol, to allow routers to process packets faster.
4. Provide better security (authentication and privacy) than current IP.
5. Pay more attention to type of service, particularly for real-time data.
6. Aid multicasting by allowing scopes to be specified.
7. Make it possible for a host to roam without changing its address.
8. Allow the protocol to evolve in the future.
9. Permit the old and new protocols to coexist for years.
To find a protocol that met all these requirements, IETF issued a call for proposals and
discussion in RFC 1550. Twenty-one responses were received, not all of them full proposals. By
December 1992, seven serious proposals were on the table. They ranged from making minor
patches to IP, to throwing it out altogether and replacing with a completely different protocol.

56

BROADBAND AND IP QOS

6.1 The Transport Service

The transport layer is the heart of the whole protocol hierarchy. Its task is to provide efficient,
reliable, cost-effective data transport from the source machine to the destination machine,
independent of the physical network or network s currently in use. The primary function of the
transport layer depends on the service or services provided by the network layer.
Some possible QoS (Quality of Service ) parameters are summarized in Fig. 6-1. Note that
few networks or protocols provide all of these parameters.
No

Parameters

Connection establishment delay

Connection establishment failure probability

Throughput

Transit delay

Residual error ratio

Protection

Priority

Resilience

Fig. 6-1 QoS parameters

6.2 The Internet Transport Protocols (TCP and UDP )
The Internet has two main protocols in the transport layer, a connection-oriented protocol
( TCP, Transmission Control Protocol ) and a connectionless protocols ( UDP, User Datagram
Protocol ) .
6.2.1 Transmission Control Protocol
Transmission Control Protocol (TCP), was specifically designed to provide a reliable end-toend byte stream over an unreliable internetwork. TCP was formally defined in RFC 793, Some
bug fixers and clarifications are detailed in RFC 1122, its Extensions are given in RFC 1323. In
short , TCP must furnish the reliability that most users want and that IP does not provide.
Figure 6-2 shows the layout of a TCP segment. Every segment begins with a fixed-format 20byte header. The fixed header may be followed by header options. After options, if any, up to
65515 ( 65535 20 20 ) data bytes may follow. Segments without any data are legal and are
commonly used for acknowledgements and control messages.
Source port

Destination port
Sequence number
Acknowledgement number

TCP

Header

Length

Checksum

Window size

Urgent pointer

57

Options ( 0 or more 32-bit words )

Data ( optional )

Figure 6-2 The TCP header

6.2.2 User Datagram Protocol
The Internet protocol suite also support a connectionless transport protocol, UDP (User
Datagram Protocol) . UDP provides a way for applications to send encapsulated raw IP dadagrams
and send them without having to establish a connection. Many client-server applications have one
request and one response use UDP rather than go to the trouble of establishing and later releasing
a connection. UDP is described in RFC 768.
A UDP segment consists of an 8-byte header followed by the data. The header is shown in
Fig. 6-3.
Source port

Destination port

UDP length

UDP checksum

Fig. 6-3. The UDP header

6.3 IP QoS technology
The Internet has been growing rapidly in recent years and it is becoming the global
telecommunications infrastructure that also covers business activities. The Internet is expected to
be the basis of services provided on all communications networks including not only the
conventional data communications networks but also real time service and reliable service
network such as public telephone networks. In order to support a wide range of applications that
have various characteristics and require service levels, the Quality of Service (QoS) support
technologies have been proposed by the Internet Engineering Task Force (IETF).
This section outlines the QoS definition for Internet services, requirements of QoS provided to
users, and currently proposed QoS technologies. The section also proposes QoS control
architecture that can control networks to provide the QoS appropriate to various network services.
6.3.1 IP QoS Services
1. QoS Demanded by Business Users
As mentioned above, "QoS" is an ambiguous statement that can have different meanings.
QoS is often used as:
A scale of network usability
Definition of concrete service characteristics and features
Traffic and service differentiation
A service level better than the "Best Efforts", and so on.
Since IP networks are becoming a business platform, the importance of QoS is increasing. If
business activities are delayed due to poor network quality, users will incur a heavy loss in their
business. To avoid such a risk, what the network users seek is predictable services of the traffic
quality. The predictable traffic quality is the QoS discussed in this section.
The QoS characteristics can be defined by the following parameters.
(1) Availability
(2) Delay
(3) Jitter

58

(4) Packet loss probability

(5) Throughput
Availability means that a user has reliable network connectivity and this is the most basic
parameter of telecommunication. Delay includes two factors, single-direction transmission delay
and round trip time (RTT). These are important factors for real-time applications such as VoIP
(Voice over IP). Jitter is variation in delay and is an important factor for real-time applications.
The packet loss probability is the rate of missing packets and is important for mission critical
applications. Throughput is the data rate that the system can transfer. This factor is important for
streaming applications.
2. SLA
The qualities demanded by users and applications are different. The service levels define these
qualities. A Service Level Agreement (SLA) is made between users and network service providers,
as well as between network providers.
Currently, the major items that a SLA of most service providers covers include, availability,
transmission delay and packet loss probability. They are mean values of one or more months, and
they often show the average performance of the provider's entire network. For example, an SLA
from UUNET (http://www.uu.net) guarantees 100% availability (except for the scheduled
maintenance time), 99% or more packet transmission, and 65-millisecond round trip time (RTT) in
the United States between hubs and routers of the backbone network. The performance in the time
band when almost no traffic flows and the values not used by each user may be included in these
SLA values. These availability and transmission delays do not define the service level actually
provided to each user. The SLA states that the service provider makes a best effort to keep the
agreed service level of the network. In other words, the current SLA usually does not guarantee
the predictable quality of services for each user.
This is because we have no method to guarantee more detailed service levels. However, the
SLA needs to guarantee the predictable quality of IP services for each user in the future. For
example, the scale of B2B E-Commerce market is estimated to rapidly increase to 1,330.8 billion
dollars in the United States in 2003 (according to Forrester Research), and the Japanese B2B
market scale is estimated to reach 6,800 billion yen in 2003 (according to the MITI). Catalogs and
settlement data are exchanged via IP networks real-time and business transactions are established
on the "eMarketPlace". This is E-Commerce over an IP Network. The users participating in an
"eMarketPlaee" will lose their business opportunity if they do not have stable quality of service on
the network.
In the future, a service level agreement can also be made between the service providers
themselves. Currently it is difficult for each provider to guarantee end-to-end services. Therefore,
multiple Internet Service Providers (ISPs), Internet exchange (IX) providers, and Application
Service Providers (ASPs) will need to agree on their predictable quality of service levels to
provide their services to end users.
As discussed above, a predictable quality of IP services is required in the future. This section
defines a SLA that can guarantee the predictable quality of services for users and explains the
mechanism that can control such qualities of service.
6.3.2 Basic QoS Technologies
59

This part describes the protocols, architecture and mechanisms that can provide QoS on the
Internet.
1. Int-Serv (RSVP)
Int-Serv is the architecture that provides QoS for a flow on the Internet. Resource Reservation
Protocol (RSVP) is a signaling protocol that requests bandwidth and other resources for the IntServ architecture. To assure a required service level, RSVP is used for end-to-end signaling to
secure network resources. Int-Serv defines two service models: Guaranteed Service and
Controlled Load. Guaranteed Service provides the mathematical upper limit of packet queuing
delay and is used for applications requiring strict QoS. Controlled Load controls load using
multiplex statistics and for applications having higher flexibility than those that require
Guaranteed Service.
As shown in Figure 6-4, the sender transmits an RSVP PATH message to the receiver before
sending the data flow. The PATH message shows the sender information and traffic characteristics
of the data flow. When a router on the data path receives the PATH message, it stores the State
information for the data flow.

Sender

Path

2
3

Receiver
R

Fig. 6-4 Int-Serv ( RSVP )

Upon reception of the PATH message, the receiver generates an RESV message, showing the
QoS request, and sends it back to the sender. This RESV message is forwarded on the same path
as the PATH message, and the resources are secured. As each router holds the State information
during data flow forwarding, the PATH and RESV messages are exchanged periodically during
end-to-end communication. However the resource reservation mechanism is required
independently from the RSVP.
Int-Serv architecture having RSVP must perform the above signaling for each data flow of
end-to-end communication. In addition, each router of the network must hold the state information
(such as session establishment information and bandwidth assignment information) of all flows
that pass through the router. Therefore, this architecture has no scalability to support a larger
network.
2. Diff-Serv
Since Int-Serv has no scalability and cannot support a large-scale network, an alternative
Differentiated Service (Diff-Serv) architecture has been studied by the IETF. Diff-Serv is intended
to provide a differentiated service class for the traffic on the Internet and to define a simple and
rough control system when compared with Int-Serv. Diff-Serv controls the QoS for each class of
aggregated flows, not for each micro flow as Int-Serv does. Therefore, Diff-Serv can provide
scalable QoS even on a large-scale network.
Diff-Serv classifies incoming packets at the edge router of the Diff-Serv domain, and assigns
60

each of them a class identifier, DSCP (Diff-Serv Code Point). The core router within the domain
reads DSCP values and schedules packet forwarding based on the PHB (Per Hop Behavior) of the
forwarding schedule definition given for each class. A group of packets to be forwarded based on
the same PHB is called the Behavior Aggregate (BA), as shown inFig. 6-5.

TR Router

The same QoS for

each hop

Domain A
ER Router
Making policy
GR Router
Classifier, policy
administration for the
data flow

Domain B

Translate for
the label

Egress edge router

Ingress edge router

Flow
classifier

EF
AF
BF

Traffic
Conditioner

ToS
classifier

EF
AF
BF

Per class scheduler

Fig. 6-5 Diff-Serv

Although Diff-Serv WG has defined the DSCP and PHB mapping (as shown in Figure 6-5),
the network operator is allowed to specify local mapping.
DSCP
DSCP
101110
001xxx
010xxx
011xxx
100xxx

CU

DS field

PHB
Expedited Forwarding (EF)
Assured Forwarding (AF) Class 1
Assured Forwarding (AF) Class 2) +Lo,Med,Hi
Assured Forwarding (AF) Class 3 Drop Precedence
Assured Forwarding (AF) Class 4
Fig. 6-6 Recommended DSCP values

The following introduces the PHB defined by the IETF.

Expedited Forwarding (EF): The EF PHB (or EF) traffic is not affected by other PHB traffic, and
the packet departure rate is guaranteed above the specified value. Like conventional leased lines,
the EF PHB can provide services of low packet loss, low delay, low jitter,, and assured bandwidth.
The EF PHB forwarding treatment is provided by accepting only the assigned volume of traffic
and by minimum queuing of the traffic. Any excess EF traffic is discarded at the incoming edge
61

router.
Since the present EF PHB definition is incomplete for its implementation, a revised definition
has been proposed., to formally define that the EF assures the PHB with a rate R (bit/sec) and
latency E (sec), the following definition has been proposed and is being studied. In this definition,
transmission time "d(j)" of the last bit of the j-th EF packet and target transmission time "F(j)"
have the following relationship:
D(j) <= F(j)+E
F(j) = max(a(j), min(d(j-1), F(j-1)))+L(j)/R, for all j > 0
F(j): Target departure time (in seconds) of the last bit of the j-th EF packet
a(j): Actual arrival time (in seconds) of the last bit of the j-th EF packet
L(j): Size (in bits) of the j-th EF packet
This expression shows that the packet departure time must not delay from the target departure
time for latency E or more in order to maintain the PHB rate.
Assured Forwarding (AF): The AF PHB provides a different level of forwarding characteristics
to IP packets. A specific amount of forwarding resources (such as buffers and bandwidths) is
assigned to each of four AF classes, and one of three different discard priorities is assigned to each
packet. The Al: PHB class allows packet forwarding with a higher probability if the total traffic
does not exceed the preset rate.
6.3.3 MPLS
As the IP network is a connection-less network, each router searches for the next hop of each
received packet based on its destination address and forwards the packet. However, the router uses
the Longest Prefix Match address search (that is, it searches for an entry whose prefix matches the
longest one), and it cannot realize high-speed packet forwarding.
The Multi-Protocol Label Switching (MPLS) system uses a frame format that has a fixed-length
label assigned to each packet, it then forwards packets based on their label values. As the
destination is determined by a fixed-length label search, MPLS can realize high-speed packet
forwarding. The packet forwarding path determined by this label is called the Label Switched
Path (LSP).
As Fig. 6-7 shows, the MPLS controls the LSP explicitly and it can determine an optimum edgeto-edge path based on the QoS required for the traffic. It can also provide traffic engineering with
load distribution of each path within the network. At present there is little mason to use MPLS for
high-speed packet forwarding. Instead, the traffic engineering on the established connection of IP
networks is becoming more popular.

LSR BE
60 Mbps
LSR B

OC3

OC3
OC3

LSR A

LSR D

Label Switching Router

OC3
OC3

LSR C

LSR AE
120 Mbps

LSR E

Fig. 6-7 MPLS-based traffic engineering

62

1. MPLS Setup Information

MPLS uses the following information for label switching of IP packets.
Forwarding Equivalence Class (FEC): MPLS identifies a group of packets requiring the
same forwarding processing as an FEC. Packets belonging to the same FEC are processed in the
same way and forwarded along the same LSP. The FEC of each packet is primarily identified by
the header information of the packet. This FEC can be a group of destination address prefix of
packets to be forwarded via the egress LSR (Label Switching Router) of the LSP end point. Also,
a specified application alone can be forwarded along the LSP with identifying FEC by the IP
address and IA port number.
Next Hop Label Forwarding Entry (NHLFE): This information is used for labeled packet
forwarding. The NHLFE has the next hop, label value to be replaced, label stack and others of
each packet.
Incoming Label Map (ILM): Used for mapping between the incoming label of a received
packet and the NHLFE. The core LSR reads the incoming label of a received packet and, based on
the ILM, selects its next hop and label to be replaced..
FEC-to-NHLFE Map (FTN): Used for mapping between each FEC and NHLFE. The edge
LSR decides the FEC of an unlabeled packet and selects its next hop and a label to be added based
on the FTN.
2. Label Distribution Procedures
The following label distribution procedures are available for LSP setup in MPLS for traffic
engineering.
CR-LDP

The Constraint-based Routing Label Distribution Protocol (CR-LDP) is a label distribution

protocol extended from the existing Label Distribution Protocol (LDP) for constraint-based
routing. As shown in Figure 6-8, the CR-LDP allows an LSP setup (passing through LSR B and C)
in the label request message. In addition, the CR-LDP allows the setup of traffic parameters such
as Peak Rate, Committed Rate, and Burst Size in the label request message. However, the actual
mechanism of QoS assurance of LSP traffic is not supported by the CR-LDP.
Label Request B,C

LSR B

Label Request C

LSR A
Ingress

LSR C
Engress
Label Mapping (17)

Label Mapping (32)

Fig. 6-7 LSP setup flow by CR-LDP

RSVP-TE

The RSVP-TE is a label distribution protocol extended from the existing RSVP. It uses several
new RSVP objects including the mandatory LABEL-REQUEST object and LABEL object. The
RSVP-TE supports the following additional functions to establish and maintain the LSP:
(1) Downstream-On-demand label distribution
(2) Instantiation of explicit label switched path
(3) Allocation of network resources (e.g. bandwidth) to explicit LSP
(4) Rerouting of the established LSP tunnel using the concept of"make-before-break"
63

(5) Tracing of the actual route traversed by an LSP-tunnel

(6) Diagnostics on LSP-tunnel
(7) The concept of nodal abstraction
(8) Preemption options, that are administratively controllable
Hop-by-hop setup by operators
Network operators use the NMS and sets up the MPLS usage information (discussed above) for
each router on the LSP path. CLI, SNMP and other protocols can be used for router setup.
3. Diff-Serv over MPLS
Although traffic engineering is realized by MPLS as discussed above, Diff-Serv is required for
scalable QoS control. However, MPLS encapsulates IP packets using the shim header that has the
label, and the core router cannot refer to the DSCP. Incompatibility with Diff-Serv posed a
problem. The IETF has proposed Diff-Serv over MPLS [MPLS-Diff] to solve this problem.
Diff-Serv over MPLS can map multiple BAs of Diff-Serv to a single LSP of MPLS. By this,
traffic on the LSP can be forwarded based on the PHB of BA. The E-LSP can be used for LSP and
BA mapping to allow assigning multiple BAs to a single LSP using the EXP field. The L-LSP can
be used to allow assigning a single LSP to a single BA (displays multiple packet discarding
priorities)..
E-LSP: The E-LSP shows the PHB of a packet using the EXP field of MPLS shim header. Up to
eight BAs can be mapped in the EXP field.
DSCP

CU

DS field

LABEL

EXP

TTL

20 bits

3 bits

8 bits

000 AF11
001 AF12
010 AF13
Fig. 6-9 Mapping example of DSCP to MPLS EXP field
L-LSP: The L-LSP determines the packet scheduling characteristics based on the MPLS label and
the packet discarding priority based on the shim header or layer-2 packet discarding mechanism.
The native ATM uses the L-LSP as it cannot use the EXP field.
As the NEs replace packet labels hop by hop, the label and DSCP mapping is difficult to
manage. The E-LSP is easier to control than the L-LSP as it can previously determine the mapping
between the EXP field and DSCP of each packet on the entire network.
6.3.4 Policy Control
A major service provider's network may have various network elements provided by different
vendors. Operators need to expend enormous effort for separate setup of sophisticated QoS
mechanisms. The Policy Control has been proposed to reduce the network setup load of operators.
As shown in Fig. 6-10, the Policy Control allows network operators to set up a policy rule,
64

which can be written in the policy repository using policy tools in the following format:
IF <Condition> THEN <Action>
Its control is determined by the Policy Decision Point (PDP) based on the policy rule. The
Policy Enforcement Point (PEP) within the network element controls the network element based
on the PDP determination. The Policy Information Model has been proposed for policy definition
in the policy repository, and the COPS protocol, for policy exchange between the PDP and the
PEP. Also, the SNMPCONF WG has proposed the SNMP-based policy control against the COPS.

Two types of QoS support technologies, Int-Serv and Diff-Serv, are available and the Policy
Control framework uses different policy control approaches (such as Outsourcing type and
Provisioning type) for those technologies.
Policy tool (GUI/API )

Policy Scheme

LDAP

Policy controller

Policy Repository
Lightweight Directory Access Point

PDP

PDP
COPS

PEP

Policy Server
Common Open Policy Service

PEP

PEP

Controlled devices
(including Routers )

Fig. 6-10 Policy Framework

Outsourcing type: In Int-Serv, each flow uses the RSVP to secure resources for the network.
When the PEP of each network element receives RSVP signaling, it asks the PDP of flow
admission. When asked, the PDP determines the admission based on the policy. This is called
Outsourcing type policy control. The RAP (Resource Allocation Protocol) WG of the IETF has
proposed the COPS-RSVP be used for the Outsourcing type policy control.
Provisioning type: In Diff-Serv, the policy server controls the DSCP and the PHB
collectively and the PDP of the policy server distributes its information to each network element.
This is called Provisioning type policy control. The RAP WG of the IETF has proposed the
COPS-PR be used for the Provisioning type policy control.
6.4 10 Hottest Technologies
The mantra for 2003 is making more out of what you've got, and no place is that more true
than in telecom. Gone are the days when service providers deploy the latest and greatest just
because it is the latest and greatest. While service provider capex has been the headline-maker the
last 18 months, reducing opex has now taken center stage.
Today's capex is being spent on technologies that extend the life and abilities of existing
equipment and in some cases, help hot technologies of days past-such as VoIP and ADSL-live up
65

to their previously hyped potential. These technologies must slide seamlessly into service provider
networks, interoperate with back-office systems and, above all, save money.
This year's technology picks are all about helping service providers maximize their existing
investments while moving forward with next-gen services. In no particular order, the 10 Hottest
Technologies for 2003 are:
Session Border Controllers
Wi-Fi n Virtual Private LAN Service
VPLS
ADSL2/ADSL2+
Data over SONET
RF/Fiber Extension
Inventory Management
Coarse WDM
Next-gen Mobile Switching Centers
TDM over MPLS
1. Session Border Controllers: VoIP's Missing Link
While VoIP has become a cost-effective mechanism for transporting long-distance voice
traffic and PRI offload, the ability to provide native VoIP peering between networks, which
requires high levels of QoS, has been all but nonexistent. Emerging to solve this problem are SCs
(session controllers), also called SBCs (session border controllers), to provide a demarcation point
between two service providers' VoIP networks, allowing them to manage signaling and control
routing for VoIP traffic.
Operating in a carrier's existing signaling infrastructure (SIP, H.323), SBCs provide Layer 5
routing and control to manage real-time traffic flows between IP networks and address network
security, signaling interoperability, call admission control, service quality, CALEA and session
routing. As a complement to already-deployed carrier routers, SBCs sit at the network edge and
core to control connections between enterprise and service provider. In addition, an SBC can
provide session routing and service level call admission control to ensure QoS and enforce carrier
and end-customer policies.
When it comes to VoIP peering, there are three elements: IP-to-PTSN, PSTN-to-IP and IP-toIP. In an IP-to-PSTN deployment, calls begin at the enterprise as VoIP and are converted to TDM
via a customer-based gateway or a media gateway within a service provider POP. On the PSTNto-IP configuration, a call begins at the enterprise as TDM and then is converted in the service
provider's core as IP. In an IP-to-IP configuration, originating VoIP calls bypass the PSTN and
peer to another service provider in a pure VoIP format. Until now, traditional premises networking
devices such as NAT (network address translation) devices and firewalls were not designed to
handle real-time voice traffic, therefore when a call is received it would be blocked by the
firewall. A key function of the SBC is to open a pinhole to engage in authorized real-time
communications sessions.
A number of emerging start-up and incumbent vendors are moving ahead with products,
including Acme Packet, Alcatel (via its Aravox acquisition), Jasomi, Netrake, Kagoor and
NexTone. Acme Packet, Alcatel and Kagoor call their products SBCs, while NexTone and Netrake
define theirs as SCs.
So besides becoming the latest telecom acronym, what's the buzz around SBCs and SCs?
66

Both will continue to gain value as enterprises continue to adopt hybrid IP/PBXs from vendors
such as Cisco, Avaya and Nortel, creating those so-called VoIP islands. As a counterattack,
providers such as WorldCom, Genuity, BellSouth and Verizon have launched their own hosted-IP
telephony services such as IP Centrex and managed PBX services.
"With SCs and SBCs, you are providing that demarcation line between the customer and the
network, or between networks, and we are starting to see the players get more specific," said Chris
Hartman, research director for Probe Research. "Some of the elements like being able to control
how many calls are in a session are not new, but are things you have to do to make packet look
like the existing TDM world." -Sean Buckley
2. Wi-Fi: Up, Up and Away
What do a fast-food restaurant, a national bookseller, a computing giant and airports have in
common? They've all jumped on the Wi-Fi bandwagon. Limited until recently to enterprise and
home networking uses (with a few failed service provider experiments), Wi-Fi is back in the
spotlight. The difference this time is the major support it's getting from some heavy hitters,
including Intel, AT&T, IBM, and a slew of smaller players. It also has a new market: wireless
LAN hot spots in your favorite hotel, the local airport, the town square and in places like
Starbucks, McDonald's and Borders Book Stores. According to the Wi-Fi Alliance, there are 2,000
to 3,000 hot spots in each the United States and Europe, and around 5,000 in Asia. Thousands
more are promised by year's end.
Wi-Fi, a.k.a. 802.11, comes in a few flavors, with more on the way-a veritable alphabet soup
of options. The most widely deployed is 802.11b, which offers 11 Mbps data rates at 2.4 GHz. The
issue: shared capacity means the more users, the slower the individual connectivity. 802.11a,
which operates in the 5 GHz spectrum at 54 Mbps, was supposed to overcome this, but hasn't seen
widespread support-perhaps because vendors are waiting for 802.11g (54 Mbps at 2.4 GHz),
expected to be ratified as a standard by the IEEE this summer and be backward-compatible with
802.11b. The good news: The Wi-Fi Alliance is trying to do away with the letters in an effort to
simplify things for users and chip makers will likely put all three standards on the same silicon
and allow equipment vendors to choose which to activate.
The computer industry is playing a very active role. Intel, for example, has set aside $150
million to invest in Wi-Fi start-ups, including Cometa Networks, a joint venture with AT&T and
IBM that aims to set up 20,000 hot spots. Intel has also launched its Centrino chip, which includes
a Pentium processor and built-in Wi-Fi antenna, and has announced deals with McDonald's and
Borders. The thinking: the more ubiquitous Wi-Fi is, the more users. The more users, the more
products sold.
The burning question for Wi-Fi recently has been security. Currently equipped with WEP
(wireless equivalent privacy) security mechanisms, Wi-Fi products will soon use WPA (Wi-Fi
protected access), a specification that offers advanced data encryption and user authentication,
which was largely missing in WEP. Next up: the 802.11i standard from which WPA is derived.
WPA should be approved by the IEEE next month; approval for 802.11i, which will be called
WPA2 and be backward-compatible with WPA, is expected by Q3. The main pieces of the 802.11i
draft not included in WPA are secure IBSS (independent basic service set), secure fast hand-off,
secure de-authentication and disassociation as well as enhanced encryption protocols.
So what's next? Many predict Wi-Fi/2.5G/3G interoperability and the advent of centralized
LAN switches for controlling multiple access points in large Wi-Fi deployments. One thing's for
67

sure: Wi-Fi's future is no longer up in the air. -Sue O'Keefe

3. VPLS: Another Choice for VPNs
It would seem that the concept of VPN is getting more and more amorphous or perhaps there
are just more options to consider. For example, MPLS technology enables IP VPNs and, by
tunneling data services such as ATM, frame relay or Ethernet across an MPLS network, Layer 2
VPNs. Now add yet another MPLS-based VPN technology--VPLS (Virtual Private LAN Service)to the choices.
The aim of VPLS is to provide customers a secure Ethernet service that interconnects
multiple customer sites in a way that they appear to be on a single LAN even if the locations are
geographically dispersed. And in this sense, VPLS would seem to be a TLS (Transparent LAN
Service), although this term is so broadly used because, for example, it also applies to stacked
VLANs, where a service provider affixes an 802.1 VLAN tag to a subscriber's 802.1Q VLAN tag.
To further complicate matters, the Metro Ethernet Forum uses the term Ethernet Private LAN
Service to describe a multipoint Ethernet service, but the forum doesn't prescribe the enabling
technology.
It is here that VPLS comes in because for the IETF and the MPLS Forum, VPLS is the
underlying technology for an Ethernet Private LAN Service. Thus, in a way, the two terms
Ethernet Private LAN Service and VPLS could be used interchangeably. "From a customer
perspective, the beauty of VPLS as a service is that it would look just like a Layer 2 switch on
their campus," notes Lindsay Newell, product marketing manager for TiMetra Networks, a
Mountain View, Calif., start-up that is building an MPLS-based router.
TiMetra, along with Riverstone Networks, has co-authored a proposal for an IETF standard
for VPLS. Because it offers an Ethernet handoff to customers, Riverstone believes that VPLS will
prove popular not only with enterprises running IP on the LAN but, unlike IP VPNs, also with
those using other Layer 3 protocols. As an access technology, Riverstone also sees VPLS being
favored by those carriers that support MPLS in the core as well as by carriers that prefer to run
Ethernet traffic over SONET, although the latter would rob a carrier of the benefits of statistical
multiplexing. In any event, the case for VPLS comes down to this: To provide a multi-point
Ethernet service a carrier don't necessarily have to build an Ethernet network. In addition to
TiMetra and Riverstone, VPLS also has the backing of heavyweights such as Cisco, Nortel and
SBC. -Sam Masud
4. ADSL2/ADSL2+: Anatomy of a Standard
While the advent of the original ADSL standard certainly gave copper providers a means to
provide broadband data services, physical limitations such as distance have prevented the
technology from reaching its mass-market potential. To capitalize on existing ADSL investments,
the ITU-based ADSL2 standard and its counterpart, ADSL2+, call for a number of improvements
that provide increased reach and power as well as add-on diagnostic features. Officially completed
in July 2002, ADSL2 is comprised of two standards: G.992.3 (G.dmt.bis) full-rate and G.992.4
(G.Lite.bis) splitterless.
ADSL2 boasts a number of improvements to the existing ADSL standard including increased
rate and reach, diagnostics, power enhancements, rate adaptation and bonding for higher data
rates. ADSL2 also features channelized VoDSL (voice over DSL) and an all-digital mode for
68

transmitting ADSL in voice bandwidth.

Right on the heels of the ADSL2 standard is ADSL2+. In January, the ITU reached consent to
bring the ADSL2+ standard, known as G.992.5, to the ADSL2 family. ADSL2+ gives providers
the ability to increase downstream data rates up to 24 Mbps on phone lines reaching 3,000 feet and
20 Mbps at 5,000 feet. To resolve crosstalk issues that often hamper data rates between the CO
and RT (remote terminal), ADSL2+ utilizes tones between 1.1 MHz and 2.2 MHz by masking sub1.1 MHz tones.
Within the ADSL2 and ADSL2+ standard are annexes or sets of specifications for certain
applications in each world region. Specifically, each of these annexes (A-L) specifies subcarriers
or tones and their respective transmission power for upstream and downstream. Perhaps one of the
most compelling annexes is annex L, a specification that has specific enhancements for both
ADSL2 and ADSL2+. Otherwise known as Reach Extended ADSL2 (RE-ADSL2), this annex,
which is expected to reach ITU consent by October, enables an ADSL2 link to incorporate a new
power spectral density mask that can enable an additional reach between 1,000 and 2,000 feet. For
ADSL2+, Annex L provides an enhanced upstream rate in a POTS environment.
As a testament to bringing ADSL2 to market, earlier this month seven silicon vendors
(Analog Devices, Aware, Broadcom, Centillium, Infineon, Samsung Electronics and Texas
Instruments) gathered at University of New Hampshire's Interoperability Laboratory to participate
in the DSL Forum's ADSL2 "plugfest" interoperability event. Based on the DSL Forum's internal
proposed Draft 013 Interoperability Test Plan for ADSL2 plugfests, PD-013 provides a suite of
basic PHY layer tests to ensure interoperability between multiple ADSL2-based chips and
systems. In May 2002, the ITU incorporated the DSL Forum's TR-048 specification into the
ADSL2 G.992.3 and G.992.4 standard recommendations.
Along with the silicon vendors, DLC/DSLAM vendors such as Alcatel, UTStarcom, Lucent,
Calix, Net to Net, XAVI and Next Level claim to be ready to integrate ADSL2/ADSL2+ elements
as silicon becomes available. More importantly, as carriers migrate to ADSL2 and ADSL2+ as
demand dictates, they won't need a forklift upgrade.
For operators like Bell Canada, the fact that ADSL2 and ADSL2+ systems are backwardcompatible with their existing base is key. "As new equipment comes in we will adapt to ADSL2.
However, we still have 1 million-plus embedded ADSL modems so we're not sure when we will
have a critical mass to turn on the ADSL2 features," said Jean Hupp, Bell Canada's director of
access technology. "For ADSL2+, a similar thing will happen as we will eventually buy new cards
and modems that will be ADSL2+ ready. In our case, if we go with VDSL, which may cover 80
percent to 90 percent of the population, I see ADSL2+ as complementary to the remaining 10
percent to 20 percent." -Sean Buckley
5. Data over SONET: The Next Generation
Nearly half a decade ago, Cerent (which was acquired by Cisco) pioneered a new type of
network equipment. And while Cerent quite possibly never used the term MSPP (multi-service
provisioning platform) to describe its box, that is now the popular label that's used to describe
next-gen SONET products from a number of other vendors including Alcatel, Fujitsu and Nortel.
Broadly speaking, these devices combine ADMs (add-drop multiplexers) that support multiple
SONET rings operating at different speeds with a cross-connect capability.
Now vendors are bringing to market a new generation of SONET-based products (Does the
69

label next, next-gen SONET sound right here?) designed to ensure that the vast, embedded
SONET infrastructure can be evolved to handle all sorts of data services. So we'll admit to coining
a term by picking Data-over-SONET as a hot technology that encompasses a handful of
techniques that would make SONET a more flexible technology for transporting and switching
data traffic.
Platforms from established vendors such as Tellabs (via its acquisition of Ocular Networks)
as well as start-ups such as Polaris Networks and Turin Networks have taken the lead to
implement some or all of the standards, or emerging standards, that enable better use of SONET
bandwidth for data services. One of the techniques these platforms support is VCat (virtual
concatenation), which, more simply put, enables the gluing together of several STS-1 or VT1.5
pipes so there is minimum wastage of a SONET network's bandwidth. For example, this enables
two GigE links to be carried inside an OC-48 channel and still leave six STS-1s available for other
traffic, as opposed to taking up an entire OC-48. Or, to use another example, a 10-Mbps Ethernet
link could be transported inside seven VT1.5s. Perhaps what sets these new platforms apart from
earlier MSPPs is the ability to transport, switch and groom (the latter ensures that trunks between
any two points are filled to capacity) traffic at the more granular VT1.5 level instead of just at the
STS level.
A new feature that makes VCat even more effective is LCAS (link capacity adjustment
scheme). LCAS enables a network operator to right-size-either up or down in bandwidth-on the
fly by, for example, adding another VT1.5 if the traffic demands without disrupting traffic. While
almost all of the new transport switches have the VT1.5 cross-connect and grooming capability,
the LCAS feature is something that many only now are in the process of incorporating. Both VCat
and LCAS are being driven by GFP (generic framing procedure), which, like its ANSI equivalent
X.86, is a method for mapping data services to SONET. Although the initial impetus for GFP was
for transporting Ethernet over SONET networks, GPF can also be used for other data services
such as Ficon, Escon, Fibre Channel and even extended ATM and frame relay. Thus in addition to
making use of the existing infrastructure, Data-over-SONET also leverages carriers' investment in
Telcordia OSSs. -Sam Masud
6. RF Distribution: Maintaining the Connection
How many times has a wireless call dropped when you enter a shopping mall, tunnel or office
building? With the onset of data and business applications, the ability to provide seamless
coverage is essential. Acquiring additional spectrum and building new base stations can be
difficult and expensive, so carriers need more effective methods to handle this problem. One
answer could be the advent of in-building and wide area RF distribution solutions, which transmit
RF signals over Cat 5, coax or optical fiber.
As its name implies, the RF/Fiber distribution approach takes RF signals and converts them
into optical signals and back to transmit to remote antenna units within a building or a remote
location in the wide area. In an in-building application, fiber or coax would be run through the
riser and delivered via distributed antennas. Vendors in this space all have a slightly different take
on this. For example, ADC's Digivance transmits signals in an all-digital format from within and
between buildings over multimode fiber or single mode fiber utilizing DWDM. Andrew's InCell
system is an analog-based, single mode fiber optical distributed antenna system that works both in
conjunction with an operator's base station or can serve as a complement to off-air repeaters to
70

provide coverage in buildings or airports.

Not everyone is following the RF/Fiber approach, however, other vendors have opted for
either a passive coax distributed antenna system (InnerWireless, Spotwave and Andrew's passive
DAS) Cat 5-only (Radio Frame) or a hybrid Cat 5/6, multimode fiber (LGC Wireless).
Hand in hand with in-building is wide area RF/fiber distribution where signals are converted
to an optical signal over long distances and back to RF at a remote location. Andrew's EOCell,
newcomer Celerica's 500 System, LGC's MetroCell and ADC's LRCS (Long Range Coverage
Solution) can help enhance wide area coverage and capacity in areas such as tunnels and
highways. Another important application is base station hoteling, where a carrier can centralize
base station equipment and deploy antennas into the necessary coverage areas. To bolster widearea links, ADC, for example, has partnered with FSO (free space optics) vendors such as fSONA
and LightPointe to enable a carrier to create connections where fiber is limited. Celerica, which
has also incorporated FSO into its system, says it can be used to extend links between base
stations and antennas. -Sean Buckley
7 . Inventory Management: A Real-world View
Service Provider A is a Tier 2 carrier in Chapter 11 bankruptcy. Service Provider B is a stable
international carrier rolling out a host of new services. What do they have in common? They both
need to know what physical and logical assets they have and how to use them to the best of their
abilities. Although it seems ironic that a bankrupt carrier would want to spend more money, to
successfully emerge from bankruptcy it needs to get a better grip on its complete asset inventory
to determine what is worth keeping and what is not. "Bankrupt companies are telling us that their
No. 1 priority is finding out what they have and where it is," says Mark Mortensen, chief
marketing officer at Granite Systems.
Coined NRM (network resource management) by analyst group RHK, inventory
management has moved to a whole new level. It's no secret: service providers' back offices are a
mess; different departments within a service provider keep records in different ways, so
engineering's view of the network isn't the same as billing's and finance's isn't the same as
customer care's. The new breed of NRM systems track and maintain information, eliminating the
need for duplicate data entry, and track more carefully the impact of changes in the network. "It's
not just tracking assets, it's understanding when a problem is reported where the problem is and
who it affects," says Larry Goldman, program director for RHK's OSS program. "It's when your
customers ask for a service, knowing what's available to provide that service. It's knowing where
to put more equipment to meet demand."
One huge piece of NRM coming into focus is auto discovery, or finding out about network
equipment whose existence was previously unknown and determining what connections and
services are running in the network. While most IP, ATM and frame relay equipment can be autodiscovered, most DWDM and SONET/SDH equipment can't, because inventory systems need
address information to find the node or a gateway node. "Auto discovery is very, very hard,"
Goldman says. "You can't discover everything, but what you can discover can be very valuable."
Newer, more intelligent network elements have a lot more ability to be discovered, he says.
The hardest part in getting the NRM market off the ground may be a service provider's
internal struggles among departments and, once those are settled, understanding that the best
solutions may not be built internally. Goldman points to Sprint and Level 3 as examples of
71

providers that have made early moves to get their NRM under control. He says others will follow;
indeed, RHK predicts that by 2005, providers will spend $1.3 billion on NRM solutions. Leading
the way for vendors are a host of smaller companies, including Granite Systems, Cramer Systems,
MetaSolv, Micromuse and CoManage, while most of the larger vendors have opted for an OEM
route. -Sue O'Keefe
8. CWDM: A Metro Dream?
We'll admit we glossed over CWDM (coarse WDM) technology when it started to make an
appearance three years ago. But in the red-hot telecom market of the time, attention was focused
on DWDM systems that could shoot lots of colors of light over increasingly long distances. The
expectation was that systems that supported 2.5-Gbps channels would soon give way to gear
delivering 10-Gbps wavelengths-and the race was on to build systems which could provide 40Gbps channels. But that was then, and in the new reality of today CWDM technology might be the
appropriate counterpoint-and even a complementary technology-to DWDM in the metro network.
Although deployment of CWDM remains thin, the technology promises to provide
bandwidth at a price point that is well below that for metro DWDM equipment, something that
should appeal to carriers keeping tight control of their capex.
While a Yugo-vs.-Lexus comparison would be unfair, the simple facts are that CWDM
systems are cheaper because they use cheaper lasers that don't have to tightly control the light and
because they're intended to operate over shorter distances (typically 40 km to 80 km). The
downside-which, in fact, is really their value proposition-is that they provide from two to 16
wavelengths at prices that won't break the bank.
Like their metro DWDM counterpart, CWDM systems from different vendors might be
designed to support point-to-point, linear add/drop, ring and mesh topologies. Likewise, some
systems may provide only 2.5 Gbps channels, while some other systems, such as those from
Germany-based Microsens and Meriton Networks, an Ottawa start-up that integrates transmission
and switching in a single platform, offer a hybrid solution that supports both 2.5- and 10-Gbps
wavelengths. In addition to these two vendors, CWDM gear today is available from a number of
established vendors including Advanced Fibre Communications, Ciena, LuxN and Nortel, with
Tellabs promising a CWDM solution late this year.
Although built for the access network, depending upon their capacity/distance requirements,
some carriers might choose to deploy CWDM in the metro core. The big benefit of CWDM is that
it offers a low-cost way for carriers to meet the transport needs of enterprise customers by
supporting multiple protocols on the same fiber.
And this exactly is what Progress Telecom did when earlier this year it announced that had
begun delivering a wavelength service to Blue Cross and Blue Shield of North Carolina. Although
Progress Telecom is largely a carrier's carrier operating its own fiber network from Miami to New
York, it's using Ciena's metro DWDM gear to support Blue Cross and Blue Shield of North
Carolina's GigE and Fibre Channel connectivity needs between various sites in the RaleighDurham area. But Progress is also in the process of installing Ciena's CWDM system as an
economical way to deliver wavelength services at the very edge of the network.
While the established carriers haven't rolled out CWDM gear as aggressively as vendors
might have hoped, some say that they'll give the CWDM market a boost when they see carriers
such as Progress Telecom lighting up fiber for customers. Meanwhile, Progress Telecom notes that
72

the economics of CWDM might give it reason to look beyond its carrier customers to the large
enterprise market. -Sam Masud
9. Next-Gen MSCs: Wireless Takes IP Baby Steps
Like the wireline network, the wireless network's journey to packet will not take place
overnight. A key enabler of this migration will be the advent of the next-gen MSC (mobile
switching center), which also promises opex savings. Similar to the wireline world, where carriers
have utilized next-gen Class 4 switches to offload Internet data and trunking applications onto
low-cost VoIP links, such a model appears to be gaining traction in the wireless world as well.
Taking on this opportunity is a host of start-up and existing vendors, all of which come from
different angles. On one hand, Winphoria purpose-built its WMS-5000 platform to leverage a
disaggregated architecture that separates bearer and control functions. Meanwhile, Santera, Sonus,
Tekelec and Telica have added new functionalities to their existing softswitch platforms to support
wireless applications. Applying the knowledge it has gained in the Class 4 VoIP market, Sonus
recently launched its SMARRT wireless solution that leverage its existing Insignus softswitch and
GSX switch. In a similar manner, Telica now offers a suite of wireless applications on its existing
Plexus 9000 switch, while Santera's broadband office exchange provides both tandem and wireless
access to the PSTN. Spatial Wireless' distributed MSC Atrium integrates its call server and
element management system with Santera's Media Gateway.
But this is not a total start-up play; Ericsson, Lucent and RAD Data Communications have
their own platforms. RAD recently launched its Vmux-2100 Voice Trunking Gateway, a compact
compressed voice solution for MSC-to-MSC interconnectivity. Lucent, which owns a large part of
the wireless market, has its own wireless softswitch. Since most wireless operators are
preoccupied with their various air interface migrations, the adoption of an all IP-based wireless
network is not the first thing on their minds. Nonetheless, there are practical applications such as
wireless gateway MSCs for transit/tandem, wireless backhaul and long distance in current 2G and
2.5G networks. Along with opex savings, a next-gen MSC gives operators a chance to offer new
IP-based services.
With the traditional inter-MSC connections tying up legacy ports and reliance on PSTN
interconnections, a wireless gateway MSC can save an operator's roaming costs by routing PSTN
calls directly to the mobile subscriber's serving MSC without touching the home MSC.
As a regional wireless transit switch, the MSC allows carriers to break their reliance on
multiple inter-machine trunks while reducing MSC routing translations. On the long-distance side,
a carrier can bypass TDM facilities by offloading long-distance traffic onto its own IP network.
Finally, when providers are ready, platforms can move with emerging 3GPP release 5 where voice
calls are converted into IP.
"Right now, it's a 'taking baby steps' approach for operators," said Peter Jarich, senior analyst
for wireless infrastructure at Current Analysis. "If you look at the latest vendor offerings, the focus
is really on opex with the notion that somewhere down the road we are providing some great tools
to move to next-gen MSCs and IP. But right now the focus is on saving money in terms of
backhaul and transport." -Sean Buckley
10. TDM over MPLS: A Tale of Two Camps
There's a strong push to make MPLS the converged network that supports all kinds of Layer
73

2 services as well as provides IP VPNs. The Martini draft, which the industry has embraced,
already defines a way to tunnel frame relay, ATM, Ethernet, PPP and HDLC services across an
MPLS network. Now there's an effort underway to come up with a standard for running TDM
services over MPLS.
Although the lead for this in the IETF has been taken up by Andrew Malis, president and
chairman of the MPLS Forum as well as chief technologist for start-up Vivace Networks, a
number of industry players including Corrigent Systems, Laurel Networks and Lucent are also
behind the effort. Strictly speaking, the proposed standard doesn't specify supporting TDM
services over MPLS; rather it says that the circuit emulation could be done over a variety of
packet-switched networks.
There are, moreover, two sometimes-overlapping camps behind this concept. While the Malis
Draft talks about doing SONET emulation over packet networks, there's a parallel effort for
supporting low-speed services, specifically DS3s and DS1s, across packet networks.
For instance, Corrigent and IC vendor Litchfield Communications back both efforts.
However, the proposed SONET-over-packet standard does not exclude running lower-speed
services over a packet/MPLS network per se because the VT1.5 channels inside an STS-1 frame
could be carrying DS1 signals. Corrigent, which makes a packet ADM that supports the transport
of SONET (as well as Ethernet) traffic across an MPLS-over-RPR network, notes that SONETover-packet also would provide carriers a cross-connect capability at no additional cost by
enabling the routing of STS-1s to their destination.
Of the two efforts, circuit emulation of SONET over packet-switched networks is expected to
become an IETF standard sooner-mostly likely by year's end-than the effort for supporting lowspeed services over packet networks. Indeed, it's quite possible that there may be more than one
RFC for the lower-speed services. The work on these specs is being done in the IETF's PWE3
(pseudo-wire edge-to-edge emulation) working group, which is the same group under whose aegis
the Martini Draft for Layer 2 VPNs was developed.

74

References:
[1] Andrew S. Tanenbaum. Computer Networks. The 4rd Ed published by Prentice Hall, Inc. 2004
[2] Wu Jichuan. Observationa and Reflections on the Current Global Telecoms Industray.
Published by Posts and Telecommunications Press. 2002
[3] William Stallings. Data and Computer Communications. The 5rd Ed published by Prentice
Hall, Inc. 1997
[4] Douglas E. Comer. Computer Networks and Internets. published by Prentice Hall, Inc. 1997
[5] IP QoS Control White Paper, Description of Basic QoS Control Technologies And Proposed
Basic QoS Services and QoS Control Methodology.

75