CHAPTER

1
THE NETWORK PUNNING
PROBLEM

1.1 INTRODUCTION

We are living in the midst of an industrial and social revolution involving the
generation, processing, and transmission of information. The topic of this volume
is a particular aspect of the technology of information transmission called
"network planning."
What is network planning? Information today is transported over a variety of
"networks"—collections of communication links and communication nodes.
These range from analog and digital wired telephone networks to cellular and
personal communication networks to satellite networks to, finally, data networks
such as the Internet, ethernets, token rings, ATM local area networks, and
metropolitan area networks. A particular organization may run a network
comprising several of these technologies.
Any organization possessing and connected to such networks would like to
be able to rely on orderly deployment, use, and upgrade of the networking
equipment (both hardware and software). This is where the "planning" comes in.
The role of network planning (and planners) is to provide an intelligent means for
an organization to meet future network needs, using the existing networks as a
starting point.
Naturally, as time proceeds, new developments occur. That is why planning
is an ongoing process that continually makes use of the latest information to
revise plans. This temporal property is a key aspect of the field of network
planning—it involves the orderly evolution of networks over time.
Network planning is most highly developed in large telecommunication
companies, which usually have distinct network planning departments for distinct
systems. The large investments involved and today's increasingly competitive
environment not only makes efficient planning a necessity; they serve to justify
the use of sophisticated planning techniques. More recently, however, organiza-
 CHAPTER 1 THE NETWORK PLANNING PROBLEM

tions with substantial network installations such as private companies, academic

institutions, and government agencies, have found some sort of planning function
to be necessary.

1.2 AN OPTIMIZATION PROBLEM

The interested reader may recognize what is being described as an optimization

problem. Genetically, one would like to minimize the cost of outlays for
equipment and operations, maximize revenues if one is considering a profit-
oriented organization, and do all this over time as technology, user requirements,
and the economic environment change. Certainly this is a tall order.
In fact, it is impossible to pose the network planning problem, in all its
baroque detail, as a single optimization problem for any reasonably sized
organization.
Why? There are many reasons. One is the absence of a single optimization
criterion. Certainly, cost, reliability, public image, capacity, and potential for
growth are all important criteria, but how does one decide to balance them?
Another reason is the size of the problem in large organizations, no matter
whether it is measured in terms of the number of constraints, the number of
people involved, and/or the number of organizational units involved. A third
reason is the role of the unexpected: free market competition, new technological
developments, and new economic developments.
Therefore what happens in reality is that the overall problem is divided into a
(possibly large) number of smaller, more manageable subproblems. The parti-
tioning of the subproblems may not be optimal, and only some of the subprob-
lems may possess optimal solutions. Still, as any good scientist or engineer
knows, only by abstracting out the essentials of a situation and posing the
problem in a tractable manner can progress be made.
In fact the network planning problem is usually divided temporally as well
into short-term, medium-term and long-term plans. These plans, naturally, should
be consistent with one another. Moreover they must be periodically updated in
what Gupta [Gupt] calls a "continual iterative process."

1.7 PLANNING FACTORS

The context in which a network planner operates is influenced by a number of

factors. According to King and Premkumar [King], these include technology
factors, business factors, organizational factors, and environmental factors. We
discuss each set of factors in turn.
SECTION 1.5 PLANNING FACTORS

1.7.1 TtchNoloqy FACTORS

A key property of technology is the periodic development of new technology that,

eventually if not initially, can provide more capacity, increased reliability, and/or
reduced cost. The network provider, whether a common carrier or a private
organization, must then determine when and how to introduce this technology
into an existing network.
Examples of the gradual displacement of older technologies include the use
of digital technology in place of analog technology and the use of fiber optic links
in place of electrical links. Some new technologies live alongside older technol-
ogies, as in the case of transoceanic satellite channels and undersea cables.
Finally, when there is no existing infrastructure—as in large parts of today's
world—there may be a technological solution such as wireless technology.
There are a plethora of issues that face the network planner as he or she
considers a network upgrade. It is obviously necessary to ascertain the point at
which the benefits of the upgrade outweigh the costs of deploying new technology.
But there are other issues, such as the economic benefits of automation, the
amount of importance attached by the network planner to integrating voice, video,
and data over a single network, and the perceived desirability of open systems and
standards [King].

1.7.2 BUSINESS FACTORS

Let's consider a common carrier service provider first. A key business factor is
access to capital for building network infrastructure. A good example of this is
the necessity to secure financing by LEO (low earth orbit) satellite providers such
as Motorola's Iridium system. While such LEO systems have a certain techno-
logical glamour attached to them, the need to pay for infrastructure development
applies equally to more mundane telecommunication systems.
There are a number of issues faced by a network planner at a common
carrier. In many cases the cost of networks must be traded against their reliability.
For instance, the widespread introduction of fiber optics has meant that a network
may be able to function with fewer transcontinental links, thus saving on
construction costs. However, the failure of one of the small number of remaining
links can be catastrophic.
The introduction of new technology, once financed, can lower costs and
increase reliability. However retiring old equipment may present a loss if an
extended time in service had been contemplated.
Free market competition presents a wild card that can upset the best laid
plans. The uncertainty in user demand for new services and the cost of deploying
new services can present a situation that is akin to asking what came first—the
chicken or the egg? This last problem gives weight to the argument for integrated
4 CHAPTER 1 THE NETWORK PLANNING PROBLEM

network technologies such as ATM. A truly integrated network could support

new services without the expense of dedicated specialized networks.
A common carrier network planner also faces issues of access to markets,
spectrum allocation for wireless services, market penetration, and revenue
maximization.
A user organization faces many of the same issues as a common carrier in
terms of trading cost against reliability and the introduction of new technology. A
user organization in today's business environment may have many common
carriers and vendors to choose from for services and equipment. Determining the
"best" offering can be a significant task in itself.
Finally, for user organizations that are profit driven, King and Premkumar
[King] have stated that various strategies for gaining competitive advantage can
be used in conjunction with properly planned telecommunications. These
strategies include product differentiation, internal cost reduction, improved acces-
sibility to markets, spawning new businesses, and changing competitive scope.

1.7.7 OnqANizATioNAl FACTORS

A great deal has been written over the years on the best ways in which
organizations should operate. This literature is not recapped here. However we
make a few specific points for common carrier service providers.
Perhaps not in the short term, but in the long term, organizations must exploit
new technologies to survive and prosper. Thus it is important to guard against the
development of a "not invented here" culture in a common carrier organization.
The skepticism with which satellite communication and packet switching were
greeted with by some in the telecommunications industry is an example of this.
More recently, with the increased competition in the telecommunications
industry, one may wonder whether telecommunications is driven by technology or
cost. A technology-driven organization can be expected to allocate substantial
resources for research and developments efforts. A cost-driven organization may
view telecommunications as a commodity that must be brought to the public,
organizationally, at the lowest possible cost. Naturally the telecommunications
world is more complex than this. Successful research and development can
reduce costs and, in the drive to reduce costs, it is sometimes necessary to use
new technologies. Still, an organization's mind-set can influence the decisions
that are made.
For a user organization [King], there are often noted similarities between
information (computer) processing and telecommunications activities. This
similarity has led to efforts at integration—both technologically and organiza-
tionally. In any case, the network planner at a user organization must be
knowledgeable about a broad mix of technologies. Resources (i.e., the amount
of money available for network operations and upgrades) form a key constraint in
the duties of such a planner. Two important concerns of network planners in user
SECTION 1.4 TYPES OF PLANNING 5

organizations are the visibility of networking in the organization and the

involvement of top managers in key decision making.

1.7.4 ENVJRONIVIENTAL FACTORS

In this context, environmental factors are taken to be regulatory, economic and

health questions. The environment a network planner at a common carrier service
provider operates in is quite complex. The regulatory environment varies from
country to country. The spectrum for wireless services in particular is heavily
regulated. Attempts to "deregulate" telecommunications in recent years have led
to a regulatory environment that is simpler, but more competitive. In the United
States it is expected that telephone and cable companies will be competing for
what was traditionally viewed as each other's core business.
The economic environment adds another layer of complexity to the network
planner's role. The economic environment affects the resources available for
network operation and expansion, the cost of money, and the money available for
customers to spend on telecommunications. The economic environment varies
from country to country, and its future evolution is hard to predict. Naturally the
economic environment is linked to the competitive environment.
Finally, there can be public health concerns such as those recently raised
concerning wireless technology and high frequency electromagnetic fields.
Many of these same issues affect—either directly or indirectly—user
organizations, where the economic environment appears in the form of budgetary
concerns. Transborder information flow regulations cause concern in multi-
national organizations. Standards are important for user organizations desiring
multiple vendor sources for critical equipment.

1.4 TYPES OF PLANNING

There are a variety of classification approaches for different types of network

planning. We partially follow the authors already cited [Gupt] [King] in listing a
number of possibilities for common carrier service providers and user organizations.

1.4.1 By LEVEI OF DETAU

• Administrative planning Goals, long-range plans, financial policy, regu-

latory efforts, forecasting, technology trends, strategies for competitive
advantage
• Fundamental technical planning Plans for network, management,
switching and routing, addressing, signaling, operations, provisioning,
and maintenance
• Engineering Detailed and immediate plans
 CHAPTER 1 THE NETWORK PUNNING PROBLEM

1.4.2 By NETWORII COMPONENTS (TeUphoNE NETS)

• Local exchanges
• Toll exchanges
• Interexchange transmission
• Loop plant
• Signaling network
• Customer premises equipment

1.4.7 By NETWonk COMPONENTS (COMPUTER NETS)

• Local area networks

• Packet networks
• PCs, workstations, minis, mainframes
• Routers, bridges, gateways
• Links (cables and wires, fiber optics, radio)

1.4.4 By NETWonk SERVJCES (TeUphoNE NETS)

• Plain old telephone service (POTS)

• Narrowband ISDN service
• Broadband ISDN service
• SMDS and/or frame relay service
• Packet service
• Video service
• Cellular telephone service

1.4.9 By NETWoak SERVICES (COMPUTER NETS)

• E-mail
• Remote login
• File transfer
• Image transfer
• Voice connection(s)
• World Wide Web
SECTION 1.5 NETWORK FEATURES 7

1.4.6 By TIME

• Long term (5-20 years)

• Medium term (2-5 years)
• Short term (1-2 years)

A temporal breakdown of ways to classify plans for a user organization

appears in [King] [Clel].

1.5 NETWORK FEATURES

Typical communication networks have a number of characteristics that are

independent of the actual implementation. These include [Gupta] the following.
1.5.1 STATISTICAI LOACIS

Demand for telecommunication services is the result of the decisions of perhaps

millions of individual users as to the time at which a service will be engaged, the
type of service, and its duration. The time scales of these activities varies widely
[Gupt], from milliseconds for e-mail to minutes for a voice call to weeks for a
leased line. There are daily variations, weekly variations, and annual variations in
traffic. Certain unpredictable events such as earthquakes or storms can also
stimulate significant demand.
There is a large body of literature available on the characterization and
prediction of this demand using statistical techniques. Among the most relevant
techniques are stochastic processes and queueing theory [Gros] [Klei 75] [Robe
94]. Stochastic processes usually deal with the evaluation of series of events over
time. Queueing theory, named for the British word for a waiting line, is the study
of things waiting in a line. These things could be packets in a switch or calls in an
exchange. Queuing theory is widely used to predict the performance of electronic
systems.

1.9.2 LARQE NuivibER of SubNETWORks ANCI SERVJCES

Most large "networks" are composed of a variety of subnetworks. This is

particularly true of the Internet, which is an amalgamation of hundreds of smaller
networks. But it is also true of large telephone networks. Here one has local
networks, regional networks, long-distance transcontinental networks, and inter-
national networks. Some portions of these networks are segmented off into
private corporate and government networks [Gupt].
Most existing large networks also offer a wide variety of services, some of
which were outlined in Section 1.4. The large variety in subnetworks and services
requires dedicated and specialized planning.
 CHAPTER 1 THE NETWORK PUNNING PROBLEM

1.9.7 GROWTII

While POTS (plain old telephone service) is a relatively mature technology with
low growth rates in developed countries, there is the potential for rapid growth in
underdeveloped countries. Moreover, while a market may be static in overall
growth, individual service providers and geographic areas may experience large
growth rates in the process of attracting new customers and residents, respec-
tively.
Certain parts of the telecommunications market have experienced rapid
growth in recent years. This is particularly true of wireless technology such as
cellular telephones. Wireless technology has experienced rapid growth in devel-
oped countries because of its convenience and in underdeveloped countries
because of the need for minimal infrastructure. During the 1980s the number of
local area networks deployed grew rapidly. It remains to be seen whether ATM
technology will follow this pattern. Internet traffic has experienced phenomenal
growth over the past decade with no end in sight.
Rapid growth in networks makes it necessary to install higher capacity
systems and is an important motivation for proper network planning.

1.9.4 HARCIWARE AN<I SOFTWARE VARIETY ANCI AdvANCE

A large network has a large variety of installed hardware and software with
differing dates of manufacture. This is because of technological change (which
makes new equipment cost effective) and the large investments involved (which
lead to the use of equipment for extended periods of time). Naturally, this
diversity of hardware and software, while economical, complicates network
management and planning.

1.5.9 TEchNoloqicAl CkANqe

New technologies are being deployed into networks at an increasing rate. The
coexistence of competing technologies is a reason for caution on the part of the
network planner. On the up side, new technologies can promise reduced operating
costs, enhanced capabilities, and increased reliability. However, it may not make
sense to invest in a new technology unless one is sure that it has staying power in
terms of market share, can be supplied by multiple vendors, and really does offer
some or all of the advantages cited above.

1.9.6 iNVEsriNq JN TIIE FUTURE

Ideally telecommunications network infrastructure should last 8 to 20 years

before requiring replacement. Thus the choice of hardware and software requires
careful consideration on the part of the network planner.
SECTION 1.6 THE REST OF THIS BOOK 9

1.9.7 STANdARds

In the network planning process the role of standards is important. Open

standards allow a variety of vendors to bring compatible hardware and software
to market at low prices. However, the timing of standards is important.
Standardization at too early a point may suffer from incomplete research and
development experience. Standardization at too late a point may be irrelevant if
large number of users are locked into proprietary products and standards. A
successful standard gives assurance to the network planner that hardware and
software will be readily available on a long-term basis.

1.9.8 ECONOMIES of SCAIE

The phenomenon of economies of scale is responsible for the tendency of larger

systems to be more cost effective than smaller systems. That is, while a larger
system will cost more than a smaller system in total investment, the rate of
increase actually decreases as systems become larger. Thus, on a per-user or per-
circuit or per-packet basis, the larger system is actually cheaper.
Naturally, this doesn't imply that organizationally, larger is necessarily better.
But in terms of specific telecommunications systems, such as transmission
facilities or switches, savings due to economies of scale can be quite real.

1.9.9 FINITE RESOURCES

The resources available to a network planner for network modernization are

usually limited and in fact can be significantly less than ideal. This serves to
encourage a conservative approach to network planning. Awareness that
resources are finite also provides a powerful incentive to try to install the right
amount and type of equipment at the right time and place. Finally, the presence of
such limitations necessitates the involvement of upper management in the
planning process.

1.6 THE REST OF THIS BOOK

This book gives a tutorial look at the fundamentals of the theory of network
planning. Chapter 2 discusses the use of mathematical programming for planning.
Mathematical programming is widely used in large telecommunications compa-
nies to reduce costs. Chapter 3 examines a variety of other algorithms for network
planning. Reliability theory is discussed in Chapter 4. This theory is used in a
variety of applications including network planning. While general text-length
treatments are available, key concepts are discussed in Chapter 4. Chapter 5
introduces key software technologies as well as a number of new optimization
10 CHAPTER 1 THE NETWORK PLANNING PROBLEM

methodologies useful for network planning. Finally, Chapter 6 covers data

analysis techniques that are network planning related.

1.7 PROBLEMS

1 What is network planning?

2 Why is network planning called a "continual iterative process"?
3 Why can network planning be viewed as an optimization problem?
4 Why is it impossible to pose the network planning problem for any
reasonably sized organization as a single optimization problem?
5 Why is the network planning problem divided into smaller subproblems?
6 What criteria must an organization balance through network planning?
7 Give some examples of technologies that have displaced older technol-
ogies.
8 What are some of the issues that face a network planner who is
considering a network upgrade?
9 Why is new technology periodically developed?
10 What are some of the business factors faced by a planner at a common
carrier? At a user organization?
11 How does the uncertainty in demand for new user services affect network
planning?
12 What are some strategies for gaining competitive advantage for organiza-
tions that are profit driven?
13 Why must organizations exploit new technologies to survive and prosper?
14 What is the difference between a telecommunications organization that is
cost driven and one that is technology driven?
15 What are some important concerns of network planners in user organiza-
tions?
16 Describe the operating environment of a network planner in a common
carrier. A user organization.
17 Differentiate types of planning by level of detail.
18 How are types of planning for network services and for network
components related?
19 Why are network loads considered statistically? What are some of the
relevant techniques used?
SECTION 1.7 PROBLEMS 11

20 Why does a large network have a multiplicity of subnetworks and

services?
21 Can there be organizational growth in a static market?
22 Why is there usually a large variety in the base of installed hardware and
software?
23 How does technological change relate to network planning?
24 How long should telecommunications infrastructure be expected to last
before it must be replaced?
25 What are some of the trade-offs between introducing technical standards
too early and too late?
26 What is the concept of economies of scale?
27 What are some of the effects of the limitations on resources available to a
network planner for network modernization?