See discussions, stats, and author profiles for this publication at: [Link]

net/publication/248386765

Real Options in Telecom Infrastructure Projects — A Tutorial

Article in IEEE Communications Surveys & Tutorials · January 2014

DOI: 10.1109/SURV.2013.062613.00126

CITATIONS READS
23 4,343

7 authors, including:

Mathieu Tahon Sofie Verbrugge

Ghent University Ghent University
23 PUBLICATIONS 171 CITATIONS 218 PUBLICATIONS 1,991 CITATIONS

SEE PROFILE SEE PROFILE

Didier Colle Mario Pickavet

IMEC - Ghent University Ghent University
641 PUBLICATIONS 9,282 CITATIONS 582 PUBLICATIONS 10,081 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Sofie Verbrugge on 15 May 2014.

The user has requested enhancement of the downloaded file.

 1

Real Options in Telecom Infrastructure Projects

- A Tutorial
Mathieu Tahon, Sofie Verbrugge, Peter J. Willis, Paul Botham, Didier Colle, Mario Pickavet,
Piet Demeester, IEEE Fellow

Abstract—The rapid technological change and uncertain future license is a straightforward example as it offers the flexibility
evolutions have a large impact on investment projects in the to decide when and where to roll out the mobile network. The
telecommunication sector. When new infrastructure networks are 4G mobile operator can start with a study period, testing the
rolled out, the initial assumptions can prove to be untrue in the
future, severely impacting the payoff. It is therefore extremely new technology in small areas. When the uptake of 4G services
important that projects offer flexibility to allow the management proves to be exceeding initial expectations, extra investments
to react to unforeseen changes. Management must, for example, can be made to speed up the rollout of the nationwide
be able to decide to speed up the project, slow it down, or even network. On the other hand, when a telecom project proves
completely abandon it. However, the standard method used to to be unprofitable, the management can decide to abandon it
evaluate investment projects, the Net Present Value analysis, is
unable to capture the value of these different flexibility options. completely. For example, only one year after its launch, British
The Real Option concept, derived from financial literature, was Telecom decided to stop its mobile broadcast TV service in
proposed as a solution and implements this flexibility in the 2007.
standard calculations. However, the Real Option Theory is only All investment problems are economically assessed before
slowly getting accepted within the telecommunication sector. In they are started. In general, this analysis consists of predict-
this paper, we introduce the basics of real options theory and
provide a practical methodology to apply real options to realistic ing the future costs and revenues of the investment project,
telecom business cases. In addition, we will indicate why the discounting them with an appropriate discount factor and then
characteristics of this sector make it very well suited to apply adding them to come to the Net Present Value (NPV) [1].
real options to investment projects. The rollout of fixed next When this NPV is positive, the project is assessed as profitable.
generation access networks offers a broad range of growth This approach is typically followed by network planners.
options to the operator, e.g. additional network upgrades or the
introduction of new services. Using real options allows one to However, conducting such a standard NPV analysis can yield
compare the flexibility value of all these options. unintuitive results. Network solutions that are thought of as
more flexible or less risky turn out to be less economically
Index Terms—Flexibility, Next generation access networks,
Real options, Techno-economics interesting according to the NPV analysis. A wireless access
network design that can be expanded or contracted for lower
cost is more flexible in handling uncertain future customer
I. W HY TELECOM REQUIRES AN EXTENDED ECONOMIC
demand, but is typically more expensive in initial deployment.
EVALUATION APPROACH
Thus, the question arises how the impact of uncertainty, risk
In the last decennia, the telecommunication industry has and flexibility can be implemented in the standard feasibility
shown rapid growth in technology, products and services, and analysis of the project. In the standard NPV analysis, two
this evolution is still ongoing. For example, the rollout of fixed drawbacks can be identified. First, the standard method does
and wireless Next Generation Access Networks (NGAN) like not indicate the impact of uncertainty on the analysis. Two
Fibre to the Home (FttH) and Long Term Evolution (LTE) is extensions covering this impact exist, scenario analysis and
currently drawing a lot of attention from operators, vendors sensitivity analysis. In a scenario analysis, the investment
and regulators. However, deployment of these NGANs is not project is assessed in a small number of possible scenarios.
yet observed or it is happening slower than expected due to While NPV analysis offers only one view on the future,
the risk associated with upfront investments. Another issue for scenario analysis compares several alternative futures. For
new technologies is the uncertainty linked with them. Doubts instance, an application provider could compare a scenario
about customer adoption, costs and technology performance of low, normal and high customer uptake. A scenario analysis
are only a few of the uncertain factors. approach can also consist of comparing different investment
However, it is untrue that this risk and uncertainty within projects to assess them based on economic feasibility. Scenario
the telecom sector cannot be managed. Managerial flexibility analysis has been applied to different cases in telecommuni-
allows the different actors in the market to respond to un- cation research [2], [3].
foreseen effects during the project lifetime. Acquiring a 4G A second extension is the sensitivity analysis [4]. While a
Manuscript received June 22, 2012. scenario analysis only studies a few possible scenarios, this
M. Tahon, S. Verbrugge, D. Colle, M. Pickavet and P. Demeester are with method analyses the impact of uncertainty in the input factors
the Internet Based Communication Networks and Services group of the Ghent on the output of the analysis. In a scenario analysis, the input
University, ([Link]@[Link], phone: +32 9 33 14891, fax: +32
9 33 14899 values only take some discrete scenario-dependent values, like
P. J. Willis and P. Botham are with British Telecom, Ipswich, UK. low and high market potential. In the sensitivity analysis, this
 2

input is extended with a statistical uncertainty distribution. It definition of real options is given in [11]. "Real options is
allows one to systematically change variables in the model to a systematic approach and integrated solution using financial
determine the effects on the final result. In techno-economic theory, economic analysis, management science, decisions sci-
research within telecoms, the sensitivity analysis has been used ences, statistics and econometric modelling in applying options
in different papers [3], [5], [6]. theory in valuing real physical assets as opposed to financial
The second drawback of the NPV method is the lack of assets, in a dynamic and uncertain business environment
flexibility [7], [8]. The project is seen as a now or never where decisions are flexible in the context of strategic capital
decision, with no possibilities for the decision makers to alter investment decision-making, valuing investment opportunities
the project during its lifetime. In a realistic business model, and project capital expenditures." As this definition states, the
this condition is not fulfilled. real option theory is based on the option concept as used in
While the previous two extensions have been proposed to financial markets. A financial option is defined as the right to
capture the value of uncertainty, only Real Option (RO) Theory buy or sell an asset for a predefined price during or at the end
captures the value of managerial flexibility in practical cases. of an agreed period. When the option can only be exercised
In addition, the concepts offered by this theory make it also of at the end of the period, it is a European option. In the other
great value for non-financial specialists, as it helps to identify case it is an American option. Hybrid options also exist; the
and catalogue intuitive notions of flexible design. option can be exercised on several dates during the agreed
Different frameworks on how to apply real option theory period. These are categorized as Bermuda options. Next to a
have already been proposed, with [8], [9] both offering a prac- differentiation between options based on the time they can be
tical approach to real options. We will show why investment exercised, they can also be divided into call and put options.
projects within the telecommunication sector are well suited While a call option is the right to buy, a put option refers to
for a real option analysis. The different examples given in this the right to sell an asset. Other additional terminology from
introduction already indicate how suited the telecom sector is option theory is the option price and strike price. The first
for applying real options. Since NGAN rollout is drawing a lot is also known as the option premium, or the price to acquire
of attention from network operators, we developed a realistic the option. The latter refers to the price to exercise the option.
business case for the upgrade of the current copper access Options on options also exist and are called compound options.
network towards Fibre to the Cabinet (FttC) and Fibre to the During the time period before the exercise price, an option
Premises (FttP) in the UK. The standard business case was can be in, at or out of the money. Assume an American call
extended with a broad range of options found in the literature. option, the right to buy a stock for a predetermined price X.
In the following section, we will introduce the theoretical In addition, consider that currently the value of the stock is S.
background and categorization of real options. An overview When S < X, the option is out of the money and it is useless
of the application domains of real options in the telecom- to execute the option today, since it is more interesting to buy
munication sector is given in Section III. Next, Section IV the stock on the market. However, this does not mean the
elaborates on a practical Real Option Analysis methodology option has no value. As long as the option is not expired, the
to evaluate the flexibility value in realistic cases. Next, we underlying asset can go up in value. This probability of S >
will introduce the example case study, namely the upgrade of X, or the option being in the money, on the final exercise date
the current copper access network towards FttC and FttP. In of the option, results in a value for the option. Obviously, the
Section V, the value of extending a standard techno-economic longer before the exercise date, the higher the probability the
evaluation with an RO analysis is demonstrated using the case option will be in the money on this date. As such, the value
study proposed above. Finally, Section VI summarizes the of an option increases with the time left to the final exercise
most important advantages of an RO analysis, together with date. At this date, either S < X, and the option will expire,
some more detailed conclusions drawn from the case study. having zero value or, when S > X, the option will be exercised
with a value of S - X. In summary, the value of an option on
II. BACKGROUND ON REAL OPTIONS exercise date equals MAX(0,S - X).
A. Real option basics Transferring the financial option concept towards business
For telecom projects, the feasibility of new project proposals investment decisions is quite straightforward. For an introduc-
is assessed through a techno-economic analysis. Verbrugge et tion to the foundations of real option theory, we refer to [8],
al. [10] propose a clear and practical methodology to conduct [9], [12], [13].
such an analysis. It consists of four steps, covering input Making an initial investment typically results in future
collection, cost and revenue modelling over the business case flexibility during the entire investment lifetime. RO analysis
analysis to its different extensions. The third step of their implements this flexibility in the previous static NPV calcula-
methodology consists of a standard business case analysis, tion. For example, the initial static NPV analysis showed the
the NPV analysis. We refer to their work for an in-depth rollout of an LTE network to be profitable under certain uptake
description of techno-economic modelling. However, we al- assumptions. This may no longer be the case after a few years.
ready indicated the flexibility shortcoming of this tool in the The static NPV analysis does not allow any flexibility here but
introduction. a RO analysis offers the possibility to abandon the project and
To implement the value of this flexibility, the real option sell the license. Conducting the RO analysis calculation will
concept was derived from financial literature. An excellent result in the value of this option.
 3

B. Real option categories has the option whether or not to go with the new technology.
Different examples of real options were already introduced For example, in Belgium Telenet decided to wait with the
in the previous sections. In general, these examples can be sub- nationwide rollout of LTE and started testing it on a small
divided into three distinct categories, namely growth, shrink site.
and learning options. The category of growth options are Next to technological uncertainty, adoption of the new
related to possible follow up investments during a later stage product is also a problematic parameter. Before introducing
in the project. When telecom projects are concerned, examples a new product, management can only make an educated guess
of growth options are the expansion of the network to adjacent about the market potential and adoption speed. A wait and
regions, a technology upgrade from ADSL to VDSL2, or even see strategy can therefore be interesting. During this period,
an extension of the product portfolio from double play to customer surveys can offer extra insight into the market. For
triple play. The shrink option category consists of the opposite example, in the wireless broadband market, only 0.40% of the
type of options. When the initial assumptions overestimated world population uses mobile broadband, but this is region
the consumer adoption or technology evolution makes some dependent [14]. Instead of hitting the national market with
products redundant, management has a disinvestment option. mobile broadband offers, an operator could use a wait and see
A project can be completely abandoned, like the mobile strategy and perform customer surveys to gain better insight
TV broadcasting service from BT. Regarding the telephony into the customer demands.
market, the ISDN product was withdrawn when it had no The last uncertain parameter that can offer study/start pos-
more potential. Learning options are a specific type of options, sibilities is the regulatory evolution. In the Fibre to the Home
where investments are postponed until extra information or (FttH) debate, uncertainty about the future regulatory actions
experience is gained. Conducting market studies or rigorous taken by the European Commission in the local loop access
testing of a new technology before its implementation are only postpones the rollout of fibre networks in Europe.
two examples.
The most well-known real options categorisation is the Description Telecom examples

7S framework by Copeland and Keenan [7]. The different Sequential investments in a

Phased network
rollout
SCALE UP later stage as market grows
real options categories are summarized in Fig. 1 and some Acquiring licenses

Switch products, process or Technology upgrade,

typical telecom examples are added. The different categories SWITCH UP plants given a shift in underlying e.g. ADSL to VDSL2
price or demand or 2G to 3G
are described in the following paragraphs. INVEST
GROW Enter another industry when
Triple play
1) Scale up and down options: The scale of the project can SCOPE UP cost-effectively possible. Link
and leverage.
Bitstream access

be expanded or reduced. A scale down option indicates that LEARN STUDY/ Delay investments until more
information and/or skills are
Trial project
Testing of LTE
START acquired Market studies
the scale of the project is reducible. During the rollout phase, Shrink or shut down a project if Adapting network rollout
opting for a slower rollout is a form of scale down option. SHRINK
SCALE DOWN new information changes the
expected payoffs
scenario
Abandoning project

The ultimate form of scale down consists of abandoning DISINVEST

SWITCH
Switch to more cost-effective Lease wavelengths
and flexible assets as new instead of dark fiber in
DOWN
the project. In this case, revenues are gained from selling information is obtained case of lower demand
Limit the scope of operations in
the infrastructure. Under positive circumstances, the scale up SCOPE
DOWN
a related industry when there is
Withdrawal from
ISDN market
no further potential
option becomes more attractive. Rollout can be sped up, or
the zone can be expanded to neighbouring regions. Fig. 1. Overview of the 7S framework [7] and telecom examples
In literature, most of these options are applied in the
telecommunication industry. Infrastructure rollout of both 3) Switch up and down options: Next to the scale of the
wired and wireless networks and the related investment costs project, several other parameters also render flexibility. In
are one of the major topics. The scale of such projects covers a production environment, managers can choose to upgrade
large areas and both rollout area and speed can be changed machine technology during the project, e.g. to produce better
to optimize investment return. Abandoning the project due to quality products. Changing technology can prove useful during
unsatisfactory results is a special case of a scale down option, the project lifetime, but results in an extra cost at the start of
where the rollout speed is reduced to zero. One must take the project. So it is important to make the trade-off between the
into account that abandoning a project results in exceptional flexibility value and the initial cost of this flexibility. Switching
revenues from the sale of the assets. After acquiring a 3G from ADSL to VDSL is an example of a switch up option in
licence, management can abandon the rollout of the 3G fixed access markets. The consumers are offered higher speeds,
network in worst case scenarios and put the licence up for but this requires an investment by the operators. Fibre needs
sale. to be brought closer to the customer, so large deployment
2) Study or start options: Another important option for investments are typically required.
telecom related projects is the study/start option. When a 4) Scope up and down options: The last possible option is
new technology enters the market, several parameters remain the scope up or down option. While a scale option changes the
uncertain. The first one is the uncertainty linked with the geographical region and the switch option allows flexibility in
technology itself. Is it efficient enough to handle high bitrates the technology, the scope option focuses on the flexibility of
over long distances? What is the mean time between failure of the product portfolio. Management can choose to offer extra
the different components? Rigorous testing of the technology, products to the customers, or reduce their offer. The move
field tests and trials can offer more insight into the techno- towards triple play is an example of operators lifting their
logical performance and after the testing period, management scope up options.
 4

C. Application domains of real options TABLE I

R EAL OPTIONS IN THE TELECOM LITERATURE
Real options have been applied to a wide range of invest-
ment projects from mine valuation to initial public offer valu- Option Flexibility Reference Uncertainty
ation [9]. Some more telecom related examples are described Scale Rollout area [9], [15], Adoption, Costs,
up/down [17], [18], Tariffs
below. A literature review of real option application to telecom [19]
examples can be found in Table I. Speed up/slow [15], [19] Adoption, Costs,
Most of the existing literature applies real options to telecom down rollout Tariffs
infrastructure rollout [9], [15], [16], [17], [18], [19]. This Abandon [9], [22] Adoption, Costs,
rollout is related to a large investment covering several years project Tariffs
and thus allows for flexibility in the rollout path. The scale of Switch up/down [17] Firm value,
Technology Adoption, Costs,
such projects covers large areas, and both the rollout area and Regulation
speed can be changed during the project to optimize the return Scope Offering bit
on investment. Abandoning the project due to unsatisfactory up/down stream access
results is a special case of a real option. A quantitative and sim- Study/start Trial project [16] Technology,
plified example to illustrate real options is the M-commerce Performance,
Market
project, describing an investment by a telecommunications
Wait and see [16], [17], Technology,
firm [9]. This is a typical example of a scale up and scale [18] Regulation
down option. During the project, management has two options,
either expanding the project scale by 60% if expectations are
exceeded, or abandoning the project completely and reaping that new entrants should also pay for the financial risk of the
the salvage value. Another paper describing scale options in incumbent since he did invest in the network infrastructure.
telecom networks is [15]. The feasibility of Mobile WiMAX LLU in fact offers a study/start option to new entrants, while
as an alternative for fixed DSL and HFC networks is analysed, the incumbent gave up his option when he invested. In [22],
with the possibility of extending the scale of the project. this problem has been discussed in more detail. Next to large
Several rollout scenarios are studied, changing the rollout infrastructure investment cases, real option valuation theory
location from nationwide to only in urban areas and with the has also been applied to service oriented cases, e.g. [23]
option to change rollout speed. applied real options to the case of the Belgian rail operator
Study/start options have also been applied extensively to offering internet services on board.
telecom network problems. In [16], the rollout of a WiMAX
network in Eindhoven is studied. Before starting the complete III. METHODOLOGY
rollout, the operator has the choice to do a field trial to analyse In this section, the methodology commonly used to perform
the technological performance. In the second phase, based on real option analyses is discussed [8], [9]. However, before
the results from the trial phase, the operator can decide to the RO analysis can be conducted, the business case must
invest or abandon the project. be assessed on three conditions. First, there needs to be
Licences for wireless networks are known to be very ex- uncertainty in the project. During the standard NPV evaluation,
pensive, so it is important to correctly evaluate the licence some assumptions influencing the future costs and revenues
investment. For example, in the UK, 35 billion dollars was have been made. However, some of these assumptions come
paid for the 3G licences. In [20], the authors try to estimate with a certain degree of uncertainty. Future customer uptake,
the value of these licences based on a real option approach. the future price of raw materials and components can only be
Buying the 3G licence resulted in acquiring a strong market estimated. When this is the case, the project meets the first
position and a broad range of options, including scale up, condition. Secondly, the project should offer some kind of
switch up and down and temporarily halting the project. This flexibility. This flexibility can easily be recognized if one of
research showed that with the correct valuation techniques, the options in the 7S framework is present in the case. Such
the value of the 3G licence was close to the price paid for flexibility allows the decision maker to counter the uncertainty.
the acquisition. Spectrum management is closely linked with The last condition concerns the timing aspect. A real option
telecom licenses. Dynamic spectrum management, with a two analysis can only be performed if the investment decision
stage assignment through the use of options was proposed in covers a two (or more) phased project. An initial decision is
[21]. The option concept allowed calculating the penalty value made at the start of the project, but extra decisions can be made
and the overbooking ratio. during later stages of the project. For example, an operator can,
A lot of research has been performed on the impact of after completion of the first part of the network, still decide in
regulation on investment decisions by network operators. later stages what his next steps will be. Will he do nothing or
Regulatory bodies imposed local loop unbundling (LLU) on extend the network to other regions? After the case has been
the incumbent operators to improve competition. For new assessed, based on these three conditions, a clear methodology
entrants, LLU has the advantage that they do not have to make needs to be followed to perform the RO analysis. In this paper,
large investments in network infrastructure before they can we use the methodology proposed in [10], which is based on
offer network services. However, fixing the price for network [8], [9]. While a standard techno-economic analysis results
access is not straightforward. One should take into account in an NPV analysis, the RO analysis methodology extends
 5

• Only incoming and outgoing cash is to be taken into

Uncertainty Phased decision
Flexibility account. It is important to notice that there exists a
Can I identify Can the final
Can I counter
different future decision be difference between cost and revenue on one side and
uncertainty?
scenarios? postponed?
income and expense on the other. The yearly depreciation
of an asset is a cost, but no expense. As this is no cash
flow, it should not be included.
• Cash flows which are independent of the project should
RO
analysis?
not be taken into account. Only the incremental or
marginal cash flows related to the project are to be
1. Execute a standard NPV analysis included.
Perform a regular techno-economic analysis, as if no
real options were present.
• Cash flows are independent of the financing of the project.
2. Identify the uncertainties As a result, interest payments or dividends are excluded.
Find out which parameters impact the result the
most, and how certain their value is. Typical numbers The cost of financing is included in the required rate of
on customer adoption come with a degree of return of the project.
uncertainty.
3. Identify the flexibility • Tax cash flows are to be included, since the expenses and
Use the 7S framework to indicate how the income influence the taxable profit.
uncertainty in the project can be countered during the
project path. Once the cash flows during the investment period have been
4. Calculate the option value
Compare the value of the case with the option with determined, the calculation of the NPV is straightforward. As
the standard case. Different calculation methods the name says, it returns the present value of the future cash
exist.
flows based on a given minimum return. This return is based
Fig. 2. Conditions and methodology to perform a real options analysis
on the return requirements for both shareholders and interest
payments for loans. More information on determining r can
be found in [1]. The formula for the NPV calculation is given
the techno-economic methodology with three extra steps. The below. All cash flows (CF) of the project are discounted with
RO analysis thus consists of four steps. First, a standard NPV the minimum return r and summed up.
analysis is conducted. It is clear that the second and third steps n
of the methodology are closely linked with the preconditions.
X CFi
NPV = (1)
In essence, the second step comes down to identifying the i=0
(1 + r)i
uncertain input parameters of the project influencing the result.
The NPV indicates the value the investment creates, since it
The third step links back to the second and third condition
reflects the total value of the future cash flows, taking into
listed above and consists of identifying the options. When
account the required return. When the NPV is larger than zero,
the management has no options to act against the changing
the investment returns, in addition to the initial investment and
parameters, performing a real option analysis is pointless. To
the required return, an extra value equal to the NPV.
identify the different options present in the studied case, the
7S framework can be used. The conditions and methodology
are summarized in Fig. 2. Before indicating how the different B. Real option valuation techniques
steps work in practice by elaborating a simple toy example, As was already introduced above, a real option analysis
more detail is given on the standard NPV analysis and the always starts from the standard NPV, which is currently
different calculation techniques for real options. used by network planners. In fact, the standard Discounted
CF approach is a special case of the real option analysis,
A. Basics of the NPV analysis evaluating the project as if no flexibility is present. It is
therefore vital to start any RO analysis with a correct standard
The goal of an NPV analysis is to indicate the viability of NPV valuation. The total value of a project is expressed by
an investment project. The question an NPV analysis answers the following formula.
is the following: "Is the investment creating value for the
company and the shareholders?" An investment is basically
P roject value = N P V + Option value (2)
an expense done today, aimed at generating income later.
Obviously, this future income should be larger than the initial Three different solution methods have been proposed to
expense and generate a required surplus return. calculate the value of real options in investment projects. We
In order to conduct the investment analysis, one should will give a short description of each of them in the following
determine the cash flows generated through the investment sections.
period. This period equals the economic lifetime of the project, 1) Black and Scholes model: Since real options are derived
the time after which the investment no longer generates cash from financial options, it is logical that the calculation methods
flows. It is clear that only cash flows directly linked to the for financial options were transferred to real option valuation.
project should be taken into account. While this is a simple The mathematical Black and Scholes model is one of the most
principle, it typically is the most difficult phase in valuing the used option valuation models in the financial sector [24]. It was
investment project. The following basic rules help to determine developed in 1973 to evaluate the value of a European option.
the cash flows in any investment project. This indicates the first underlying assumption of the model,
 6

namely the option can only be exercised at the end of the time σ = project uncertainty
period. Most of the parameters of the mathematical model
are straightforward but others cannot be directly transferred rf = risk free interest rate
to investment projects. Calculating the static NPV refers to
the first step of the proposed methodology. Parameters like N = cumulative normal distribution
the exercise price and lifetime can also be directly linked
to the investment problem. The exercise price of the option The Black and Scholes formula to calculate the value of a call
for an investment project is the income from exercising the option through is shown above. When having a closer look
option, and can again simply be calculated using the standard at the two terms of the equation, the two important parts of
NPV analysis. The lifetime equals the time period (in years) the Black and Scholes model can be observed. The first term
during which the company has the opportunity to execute in returns the expected benefit of doing the investment right away,
the option. For the risk free interest rate, the return on assets while the second term reflects the value of paying the exercise
that are considered risk free is typically used. Examples of price on the expiration date, weighted by the probability of
such assets are German or US government bonds. However, exercising the option. The formula also indicates the impact
the parameter posing most problems is the project uncertainty of time on the option value. Increasing t will result in a higher
(σ), expressed in percentage terms. For financial assets, this is d1 and a smaller d2 , resulting in a higher option value.
linked with the volatility of the underlying asset, e.g. stock or In order to calculate the value of a put option through
oil prices. As these options, or the underlying assets are traded Black and Scholes, the concept of call-put parity for European
on financial markets, it is easy to calculate this volatility. options can be used. This parity states that the sum of the
For real options, where this market is absent, this calculation value of a call option and the present value (PV) of the strike
cannot be made and should be estimated. Estimating this value price equals the sum of the value of a put option and the
for an investment project is not that straightforward. What current value of the underlying asset. For more background
is for example the project uncertainty of a wireless license on financial option valuation information, we refer to [1].
purchase and the investment in base stations? Another assump-
tion of the Black and Scholes model is that the logarithm of
the NPV follows a Brownian motion. Again, for stocks this C + P V (X) = P + S (4)
is a reasonable assumption, but not for investment projects.
These drawbacks make this calculation method less suited for Application of Black and Scholes to real option valuation
realistic business cases. As a result, the Black and Scholes
model outcome overestimates the value of the real option. In this simplified illustrative example, the Black and Scholes
In addition, there is an important difference between finan- formula will be applied to the valuation of a put option. A
cial and real options, which results in Black and Scholes being telecom operator bought a license for e3.1 million, valid for 5
less accurate for real option valuation. Financial options are years. The expected future cash flows during this period can be
by definition independent of each other. Exercising a call or found in Table II. Conducting the NPV analysis with a required
put option has no influence on the value of other options, or return of 10%, results in cumulative future expected cash flows
on the value of the underlying asset. Real options typically of e2.975.339, insufficient to cover the initial expense, and
do interact. In a simple example, a company has a scale up thus a negative NPV of e-124.661. According to this analysis,
option to expand a factory and a scale down option where the project would not be executed. However, the operator has
the factory is sold. When executing the scale down option, the option to sell the license back after one year for e2 million.
the scale up option loses its value. As Black and Scholes As stopping the project is a clear put option, both the Black
calculates the value of an option portfolio as the sum of the and Scholes formula for a call option and the call-put parity
values of the independent options, this cannot be translated to will be applied here.
a real option portfolio. These drawbacks make the Black and In the first step, the required parameters are calculated or
Scholes formula less suited for real option valuation. estimated (Table III). Above, the expected future cash flows
(S), the lifetime of the option (t) and the exercise price (X)
were already given. In addition, the formula requires the risk-
Option value = S · N (di ) − X · e−rf t · N (d2 ) (3) free interest rate (Rf ) and the volatility of the underlying cash
flows (σ 2 ). As already indicated, for Rf the return on risk free
With: government bonds is typically used. However, the volatility of
2
S
ln X + (rf + σ2 )t the expected cash flows is much harder to estimate. Here, a
d1 = √
σ t
√ value of 50% is used. Choosing a high value indicates that
d2 = d1 − σ t the project is very risky and the prediction comes with a
S = future cash flows large degree of uncertainty. With these parameters, the Black
and Scholes formula for a call option returns a value of
X = exercise price e1.191.295.
Using the call-put parity, the value of the put option in this
t = option lifetime example can be calculated straightforwardly. The PV of the
exercise price is e1.902.459, resulting in a put option value
 7

TABLE II TABLE IV
YEARLY CASHFLOWS FROM LICENSE COMPARISON OF VALUATION METHODS

year Cash flow

Year 1 e582.000 Technique Pro Con
Year 2 e687.000 Simple to use Parameter estimation (volatility!)
Black and
Year 3 e821.000 Scholes Spreadsheet calculation Option portfolio valuation
Year 4 e929.000 Discrete choices No continuous uncertainty
Year 5 e1.010.000 Binomial Spreadsheet calculation What about additional
tree Intuitive uncertain parameters?
TABLE III Realism
BLACK AND SCHOLES INPUT PARAMETERS Based on typical spread-
Requires advanced software
Monte Carlo sheet model
Estimating uncertainty
Parameter Value Option portfolio valuation
S e2.975.339 Intuitive results
X e2.000.000
t 1
σ 50 % down to maximizing payoff, this is quite straightforward. After
Rf 5% indicating all uncertain input parameters with an appropriate
probability distribution, the Monte Carlo simulation can be
conducted. Choosing these probability distributions for the
of e260.067. The total value of the project now equals the input parameters is the most delicate task in the Monte
sum of the NPV and the option value, e135.406. Carlo simulation. For every simulation, the input parameter
It is important to notice that the estimation of the project is randomly sampled from the defined probability distribution
uncertainty has a major impact on the option value. If the and the best project path is selected. The NPV is calculated
future cash flows are assessed as less uncertain, and the for thousands to hundreds of thousands of possible combi-
operator uses a volatility of 25%, the option value drops to nations of input parameters within the predefined distribution
e118.415. With this option value, the total project remains boundaries. As indicated, the model automatically selects the
value destroying. best option in each scenario. The result from a Monte Carlo
The effect from timing on the option value was already analysis is a probability distribution of the expected payoff.
indicated above. The longer the time before expiration, the From this distribution, an extended NPV can be derived, to-
higher the probability of the option becoming in the money. gether with the option value for the studied case. This extended
In this example, if the operator can wait two years instead of NPV is the average of the probability distribution, while the
one before he has to make the decision to abandon or continue option value is the additional value of this average compared
the project, the put option value rises to e432.787. to the standard NPV. Additional information that can be drawn
2) Binomial tree model: The binomial tree model is a from such a probability distribution is the impact of the option
discrete time model. A binomial tree model is applicable to on the risk associated with the project. Typically, an option
simple processes. The main assumption is that the uncertain decreases the probability of a low payoff, and increases the
input can only take discrete values. This allows modelling the probability of more positive result. Existing software solutions
problem by a tree structure. The main assumption results in exist that allows extending an existing spreadsheet techno-
both the greatest advantage and disadvantage of the model. economic analysis with specific uncertainties and conduct the
An uncertain parameter only taking discrete values largely Monte Carlo analysis [26]. More information on Monte Carlo
simplifies the analysis, but realistic cases are generally sub- basics can be found in [27].
ject to continuous uncertainty. Detailed examples using the 4) Comparison of the valuation methods: The three valua-
binomial tree method can be found in [9]. The toy example tion methods introduced above each have their advantages and
used to indicate the methodology below is an application of disadvantages. The most important are listed in Table IV.
the binomial tree model method.
3) Monte Carlo simulation: The Monte Carlo simulation is
the last calculation method we will discuss. While the two pre- C. The methodology in practice: a simple example
vious models allow for a simple option value calculation, they Before moving to a realistic application of RO analysis, the
both have their own drawbacks. Their underlying assumptions methodology is applied to a toy example, to allow the reader
do not always match reality. A Monte Carlo simulation solves to become familiar with the different concepts. By following
these problems but results in a more complicated calculation the four-step methodology, the toy example will indicate how
method. Sawilowsky defines the Monte Carlo simulation as uncertainty and flexibility can be identified, categorized and
a repeated sampling to determine the properties of a phe- quantified. The following investment project is considered. An
nomenon [25]. entrepreneur has to decide today if he starts an online business
To perform a Monte Carlo analysis, spreadsheet based so- or not, but due to uncertain market perspectives, he does not
lutions exist. In general, these consist of extending a standard know exactly how many customers will be willing to buy
NPV analysis with the existing options. Since an option comes his product online. The entrepreneur believes the probabilities
 8

of a small or large market equal 50%. To host all his client e195. It is now straightforward to get the option value from
data our entrepreneur has the opportunity to keep his current this analysis. Compared with the standard NPV case analysis,
slow server or buy a fast server. Notice that in this short case the RO analysis returns a RO Value which is e5 higher. This
description, two of the three different conditions to perform is exactly the value of the option to wait.
a RO analysis are present. We will discuss all conditions in
more detail.
• Uncertainty
The entrepreneur is not sure about his customer potential
and the revenues related to these customers. He estimates
that there is a 50% chance of high sales and a 50% chance
of low sales.
• Flexibility
The entrepreneur has two choices. Either he buys a new
server for e60, or he keeps his current slow one.
• Phased process
Looking at this case, we do not see two phases in
the investment process. However, nothing forces our
entrepreneur to decide today if he buys the new server.
He can decide today to start with the online business and
only invest in a faster server next year.
The results of the first step of the methodology, the standard Fig. 3. A simple example - Step 1: NPV analysis.
NPV analysis, are presented in Fig. 3. Where the entrepreneur
installs the new server, his payoff is the weighted average of
e40 and e340, or e190. The additional cost for the server was
already subtracted from the expected revenues. In the other
case he will only gain e100. The standard NPV analysis thus
indicates that the entrepreneur should buy the fast server today,
since this maximizes his payoff. Notice that in order not to
overcomplicate the toy example, the required return was set
to zero.
The second step, identifying the uncertainties in the case,
was performed when describing the three preconditions. The
entrepreneur is uncertain about the customer uptake of his
service and on the type of server to install. When checking the
third condition, the entrepreneur has the flexibility to wait and
postpone his investment decision until he has more information
on the customer uptake. For the investment decision, he has
the choice between keeping his slow server and switching to a
faster one. The value of the real option can now be calculated. Fig. 4. Step 4: Value of the option to wait
We start by analyzing the different scenarios under the
uncertainty. In case there is a low customer uptake, not
investing in the new server has the best payoff. In case of IV. MIGRATION TO FIBRE: STANDARD BUSINESS
high customer uptake, installing the fast server clearly returns CASE ANALYSIS
the best result. To indicate the power of real options on realistic business
Now remember this project consists of two stages. When cases, the RO analysis technique will be applied to a telecom
identifying the flexibility, we indicated that the entrepreneur infrastructure network project. The studied case consists of
had the option to postpone his server investment decision until the rollout of a fibre access network in the UK [28]. FttH
he had gained extra information on the customer uptake. What networks are the final stage in the continuous upgrade of the
is now the value of the option to wait? If he waits, he will copper access networks. However, many networks still require
be able to better assess the customer uptake on the day he upgrading towards FttC networks. It is this infrastructure
makes the investment decision for the server. Waiting ensures investment that is considered in this paper.
the entrepreneur will make the best decision in the future. An incumbent currently possesses a nationwide copper ac-
If he notices a low uptake he will keep his original server, cess network, which has already been upgraded towards Fibre
in the other case he will buy the fast one (Fig. 4). In both to the Central Office. This allows offering ADSL services
the low and high uptake case, he chooses the scenario having to its customers. In order to offer higher access speeds to
the highest payoff. With the option to wait, our entrepreneur its end customers, the incumbent has decided to upgrade its
knows he has a 50% chance on a payoff of e340 and 50% network towards FttC, allowing it to offer VDSL services.
chance on a payoff of e50, or a total value of the project of Two important upgrades are necessary in the access network to
 9

migrate towards an FttC network. First, fibre has to be installed 1) Service adoption modelling: Modelling the adoption of
between the central offices and the street cabinets. Secondly, the offered services is an important aspect of the standard
the cabinets need to be replaced and Digital Subscriber Line business case analysis. While several mathematical models
Access Multiplexers (DSLAMs) are required in these street have been proposed to estimate the adoption of services and
cabinets. At the start of the project, the operator first has to technologies, [30] has indicated the Gompertz adoption curve
decide on the cabinet size. The operator can decide to deploy as the most appropriate approach to model the adoption of
cabinets which are large enough to host a connection for each telecom business cases as a function of time. Three parameters
household in the cabinet area. Or he can decide to deploy need to be estimated in the mathematical formula, inflection
smaller (and cheaper) cabinets initially, only dimensioned for point (a), slope (b) and market size (m). The inflection point
an estimated uptake percentage of 30%. in a Gompertz curve is at 37%, and indicates the time at which
curve shifts from convex to concave. The higher a, the more
A. Technology overview stretched the adoption curve is. Slope indicates the pace of
adoption. The higher b, the faster adoption will occur, with
Before the business case is introduced in detail, a short
b [0,+ ]. For telecom cases, values of 4 (a) and 0.3 (b) have
introduction to FttC and FttH networks is given. Research con-
been found realistic [29]. The market potential parameter of
cerning these technologies is still ongoing, with Wavelength
20% used in the case is based on industry insight [31].
Division Multiplexing - Passive Optical Network (WDM-
PON) as one of the most recent technological evolution. −b(t−a)

However, the focus in our case is clearly on the passive S(t) = m · e−e [32] (5)
network infrastructure and its related costs and revenues. Since 2) Network dimensioning: For the rollout of an FttC net-
research has shown that most of the costs for the deployment work, fibre needs to be deployed from the local exchange
of new networks are related to the initial installation and in towards the cabinets. Each cabinet has a unique fibre section
particular the physical installation of the cables in the access and a shared section with the other cabinets. Based on the
network, we will focus on the topology design of FttC and duct length, fibre cable cost and installation cost per meter,
FttH networks [29]. the initial deployment cost can be calculated. The specific cost
The fixed telecom access network (both for FttC and FttH) parameters can be found in [33]. From the customer adoption,
can be represented by a tree structure, with the local exchange the necessary amounts of equipment in the local exchange and
as the source node. From this local exchange, cables towards for each cabinet can be derived.
the street cabinets depart. At each street cabinet, there are
3) Modelling costs and revenues: A detailed cost and
again cables running towards distribution points (DPs) and
revenue model is built to conduct the economic analysis of
finally to single households. For FttC, VDSL cabinets are
the small and large cabinet scenario. The costs are divided
installed on the current cabinet locations, together with the
into Capital Expenditures (CapEx) and Operational Expendi-
necessary fibre and ducts. For a local exchange, line cards
tures (OpEx). The revenues are based on the adoption model
towards the cabinets and towards the core are dimensioned,
described above. It is important to incorporate both costs
together with an optical distribution frame (ODF). In the
and revenues in the techno-economic analysis. When rolling
cabinets DSLAM line cards will be installed. When more
out a fibre network, previous research already focussed on
customers connect, extra equipment is only installed when
a minimum-cost design [34], but it is important to link the
necessary, in order to follow operational practices. For FttH,
design to the expected revenues, as has been shown in [35].
passive splitters are installed at the same location as the FttC
a) Capital expenditures: CapEx are expenditures creat-
cabinets and at the distribution points. Comparable to the
ing future benefits and are incurred when the company spends
FttC network, the equipment in the central office and the
money to buy fixed assets or upgrade existing fixed assets.
Customer Premises Equipment (CPE) are only provisioned
According to this definition, CapEx costs were subdivided into
when required.
cable and duct, local exchange, cabinet and CPE costs.
In the rollout of FttC or FttH networks, cable and duct
B. Basic business case: migration to FttC costs are generally the largest expense [29]. To dimension
As already discussed, the studied case is the migration of the the initial installation, assumptions on the uptake were made.
current copper access network towards an FttC network. The Ducts are installed to host the fibre cables and it is estimated
standard techno-economic analysis will follow the methodol- that 80% of the existing ducts can be reused. The installation
ogy proposed in [10]. For the London area, a representative cost for cables depends on the installation location, with a
exchange is modelled, taking into account average line length, buried installation being the most expensive (100 GBP/m)
number of cabinets, drop points and the amount of lines per and aerial installation the cheapest (15 GBP/m). Footpath and
exchange. As indicated, the operator can choose to install large grass installation have a cost between these two extremes. The
cabinets, offering VDSL access for all customers in the cabinet installation locations were taken from [33]. The dimensioning
area, or small cabinets, only dimensioned to provide access of the fibre cables depends on the rollout scenario chosen
for a percentage of the households. Based on the customer by the operator. For the small cabinet scenario, the operator
adoption, it will be possible to calculate the necessary amounts estimates that migrating 30% of the lines to FttC will suffice.
of equipment in the local exchange and for each cabinet As uptake is only expected to be 20%, these would suffice.
separately. For the large cabinet scenario, 60% of the lines are migrated,
 10

TABLE V £6.00

NPV (millions)
COST OF THE DIFFERENT HARDWARE COMPONENTS [33]
£5.00

Hardware component Cost (GBP)

£4.00
Street Cabinet
Small 1250 £3.00

Large 1375 Small cabinet

£2.00 installation
ODF (1440 subscribers) 935
GigE card (24 DSLAMs) 3500 Large cabinet
£1.00
installation
Chassis (16 cards) 6000
£-
DSLAM 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cost per port in small cabinet 60 -£1.00
year
Cost per port in large cabinet 120
CPE 200
Installation cost CPE 100 Fig. 5. NPV results for both installation scenarios for different geotypes

since this is the upper level of expected uptake. Other general input parameters were added, like cost erosion
To offer VDSL service to customers, the operator needs for optical and electronic equipment and a discount factor
to install equipment in the local exchange and street cabinet. of 10%, which is a standard discount factor for technology
In the local exchange, an ODF is provisioned, together with a projects. The cash flows for each year are calculated, dis-
chassis to host the GigE cards towards the cabinet and the core counted with the discount factor and summed up.
network. One GigE card has 24 ports, with each port capable With an NPV over 5 million GBP (e5.3 million) for the
of hosting one connection towards a DSLAM line card in a London geotype, both choices prove to be highly profitable,
street cabinet. These DSLAM line cards can in turn host 32 but the small cabinet FttC rollout scenario turns out to be the
connections to separate households. For the initial installation, best choice (Fig. 5). Due to the lower cost for the small cabinet
street cabinets are installed, together with one DSLAM line and the incremental cost per customer connecting, the total
card to host the first connections. When customer adoption NPV over the 15 year investment period is above the NPV of
takes off, additional cards are installed when necessary. the large cabinet scenario. Since this difference is limited, the
CapEx at the customer side comprises both hardware cost curves remain close together. The initial investment in cabinets
for the CPE and initial installation cost. These are also only results in a negative NPV in year 0, but when customers start
installed when necessary. An overview of the cost figures, connecting to the cabinets, the yearly cash flow from customer
based on industry insight, can be found in Table V. revenues improves the NPV over time. Between year 3 and
b) Operational expenditures: Operational expenditures 4, the NPV becomes positive. From this point on, the initial
are recurring and on-going costs to keep the business running. investment is paid back and the project starts generating value.
Generally, expenses like sales and administration and research Under the static assumptions, the small cabinet is dimensioned
and development are categorised as OpEx. In this model we large enough to host all connections in the future, while it is
have chosen a fractional approach to quantify OpEx. A more cheaper than the large one. This concludes the standard NPV
detailed quantification for OpEx is possible and has been analysis. Management should choose to install small cabinets
conducted in several publications [36], [37] but the fractional to offer VDSL services to its customers.
approach was chosen so as not to overcomplicate the analysis. However, this conclusion is completely dependent on the
The two OpEx categories that were identified are electronic initial assumptions concerning customer adoption, duct reuse,
equipment and other operations. OpEx for electronics is es- etc. In the next sections, we will question these input pa-
timated as a 10 percent fraction of the total CapEx for the rameters and indicate how they impact the NPV analysis by
electronic equipment, like DSLAM line cards and CPEs. For conducting a scenario and sensitivity analysis. However, these
the non-electronic fixed assets, a yearly one percent fraction evaluation methods do not allow us to calculate the value of
is taken into account as OpEx. managerial flexibility. Therefore, we extend the standard NPV
c) Revenues: Revenues in this model are based on the analysis towards a RO analysis. The value of managerial flex-
adoption assumptions from above. For each cabinet the adop- ibility will be implemented using the practical methodology
tion curve is modelled and the yearly number of customers introduced above.
is estimated. The yearly average revenue per user (ARPU) is
estimated at 500GBP. V. INTRODUCING FLEXIBILITY - REAL OPTION
4) Business case evaluation: standard NPV analysis: The ANALYSIS
input parameters from the previous paragraphs suffice to build After conducting the standard NPV analysis, it was con-
the business model for the FttC infrastructure rollout. Based cluded that the installation of small cabinets was the most
on adoption percentages and the geotype input parameters, profitable investment. However, it should be clear that basing
a network dimensioning model was built for both the small the investment decision on the outcome at this point of the
cabinet and the large cabinet scenario, calculating the required analysis could prove to be suboptimal. It was indicated that
quantities of equipment in each year of the 15-year project. three extensions have been proposed to the NPV analysis,
 11

and each of them will be applied to the case. A scenario different parameters ex-ante. This is especially true for the
analysis will be conducted comparing the initial installation market potential, since this can typically only be quantified
of small and large cabinets under different customer uptake through market studies. Another difficulty with the adoption
scenarios. This scenario analysis will be extended with a of the service for a larger area like London is that the average
sensitivity analysis to indicate the impact of uncertainty on market potential estimated for the entire area might be correct,
the outcome and as such on the final decision made. Thirdly, but large differences can exist between different subareas. For
we will indicate the different real options present in this case example, one cabinet could have an FttC uptake over 40%,
and show how they can be quantified using the methodology while another cabinet only has a final uptake potential of
presented in the previous section. Indeed, when checking for 5%. A scenario analysis was conducted for the small and
the three conditions necessary to make a business case eligible large cabinet scenario with changing market potential. The
to extend with a RO analysis, we find all three of them present. results can be found in Fig. 6. As long as the market potential
Firstly, there is undoubtedly a large degree of uncertainty stays below 30%, the small cabinets are clearly the correct
present in this case. The revenues of the project are based on rollout choice. Once a higher market potential is achieved,
a mathematical model, where the chosen parameters are in the large cabinets result in a higher payoff. This is of course a
best case "guesstimates". Secondly, there is flexibility present logical conclusion, since for all uptakes below 30% the large
in the choice of cabinet to install. Thirdly, the network operator cabinet is over dimensioned.
is not obliged to stick with the small cabinets when they cannot When we extend the scenario analysis towards a sensitivity
host extra connections under larger customer adoption. This is analysis, a probability distribution on each cabinet potential
where the timing condition can be found. On a later date in uptake between 0 and 60 percent, with 20 percent as the most
the project, the network operator can decide to expand the likely was added.
capacity or to start offering extra services. To check the impact of these parameters on the total
This business case lends itself to a real option analysis, since outcome of the business case, a Monte Carlo analysis was
all three conditions are met. Additionally, most option types conducted. Crystall Ball, a commercial tool, was used to
from the 7S framework can be identified. Once a cabinet is perform the simulations. This tool allows one to indicate
full, a simple scale option exists in placing a second cabinet the different uncertainties in a spreadsheet. After selecting
to host the extra connections. A switch option for full cabinets the cells to forecast, the tool runs a predefined number of
is the installation of an FttH solution for the extra customers. simulations, resulting in a distribution of the value of the
When a scope option is considered, installing extra equipment forecasted cells, based on the uncertainty distributions added
in the local exchange to upgrade your internet service portfolio to the assumption cells [25]. The small and large cabinet
towards IPTV and gaining extra revenues is possible. Since rollout scenarios are compared, now with uncertainty added
realistic business cases generally possess a wide variety of on some input parameters. Before the rollout of the FttC
real options, we also introduce compound real options. It will network starts, the management has to decide which scenario
also be shown how they influence the decision process in the it will choose. However, we would like to indicate that no
standard case. options are implemented in the model yet, so once a small
cabinet is full, no extra customers can be connected to this
cabinet, resulting in a loss of potential revenues. While the
A. Impact of uncertainty on the final decision
previous static NPV analysis results in a fixed number, the
The standard NPV analysis conducted identified the instal- sensitivity analysis returns two distributions to compare (Fig.
lation of small cabinets for the London geotype as the most 7). It is immediately clear that the best rollout option is the
profitable scenario. Step two of the RO analysis methodology installation of large cabinets in all areas. On average, the large
requires identifying the uncertainties present in the case. As cabinet scenario outperforms the small cabinet scenario with
always in long term infrastructure projects, all input param- over 260.000 GBP (+3.88%).
eters are uncertain, especially in the long run. In order not This sensitivity analysis clearly indicates the impact that
to overcomplicate the analysis, we have chosen to select two uncertainty has on the decision process. However, a sensitivity
major uncertainties, user adoption and duct reuse. analysis alone does not allow us to implement the flexibility
User adoption is typically the most uncertain factor in an in the project. The assumption that the operator will not act
economic analysis. Before the introduction of a new product when the small cabinets are full does not hold in reality. In the
or service, it is very hard to estimate how many consumers following section, the different options are identified and the
will buy it. On the other hand, it is a factor with a high last two steps of the RO analysis methodology are conducted.
impact on the final economic assessment [3]. Research into
the adoption of new services and products has indicated that
consumer adoption generally can be modelled using a bell B. How managerial flexibility impacts the result - a real option
curve, with the Rogers’ bell curve as the most well-known for analysis
technology adoption [38]. These models have been translated It is straightforward that when a small cabinet is installed,
to mathematical S-curve penetration models, from which we the option to expand is available. This expansion option can
have chosen the Gompertz curve in this business case [32], be broken down into three different options from the 7S
[39], [40]. However, while these models ex-post show a good framework. When the first cabinet is full, the operator can
fit with the observed adoption, it is hard to estimate the decide to go for a standard scale up option by installing a
 12

2500
£18.00
Millions

£16.00 2000

Scale up option
£14.00 1500

Frequency
LARGE Static case

£12.00 1000 Mean Scale up

option
Mean LARGE
£10.00
500 Static case
NPV

Small cabinets
£8.00
Large cabinets 0

£6.00

£4.00 NPV

£2.00

Fig. 8. Overlay chart large scenario and small scale case

£-
10% 20% 30% 40% 50% 60%
Uptake
3000

2500
SMALL Static case
Fig. 6. Small and large cabinet market potential scenario analysis
2000
LARGE Static case

Frequency
3000 1500
Switch up option

2500 1000
Mean SMALL
Static case
2000 SMALL Static 500
case
Frequency

Mean LARGE
LARGE Static case Static case
1500 0

Mean SMALL Mean Switch up

1000 Static case option
Mean LARGE
Static case NPV
500

Fig. 9. Improvement of the business case with a switch up option

NPV

TABLE VI
Fig. 7. Impact of uncertainty on the business case outcome COST PARAMETERS FOR AN FTTH DEPLOYMENT [33]

Hardware component Cost (GBP)

second small cabinet on the location to offer FttC services [Link] card (256 customers) 6.000
to the new customers. However, he could also choose a Cost per [Link] port 500
technology upgrade by connecting the extra customers via Passive splitter 210
a more future-proof technology over an FttH network. A CPE 80
third expansion option is raising the ARPU per customer by CPE installation cost 100
offering extra services to the existing customers, for example
by starting to offer IPTV services. In this section, we will
apply these three options to the business case and show how there is an extra scale option value in the small cabinet
they impact the previous results. scenario. The scale up option offers the operator the possibility
1) Scale up: installing extra small cabinets: The first identi- to initially install the cheaper small cabinets and only invest
fied flexibility is the scale up option. Once the small cabinet is in additional capacity when necessary. Large cabinets offer
full, the operator can install a second cabinet to host the extra enough capacity to host all connections, but in most cases this
connections. Of course, this comes with an additional capital capacity is never used.
expenditure for a small cabinet and DLSAM line cards. The 2) Switch up: additional capacity through a more future
business case presented above was extended with this scale proof network: It was already indicated that the operator could
option. In the small scale scenario, the extra customers on a migrate the extra customers on a full cabinet towards a more
full cabinet are now connected to an additional small cabinet, future proof FttH network. To implement this switch up option
and their ARPU is added to the business case. However, in the business case, some extra additions are necessary in the
the option will only be executed when it is economically model. Fibre cables need to be installed in the last mile and
interesting. It is as such a simple maximisation function of extra [Link] equipment provided in the local exchange. In
the small static case and the scale up case. the access network, passive splitters ensure the connectivity.
When comparing the results for the large cabinet and the The cost parameters are based on industry insight and can be
small scale scenario, we notice that the small scale case is the found in Table VI.
most interesting for the operator. It yields an average payoff The impact of the switch option can be seen in Fig. 9. Again
which is 2.23% higher than the large cabinet scenario, which the business case for the small cabinet scenario is greatly
was initially the best choice after the sensitivity analysis (Fig. improved. However, when comparing with the scale option,
8). Compared to the small static case, it is even 6% higher. we see that an FttH extension is more expensive and does not
While the large cabinet scenario is definitely the most future improve the small scenario enough to outperform the large
proof option in the scenario analysis, the RO analysis indicates cabinet scenario.
 13

3000
TABLE VII
IPTV COST PARAMETERS [33] 2500

2000 Scope up option

Component Cost (GBP)

Frequency
1500 SMALL Static case
Video server 20.000
Mean SMALL Static
1000
CPE 250 case
Scope up option
500
CPE installation cost 200
Extra ARPU 100 0

NPV
TABLE VIII
IPTV GOMPERTZ ADOPTION PARAMETER ESTIMATES
Fig. 10. Scope up option
Parameter Value
a 4.667 3000
SMALL Static case

b 0.366 2500

Scale Up option
m 0.405 2000

Frequency
1500 Combined scope-
switch option

1000
3) Scope up: offering extra services over the existing infras- Mean SMALL
Static case

tructure: One of the examples given as a scope up option is 500

Mean Scale up

the extension of the typical telephone and internet incumbent 0 option

product portfolio towards triple play. We will indicate in this Mean Combined
scope-switch
option
section how offering an IPTV service can be implemented in NPV

the business case as a real option. The operator can decide to

Fig. 11. Compound options: switch up and scope up combined
extend his product portfolio towards triple play in the fourth
year of the project.
Before the standard business case is extended, some extra upgrade its network capacity with an FttH network once the
input parameters are required. Offering IPTV to end customers first cabinet is full, they can start offering IPTV to these
will require extra equipment in the local exchange. In order to customers. This is in fact a compound option, or an option
avoid unnecessary complexity in the business case, we added on an option. The same adoption curve as before was used
a fixed cost for video server per local exchange and other but is now only applied to the customers who are connected
equipment to the scope up business case extension. At the via the FttH network. In contrast with the single scope up
customer’s premises, a new CPE needs to be installed (Table option, this now results in a positive option value. The total
VII). NPV for this case results in an even larger payoff compared
Typically, the adoption of such a service will again follow with the small cabinet scenario with the scale option (Fig. 11).
the S-shaped adoption curve. Since it is not straightforward
to translate this adoption into a mathematical model, we
have combined given adoption percentages [14] together with C. Rollout of an FttC network - case conclusion
statistical software to fit the historic adoption percentages to The considered business case evaluates the economic fea-
the mathematical Gompertz model. All parameters were found sibility of the rollout of an FttC network in the UK. To
to be statistically significant. The resulting parameters are achieve the FttC coverage, the network operator can follow
summed up in Table VIII. two investment paths, full coverage or coverage following
The results of this RO analysis can be found in Fig. 10. demand. The traditional NPV analysis showed the coverage
Apparently, the extension of the product portfolio towards following demand scenario was the most profitable. However,
triple play services has only marginal value compared with the when checking the requirements for an RO analysis, it was
small cabinet scenario. This means that offering triple play to clear that uncertainty surrounding the initial assumptions most
customers will be not interesting in this scenario, so the option likely would have an impact on the analysis results.
is almost never executed. Therefore, we followed the proposed four-step methodology
4) Combining options: scope and switch up combined: to extend the case with the real option approach. After every
In the previous analysis only single options were presented. step, we indicated the impact on the results of the analysis.
However, realistic business cases generally possess a wide The effect of uncertainty on the business case was assessed by
spectrum of different options. Consider the three options both a scenario and sensitivity analysis. The initial results from
discussed above. It is clear that the scale and switch up the traditional approach were contradicted when the customer
options are mutually exclusive. If a given area is extended uptake assumptions were questioned. When flexibility was
with an extra cabinet, the FttH switch up option will become implemented in the business case, to allow the decision makers
redundant. However, the scale and switch up option can be to react to this uncertainty, all option categories were identified
easily combined with the scope up option. in the following demand scenario of the business case. The
Consider the case where the triple play services require initial network could be expanded with extra capacity, either
a network with extra capacity. If the operator chooses to via a scale up option or a switch up option. The scale up option
 14

installs more of the same, while the switch up option gradually • Conduct a standard NPV analysis
migrates the existing network towards an FttH network. The Real options extend the standard NPV analysis, and their
third scenario investigated offering extra services, following a valuation starts with a clear understanding of the value
scope up option path. of the project in absence of uncertainty and options.
In this case, a scale up option is the most designated, • Identify the uncertainties
as it improves the result of the small cabinet scenario to a Future uncertainty is a condition for RO analysis, and
level higher than the large cabinet scenario. The switch up should therefore be identified before the quantification of
option also improves the small cabinet scenario. However, the real option value. The impact of uncertainty on the
when compared to a static large cabinet scenario, the latter investment project can be checked through a scenario or
remains more interesting. It has also been shown that offering sensitivity analysis.
extra services adds only marginal value to the business case • Identify the flexibility
for the small cabinets and is thus almost never executed. Without the ability to react against uncertainty, no real
However, in realistic business cases, options almost never options are present. Here, the 7S framework can be a
occur in isolation. This tutorial extended the example case with guideline to formalize the intuitive notions on managerial
a compound option by implementing the scope up option on flexibility present during the project lifetime.
the switch up option. Using the proposed methodology, this • Calculate the option value
remains a rather straightforward exercise. While the Monte Carlo analysis is designated for ex-
tended techno-economic cases, a more simple back of the
VI. LESSONS LEARNED, SHORTCOMINGS AND
envelope binomial tree analysis can offer initial insights.
POSSIBLE EXTENSIONS
A. Real option analysis guidelines
B. Pitfalls of RO analysis
The practical RO analysis methodology described in this pa-
per bridges the gap between the financial and technical world A RO analysis is a helpful tool to value inherent flexibility
in telecommunication firms. In all areas of telecommunication, in typical telecom investment projects. However, when con-
decisions on new investments need to be made. Design of ducting a RO analysis, several things should be kept in mind.
passive infrastructures by network planners, testing of new First, the value of an option is a function of the uncertainty
technology before implementing it, etc. These decisions are attributed to the different input factors. The higher the uncer-
not only based on cost optimizations, but intuitive notions on tainty, the higher the option value. This effect can easily be
the capability of the chosen design or technology to counter observed in the Black and Scholes formula. Estimating this
uncertainty, the inherent flexibility of the design, or the oppor- uncertainty remains a difficult exercise and should be handled
tunities it offers to reduce risk related to future costs are also with care. Although commercial software allows the user to
taken into account. Real options translate this intuition into attribute uncertainty to all input parameters, it is important to
the language of the financial department. The three conditions focus on the uncertainties with the highest expected impact,
required for a RO analysis can help the network planners to e.g. customer adoption, lifetime of the technology, etc.
identify the presence of real options in their network plan. Secondly, the results of a RO analysis give an indication of
Additionally, the 7S framework can help to categorize these the average extra value the option generates. In a Monte Carlo
options. With the four step methodology, the standard NPV analysis, thousands of possible futures are calculated and,
analysis executed today can easily be extended towards a full from the resulting probability distribution, several conclusions
RO analysis. In order to identify real options and quantify can be drawn. It indicates how the option impacts the risk
their impact, the following guidelines are offered. First, when associated with the investment. An option typically reduces the
making network design decisions, ask three questions. risk of a low payoff, but it does not guarantee a positive payoff.
For example, a project with a scale option can in the future
• How uncertain is the future?
still turn out to be unprofitable, since the customer uptake is
Uncertainty surrounding future conditions will typically
much lower than expected, turning the scale option value to
have a large impact on your decision. Identifying several
zero.
future scenarios can help to see how this uncertainty
Finally, a RO analysis always attributes value to waiting. In
would impact your decision.
a project with an option to wait, the further the investment can
• Where is the flexibility?
be postponed, the more value the option typically generates.
If uncertainty is present, different actions may exist to
This effect was, for example, indicated in the Black and
counter it. Indicating how to alter the initial project path
Scholes formula. In practice, decision makers do not have the
under different conditions helps to identify the different
option to postpone decisions forever. The threat of competitive
options. The 7S framework can be a guideline.
entry pushes decision makers to move as fast as possible. In a
• When do I have to decide?
competitive environment, the value of waiting erodes quickly,
Flexibility and uncertainty are not sufficient to have op-
since there is typically a first mover advantage.
tions in the investment case. Flexibility is only interesting
if it can be executed in later phases of the project.
When these three conditions are met, real options are present C. Future work
in the investment project. It is then important to check how One of the most important future research directions in real
they impact your decision. option analysis is its interaction with competition. Typically,
 15

the impact of competition is analysed through a game theoretic preconditions are met, the calculation of the option value is
analysis. Game theory is another extension of the standard quite straightforward. In the standard NPV analysis model,
techno-economic analysis. For a clear case study on the impact both the uncertainties and flexibilities need to be identified
of competition on an economic assessment of a realistic and added to the model. Executing a Monte Carlo simulation
business case, the authors refer to [41]. on this extended model then results in the real option value of
However, as was already indicated above, in realistic busi- the project.
ness cases, the decision maker has to take both his options To indicate the strength of such a practical framework,
and competition into account. There is a trade-off between we applied it to the investment project for a next generation
the value of waiting and reducing uncertainty, and the first fixed access network rollout. The migration towards an FttC
mover advantage. While waiting typically increases value in network in the UK was studied. It was indicated that the
option theory, it decreases value when competition comes into operator had two rollout choices, either installing small or
play. large cabinets. From the traditional NPV analysis, the small
Steps towards this integration have already been made. cabinet installation proved to be the most profitable. However,
Option games allow for such an integration, but they have uncertainty surrounding the several input parameters could
currently only been applied to more simple illustrative ex- have an important impact on the final outcome. Therefore, the
amples [42]. In option games, basic binomial tree games standard NPV analysis was extended with a scenario, sensitiv-
are extended by a game theoretic analysis of the end nodes. ity and real option analysis. While the scenario and sensitivity
While this can indicate the value of a combined option-game analysis allowed adding uncertainty to the investment project,
thinking, it should be extended towards more realistic settings, it is only the RO analysis that includes the value of managerial
to increase its applicability in day-to-day decision making. flexibility in the decision process. Four different options were
Another approach is through sensitivity games [43] where identified in the case, each having a different impact on the
the impact of uncertainty is assessed on the game theoretic decision process of the management.
analysis and the resulting equilibria. Extending the underlying In this tutorial, we indicated the importance of real options
business cases with the value of options would offer new in the telecommunications sector. The authors hope to speed
insight into the dynamic interplay between options and games. up the application of real options within this sector by pro-
viding some clear real-life examples of firms executing their
options and by offering a complete real option analysis of an
VII. CONCLUSION
existing business case.
The broad range of uncertainties concerning future tech-
nological evolution, customer adoption and regulation which ACKNOWLEDGMENTS
is characteristic of the telecommunication sector definitely
This research was partly carried out as a collaboration
requires managerial flexibility in large investment projects
between BT and iMinds.
in this field. However, the traditional economic evaluation
The work leading to these results has received funding
methods cannot capture the value of this flexibility. Different
from the European Union’s Seventh Framework Programme
extended evaluation models have been proposed to solve this
(FP7/2007-2013) under grant agreement n◦ 249025.
problem. In particular, the real option theory has shown great
This research was partially supported by the iMinds TER-
potential to integrate managerial flexibility with the standard
RAIN project. This project is co-funded by iMinds, IWT and
evaluation methods. However, this extended model is only
Acreo AB, Alcatel-Lucent, Comsof, Deutsche Telekom Lab-
slowly finding acceptance. To indicate the importance of real
oratories, Digipolis, FTTH Council Europe, Geosparc, Stad
options for telecom investment projects, a wide range of
Gent, TMVW, TE Connectivity, UNET and WCS Benelux BV.
realistic examples was introduced showing the broad array of
options existing in all telecom sectors. The abandon option
R EFERENCES
in the mobile broadcast TV service of British Telecom or the
testing periods of LTE by Telenet in Belgium are just two [1] S. A. Ross, R. W. Westerfield, and J. F. Jaffe, Corporate Finance.
McGraw-Hill, 2006.
examples. [2] C. Lange, T. Monath, E. Weis, M. Kind, M. Adamy, and N. Gieschen,
It may be clear that the application of RO Theory in “Techno-Economic Comparison of Passive Optical Network Scenarios,”
telecommunication projects should be a logical extension to in BroadBand Europe, pp. 3-5.
[3] J. Harno, D. Katsianis, T. Smura, T. G. Eskedal, R. Venturin, O. P.
the traditional evaluation methods. However, a common com- Pohjola, K. R. R. Kumar, and D. Varoutas, “Alternatives for mobile op-
plaint regarding this theory is the lack of a practical framework erators in the competitive 3G and beyond business,” Telecommunication
for realistic cases. In this tutorial, we extended the overview of Systems, vol. 41, no. 2, pp. 77-95, Apr. 2009.
[4] A. Saltelli, M. Ratto, T. Andres, F. Campolongo, J. Cariboni, D. Gatelli,
real option basics and the application domains in telecom with M. Saisana, and S. Tarantola, Global Sensitivity Analysis: The Primer.
a practical approach to extend realistic business cases with a John Wiley & Sons, 2008.
real option analysis. We stressed the importance of the three [5] T. Rokkas, D. Katsianis, T. Kamalakis, and D. Varoutas, “Economics of
Time and Wavelength Domain Multiplexed Passive Optical Networks,”
requirements for a RO analysis and the four-step methodology Optical Communications and Networking, IEEE/OSA Journal of, vol.
to implement it. Before a business case is eligible for a RO 2, no. 12, pp. 1042-1051, Nov. 2010.
analysis, uncertainty surrounding the project should be present. [6] K. Casier, S. Verbrugge, J. Van Ooteghem, D. Colle, R. Meersman, M.
Pickavet, and P. Demeester, “Impact of sensitivity and iterative calcula-
This uncertainty can however be handled by the managerial tions on cost-based pricing,” in 6th Conference on Telecommunication
flexibility in the project at a later point in time. When these Techno-Economics (CTTE), 2007.
 16

[7] T. E. Copeland and P. T. Keenan, “How much is flexibility worth?,” contingencies,” Philosophical Transactions of the Royal Society, vol. 36,
McKinsey Quarterly, no. 2, pp. 38-50, 1998. pp. 513-585, 1825.
[8] T. E. Copeland and V. Antikarov, Real Options - a practitioner’s guide. [33] Analysys Mason, “The costs of deploying fibre-based next-generation
Texere, 2003. broadband infrastructure,” 2008.
[9] W. De Maeseneire, The real options approach to strategic capital [34] M. J. O’Sullivan, C. G. Walker, M. L. O’Sullivan, T. D. Thompson, and
budgeting and company valuation. Financiële cahiers, Larcier, 2006. A. B. Philpott, “Protecting local access telecommunications networks:
[10] S. Verbrugge, K. Casier, J. Van Ooteghem, and B. Lannoo, “Practical Toward a minimum-cost solution,” Telecommunication Systems, vol. 33,
steps in techno-economic evaluation of network deployment planning,” no. 4, pp. 353-376, Dec. 2006.
IEEE Globecom, Tutorial, vol. 11, 2008. [35] K. Casier, J. Van Ooteghem, M. Sikkema, S. Verbrugge, D. Colle, M.
[11] J. Mun, “Real options and Monte Carlo simulation versus traditional Pickavet, and P. Demeester, “Influence of geomarketing on the rollout
DCF valuation in layman’s terms,” in Managing enterprise risk: what of new telecom network infrastructure,” in Telecommunication, Media
the electric industry experience implies for contemporary business, K. and Internet Techno-Economics (CTTE), 10th Conference of, 2011, pp.
B. Leggio, D. B. Bodde, and M. L. Taylor, Eds. Elsevier, 2006, pp. 1-7.
75-100. [36] S. Verbrugge, D. Colle, M. Pickavet, P. Demeester, S. Pasqualini,
[12] L. Trigeorgis, Real options: Managerial flexibility and strategy in A. Iselt, A. Kirstädter, R. Hülsermann, F.-J. Westphal, and M. Jäger,
resource allocation. The MIT Press, 1996. “Methodology and input availability parameters for calculating OpEx
[13] B. Chevalier-Roignant, C. M. Flath, A. Huchzermeier, and L. Trigeorgis, and CapEx costs for realistic network scenarios,” Journal of Optical
“Strategic investment under uncertainty: A synthesis,” European Journal Networking, vol. 5, no. 6, pp. 509-520, 2006.
of Operational Research, vol. 215, no. 3, pp. 639-650, Jun. 2011. [37] K. Casier, S. Verbrugge, R. Meersman, D. Colle, and P. Demeester, “On
[14] Point Topic, “The leading websource on DSL and broadband,” 2013. the costs of operating a next-generation access network,” in Proceedings
[Online]. Available: [Link] [Accessed: 25-Jan-2013]. of CTTE2008, the 7th Conference Telecom, internet and media Techno-
[15] B. Lannoo, S. Verbrugge, J. Van Ooteghem, B. Quinart, M. Casteleyn, Economics (CTTE), 2008.
D. Colle, M. Pickavet, and P. Demeester, “Business Model for a Mobile [38] G. M. Beal, E. M. Rogers, and J. M. Bohlen, “Validity of the concept
WiMAX Deployment in Belgium,” in Mobile WiMAX, no. June 2004, of stages in the adoption process,” Rural Sociology, vol. 22, no. 2, pp.
K.-C. Chen and J. R. . de Marca, Eds. John Wiley & Sons, 2008, pp. 16-168, 1957.
349-369. [39] J. Fisher and R. Pry, “A simple substitution model of technological
[16] B. Sadowski, M. Verheijen, and A. Nucciarelli, “From experimentation change,” Technological Forecasting and Social Change, vol. 3, pp. 75-
to citywide rollout: Real options for a municipal WiMax network in the 88, 1971.
Netherlands,” Communications & Strategies, no. 70, pp. 101-123, 2009. [40] F. M. Bass, “A new product growth for model consumer durables,”
[17] V. P. Tanguturi and F. Harmantzis, “Migration to 3G wireless broadband Management Science, vol. 15, no. 5, pp. 215-227, 1969.
internet and real options: The case of an operator in India,” Telecom- [41] M. Tahon, B. Lannoo, J. Van Ooteghem, K. Casier, S. Verbrugge, D.
munications Policy, vol. 30, no. 7, pp. 400-419, Aug. 2006. Colle, M. Pickavet, and P. Demeester, “Municipal support of wireless ac-
[18] J. J. Alleman and E. M. Noam, The new investment theory of real op- cess network rollout: A game theoretic approach,” Telecommunications
tions and its implications for telecommunications economics. Springer, Policy, vol. 35, no. 9-10, pp. 883-894, Jun. 2011.
1999. [42] B. Y. N. Ferreira, J. Kar, and L. Trigeorgis, “Option Games The Key to
Competing in Capital-Intensive Industries,” Harvard Business Review,
[19] M. Tahon, S. Verbrugge, B. Lannoo, J. Van Ooteghem, P. De Mil,
no. March, pp. 101-108, 2009.
M. Pickavet, and P. Demeester, “Parking sensor network: economic
[43] K. Casier, B. Lannoo, J. Van Ooteghem, S. Verbrugge, D. Colle, M.
feasibility study of parking sensors in a city environment,” in Telecom-
Pickavet, and P. Demeester, “Game-theoretic optimization of a fiber-to-
munication, Media and Internet Techno-Economics, 9th Conference,
the-home municipality network rollout,” Journal of Optical Networking,
Proceedings, 2010, pp. 1-8.
vol. 1, no. 1, p. 30, Jun. 2009.
[20] M. Basili and F. Fontini, “The option value of the UK 3G Telecom
licences: Was too much paid?,” info, vol. 5, no. 3, pp. 48-52, 2003.
[21] L. Mastroeni and M. Naldi, “Options and overbooking strategy in the
management of wireless spectrum,” Telecommunication Systems, vol.
48, no. 1-2, pp. 31-42, May 2010.
[22] R. S. Pindyck, “Mandatory unbundling and irreversible investment in
telecom networks,” 2004.
[23] J. Van Ooteghem, B. Lannoo, D. Colle, M. Pickavet, I. Moerman, and P.
Demeester, “Internet service option on trains: an application in Belgium
Railway,” in Annual international conference on Real Options, 12th,
Presentations, 2008.
[24] F. Black and M. Scholes, “The pricing of options and corporate Mathieu Tahon Mathieu Tahon graduated as Master
liabilities,” The journal of political economy, vol. 81, no. 3, pp. 637-654, in Applied Economics, option Finance in 2009 at
1973. Ghent University. He joined the IBCN research
[25] S. S. Sawilosky, “You think you’ve got trivials,” Journal of Modern group of the Department of Information Technol-
Applied Statistical Methods, vol. 2, no. 1, pp. 218-225, 2003. ogy as PhD student in August 2009, at the same
[26] Oracle, “Crystall Ball Decision Optimizer.” [Online]. Available: university. He is affiliated with the techno-economic
[Link]/us/products/applications/crystalball/[Link]. research cluster, where he performs research on the
[Accessed: 04-Dec-2012]. application of advanced evaluation techniques in
[27] M. H. Kalos and P. A. Whitlock, Monte carlo methods. Wiley-VCH, broadband network rollout (both fixed and wireless)
2009. and new service introduction (smart meters, electric
[28] M. Tahon, S. Verbrugge, P. J. Willis, P. Botham, D. Colle, and M. vehicles). He specifically focuses on the use of real
Pickavet, “Migration to Next Generation Access Networks?: a Real option and competition modeling in the economic evaluation.
Option Approach,” in Proceedings of ROC 2012, 2012, p. 11. He was involved in various national and European research projects,
[29] K. Casier, “Techno-economic Evaluation of a next including the European FP7 ICT-STRONGEST project and the national IBBT-
generation access network deployment in a competitive ICON TERRAIN, where he led the work package on extended evaluation
setting (Doctoral dissertation),” 2009, p. 283. Available: techniques. He is currently working within the European FP7 project ICT-
[Link] TREND, studying cooperative models in wireless networks in order to reduce
[30] A. Cárdenas, M. García-Molina, S. Sales, and J. Capmany, “A New cost and energy consumption.
Model of Bandwidth Growth Estimation Based on the Gompertz Curve?:
Application to Optical Access Networks,” Lightwave Technology, vol.
22, no. 11, pp. 2460-2468, 2004.
[31] The information society EURIM, “Making broadband investment mar-
kets work - summary,” 2011.
[32] B. Gompertz, “On the nature of the function expressive the law of
human mortality, and on a new method of determining the value of life
 17

Sofie Verbrugge Sofie Verbrugge received an MSc Mario Pickavet Mario Pickavet received an [Link].
degree in computer science engineering from Ghent and Ph.D. degree in electrical engineering, special-
University (Ghent, Belgium) in 2001. She obtained ized in telecommunications, from Ghent University
the PhD degree from the same university in 2007 in 1996 and 1999, respectively. Since 2000, he is
for her thesis entitled "Strategic planning of optical professor at Ghent University where he is teaching
telecommunication networks in a dynamic and un- courses on discrete mathematics, multimedia net-
certain environment". works and network modeling.
She is currently working as a researcher affiliated He is co-leading the research cluster on Net-
to iMinds, a research institute to stimulate ICT inno- work Modeling, Design and Evaluation (NetMoDeL)
vation in Flanders, where she is a coordinator for the covering 4 research topics: Fixed internet architec-
techno-economic research within the Internet Based tures and optical networks, techno-economic studies,
Communication Networks and Services group (IBCN). Her main research green-ICT and design of network algorithms (DNA). In this context, he is
interests include infrastructure as well as operational cost modeling, telecom currently involved in several European and national projects.
service and network deployment planning, advanced evaluation techniques He has published about 300 international publications, both in journals
including real options and game theory. She has been involved previously (IEEE JSAC, IEEE Comm. Mag., Journal of Lightwave Technology, ...) and in
in several European as well as national research projects in these domains, proceedings of conferences. He is co-author of the book ’Network Recovery:
including the COST-action Econ@tel. She led the work package on "Business Protection and Restoration of Optical, SONET-SDH, IP, and MPLS’.
modeling" with the European FP7 project ICT-OASE and is currently working
within the FP7 project ICT-Flamingo.

Peter Willis Peter Willis is Chief Data Networks

Strategist, Research & Innovation at British Tele-
com. Peter is researching the future of networks
covering everything from photonics to network ap-
plications. He is now working on Network functions
Virtualisation, which is implementing network func-
tions as software running on high volume standard
servers. Peter previously worked on the development
of PBB-TE, BT’s 21st Century Network Architec-
ture and BT’s Internet service. He has a long term
interest in developing business cases for introducing
new technologies into networks which lead to the work on applying Real
Options Analysis techniques to network investments.

Paul Botham Paul Botham joined BT with a

[Link]. in theoretical physics from Oxford Univer- Piet Demeester Piet Demeester is professor in the
sity. Within Research and Innovation, he has worked faculty of Engineering at Ghent University. He is
in a variety of network modelling roles, applying head of the research group "Internet Based Com-
mathematical techniques to develop software tools munication Networks and Services" (IBCN) that is
for designing least-cost, resilient BT networks. This part of the Department of Information Technology
has involved a wide range of technologies in both (INTEC) of Ghent University. He is also leading the
core-transport and access environments. Paul is cur- Future Internet Department of the strategic research
rently working on optimised deployment of next- center iMinds. He is Fellow of the IEEE. The IBCN
generation network technology, incorporating risk group is focusing on several advanced research
analysis for BT business cases. topics: Network Modeling, Design & Evaluation;
Mobile & Wireless Networking; High Performance
Multimedia Processing; Autonomic Computing & Networking; Service En-
gineering; Content & Search Management and Data Analysis & Machine
Learning."
Didier Colle Didier Colle received a M. Sc. degree
in electrotechnical engineering (option: communi-
cations) from the Ghent University in 1997. Since
then, he has been working at the same univer-
sity as researcher in the department of Information
Technology (INTEC). He is part of the research
group INTEC Broadband Communication Networks
(IBCN) headed by prof: Piet Demeester. His research
lead to a Ph.D degree in February 2002. He was
granted a postdoctoral scholarship from the Institute
for the Promotion of Innovation through Science and
Technology in Flanders (IWT-Vlaanderen) in the period 2003-2004. Currently,
he is co-responsible for the research cluster on network modelling, design and
evaluation and is coordinating the research on fixed internet architectures and
optical networks. His research deals with design and planning of commu-
nication networks. This work is focussing on optical transport networks, to
support the next-generation Internet. Up till now, he has actively been involved
in several national and international research projects. His work has been
published in more than 250 scientific publications in international conferences
and journals.

View publication stats