Articulo cientifico

1. ISBN: 978-1-5386-5447-7/18/$31.00 ©2018 IEEE
Evaluation of the National Optical Fiber Backbone
Network Model for the Universal Internet access in
Peru
Martin M. Soto-Cordova
Universidad de Ciencias y Humanidades
Lima, Peru
msoto@uch.edu.pe
Carlos Sotelo-Lopez
Universidad Nacional Mayor de San Marcos
Lima, Peru
carlos.sotelo.lopez@gmail.com
Abstract—Currently the National Optical Fiber Backbone
Network is the main data network infrastructure financed by
the Peruvian government, which has been granted to Azteca
Comunicaciones Peru —a subsidiary of Azteca—to provide
broadband carrier services for eighteen years. The purpose of
this project is to interconnect one hundred and eighty
provincial capital cities, with the aim of reducing the big gap
between the capital city of Lima and the provinces in Peru, in
terms of accessibility and affordability to the Internet. This
paper analyzes the problems of the current technical-economic
model of the network, determines which aspects should be
improved and proposes a new model, aimed at enabling a
greater development of broadband in Peru.
Keywords— broadband, optical fiber, backbone, DWDM
I. INTRODUCTION
The goal of national backbone network is to constitute
the main communications platform throughout a country
reaching the smallest geopolitical division, in the case of
Peru the district cities [1]. In general, national backbone
networks are financed by the Government and do not have as
objective a direct monetary profitability but the increase of
the coverage and the number of subscribers of the broadband
service generating a greater competition in telecom services
and increasing the penetration of broadband services that
would allow the increase of GDP [2].
Studies on the National Optical Fiber Backbone Network
of Peru were initiated in 2008 at the initiative of the FITEL-
Government telecommunications investment fund-and was
awarded to the Azteca Comunicaciones Peru company in
2013, which started installations in July 2014 and begun
operations in mid-2016. To date, having already
implemented the project and being almost two years since its
implementation, it is noticed that the use of this transport
network is insignificant in relation to what was expected, not
fulfilling the purpose for which it was created in accordance
with Law 29904 [3, 4].
II. THE BACKBONE IN THE CONTEXT OF NATIONAL
BROADBAND
Fig. 1 shows the national backbone network, which
includes 24-wire fiber optic network along the Peruvian
territory with an approximate length of 13,500 km. Its
operation by Azteca will be for 18 years and has the scope to
connect 22 regional capitals and 180 capitals of provinces. It
consists of three Network Access Point (NAP) and a
Network Operations Center (NOC) located in Lima and
forms five rings in the core network. The links of the core
network use DWDM at 100Gbps for each wavelength with
capacity of 80 wavelengths of 10/100Gbps, having initially
implemented 40 wavelengths. The provincial capital cities
have Metro Ethernet 10Gbps links with the Core Network. It
has 136 nodes along its route.
This network has four international interconnection nodes
and includes Core Nodes, each located in eight regional
capital cities; Aggregation Nodes, each located in twenty two
regional capital cities; Distribution Nodes, each located in
the one hundred and eighty provincial capital cities;
Connection Nodes, located in the route of fiber laying.
Fig. 1. Backbone Network Architecture. (Source: FITEL)
There are some limitations on backbone network as: It
cannot market Internet traffic to other countries; It cannot
market Internet traffic to other operators, only link provincial
capital city to Lima; There is a fixed rate for five years with a
high price; The profits of the operation is only 20% for the
private and 80% for government.

2. III. A TECHNICAL-ECONOMIC EVALUATION
The backbone network cannot be analyzed in isolation
but within a context of integration with other regional and
local networks to demonstrate its effect on the Internet price
to the end user. Peru is divided geopolitically from 24
regions, each region into provinces, each has districts. In
order to have communications of high capacity and
affordable prices in all districts nationwide, it is necessary to
have transportation networks to the district capital cities.
This network only reaches the capitals of 180 provinces
and according to a contract between the Peruvian
Government and Azteca, the fixed rate for five years is $23
per Mbps, regardless of distance and the contracted capacity.
To reach the 1871 district capital cities the use of other
regional transport networks is required. These regional
transport networks have a maximum rate of $23 per fixed
Mbps. This rate was established by OSIPTEL, which is the
Peruvian Telecom Regulator. All the rates mentioned do not
include taxes. However, the length of the regional transport
network in most cases only reaches 5% of the length of the
backbone network. This set the rate of transport from Lima
to a district capital up to $46. Therefore, the join of the
backbone network with the regional transport networks is
necessary, not to create economies of scale, but because
otherwise it would not be possible to have accessible rates
for the transport network that reaches the 1871 district
capitals where local broadband regional access networks are
being developed financed by the Peruvian Government.
The current model of this network is shown in Fig. 2.
Internet rate for the end user is defined in (1):
IST = IL + DN + RT + AO (1)
Where:
IL: Internet cost in Lima
DN: transport cost over backbone network
RT: regional transportation cost
AO: access network fee for regional projects of FITEL
Fig. 2. Backbone Network model.
Fig. 3 shows the cost components for an Internet service
outside Lima. In the present model, the Internet must be
purchased in Lima from an international carrier paying an IL
price, which according to the OSIPTEL estimate, is $19.50
for 1Mbps without oversubscription. If this Internet is to be
brought to a district to be commercialized, the cost is
increased by up to $46 ($23 due to backbone network
transport and $23 due to regional transportation network).
Therefore, at the entrance of the network of a regional access
operator the price is $65.00 per Mbps. Thus, a local operator
in a district outside Lima would have a cost of 336% higher
than the Internet purchase price of a local provider in Lima.
According to an analysis carried out by OSIPTEL,
fixation of the maximum fee referred to the transportation
and Internet access services provided through the regional
network [5], for regional operators subsidized by the
government the maximum fee for 2Mbps and 4Mbps are
shown in Table I. The rates established by the OSIPTEL
correspond to government entities for regional broadband
projects where the government subsidizes the total CAPEX
and part of the OPEX through adjudications through public
tender. In any other case there is no tariff ceiling, both for
projects subsidized by the State and for private local
initiatives. However, higher rates are expected due to the fact
that in the previous cost structure of Table I, the recovery of
the investment and a reasonable profit are not being
considered. Likewise, there are local initiatives as cable
operators that provide Internet access. For these operators, in
addition to the cost of personnel and operating expenses, a
cost of reasonable investment and utility recovery must be
considered. Thus, according to experience, the value that
includes both concepts, which we will call connection cost
per client, is estimated at $20.
Fig. 3. Costs for a local provider in district outside of Lima.
TABLE I. COST STRUCTURE OF OSIPTEL FOR 2 AND 4 MBPS
FOR A REGIONAL OPERATOR SUBSIDIZED BY GOVERNMENT
Cost structure Mbps Cost ($)
A Dorsal Network Cost without taxes (1:1) 23.00
B
Internet cost per Mbps (purchased in
Lima) (1:1)
19.50
1
Components by capacity ($ without
taxes)
2 Mbps 4 Mbps
1.1 International Internet (1:10) 3.90 7.80
1.2 Dorsal Network Transport (1:10) 4.60 9.20
1.3 Regional Transport (1:10) 4.60 9.20
Sub Total (I) 13.10 26.20
2
Other operating expenses and fess ($
without taxes)
2 Mbps 4 Mbps
2.1
Other maintenance and personnel
operating expenses
15.63 15.63
2.2
Rates of FITEL (1%), OSIPTEL (0.5%)
and MTC (0.5%)
0.57 0.84
Sub Total (II) 16.20 16.47
Total access fee ($ without taxes) I+II 29.30 42.67
3 Access Fee ($ without taxes) 2 Mbps 4 Mbps
3.1 Taxes (18%) 5.27 7.68
Access Fee including taxes ($) 34.58 50.35
Access Fee including taxes (S/) 97.50 142.00

3. For local Internet operators not subsidized by the
government, the price to the end user is very high, generating
a gap between subscribers from Lima and province due to
the inefficiencies of the model. If we replace in the previous
table the amount of $15.63 that corresponds to subsidized
projects for $20 that would correspond to a private initiative
of broadband service without subsidy, the estimated service
fee with a containment of 1:10 would be for 2Mbps $39.80
and 4Mbps $55.60. As can be seen, a local initiative to
provide the Internet would not be feasible using the
backbone network and the regional transport networks.
As we know, the demand for Internet connection requires
higher capacities than 2 and 4 Mbps, even more considering
that with 5G communications, there will be speeds of
10Gbps in a user terminal. Therefore, it is considered
convenient to analyze what would be the price for capacities
of 6, 12, 20 and 120 Mbps, results that we can compare with
the rates published by OSIPTEL for the dominant operator in
the Peruvian telecom market as shows the Table II.
TABLE II. COMPARATIVE COSTS BETWEEN LOCAL OPERATOR USING
BACKBONE NETWORK AND REGIONAL PROJECT VS DOMINANT OPERATOR
OPERATOR
RATES
(MBPS)
CURR. 2 6 12 20 120
LOCAL
OPERATOR
$ 40.00 68.00 105.00 140.00 729.00
DOMINANT
OPERATOR
[6]
$ 15.00 21.00 24.00 30.00 96.70
Both the internet cost in Lima ($19.50) and transportation
costs of the backbone and regional transportation networks
are high. In these conditions any local operator connected to
the backbone network will be economically harmed.
If backbone network rate ($23) remain for 5 years and the
transportation network rate ($23) set by OSIPTEL remain
too, even when the purchase price of the Internet in Lima in
hypothetical case fall to $1, the price to the end user would
remain high. If with the same model we simulated a case of
100Mbps with oversubscription 10 to 1, with an Internet
purchase cost in Lima of $1, the resulting rate would be $582
which is very high compared with other current prices.
In the short term it is possible to have reductions in the
purchase price of Internet in range of $19.50 to $10.00.
Maintaining the fee of the backbone network and the
regional transportation network for a regional operator
subsidized by government, the fee to the end user, for
2Mbps, it would vary from $32.47 to $30.70, and for 4Mbps
from $48.30 to $43.50, which can be observed in Fig. 4.
In case of small local Internet provider without subsidy,
the connection cost is approximately $20, by performing a
sensitivity analysis for a variation of the Internet purchase
price in Lima in range $10 to $19.5, or 2Mbps the price to
the end user would vary between $52 to $49 and for 4Mbps
between $67 to $59 which can be seen in Fig. 5.
In current model, a reduction of internet price in Lima
does not have a significant impact on the reduction of the
final fee due to the high cost of the backbone network ($23)
and the regional transportation network ($23). Also, this
model limits to small local operators although the regulation
does not prevent it. The difference cost at market resulting
from the current model $40 for 2Mbps for a small local
operator and $13 for the dominant operator.
Fig. 4. Fees for regional provider subsidized by government.
Fig. 5. Rates for regional provider without subsidy.
At this point it is clear that the transport price of the
backbone network and the transport of the regional network
are very high. A question that arises is: Is it possible to
estimate the top price of the backbone network plus the
regional network by Mbps? The answer is affirmative.
IV. FEE ESTIMATION OF THE BACKBONE NETWORK
AMONG THE REGIONAL TRANSPORT
To estimate the top fee of the backbone network is
necessary to specify certain considerations:
A. The costs per Mbps of the dominant operator are low,
where the Internet service fee is equal except jungle.
B. Dominant operator has already recovered its
investment costs, thus exists O&M based costs.
C. The backbone network objective is to mass the Internet.
Consequently, the rate should not include the recovery
of the investment, only the O&M costs.
In Table II there is a big price difference between the
dominant operator's fee and that of a local operator that uses
the backbone and the regional transport network. This fees
are the result of two input variables: a) Internet cost in Lima
and b) transportation cost of the backbone network plus
regional transportation network. To estimate maximum fee,
we assume that the dominant operator uses the backbone and

4. the regional transport network and we look for the values (a)
and (b) that allow us to obtain the values of their current fees.
The international Internet transport maximum cost for a
provider as the dominant operator is in the order of $2;
consequently, the cost in Lima for the dominant operator is
set to $2. Replacing this value and looking for the transport
price that result in the closest to the current fees of the
dominant operator. Following an iterative process, we
estimate the top rate of the backbone network together with
the regional transportation network which is $3 per Mbps.
The fee determination of backbone network is not the
objective of this paper since estimating the exact value
requires further analysis [7]. The exercise performed together
with analysis, evidence the following:
A. The price of $23 per Mbps without VAT is very high.
D. The maximum fee of $23 established by OSIPTEL for
the regional transport network is higher than backbone
network cost. Also, in most cases regional transport
network do not exceed 5% of length of backbone
network.
B. The division of the national transport in two networks
excessively raises the price to $46, which contrasts with
the $3 maximum transport cost for the dominant
operator.
C. The Internet purchase price in Lima, of $19.5 estimated
by OSIPTEL, has fallen by competition effect and it is
now possible to find it from $10 up to $15, depending on
the volume of purchase.
D. It is clear that backbone network should no longer sell
capacity in Mbps units but at least in units of 100Mbps,
with volume discounts for purchases of 1Gbps or more.
(i)
As there is a greater demand, one of the questions is:
“Does the O&M costs of the backbone network allow us to
provide competitive transport prices?” The answer is yes.
The architecture of the backbone network is fully
scalable, so it is feasible to go from 10Gbps to 100Gbps
without significantly increasing the CAPEX of the Network
and without increasing the OPEX. If we take into account
that all investment, O&M costs of the backbone network are
guaranteed by government, and that the annual maintenance
cost amounts to $14.5 million, taking into account 180
provincial capitals with 10Gbps per each province, it is
possible perform an analysis to determine the cost per Mbps.
Taking into account that the fiber optic cable of the
network has 24 fibers and that the equipment located in the
distribution nodes are Service Access Switch with high
throughput it is possible to add optical network elements to
go from 10Gbps current capacity to 100Gbps, thus the
100Gbps modules on both ends of the link and using only
two fibers of the 24 available fiber optic cable of the
backbone network. The cost per Mbps for the case of
100Gbps in the provincial nodes is shown in Table III.
In the future, using 20 fibers it will be possible to pass
from 10Gbps to 1Tbps. For this scenario, the cost per Mbps
would be $0.07. According to the results, we observed that it
is possible reduce price per Mbps of the backbone network
and backbone network architecture allows to scale capacities.
TABLE III. COSTS FOR HAVING 100GBPS IN PROVINCIAL CAPITALS
10 Gbps for each province
1 000 Mbps for each province
180 Province
180 000 Mbps for each province
14 500 000 O&M Annual USD fee
80.56 Annual cost per optical fiber 10Gbps
8.06 Annual cost considering 100Gbps
0.67 Cost per Mbps (US$) Monthly
V. REMARKS ON BACKBONE NETWORK
Although the core network is prepared to meet the
current demand, restricted by the high costs generated by the
current model, the network is prepared to increase its
capacity up to 100λ with speeds of 10/100Gbps, however the
capacity to connect to the provincial capitals is only 10Gbps,
which should be increased to 100Gbps. The Fig. 6 shows the
proposal for a new model with the following criteria:
A. A single transport network to the district capital
transporting greater capacities
B. Marketing of 100Mbps units. It would allow equal fees
across country
C. Allow to network operator to sell Internet to other
providers, generating more competition in the market
D. Eliminate the restrictions that the backbone network
currently has, that limit its income and efficient use
Fig. 6. New model for backbone and regional Internet access networks..
VI. CONCLUSIONS
The regulatory model of backbone network must be
modified to enable an efficient use of the network. Also, the
backbone network must reach to the district capital cities.
The cost of $23 for the backbone network is very high at
have two transport networks to reach the district capital city.
The operation cost of transport is not sensitive to the
increase in capacity over a wide range, so that, the cost-per-
Mbps scheme should be changed.
It is necessary to eliminate some restrictions in backbone
network, for example, allow to market the Internet with
others international operators, which would allow it to adjust
efficiently to the current dynamics of the market.
A massive use of the backbone network will lead to
greater social benefits, boosting the economy of the country.

5. REFERENCES
[1] Karina Montoya, “Red de Redes a la Peruana”, Revista Semana
Económica , N° 1475, pp. 8-10, Junio 2015.
[2] A. García Cevallos y E. Iglesias Rodríguez, “Informe Anual del
Índice de Desarrollo de la Banda Ancha en América Latina y el
Caribe (IDBA 2016)”, 2017.
[3] Ley Nº 29904, “Ley de Promoción de la Banda Ancha y Construcción
de la red Dorsal Nacional de Fibra Óptica”, Julio 2012.
[4] MTC, “Plan Nacional para el Desarrollo de la Banda Ancha en el
Perú, Ministerio de Transportes y Comunicaciones”, 2011.
[5] OSIPTEL, “Informe Nº 007-GPRC/2015”, Lima, 2015
[6] https://www.osiptel.gob.pe/documentos/oferta-comercial-2018.
[7] Patrick Maillé, Bruno Tuffin, “Telecommunication Network
Economics From Theory to Applications”, Cambridge University
Press, April 2014