10 8320 Final Model With Tracks Mobile

8/4/2019 10 8320 Final Model With Tracks Mobile

1/83

Report for the National Post and TelecomAgency (PTS)

New mobile long-run incremental

cost (LRIC) model

Documentation for the final cost model

16 May 2011

Ref: 13392-86b

.


2/83

Ref: 13392-86b .

Contents

1

Introduction 1

1.1 Structure of this document 12 Background to the model 32.1 Motivation for the new cost model 32.2 Summary of the new cost model 42.3 Overall flow of the new model 53 Operator input template 84

Market calculations 10

4.1 Voice traffic 104.2 Data traffic 124.3 Worksheet MarketDemand 155 Demand calculations 165.1 WorksheetNetworkLoad 165.2 WorksheetNetworkShare 196 Network calculations 216.1 Network design inputs 6.2 Radio network 246.3 LMA 276.4 Hub to core transmission 286.5 BSCs and RNCs 296.6 Remote BSC and remote RNC to core transmission 306.7 Core-to-core transmission 316.8 Switches and support systems 317 Expenditure 337.1 WorksheetInAsset 337.2 WorksheetFullNw 347.3 WorksheetNwDeploy 347.4 Worksheet CostTrends 357.5 Worksheet UnitCapex 367.6 Worksheet UnitOpex 367.7 Worksheet TotalCapex 367.8 Worksheet TotalOpex 378 Depreciation 388.1 Overview of economic depreciation 38


3/83

New mobile long-run incremental cost (LRIC) model

Ref: 13392-86b .

8.2 WorksheetRFs 408.3 WorksheetNwEleOut 418.4 WorksheetDF 418.5 WorksheetED 429 Results 449.1 Calculation of LRAIC(+) 449.2 Calculation of pure LRIC 469.3 WorksheetResults 4910 Supplementary worksheets 5010.1 WorksheetLists 5010.2 WorksheetAreaToPop 5010.3 WorksheetErlang 5111 How to use the new model 5211.1 Basic operation 5211.2 Adding additional operators 5311.3 Worksheet Ctrl 53

Annex A Acronyms 55

Annex B

Changes made to the draft model during finalisation 59

Annex C Source of the inputs used in the model 62


4/83

New mobile long-run incremental cost (LRIC) model

Ref: 13392-86b .

Copyright 2011. Analysys Mason Limited has produced the information contained herein

for PTS. The ownership, use and disclosure of this information are subject to the

Commercial Terms contained in the contract between Analysys Mason Limited and theNational Post and Telecom Agency (PTS).

Analysys Mason Limited

St Giles Court

24 Castle Street

Cambridge CB3 0AJ

UK

Tel: +44 (0)845 600 5244

Fax: +44 (0)1223 460866

[email protected]

www.analysysmason.com

Registered in England No. 5177472


5/83

New mobile long-run incremental cost (LRIC) model | 1

Ref: 13392-86b .

1 Introduction

The National Post and Telecom Agency (Post-och telestyrelsen, or PTS) has commissioned

Analysys Mason Limited (Analysys Mason) to develop a long-run incremental cost (LRIC)

model for the purposes of understanding and regulating the cost of mobile voice termination in

Sweden. This wholesale service falls under the designation of Market 7, according to the European

Commission (EC) Recommendation on relevant markets.1

Analysys Mason and PTS have agreed a process to deliver the LRIC model, which will be used by

PTS to inform its regulation for mobile termination. This process presents industry participants

with the opportunity to contribute at various points during the project.

The first phase of the project to set out the specification for the cost model has been completed,

and a final model specification was issued on 17 January 2011. This paper also sets out the

background to the process.

The draft cost model has been developed, reflecting this model specification, and exploring a

number of critical modelling aspects which have been discussed with interested industry parties at

a meeting in Stockholm (10 February 2011). This document accompanies the final model released

after consultation with industry parties.

1.1 Structure of this documentThe remainder of this document describes the new mobile LRIC model and is structured as

follows:

Section 2 summarises the background to this modelling work

Section 3 describes the operator input template, which allows alternative network

configurations to be considered within the model

Section 4 describes the market-related calculations

Section 5 describes the demand-related calculations

Section 6 describes the network design calculations

Section 7 describes the expenditure calculations

Section 8 describes the depreciation calculations

1

Seehttp://ec.europa.eu/information_society/policy/ecomm/doc/implementation_enforcement/article_7/recom_term_rates_en.pdf


6/83


Ref: 13392-86b .

Section 9 describes the display of results in the model

Section 10 describes a small number of supplementary worksheets in the model

Section 11 describes how a user can operate the model.

A supplementary annex includes a list of the acronyms used within this document.

Additional annexes describe the changes made in finalising the model and additional descriptions

on the inputs to the model.


7/83


Ref: 13392-86b .

2 Background to the model

This section summarises the background to the modelling work, as follows:

Section 2.1 describes the motivation for designing the new cost model

Section 2.2 summarises the principles of the new cost model

Section 2.3provides an overview of the flow of information in the new cost model.

2.1 Motivation for the new cost modelIn 2004, Analysys Consulting Limited built a bottom-up mobile LRIC model for the PTS, with the

aim of calculating the cost of voice termination for the GSM mobile operators in Sweden. In

2007/2008, an upgrade process was undertaken so that UMTS networks could be included within

the model. Quality checks were subsequently undertaken of this upgraded model in 2009 and

2010. The latest version of this model (v6.3) was released in June 2010.

The previous approach calculated the costs of seven actual networks and blended the costs of these

networks together into the actual costs of the operators, based on the infrastructure-sharing

relationships present between the four major mobile network operators (MNO) in Sweden. This

resulted in the modelling of effectively seven separate networks. Due to limitations on the flow of

information from the shared network joint ventures to their parent companies, it was difficult to

demonstrate full reconciliation transparently to all MNOs.

We note that this lack of flow of information has intensified since the previous model upgrade in

2008, with other joint ventures now present in the market (e.g. the Net4Mobility (N4M) GSM joint

venture between Telenor and Tele2). This increasing complexity is summarised in Figure 2.1.

Figure 2.1: Increasing complexity of mobile network infrastructure sharing in Sweden [Source:

Analysys Mason]

Relevant networks in

the 2003 original model


the 2007/2008 upgrade


the 2010/2011 upgrade

GSM network UMTS network LTE networkKEY:

Tele2 Telenor

TeliaSonera

Tele2

TeliaSonera

SUNAB

Tele2 Telenor

TeliaSonera

3GIS

Hi3G

Telenor

SUNAB

3GIS

Hi3G

N4M

N4M


8/83


Ref: 13392-86b .

A new LRIC model has thus been proposed for the current modelling process, which no longer

captures actual operators explicitly, but seeks a simpler approach.

2.2 Summary of the new cost modelAnalysys Mason has developed a new mobile LRIC model for PTS, to provide cost-based

information for future wholesale termination regulation in Sweden. This bottom-up model has

been developed using demand and network parameter information submitted by Market 7

stakeholders in Sweden, combined with estimates and calculations performed by Analysys Mason.

The three broad types of inputs that feed into the LRIC model calculation are related to network

design, service volumes and costs, as shown below in Figure 2.2.

Figure 2.2: Overview of

the new mobile LRIC

model [Source: Analysys

Mason ]

The model then calculates long-run incremental costs for mobile network operations in Sweden.

These service costs are derived using both long-run average incremental cost (LRAIC) and pure

long-run incremental cost (pure LRIC) principles. The latter is in accordance with the EC

Recommendation, as referenced in in Section 1. This requires the LRIC model to be run twice,

under different situations, as shown in Figure 2.3.

Network cost

model:

Schedules of asset

volumes, total service

output, total capex,

total opex

Network designinputs

(e.g. technologies,

coverage)

Traffic inputs

(e.g. volumes

carried by service,

busy-hour

characteristics)

Cost inputs

(e.g. unit capex,unit opex, cost

trends, asset

lifetimes)


9/83


Ref: 13392-86b .

Figure 2.3: Costing approaches within the new LRIC model [Source: Analysys Mason]

A variety of operator network configurations can be defined by choosing the input parameters

appropriately in the model. The model has been set up to calculate costs for a generic Swedish

operator, but it is also capable of reflecting different configurations through inputs on market

share, spectrum and coverage, including configurations similar to the actual MNOs.

For a configuration defined by a given set of inputs, the model derives the assets in a forward-

looking manner and then determines the costs of these assets over a specified timeframe (up to 50

years).

These costs are then recovered by the services assumed to be conveyed over this network during

its lifetime using an economic depreciation calculation. Capital costs are determined using a

weighted average cost of capital (WACC) determined by PTS in a separate workstream. No

remaining terminal value is applied within the LRIC model at the end of the cost recovery period.

The model applies the scorched-node principle, as described in the final model specification

referenced in Section 1. This allows some top-down validation of the bottom-up asset calculation.

In particular, based on operator information, we have:

compared the modelled number of radio sites with the actual number (by geotype)

used typical average numbers of switch locations to identify a reasonably efficient, typical

network structure for a modern national operator.

In addition, the overall expenditures in the model have been checked in aggregate against the total

top-down expenditure information submitted to us by the mobile operators.

2.3 Overall flow of the new modelThe overall flow of the new LRIC model is shown below in Figure 2.4.

Network cost

model:

Schedules of assetvolumes, total

service output, total

capex, total opex

Network design

inputs

(e.g. technologies,

coverage)

Traffic inputs

(e.g. volumes carriedby service, busy-hour

characteristics)

Cost inputs

(e.g. unit capex, unit

opex, cost trends,

asset lifetimes)

LRAIC calculation

(as in the previous model)

Pure LRIC calculation

(calculate difference in the

two cases, as in the EC

Recommendation)

Run network cost

model with all

traffic

Run network cost

model with all

traffic except

termination

increment volume


10/83


Ref: 13392-86b .

Figure 2.4: Overview of the model calculation flow [Source: Analysys Mason]

The model uses PTS market information (from 2008 onwards) as inputs, and projects these market

parameters over time in order to have a long-term forecast for the model calculations. Demand and

network inputs are defined, either as universal standard parameters or for specific operator

definitions (e.g. generic average operator). Maximum utilisation factors are applied to various

network element capacities in order to reflect realistic and design loading.

The network requirement is combined with cost inputs which determine how much capital and

operating expenditures (capex and opex) are required for the network, including the ongoing

replacement of assets. The model depreciates the expenditures over time, using an economic

depreciation algorithm which takes into account network output (based on LRAIC routeing

factors), price trends, and a discount rate to reflect the return on capital employed (i.e. the time-

discounting of cost recovery relative to expenditure outflow).

Finally, the model produces two sets of outputs:

the costs of termination according to the LRAIC+ methodology

a pure LRIC of termination which is derived by running the model twice (once with, and once

without, wholesale termination traffic).

In this model documentation we denote the source of various inputs as follows:

[1] Analysys Mason estimate

[2] Analysys Mason estimate informed by operatorinputinformation or data

Input Calculation Output

4. Costs 5. Depreciation 6. Results

PTS market

information

Total market Network drivers

Operatorspecification/

market share

Network

design inputs

Network

requirement

Projections Demand drivers Maximum

utilisation %

Asset inputs

Unit costs and

cost trends

Capex and opex

LRAIC routeing

factors

Annualised

economic costs

Network

common costs

and EPMU

Discount rate

1. Market 2. Demand 3. Network

LRAIC+

(as previous

model)

Pure LRIC

(difference in the

two cases)

Run network cost

model with all traffic

Run network cost

model with all traffic

except MT volume

Key:


11/83


Ref: 13392-86b .

[3] Analysys Mason estimate informed by operatoroutputinformation or data (e.g. scorched-

node reference to total amounts of operator equipment, or reconciliation reference to total

amounts of opex)

[4] Swedish market average based on operator data (rounded or standardised where

appropriate)

[5] standard technical parameter

[6] operator-specific input or choice.


12/83


Ref: 13392-86b .

3 Operator input template

The model is set up so that a subset of inputs is defined in an operator template. This template

forms a separate worksheet in the LRIC model. Through this method, additional operator

templates can be added to the model by:

duplicating the template worksheet and renaming the worksheet to beInput_(new name)

ensuring that the new operator worksheet name is added to the list of operators, in the Lists

worksheet, column Z

selecting the new worksheet name from the operator selector in the model control panel.

The structure of the inputs on theInput_(new name) worksheet is summarised below.

1. Share of market Specifies the share of the national market for GSM, UMTS, HSPA

(mobile broadband) and LTE (mobile broadband) traffic.

Specifies where the operator has network deployed (e.g. it can be used

for 3GIS to reduce network deployment in urban areas).

2. Coverage and

spectrum

Specifies the population covered by each of the different technologies

for each year the model is running.

Specifies the number of urban micro sites for coverage.

Specifies the frequency used to deploy coverage for each network. This

is used to calculate the number of coverage sites required based on a

predetermined cell radius by frequency and geotype.

Specifies the amount of paired spectrum by technology and whether this

spectrum is used for coverage or capacity.

Specifies the number of UMTS channels set aside for UMTS rather thanHSPA traffic.

3. Network design

parameters

Specifies the proportion of links that are leased and the transmission

protocol they use in each geotype.

Specifies the proportion of sites collocated with hubs and the

transmission protocol they use in each geotype.

Specifies the proportion of sites connected via a hub to the core

network, rather than being connected directly to the core network, thenumber of sites per hub, and the number of hubs per hub-core


13/83


Ref: 13392-86b .

transmission link, in each geotype.

Specifies the number of locations where base station controllers (BSC)

and radio network controllers (RNC) are deployed, and the share of

radio traffic in the suburban or rural geotypes that is handled by a BSC

or RNC in the same geotype rather than being transferred to a BSC or

RNC in the urban geotype.

Specifies the transmission protocol used by BSC/RNC to core nodes for

voice and data.

Specifies the number of core sites in each geotype (by default 0 except

in the urban geotype).

Specifies the proportion of voice and data conveyed across core-core

links, and the transmission protocol.

4. Adjustment

factor for operator

assets

By default, all adjustments are set to 100%. However it is possible to

use this input to remove or reduce various assets from the cost base of

individual operators.


14/83


Ref: 13392-86b .

4 Market calculations

The model uses PTS statistics on the total market in Sweden to drive the forecasts for both mobile

market subscribers and traffic. This market information is then rearranged to suit the categories

used in the model. Three subscriber types are modelled: voice-only handset, voice+data handset,

and mobile broadband laptop/dongle. Voice and data traffic are treated separately. Both are split

into sub-categories: incoming, outgoing and on-net traffic for voice; handset data usage and

mobile broadband data usage for data. Both are also split into the different access technologies

used. SMS is modelled as voice-equivalent traffic, but has very little impact on the large-scale

network.

An outline of the market calculation is shown in Figure 4.1.

Figure 4.1: Market calculation steps [Source: Analysys Mason]

The rest of this section describes the voice traffic (Section 4.1) and data traffic (Section 4.2)

captured in the model, and concludes with a summary of the structure of the MarketDemand

worksheet (Section 4.3).

4.1 Voice trafficHistorical total voice traffic and number of subscribers from 1H 2008 to 2H 2010 are used to

derive a forecast for the duration of the model. Originated traffic from mobiles (including on-net)

and incoming traffic both increase until 2013. Originated traffic is assumed to increase at a faster

PTS market

information

20082010

Mobile-originated

minutes per

subscriber

Mobile-terminated

minutes per

subscriber

Projected growth

201020152020

Projected growth

201020152020

Mobile-originated

SMS per

subscriber

Mobile-terminatedSMS per

subscriber

Projected growth

201020152020

Projected growth

201020152020

Proportion of

people who

use data

Proportion of

people who have

broadband

Projected growth

201020152020

Projected growth

201020152020

Data traffic per

handset user

Data traffic per

broadband user

Total market

200820102010 = 1H + estimated 2H, to be updated in May 2011


15/83


Ref: 13392-86b .

rate than incoming traffic. Usage per subscriber is then assumed to have reached a steady state,

remaining constant from 2013 onwards. This evolution is shown in Figure 4.2.

Figure 4.2: Evolution of

voice usage in Sweden

[Source: PTS, Analysys

Mason]

From this traffic by user, and assuming the number of voice users remains constant from 2010

onwards, the total voice traffic is calculated for on-net, outgoing (excluding on-net) and incoming

traffic. These three categories are added up in Figure 4.3, showing that total voice traffic is

forecast to increase from 32 billion minutes in 2010 to 36 billion minutes in 2013.


total voice usage in

Sweden [Source: PTS,

Analysys Mason]

In the previous model, the share of voice traffic carried by the GSM network decreased until

0

50

100

150

200

250

2008 2010 2012 2014 2016 2018 2020

Mobile originated minutes per user per month

Incoming minutes per user per month

ActualForecast

0

5

10

15

20

25

30

35

40

2008 2010 2012 2014 2016 2018 2020

Billions

Mobile incoming minutesMobile outgoing minutesMobile on-net minutes

Actual Forecast


16/83


Ref: 13392-86b .

disappearing in 2016. In this model, it is no longer assumed that the GSM network will be shut

down. The main reason for this new assumption is that it is now known that Telenor and Tele2 are

jointly deploying a new GSM network under the N4M joint venture. The share of voice traffic

carried on the 2G network is now assumed to decrease to 40% in 2013, remaining constant

thereafter. Figure 4.4 illustrates this new forecast.


the share of voice traffic

by technology [Source:

PTS, Analysys Mason]

4.2 Data trafficThe model continues the strong growth of handset data users, associated with the increasing

penetration of smartphones. From 54% in 2010, the proportion of handset data users is forecast to

reach a steady state of 74% of voice subscribers from 2017, as illustrated in Figure 4.5. Mobile

broadband (i.e. dongle) users are much fewer in number, representing only 16% of voice

subscribers in 2010. This proportion is assumed to increase faster than handset data users to reach

24% of voice subscribers in 2019, remaining constant thereafter.

0%

10%

20%

30%

40%

50%

60%

70%

80%

2008 2010 2012 2014 2016 2018 2020

Percentage of total voice traffic on the 2G network

Percentage of total voice traffic on the 3G network

Actual Forecast


17/83


Ref: 13392-86b .

Figure 4.5: Proportion of

voice subscribers who

are also data users


Mason]

Data usage per handset data subscriber is forecast to remain constant from 2010, as illustrated

below in Figure 4.6.


data usage for handsets

and dongles [Source:


On the other hand, data usage from dongles (or broadband) is assumed to start at a much higher

use per subscriber than handset usage, and it is forecast to approximately double between 2010 and

2014. As a result, the vast majority of data traffic is expected to originate from dongles, as shown

in Figure 4.7.

0%

10%

20%

30%

40%

50%

60%

70%

80%

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Proportion of users who are handset datausers

Actual Forecast

0

1,000

2,000

3,000

4,000

5,000

6,000

0

20

40

60

80

100

120

2008 2010 2012 2014 2016 2018 2020

Broa

dban

dusers

Han

dse

tusers

Mbytes per handset data user per month

Mbytes per broadband data user per month

Actual Forecast


18/83


Ref: 13392-86b .


total data usage for

handsets and dongles


Mason]

Between 2008 and 2010, HSPA is assumed to carry almost all of the data traffic (HSDPA for the

downlink traffic and HSUPA for the uplink traffic). R99 is forecast to decline quickly to comprise

only a small proportion of data traffic in 2011. GPRS and EDGE are marginal throughout the

modelling period, and LTE has not yet reached significant volumes. From 2011, LTE is assumed

to grow steadily to account for 45% of the total data traffic by 2014, whilst HSDPA and HSUPA

are projected to decline to 43% and 12% of data traffic, respectively, as shown below in Figure

4.8. This decline in the share of HSPA traffic is not linked to a decline in data volumes carried, as

data volumes are assumed to quadruple between 2010 and 2018, but rather indicates that most of

the increased traffic is carried on LTE networks rather than HSPA networks.

0

20

40

60

80

100

120

140

160

180

2008 2010 2012 2014 2016 2018 2020

Billions

Total Mbytes of broadband data Total Mbytes of handset data

Actual Forecast


19/83


Ref: 13392-86b .


the share of data traffic

by technology [Source:


4.3 Worksheet MarketDemand1. PTS source

market information

Collates data from the PTS market statistics from 1H 2008 to 2H 2010.

Reorders the data into the categories used in the model and only keeps

end-of-year values.

2. Forecast marketinformation

Derives forecasts for the whole duration of the model (until 2058),starting from the existing market information. [1, 4]

Splits the traffic by technology (2G/3G for voice traffic and SMS,

GPRS/EDGE/R99/HSDPA/HSUPA/LTE for data traffic) and by device

(handsets or dongles). [2]

3. Total market

volumes

Calculates the volume of traffic by technology and service for the whole

duration of the model.

4. Output total

market volumes

Calculates the total volume of traffic by technology and service after

applying a sensitivity multiplier.

0%

10%

20%

30%

40%

50%

60%

70%

80%

2008 2010 2012 2014 2016 2018 2020

GPRS EDGE R99

HSDPA HSUPA LTE

Actual Forecast


20/83


Ref: 13392-86b .

5 Demand calculations

The demand calculations are used to determine the traffic measures that dimension the network of

the modelled operator. They determine, from the whole market and the market share of thisoperator, what is the peak traffic load that the network needs to be able to handle. This is

calculated based on the share of traffic in the busy hour, the average duration of voice calls, and

the proportion of data traffic in the busiest data path (uplink or downlink).

The remainder of this section is structured as follows:

the calculation of network loading on theNetworkLoadworksheet (Section 5.1)

the spreading of this load across the modelled geotypes (Section 5.2).

5.1 Worksheet NetworkLoadThis worksheet calculates the loading at the various levels of the network based on the traffic

throughput.

1. Market share Links in the total market and the operatorsmarket share.

Calculates the average number of voice/voice+data subscribers and

mobile broadband subscribers.

2. Total volumes

for the network

Multiplies the total market by the operators market share to obtain the

total volume of traffic carried by the selected operator.

3. Load

calculations

Specifies inputs for busy days, busy-day traffic and busy-hour traffic.

[2, 4]

Calculates busy-hour Erlangs (BHE) for each voice service.

Specifies inputs for call attempts, ring minutes per call and radio

loading factors. [2, 4]

Specifies inputs for average call duration. [4]

Calculates BHE in the radio network for each voice service.

Calculates SMS in the busy hour.

Specifies inputs for the proportion of data service traffic in the uplink

versus the downlink. [1,2]


21/83


Ref: 13392-86b .

Calculates data Mbit/s in the busy hour.

4. Radio load

for voice

Calculates total 2G and 3G BHE in the radio network.

5. Network load for

GPRS and EDGE

Calculates downlink busy-hour Mbit/s for voice and data traffic in their

respective busy hours.

Determines total GPRS and EDGE Mbytes, using a conversion factor

for EDGE traffic [1].

6. Network load

for UMTS and R99

data

Calculates BHE for voice and data traffic in their respective busy hours,

converting R99 UMTS data into voice-equivalent channels using a

conversion factor based on the assumed CE rate for R99 data [1].

Calculates the peak BHE, by taking the maximum of the voice busy

hour and the data busy hour.

7. Network load

for HSPA

Links in total HSDPA and HSUPA load in the data busy hour.

8. Network load

for LTE

Links in total downlink LTE load in the data busy hour.

9. Network load for

traffic from radio

layer into core/ring

network

Defines the amount of provisioned bandwidth for supporting the busy-

hour Mbit/s in the radio network for each data bearer [5].

Calculates the provisioned downlink data in the voice and data busy

hours for data traffic.

Defines the amount of provisioned bandwidth for supporting the voice

BHE in the radio network [5].

Calculates the provisioned upstream/downstream data for voice traffic

in both the voice and data busy hours.

Calculates the total load in both the voice and data busy hours, and the

peak load, by taking the maximum of the voice and data busy hour.

10. Network load

for BSC traffic

Defines the amount of provisioned bandwidth for supporting BSC-core

data traffic [5].


hours for data traffic.



22/83


Ref: 13392-86b .

BHE in the radio network in BSC-core links [5].

Calculates the provisioned duplex data in the voice and data busy hours

for voice traffic.

Calculates the total load in the voice and data busy hours, and the peak

load, by taking the maximum of the voice and data busy hour.

11. Network load

for RNC traffic

Defines the amount of provisioned bandwidth for supporting the busy-

hour Mbit/s in the radio network. This is undertaken for each data bearer

and is calculated in terms of the RNC throughput. [5]


hours for data traffic through the RNC.


BHE in the radio network through the RNC. [5]

Calculates the provisioned duplex data in the voice and data busy hours

for voice traffic through the RNC.

Calculates the total load in the voice and data busy hours, and the peak

load, by taking the maximum of the voice and data busy hour.

12. Network load

for core-core traffic

Calculates the network BHE for voice traffic in the voice and data busy

hours.

Calculates the amount of core-core busy-hour Mbit/s for voice traffic in

the voice and data busy hours, by applying the proportion of voice

traffic that is conveyed between core sites.

Calculates the amount of core-core busy-hour Mbit/s for data traffic in

the voice and data busy hours, by applying the proportion of data traffic

that is conveyed between core sites.

Calculates the peak core-core Mbit/s, by taking the maximum of the

voice and data busy hour.

13. Network load

for switches and

servers

Calculates the load on the data servers using the number of data

subscribers and inputs for active packet data protocols (PDP)

contexts [1] and simultaneous active users (SAU) [1].

Calculates the number of minutes in a busy day for the wholesale billing

system.

Calculates the number of 2G and 3G call attempts in the busy hour.


23/83


Ref: 13392-86b .

Calculates the number of SMS in the busy hour.

For each server in the list, calculates or links in the individual load

amount.

5.2 Worksheet NetworkShareThis worksheet splits out network loading by geotype.

1. Traffic by

geotype

Determines input for the proportion of traffic by geotype, if full network

coverage. [2,3]

Determines input for the proportion of national traffic which occurs on

micro sites. [2,3]

Links in the network location inputs.

2. Coverage by

geotype

Links in the population coverage by technology, and micro sites for

coverage.

Calculates area coverage by geotype.

Calculates the proportion of population covered by each technology in

each geotype.

Calculates the actual distribution of traffic within the covered areas.

3. Network GSM

voice traffic by

geotype

Links in 2G voice BHE in the radio network.

Calculates 2G voice BHE in the radio network by geotype.

4. Network UMTS

R99 voice traffic by

geotype

Links in UMTS R99 BHE in the radio network.

Calculates UMTS R99 BHE in the radio network by geotype.

5. Network HSPA

traffic by geotype

Links in HSDPA and HSUPA busy-hour Mbit/s of the radio network.

Calculates HSDPA and HSUPA busy-hour Mbit/s of the radio network

by geotype.

6. Network LTE

traffic by geotype

Links in downlink LTE busy-hour Mbit/s of the radio network.

For each geotype, calculates the LTE busy hour Mbit/s, in the downlink,

of the radio network.


24/83


Ref: 13392-86b .

7. Network traffic

into RNC/BSC

core nodes

Links in peak Mbit/s load passing into the core network.

Calculates peak Mbit/s load passing into the core network by geotype.

8. Network traffic

for RNC Mbit/s

Links in peak RNC load in Mbit/s.

Calculates peak RNC Mbit/s passing into the core network by geotype.


25/83


Ref: 13392-86b .

6 Network calculations

The network calculations within the model take the demand drivers and other network inputs and

compute the number of each network element that is needed. The structure and nature of the

network design inputs is described in Section 6.1. These network design calculations cover the full

range of layers in the network hierarchy, as follows:

network design inputs and utilisation factors (Section 6.1)

radio network (Section 6.2)

last-mile access (Section 6.3)

hub to core transmission (Section 6.4)

BSCs and RNCs (Section 6.5)

remote BSC and remote RNC to core transmission (Section 6.6)

core-to-core transmission (Section 6.7)

switches and support systems (Section 6.8)

6.1 Network design inputsNetwork design inputs are either operator-specific or universal. Operator-specific inputs are linked

(using an INDIRECT function) from the relevant operator input template. Universal network

design inputs are entered in this part of the model.

6.1.1Worksheet NetworkDesignInputs1. Coverage Cell radius for outdoor coverage [1].

Cell pi which is used to calculate the cell area covered [5].

Frequency used for coverage added in each year, linked from the

selected operator [6].

2. Spectrum Amount of paired spectrum in each coverage and capacity layer, linked

from the selected operator. [6]

Size of a radio channel, in MHz. [5]

Calculation of the number of channels available.


26/83


Ref: 13392-86b .

Number of UMTS channels reserved for voice and low-speed R99 data

(not HSPA). [6]

Number of channels available for traffic load.

3. GSM capacity Input of cell reuse factor. [5]

Input of the average sectorisation of GSM sites. [2]

Input of physical TRX per sector limit, along with the calculation of the

effective limit on average by geotype. [2]

Calculation of the maximum number of TRX per sector, either by

spectrum or by geotype.

Input of the number of GSM channels reserved for GPRS/EDGE packetdata and for signalling. [1, 5]

Input of GSM channel rates. [5]

Input of GSM blocking probability. [1, 5]

Calculation of Erlang capacity per site.

4. UMTS capacity Input of R99 channel rate in Mbit/s. [1, 5]

Input of the average sectorisation of UMTS sites. [2]

Input of soft- and softer-handover overheads. [1, 5]

Input of the number of R99 signalling channels per carrier, minimum

and maximum R99 carriers per carrier (pooled at the NodeB). [1, 5]

Input of UMTS blocking probability. [1, 5]

Calculation of Erlang capacity per carrier (pooled at the NodeB).

Calculation of Erlang capacity per site.

Input of channel kit size (in CE). [5]

5. HSPA capacity Input of the cell peak to effective rate for data throughput. [1]

Specification of the HSDPA and HSUPA rate ladder. [5]

6. LTE capacity Input of the cell peak to effective rate for data throughput. [1]


27/83


Ref: 13392-86b .

Specification of the LTE rate ladder. [5]

7. Physical sites Input of percentage for sites deployed as single technology or co-located

sites. [1]

Input of percentage of sites which are deployed on third-party

infrastructure. [1]

8. LMA and hub

to core

Specification of the LMA and hub-core rate ladders. [5]

Linked operator inputs for site transmission choice, hub co-location,

leased LMA, and hub-core link parameters for rings or point-to-point

hub-core transmission. [6]

9. RNC and BSC Linked operator inputs for the number of BSC/RNC locations, and the

proportion of load served in each geotype. [6]

Specification of the BSC and RNC capacity ladders. [5]

10. BSC-core

traffic

Specification of the remote BSC-core rate ladder. [5]

Input for the redundancy in BSC-core links. [1, 5]

11. RNC-core

traffic

Specification of the remote RNC-core rate ladders. [5]

Input for the redundancy in RNC-core links. [1, 5]

Linked operator input for the protocol used for voice and data

interfaces. [6]

12. Core-core

traffic

Linked operator input for the number of core sites, proportion of traffic

conveyed across the core, and transmission protocol for voice and data

layers. [6]

Specification of the core-core rate ladder, and number and distance of

hops in the dark-fibre core network. [1, 3]

13. Switches and

servers

Input of capacity for each network element in the list. [1, 2, 3]

Input of the minimum number and redundancy multiplier for each

network element in the list. [1]

14. Specify scope of

operator assets

Linked operator input for the specific assets which are included in each

operator network. [6]


28/83


Ref: 13392-86b .

6.1.2Worksheet NetworkUtil1. Network capacity

utilisation factors

for calibration

Maximum utilisation factors for network capacity for each set of

network elements. [1, 3, 5]

6.2 Radio networkThe network design for the radio layer considers the three technologies (GSM, UMTS and LTE)

with radio capacity upgrades, as well as the physical site requirements (single technology sites, co-

located sites, own tower sites and third-party installations). The network design first considers sites

for coverage and then considers the radio interface traffic loading to calculate the additional assets

required to carry this loading.

Figure 6.1: Overview of the modelled radio networks [Source: Analysys Mason]

6.2.1Worksheet NwDesRadioCov1. GSM radio

network coverage

Links in the area to be covered.

Calculates area coverage added in each year.

Links in area per site.

Calculates the number of sites added for coverage in each year.

Calculates the total number of sites for coverage.

2. UMTS radio

As above but for UMTS.

BTS

TRX

Ancillary

power

Tower

Rooftop or

third-party

site

NodeB

R99-CK

Ancillary

power

eNodeB

LTE

Ancillary

power

HSPA

GSM

UMTS

LTE

Shared

Not costed

Site acquisition and

preparation

R99 CK = 16 duplex CEHSDPA

upgrades in

Mbit/s per25MHz:

1.8

3.6

7.2

10.1

14.1

21.1

HSUPA

upgrades in

Mbit/s per25MHz:

0.73

1.46

2

2.93

5.76

11.5

LTE

upgrades in

Mbit/s per25MHz:

10.8

16.2

21.6

32.4

43.2

86.4

own tower

or third

party


29/83


Ref: 13392-86b .

network coverage

3. HSPA radio

network coverage

As above but for HSPA.

4 LTE radio

network coverage

As above but for LTE.

6.2.2WorksheetNwDesLoadPart of this worksheet contains the radio network calculation, for each technology.

1. GSM capacity

calculation

Links in sites for coverage and voice BHE.

Calculates capacity of the coverage deployment.

Calculates BHE which cannot be supported by the coverage deployment

and must be supported by capacity upgrades.

Calculates the number of capacity BTS layers which must be added to

coverage sites.

Calculates BHE which cannot be supported by upgraded coverage sites,

and must have new sites deployed.

Calculates the number of new (capacity) sites needed to support

remaining BHE.

Calculates the total number of GSM sites and BTS.

Calculates the number of TRX in the coverage layer of coverage sites.

Calculates the number of TRX in the coverage layer of capacity sites.

Calculates the number of TRX in the capacity layers.

Calculates the number of TRX in total.

Checks whether the reservation of channels for GPRS is sufficient for

the average throughput required.

2. UMTS capacity

calculation

Links in sites for coverage and R99 BHE.

Calculates capacity of the coverage deployment.

Calculates BHE which cannot be supported by the coverage deployment


30/83


Ref: 13392-86b .

and must be supported by capacity upgrades.

Calculates the number of capacity carrier layers which must be added to

coverage sites.

Calculates BHE which cannot be supported by upgraded coverage sites,

and must have new sites deployed.

Calculates the number of new (capacity) sites needed to support

remaining BHE.

Calculates the total number of UMTS sites.

Calculates the total number of UMTS R99 NodeBs.

Calculates the total number of R99 carriers and CK in the coveragecarriers of NodeBs.

Calculates the total number of R99 carriers and CK in the additional

capacity carriers of NodeBs.

3. HSDPA ladder

calculation

Links in sites for coverage and BH Mbit/s.

Calculates BH Mbit/s per site.

Calculates maximum capacity based on the rate ladder and the numberof carriers (spectrum) available.

Checks that there are sufficient UMTS sites deployed to support the data

upgrades.

Calculates BH Mbit/s per site.

Calculates the rate needed in the first, second, third and fourth carrier

upgrade (if needed).

Calculates the number of sites at each step of the rate ladder.

4. HSUPA rate

ladder

As above except for HSUPA.

5. LTE rate ladder As above except for LTE.

6. Physical sites Calculates the number of GSM, UMTS and LTE sites on single-

technology sites, using leased or microwave LMA.

Calculates the number of multi-technology sites which are co-locating


31/83


Ref: 13392-86b .

multiple radio layers, using leased or microwave LMA.

Calculates the number of sites on own towers and on third-party sites.

6.3 LMAThe LMA network is common for all three radio network technologies. It considers two

transmission protocols (ATM/SDH/PDH and Ethernet) with capacity upgrades, as well as the

physical transmission infrastructure (which can be either leased lines or microwave links).

Figure 6.2: Overview of the modelled LMA networks [Source: Analysys Mason]

6.3.1WorksheetNwDesLoad7. LMA Calculates the LMA capacity requirement for single-technology sites.

Calculates the LMA capacity requirement for multi-technology sites.

Determines the actual capacity of LMA links by geotype according to a

predefined ladder of options.

Calculates the number of leased-line LMA links and self-provided/

microwave LMA links by rate according to that same rate ladder.

Sites (single or multi-technology)

Next node in the network (site, hub, BSC/RNC, core)

Leased line

Microwave link

Traditional

for GSM

Mbit/s

2

4

8

16

32

155

622

Ethernet

for others

Mbit/s

10

30

100

150

200

300

1000

Traditional

for GSM

Mbit/s

2

4

8

16

32

155

622

Ethernet

for others

Mbit/s

10

30

100

150

200

300

1000

x% single

technology

y% multi-

technology

10% leased

lines for

outdoor sites

(100% indoor)

90%

microwave foroutdoor sites

2% sited on ahub (therefore

no LMA)

90% to hubs,

10% direct to core

10 sites per hub

GSM

UMTS

LTE

Shared

Not costed


32/83


Ref: 13392-86b .

6.4 Hub to core transmissionThe hub to core transmission network is also common for all three radio network technologies.

There are again capacity upgrades, and the physical transmission infrastructure can at this level be

in rings (for leased lines) or point-to-point (for microwave links).

Figure 6.3: Overview of the modelled transmission between hubs and the core network [Source:

Analysys Mason]

6.4.1WorksheetNwDesLoad8. Hub to core

transmission

Calculates the number of radio sites connected via a hub.

Calculates the number of hubs and related point-to-point links and rings

to the core network.

Calculates the total network traffic at the hub layer, and split it by point-

to-point link or ring.

Determines the actual capacity of hub to core point-to-point links or

rings by geotype according to a predefined rate ladder.

Calculates the number of point-to-point links and rings by rate

according to that same rate ladder.

Calculates the number of hubs on rings by rate.

Hub

Microwave point-to-point

Traditional

Mbit/s

32

155

622

2488

Ethernet

Mbit/s

100

1000

2500

10000

BSC/RNC or core site

Access

point

Access

point

Multiple hubs

Leased dark fibre

520km

length per

ring by

geotype

OR

100% of sites

on four-hub

rings

GSM

UMTS

LTE

Shared

Not costed

Traditional

Mbit/s

32

155

622

2488

Ethernet

Mbit/s

100

1000

2500

10000

OR


33/83


Ref: 13392-86b .

6.5 BSCs and RNCsBSCs and RNCs aggregate the 2G and 3G traffic respectively. In both cases, all the urban radio

traffic is routed through BSCs/RNCs in the urban geotype, but only a percentage of the suburban

radio traffic and a percentage of the rural suburban radio traffic is routed through BSCs/RNCs in

their respective geotype, the remaining share being sent to the urban geotype. There are capacity

upgrades implemented in the model for this level as well.

Figure 6.4: Overview of the modelled BSCs and RNCs [Source: Analysys Mason]

6.5.1WorksheetNwDesLoad9. BSC

(to total number of

BSC by capacity)

Reallocates the number of TRX needed in each geotype according to the

load served locally or sent to switches in the urban geotype.

Calculates the BSCs capacity requirement per location in each geotype.

Calculates the number of BSCs required in each geotype according to

the capacity requirement and the unit capacity of a BSC in each

geotype.

Calculates the number of BSCs by capacity according to a predefined

ladder.

10. RNC

(to total number of

RNC by capacity)

As above but for RNCs.

Rural

BSC site

Urban

BSC site

Suburban

BSC site

Rural

RNC site

Urban

RNC site

Suburban

RNC site

Rural TRXUrban TRXSuburban

TRX

Urban

micro TRX

BSCBSC BSC

Rural Mbit/sUrban Mbit/sSuburban

Mbit/s

Urban micro

Mbit/s

RNCRNC RNC

BSC capacity in TRX

384

512

2048

RNC capacity in Mbit/s

196

450

2600

Most load is carried back to large switch sites, a proportion of suburban and rural load

is served locally before being trunked back to core sites

50% 50%10% 10%

GSM

UMTS

LTE

Shared

Not costed


34/83


35/83


Ref: 13392-86b .

6.7 Core-to-core transmissionThe core network is assumed to be a ring linking Stockholm to Gteborg to Malm, through eight

other large cities, and two links (for redundancy) linking Stockholm to Lule. It carries a

proportion of the data traffic and a proportion of the voice traffic. The ATM/SDH/PDH and

Ethernet protocols can be used for voice and data layers, or all traffic carried in a converged

Ethernet network. The capacity of links follows a predefined hierarchy of options.

Figure 6.6: Overview of modelled transmission within the core network [Source: Analysys Mason]

6.7.1WorksheetNwDesLoad11. Core-to-core

rings

Calculates the core-to-core traffic load (including the utilisation factor),

separately for ATM/SDH/PDH and for Ethernet.

Determines the actual capacity of the core-to-core links according to a

predefined rate hierarchy (separately for ATM/SDH/PDH and for

Ethernet).

Calculates the number of core nodes by speed.

6.8 Switches and support systemsDifferent types of switches are necessary to ensure the network of the operator modelled is able to

function as planned to offer mobile services. Figure 6.7 presents these switches and states the

minimum number required in any network. The traffic load on the network may then require larger

numbers of units to be deployed. Some switches are assumed to have redundant deployments.

Remote RNC

Main

switching site

Remote BSC

Remote RNC

Remote BSC

Main

switching site

Main switching

site

Remote RNC

Remote BSC

Remote RNC

Remote BSC

Stockholm,

Gothenburg

and Malmo

Upsala, Vasteras,

Orebro, Karstad,

Kristianstad, Karlskrona,

Jonkoping, Norkoping

Sundsval,

Lulea

900km dark

fibre pair

1645km dark

fibre pair

Voice and

data layers

(or all IP)

Traditional

for voice

Mbit/s

155

622

2488

9952

Ethernet

for data

Mbit/s

1000

2500

10000

40000

15% of datatraffic

2515% of

voice traffic

GSM

UMTS

LTE

Shared

Not costed


36/83


Ref: 13392-86b .

All switches are shared between the different network technologies (GSM, UMTS, LTE), except

the GSM MSCs used only in the GSM network and the MMEs and SGWs are used only in the

LTE network.

Asset Assumed

capacity [1, 2]

Minimum

deployment [1]

Asset Assumed

capacity [1, 2]

Minimum

deployment [1]

GSM MSC 400 000 BHCA 2 AUC 1 000 000 subs 1

MSS 800 000 BHCA 2 EIR 1 000 000 subs 1

MGW 20 000 BHE 2 2 (for

redundancy)

MNP 1 per operator 1

SGSN 1 400 000 SAU 2 IN 500 000 subs 1

GGSN 80 000 PDP 2 VMS 50 000 subs 1

MME 10 Gbit/s 2 SMSC 1 000 SMS per

second

2 2 (for

redundancy)

SGW 10 Gbit/s 2 MMSC 1 per operator 1

HLR 1 800 000 subs 2 VAS 500 000 subs 1

PoI 2 422 BHE 2

Figure 6.7: Overview of the switch capacity assumptions [Source: Analysys Mason]

6.8.1WorksheetNwDesLoad12. Servers Calculates the number of each type of servers required to handle the

traffic determined in the NetworkLoad worksheet and applying to it

predetermined specifications about their capacity and utilisation factor.

6.9 Switch portsA number of port upgrades network elements are present in the model, for a variety of switches.

These network elements reflect the upgrade costs to connect links into switches.

6.9.1WorksheetNwDesLoad13. Switch ports Calculates the number of site-facing and core-facing ports for BSC and

RNC switches, using E1 or 10Mbit/s port units, and for voice and data

traffic heading into the core where applicable (to MGW or SGSN

accordingly).


37/83


Ref: 13392-86b .

7 Expenditure

The network design algorithms compute the assets (network elements) that are required to support

a given demand in each year. A series of steps are then undertaken in order to arrive at the

schedule of capex and opex over the modelling period. These steps are detailed below and

summarised in the remainder of this section:

defining the list of assets on theInAssetworksheet (Section 7.1)

summarising the assets required in the network over time on the FullNw worksheet (Section

7.2)

determining the assets purchased in each year on theNwDeploy worksheet (Section 7.3)

calculating the unit cost trends for each asset over time on the CostTrends worksheet (Section

7.4)

calculating the unit capex over time on the UnitCapex worksheet (Section 7.5)

calculating the unit opex over time on the UnitOpex worksheet (Section 7.6)

calculating the total capex over time on the TotalCapex worksheet (Section 7.7)

calculating the total opex over time on the TotalOpex worksheet (Section 7.8).

7.1 WorksheetInAsset1. Standard cost

inputs

For a given set of cost input categories, specifies an assumed lifetime,

planning period, proportion of asset replaced per annum and opex as a

proportion of capex for each category [1,2].

2. Inputs by asset For each asset, specifies:

asset name

cost category

cost input category (from the list in the table of standard cost

inputs)

year of first possible deployment of asset [1]

first year of cost recovery of asset [1]

final year of capex (after which no further replacement capex is


38/83


Ref: 13392-86b .

incurred) [1]

final year for cost recovery [1]

retirement period2

lifetime, planning period, proportion of assets replaced per

annum and opex as a proportion of capex based on the cost

input category [1,2]

[1]

unit capex in real 2010 SEK [1, 2, 3]

unit opex in real 2010 SEK [1, 2, 3].

Some additional cost inputs [1, 2, 3] are placed to the right of these

columns.

7.2 Worksheet FullNw1. Network

elements by year

Pulls together the assets required in the modelled network for each year

in the modelling period.

2. Network

elements,

accounting fornetwork activation

This switches off assets that are being specified either:

out of the scope of the modelled network configuration

outside the network lifetime.

7.3 WorksheetNwDeployThe network design algorithms compute the network elements that are required to support a given

demand in each year. In order for these elements to be operational when needed, they need to be

purchased in advance, in order to allow provisioning, installation, configuration and testing before

they are activated. This is modelled for each asset by inputting a planning period between 0 (noplanning required) and 18 months. The number of assets purchased in each year is derived on this

worksheet, accounting for:

additional assets required to provide incremental capacity

equipment that has reached the end of its lifetime and needs to be replaced

2

By setting the value to 0, 1 or 2, the model will remove the assets as traffic reduces, either in the same year, oneyear later, or two years later respectively. By setting the value to 100, the model will retain the asset in the networkuntil the last year of operation.


39/83


Ref: 13392-86b .

advanced purchase in both cases based on the assumed planning period.

The steps taken are described below.

1. Required units

in full network

Links in the network elements, accounting for network activation from

theFullNw worksheet.

2. Deployed assets

with retirement

algorithm

Determines the maximum number of units required of each asset and the

first year in which this maximum is reached.

3. Annual

activation

(including

replacement)

Calculates the difference between the number of units required and the

number of units previously deployed that are still active (this does not

remove assets before the end of their lifetime even if they are no longer

required).

4. Direct equipment

purchases (incl.

replacement)

Determines the equipment required across all replacement cycles,

purchased prior to activation based on the planning period (fractional

units of purchase are permissible on the basis that they reflect phasing

of purchase over each modelled year).

5. Direct equipment

purchases (for

network

regeneration only)

Determines asset replacement (where activated for a given asset) on the

basis that equipment is purchased as part of the constant renewal of

parts of the network, rather than using the asset lifetime as the trigger

for replacement.

7.4 Worksheet CostTrendsThe cost of purchase for network assets varies over time. In the economic costing approach, the

modern equivalent asset (MEA) provides the appropriate cost basis for purchase. Real-term unit

asset cost trends are applied to 2010 unit asset costs to reflect the evolution of the modern

technology unit asset costs over past and future time. The evolution of MEA unit asset costs also

provides an important input into the economic depreciation calculation, as described in Section 8.

Certain quantities for the economic depreciation calculation, including the capex/opex indices, arealso calculated on the CostTrends worksheet.

These calculations are described below.

1. Equipment capex

trends

Specifies the year-on-year change in capex trends over time for a set of

specified categories [1, 2].

Determines the year-on-year change in capex trends for each asset,

based on a specified category.

Calculates the cumulative year-on-year change in capex trends for each


40/83


Ref: 13392-86b .

asset, indexed with the first modelled year set to be 1.

Multiplies this capex index by the network element output, which is

described in Section 8.3, to give the capex cost-weighted output.

2. Equipment opex

trends

Specifies the year-on-year change in opex trends over time for a set of

specified categories [1].

Determines the year-on-year change in opex trends for each asset, based

on a specified category.

Calculates the cumulative year-on-year change in opex trends for each

asset, indexed with the first modelled year set to be 1.

Multiplies this opex index by the network element output, which is

described in Section 8.3, to give the opex cost-weighted output.

7.5 Worksheet UnitCapex1. Unit capex per

network element

Calculates the unit capex by asset in each modelled year, using the

MEA capex index, scaled by the capex index value in 2010. This

ensures that the unit capex is determined relative to the base year of the

inputs, which is 2010.

2. Shut-down capex

profile

Determines a binary multiplier, which is zero where an asset is assumed

to no longer incur replacement capex; otherwise, the binary multiplier is

one.

7.6 Worksheet UnitOpex1. Unit opex per

network element

Calculates the unit opex by asset in each modelled year, using the MEA

opex index, scaled by the opex index value in 2010. This ensures that

the unit opex is determined relative to the base year of the inputs, whichis 2010.

2. Shut-down opex

profile

Determines a binary multiplier, which is zero when an asset has been

assumed to be completely removed from the network; otherwise, the

binary multiplier is one.

7.7 Worksheet TotalCapex1. Total annual

Multiplies the unit capex derived in the UnitCapex worksheet by thenumber of assets purchased in each year, calculated in the NwDeploy


41/83


Ref: 13392-86b .

capex worksheet.

The capex is set to be zero for those assets in those year when the shut-

down profile for capex from the UnitCapex worksheet is zero.

2. Category totals Aggregates the total capex by asset derived above by cost category.

Cumulates the capex by cost category over time, starting in the first year

of the modelling period.

7.8 Worksheet TotalOpex1. Total annual

opex

Calculates the working capital allowance in each year (currently

assumed to be 30/365 of the weighted average cost of capital (WACC)).

Multiplies the unit opex derived in the UnitOpex worksheet by the

number of assets active in the network in each year, calculated in the

NwDeploy worksheet.

The opex is set to be zero for those assets in those year when the shut-

down profile for opex from the UnitOpex worksheet is zero.

The opex is also uplifted by the working capital allowance.

2. Category totals Aggregates the total opex by asset derived above by cost category.


42/83


Ref: 13392-86b .

8 Depreciation

This section describes the implementation of the economic depreciation algorithm used in PTSs

new mobile LRIC model. We describe this algorithm in several stages:

overview of the conceptual approach and the principles of the implementation (Section 8.1)

description of the location of the key inputs to economic depreciation (routeing factors,

network element output and discount rates respectively) (Sections 8.2, 8.3 and 8.4)

description of the calculation steps implemented to derive economic costs (Section 8.5).

8.1 Overview of economic depreciationBelow we describe the conceptual approach and the implementation principles of economic

depreciation.

8.1.1Conceptual approachAn economic depreciation algorithm recovers all efficiently incurred costs in an economically

rational way by ensuring that the total of the (cost-oriented) revenues generated across the lifetime

of the business are equal to the efficiently incurred costs, including cost of capital, in present value

(PV) terms. This calculation is carried out for each individual asset class, rather than in aggregate,in order to allow the price trends and opex cost trends for each asset to be reflected.

The calculation of the cost recovered needs to reflect the time value of money. This is accounted

for by the application of a discount factor on future cashflows, which is equal to the WACC of the

modelled operator.

The business is assumed to be operating in perpetuity, and investment decisions are made on this

basis. This means it is not necessary to recover specific investments within a particular time horizon

(e.g. the lifetime of a particular asset), but rather throughout the lifetime of the business. In the

economic depreciation model, this situation is approximated by explicitly modelling a period of50 years. At the real discount rate applied (which is derived using the WACC), the PV of the

cashflows in the last year of the model is very small and thus any perpetuity value beyond 50 years

is regarded as immaterial to the final result.

The constraint on cost recovery (NPV of costs = NPV of output calculated unit costs) can be

satisfied by (an infinite) number of possible cost-recovery profiles. However, it would be

impractical and undesirable from a regulatory pricing perspective to choose an arbitrary or highly


43/83


Ref: 13392-86b .

fluctuating recovery profile.3 Therefore, we choose a cost-recovery profile that is in line with

revenues generated by the business. In a competitive and contestable market, the revenue that can

be generated is a function of the lowest prevailing cost of supporting that unit of demand, thus the

price will change in accordance with the costs of the MEA for providing the service.4

The efficient expenditure of the operator comprises all the operators efficient cash outflows over

the lifetime of the business, meaning that capex and opex are not differentiated for the purposes of

cost recovery. As stated previously, the model considers costs incurred across the lifetime of the

business to be recovered by cost-oriented revenues across the lifetime of the business. This

principle implies that the treatment of capex and opex should be consistent, since they both

contribute to supporting the cost-oriented revenues generated across the lifetime of the business.

The unit cost

is therefore assumed to follow the MEA unit asset cost trend for that asset class. The cost-recoveryprofile for each asset class is the product of the demand supported by the asset (i.e. its economic

output) and the MEA unit asset cost trend. This gives a unique solution.

8.1.2Principles of implementationThe PV of the total expenditure is the amount which must be recovered by the revenue stream. The

discounting of revenues in each future year reflects the fact that delaying cost recovery from one

year to the next accumulates a further year of cost of capital employed. This leads to the

fundamental equation of the economic depreciation calculation that is:

PV (expenditures) = PV (unit cost output)

The unit cost output which the operator gains from the service in order to recover its

expenditures plus the cost of capital employed is modelled as output year 1 unit cost MEA

price index. This quantity is discounted because it reflects future cost recovery. (Any costs

recovered in the years after a network element is purchased must be discounted by an amount

equal to the WACC in order that the cost of capital employed in the network element is also

returned to the operator.)

output the service volume carried by the network element

MEA price index the cumulated input price trend for the network element which

proportionally determines the trend of the unit cost that recovers the expenditure (effectively,the percentage change to the cost of each unit of output over time).

This leads to the following general equations:

cost recovery (year n) = unit cost in year 1 output MEA price index

3For example, because it would be difficult to send efficient pricing signals to interconnecting operators and their

consumers with an irrational (but NPV=0) recovery profile.

4 In a competitive and contestable market, if incumbents were to charge a price in excess of that which reflected the

MEA prices for supplying the same service, then competing entry would occur and demand would migrate to theentrant which offered the cost-oriented price.


44/83


Ref: 13392-86b .

Using the relationship from the previous section, the above equation is equal to:

PV (expenditure) = PV (unit cost in year 1 output MEA price index)

This equation can be rearranged as follows:

unit cost in year 1 = PV (expenditure) / PV (output MEA price index)

Then, returning to the original equation for cost recovery in yearn, the yearly price over time is

simply calculated as:

yearly unit cost over time

This yearly price over time is calculated separately for the capex and opex components in one step

in the model.

= unit cost in year 1 MEA price index

8.2 WorksheetRFsRouteing factors determine the amount of each elements output required to provide each service.

The routeing factors used in the model are average traffic routeing factors and are converted into

equivalent traffic measures using a number of derived conversion factors. All of these inputs can

be found on this worksheet.

1. Source

calculations

Links in a series of standard technical parameters [1, 5].

Calculates factors for conversion of the following quantities on the airinterface into minute equivalents:

SMS, separately for GSM and UMTS

GPRS megabytes, separately for handset/mobile broadband

traffic

EDGE megabytes, separately for handset/mobile broadband

traffic

R99 megabytes, separately for handset/mobile broadband traffic

HSDPA megabytes, separately for handset/mobile broadband

traffic

LTE megabytes, separately for handset/mobile broadband

traffic.

Calculates factors for conversion of data traffic on transmission links.

2. Routeing factor For a list of asset measure options, derives a routeing factor for that


45/83


Ref: 13392-86b .

options option for each of the modelled services.

3. Full routeing

factor table

For each asset and each modelled service, identifies the routeing factor

from the above table based on the asset measure option for that asset.

[1]

8.3 WorksheetNwEleOutThe quantity of network element output, by asset over time, is used as the basis on which to derive

economic costs. This quantity is taken to be the annual sum of service demand produced by the

asset, weighted according to the routeing factors of that asset for the modelled services. Network

element output is calculated on theNwEleOutworksheet.

1. Service demand

for the whole

market

Links in the service volumes for the modelled network over time from

theNetworkLoadworksheet.

2. Service routeing

factors

Links in the full routeing factor table from theRFworksheet.

3. Recovery profile Currently set to be 0% before cost recovery is assumed to start and after

cost recovery has ended, 100% otherwise.

4. Recovery profile

in binary form

Currently set to be 1 if the corresponding entry in the recovery profile

above is nonzero, and zero otherwise.

5. Network element

output

Calculated as:

service volumes routeing factors binary profile

8.4 WorksheetDFThe model operates in real terms and hence requires a real discount rate with which the modelled

cashflows can be discounted when deriving present values. This is derived using the real cost of

capital, specified on the Ctrlworksheet.

1. Discount rate

data

Links in the real discount rate (WACC) [4].

Derives the real discount rate multiplier.

Derives the real discount rate divider.

Derives the inflation multiplier from the retail price index [4].


46/83


Ref: 13392-86b .

8.5 WorksheetEDThis worksheet is where the economic costs of capex/opex are calculated over time, using the

above inputs and the unit asset cost trends from the CostTrendworksheet, described in Section 7.4.

1. Capex per

unit output

Calculated separately for each asset across the modelling period.

Derived as the capex index over time scaled by a constant factor.

This factor is the ratio of the cumulative discounted asset capex and the

cumulative discounted capex weighted output (referred to as

PV(expenditure) / PV(outputMEA price index) above).

2. Opex per

unit output

Calculated separately for each asset across the modelling period.

Derived as the opex index over time scaled by a constant factor.

This factor is the ratio of the cumulative discounted asset opex and the

cumulative discounted opex weighted output (referred to as

PV(expenditures) / PV(outputMEA price index) above).

This is calculated separately to the capex per unit output since the asset

unit capex trend could differ to the asset unit opex trend.

3. Total cost per

unit output

Calculates the sum of the capex per unit output and the opex per unit

output, multiplied by the binary recovery profile.

4. FAC per

service unit

Calculates the multiplication of the cost per unit output matrix and the

routeing factor matrix to give unit fully allocated costs (FAC) by

service.

5. Total economic

costs

Calculates the total cost per unit output multiplied by the network

element output.

Calculates the total economic costs over time.

Calculates the total discounted economic costs over time.

Calculates the cumulative discounted economic costs over time.

Calculates the present value of the economic costs.

6. Total costs

recovered by FAC

Multiplies the FAC per service unit by the modelled network service

volumes.

Calculates the total discounted FAC.


47/83


Ref: 13392-86b .


48/83


Ref: 13392-86b .

9 Results

The model calculates service costs using both LRAIC and pure LRIC principles. The outputs of

these calculations can be found on theResults worksheet.

The remainder of this section describes these calculations, as set out below:

Section 9.1 describes how the LRAIC (and LRAIC+) are derived in the model

Section 9.2 describes how the pure LRIC is derived in the model

Section 9.3 describes where to find the major outputs of the model.

9.1 Calculation of LRAIC(+)This calculation takes the total economic costs for each network asset, and applies a proportion to

that asset, derived on the Common worksheet. This proportion is the number of assets assumed to

be common to all services in the network, expressed as a percentage of the total assets. These costs

are then entirely included within the common cost base.

On theED worksheet, the cost per unit output calculated for each asset is separated into common

and incremental components using the common cost proportions derived on the Common

worksheet. Incremental service costs are derived by multiplying the incremental cost per unit

output by the routeing factors according to the following equation:

),()(___cos)( kiassets

ik serviceassetctorRouteingFaassetoutputunitpertServiceCost =

Total common costs (network common costs and business overheads) are then marked up onto

each incremental service cost in an equi-proportional manner, according to the ratio of common to

incremental network costs, resulting in the LRAIC+. This approach is illustrated below in Figure

9.1.

Figure 9.1: Illustration ofLRAIC+ costing

approach [Source:

Analysys Mason ]

Incremental cost ofall traffic

(MSC, BSC, radio sites, etc.)

Network share of business overheads

Subscriber

SIM

Network common costs

(some coverage, spectrum)

E

P

M

U

E

P

M

U


49/83


Ref: 13392-86b .

Below we describe the derivation of the common cost proportions on the Common worksheet,

followed by a description of the remaining calculations required for the LRAIC (and LRAIC+) on

theLRAIC+ worksheet.

9.1.1Worksheet CommonThe common cost proportions are derived on this worksheet, so that incremental and common

costs can be separated within the model.

1. Source

calculations

Links in GSM and UMTS coverage sites separately.

Determines the number of common coverage sites in each year [1].

2. Full table of

common number

of assets

Calculates the number of common assets (i.e. used by both GSM and

UMTS networks).

3. Full table of

common cost

proportions

For each asset, calculates common assets as a proportion of total assets

over time [1].

9.1.2WorksheetLRAIC+

On this worksheet, economic costs are mapped to services and mark-ups are applied.

1. Total economic

costs

Links in the total economic costs by asset over time from the ED

worksheet.

2. Total common

costs

Multiplies the total economic costs by the common cost proportions

from the Common worksheet.

3. Total

incremental costs

Derives the difference of the economic costs and common economic

costs.

4. Calculation of

mark-ups

Calculates common economic costs as a proportion of incremental

economic costs in order to arrive at the equi-proportional mark-up

(EPMU).

5. Calculation of

unit LRAIC

Calculates the incremental cost per unit output by multiplying the total

cost per unit output from the ED worksheet by the common cost

proportions.

Multiplies the incremental cost per unit output matrix and the routeing

factor matrix to arrive at the unit LRAIC by service.


50/83


Ref: 13392-86b .

Multiplies the unit LRAIC by service by the network service volumes to

derive the total LRAIC by service.

For a selected year, calculates the breakdown of network service unit

costs by asset and service, by multiplying the incremental cost per unit

output in that year by the routeing factors.

Aggregates this breakdown by cost category.

Calculates the total network service costs by asset and service for the

selected year, and aggregates this breakdown by cost category.

6. Calculation of

unit LRAIC+

Applies the derived EPMU to the unit LRAIC by service to derive the

unit LRAIC+ by service.

Derives the total LRAIC by service.

Multiplies the unit LRAIC+ by service by the network service volumes

to derive the total LRAIC+ by service.

Calculates the discounted LRAIC+ by service and the total present

value of the LRAIC+.

7. Calculation of

cost recovery

Calculates LRAIC+ by service group.

Calculates the total cumulative LRAIC+ by service group.

9.2 Calculation of pure LRICThis requires that the model is run in two different states: with and withoutmobile terminated

traffic on the modelled network. Clicking on the Run Pure LRIC and LRAIC+ macro button on

the Ctrl worksheet will result in the model calculating twice the total capex and total opex

required by asset over time in each case is then pasted on thepureLRICworksheet. The pure LRIC

of termination is then calculated as shown below in Figure 9.2.


51/83


Ref: 13392-86b .

Figure 9.2: Calculation of pure LRIC [Source: Analysys Mason]

The difference in both capex and opex (the avoidable costs) is determined from the two model

calculations, and economic depreciation is then applied to this difference. This is run separately for

capex and opex, in order to use their respective unit asset cost trends. The pure LRIC of

termination in each year is then calculated as the total economic cost in that year divided by the

total terminated minutes.

Another option for the pure LRIC calculation is also included within the model. The above

approach is the economic cost of the difference in expenditure. An alternative approach is to

simply take the difference in the economic costs of the two modelling states. This is derived at

the bottom of thepureLRICworksheet.

A comparison of the avoidable cost base within the pure LRIC costing approach compared with

the average incremental cost base is shown below in Figure 9.3.

Figure 9.3: Comparison of LRAIC+ with the pure LRIC approach [Source: Analysys Mason]

These calculations are all undertaken on thepureLRICworksheet, as described below.

Run model

with all

traffic

Run model

with all

traffic

except

termination

increment

volume

Expenditure with

voice

termination

(asset, time)

Output profile

with voice

termination

(asset, time)

Expenditure

without voice

termination(asset, time)

Output profile

without voice

termination

(asset, time)

Difference in

expenditure

(asset, time)

Difference in

output (asset,time)

Capex and opex

cost trends

(asset, time)

Economic cost

of difference in

expenditure(asset, time)

Total economic

cost of the

difference in

expenditure (time)

Pure LRIC per

minute (time)

Voice

termination

traffic minutes

(time)

Key Input Calculation Output

Incremental cost ofall traffic

(MSC, BSC, radio sites, etc.)

Network share of bus

10 8320 Final Model With Tracks Mobile

Documents

Transcript of 10 8320 Final Model With Tracks Mobile