Operators meeting 21 st September 2017 Post-och telestyrelsen (PTS) - - PowerPoint PPT Presentation

TERA Consultants

39, rue d’Aboukir 75002 PARIS Tél. + 33 (0) 1 55 04 87 10

• Fax. +33 (0) 1 53 40 85 15

S.A.S. au capital de 200 000 € RCS Paris B 394 948 731

Post-och telestyrelsen (PTS)

Presentation of the draft BU-LRIC+ Cost models for fixed network services Operators meeting 21st September 2017

21st September 2017

Presentation to the industry

This slideshow illustrates the models and documents issued by PTS. In case of discrepancies with the models, the model reference paper, the model specifications or the model documentation, statements within this slideshow should be disregarded.

PTS – Operators meeting 21 September 2017

Agenda

2

• 1. Context
• 2. Main modelling assumptions
• 3. Modelling approach
• 4. Model implementation and usage

PTS – Operators meeting 21 September 2017

Team

• Marc LAMELOISE, Project leader
• 9-year experience in cost modelling;
• Involved in several similar fixed network cost modelling projects

(Ireland, Croatia, Denmark, New Zealand, France, Luxembourg, Kingdom of Bahrain, Gibraltar… ).

• Mohammed EL HIMDY, Consultant (Cost model)
• Involved in a similar fixed network cost modelling project in Ireland.
• Alexandre JOURNO, Consultant (Cost model)
• Involved in similar fixed network cost modelling projects in France,

New Zealand and Ireland.

• Martin ROUNDILL, GIS expert
• Involved in a similar project in New Zealand.

3

PTS – Operators meeting 21 September 2017

Context

• As part of its role, PTS has imposed that Telia should provide a set of products

and services on the wholesale market for local access at a fixed location (Market 3a) and the wholesale market for fixed call termination (Market 1) on a cost-

• riented basis.
• To calculate these cost-oriented prices, PTS have until now used a cost model, the

Hybrid model v10.1 (HY model).

• TERA Consultants has been instructed by PTS to develop a new BU-LRIC+ model

based on the Swedish Road network and the exact building location to better assess the price of these services.

• This presentation is a technical presentation of the draft BU-LRIC+ model and is

structured as follows:

• Description of the main modelling assumptions;
• Description of the modelling approach;
• Description of the models implementation and how to use them.

4

PTS – Operators meeting 21 September 2017

Agenda

5

• 1. Context
• 2. Main modelling assumptions
• 3. Modelling approach
• 4. Model implementation and usage

PTS – Operators meeting 21 September 2017

The model follows a set of general rules arising from the MRD

Main model assumptions

• The modern efficient network for the fixed access network is based on FttH (point-

to-point), and all-iP (NGN) for the core network

• All network is 100% underground with no utilisation of FWA
• In the hybrid model, part of the network was overhead and FWA was used at the edge of the

network

• Costs are valued using optimised replacements costs
• Costs are depreciated using a tilted annuity
• OPEX are derived from industry inputs:
• They are adjusted depending on the size of the network (depending on the network dimension for

network OPEX or FTE count per service for non-network OPEX)

• The existing civil engineering infrastructure that can be reused is considered:
• A 15% re-use factor has been used;
• The reusable civil engineering assets are valued according to their book value, depreciated

through their remaining lifetime.

• For copper services costs, an economic adjustment to the fibre cost results is

performed:

• Equivalent copper equipment’s unit costs, price trends are considered.

6

New New New

PTS – Operators meeting 21 September 2017

The network structure was optimized for the calculation

Main model assumptions

7

• The modified scorched node approach has been followed:
• the network roll-out follows the road network (when average data by access nodes type was used in the HY model);
• existing nodes of the copper access network (access nodes) are the starting points of the modelling (smaller nodes

as FOS are dimensioned depending on the calculated demand and are outputs of the model)

• The node structure has been cleaned in order to remove redundant nodes and the nodes to be dismantled by 2018

(~1700 nodes).

• Following the Voronoï approach, end-users are connected to the closest access node following the road network.
• The network is optimized in order to minimize the distance of each line using road network and

buildings locations as a starting point.

Access node 1 Access node 2

Voronoi’ polygon’s boundaries Real boundaries of the network

1 2

Comparison of real access node coverage area and

• ptimized access node coverage area

Buildings

Access node

1F 3F 4F 4F 3F 4F 2F 2F 4F 14F Premises 2F

Shortest path from the access node to buildings

New

PTS – Operators meeting 21 September 2017

A set of assumptions have been followed for demand

Main model assumptions

8

Access model

• Passive demand (lines passed) depends on the dwellings/businesses to be

connected and is considered flat from year 2016

• The active demand is set for a national HEO which would deploy a nationwide

network excluding the most costly lines:

• 60% market share in urban areas;
• 100% market share un rural areas.
• No take-up is considered
• The demand then evolves in line with the number of active ports in the Core model.

Core model

• Derived to a large extent from PTS statistics (traffic, customer based)
• Representative of a 100% footprint
• Market share is consistent with the access

New New

PTS – Operators meeting 21 September 2017

The footprint is restricted to the lines that a commercial

• perator would deploy

Main model assumptions

• The footprint to cost the network of the modelled operator shall be national and be established

in three steps:

• Establishing all buildings that are relevant to connect to the network comprising residential apartments, relevant

business locations, industrial and public buildings (agricultural and other buildings are not taken into account) as well as secondary homes. This will determine a national network with 100 percent coverage.

• Then, the footprint is restricted after excluding the most expensive lines by removing 15% of lines passed that

have the highest cost to connect to the modern network (Number of lines passed is used as the control variable).

• Besides a further reduction of the footprint is performed to take into account the sites that would not be deployed

because the economies of scale at the access node level are limited due to the low number of active lines.

9 Full Network Restricted footprint for all lines to be passed Restricted footprint for which only sites with active demand are deployed Restricted footprint for which only sites with active demand > 50 lines are deployed Total number of lines passed

5 647 131 4 799 995 4 726 084 4 554 409

Total number of buildings passed

2 650 393 1 844 323 1 780 260 1 634 363

Number of active access nodes

6 402 5 077 3 046

Network footprint depending on the active threshold scenario

• 15%

New

PTS – Operators meeting 21 September 2017

Costs are allocated either using a capacity based allocation or an EPMU

Main model assumptions

• The capacity based allocation approach is used to allocate network costs:
• The EPMU approach is used to allocate non-network costs.

10

Asset class Capacity driver Trenches Ducts Ducts Surface of the cables inside of the ducts Fibre cables Fibres Fibre access Switches Active customers Edge and IP Core switches Traffic per node

New

PTS – Operators meeting 21 September 2017

Other model assumptions

Main model assumptions

• Access model assumptions:
• NTPs and BDFs are excluded from the network cost to be recovered.
• Final drop infrastructure that are part of the private domain (Vertical Trenches and sub-ducts) are

recovered through the one-off charge, when the remaining network assets (including final drop cables and horizontal infrastructure are recovered through the monthly rental charge.

11

ODF

Network termination point

FOS

Distribution Final drop

Architecture of the local access fibre network

PTS – Operators meeting 21 September 2017

Agenda

12

• 1. Context
• 2. Main modelling assumptions
• 3. Modelling approach
• 4. Model implementation and usage

PTS – Operators meeting 21 September 2017

Access network modelling approach

Modelling approach

13

Engineering rules, Efficiency algorithm, Demographic data, Shortest path algorithm Step 4 – Current asset prices Step 6 – Depreciation Step 7 – OPEX calculation Step 8 – Cost results Step 5 – CAPEX Step 1 – Node location and coverage Step 2 – Network deployment at the street level Step 3 – Full network deployment

Network dimensioning Network costing Network cost allocation

WACC, asset lives, price trends

PTS – Operators meeting 21 September 2017

Each section is specified either urban or rural, from which depend the trench type used

Modelling approach

14

Network dimensioning Location Type of trench Cross-trenches Asphalt Urban trenches Bicycle Rural trenches Grass Final drop Ploughing

Legend Section by trench classification Urban Rural

For illustrative purposes only

PTS – Operators meeting 21 September 2017

Road network and exact location of all Swedish buildings was used for calculation

Modelling approach

15

For illustrative purposes only

Network dimensioning

PTS – Operators meeting 21 September 2017

Network roll-out follows a shortest path algorithm

Modelling approach

16

1 Starting point: the location and the coverage areas of the existing copper exchanges are used as the location and the coverage of the modelled fibre network 2 Each building is linked to its parent exchange using the shortest path algorithm

ODF

For illustrative purposes only Cables deployed following the shortest path algorithm

Network dimensioning

Legend: Warmest color corresponds to bigger cables

PTS – Operators meeting 21 September 2017

The access network is dimensioned section per section

Modelling approach

• The access network is dimensioned section per section (a section is a part of a road/street

between two consecutive intersections), knowing:

• The demand of the section: the number of lines on the section plus on its rear area;
• The features of the section: length, buildings location, number of dwellings per building, etc.

17

• First,

the number and location

• f

the distribution points are derived.

• Then, the assets dedicated to each building are

dimensioned:

• The final drop cables (length and size);
• The dedicated civil engineering.
• Finally, the assets shared between the different

buildings of the section are dimensioned:

• The cables and joints (size, length), according

to the local and rear demand and section configuration;

• The civil engineering, shared among access

network and core network.

Final drop cable

FOS FOS

Distribution point Dedicated trench Dedicated subduct Distribution cable Final drop cable Core cable Joints Rear area Trench Duct

Network dimensioning

PTS – Operators meeting 21 September 2017

A set of engineering rules is followed for the access model dimensioning

Modelling approach

18

• The access network dimensioning follows a set of engineering rules that either stem from
• perators, the hybrid model or are cost-effective:
• The section is trenched on one side or on

both sides according to the cost-efficient solution.

• Joints are deployed at the intersection with other

sections and along the section according to the cables’ standard drum length.

• Distribution points (FOS) are dimensioned given

their capacity and are deployed uniformly along the section.

• The final drop cable is then deployed from the

building to the closest distribution point

• The

distribution cable is dimensioned to meet the local demand and the rear area demand

Access node

Jnt FOS FOS Jnt FOS Jnt

Rear area Drum length

Jnt

Joint at the intersection with the minor side

Network dimensioning

PTS – Operators meeting 21 September 2017

Civil engineering is shared between different type of links

Modelling approach

19

For illustrative purposes only

• The access network shares its trenches

with other networks:

• The core network:

– Inter Exchange links, – Core-IP links – Submarine links;

• Other utility networks (Cable TV, Electricity,

Water, other operators)

• The routes used by the HEO’s networks

have been modelled in order to capture the relevant economies of scale/scope.

• Besides, sharing rates with other utilities

is differentiated between Urban/Rural areas.

Core- links modelling

Legend Core Links IP- Red IP IP-Blue Edge Common Metro

Horizontal trench Duct Vertical trench Urban 16% 1% 24% Rural 15% 1% 11%

Sharing rates with other utility networks depending on the infrastructure Network dimensioning

PTS – Operators meeting 21 September 2017

The active demand is set for a national HEO based on PTS’ market data

Modelling approach

20

PTS market statistics (fibre, cable) Active copper line Active access demand All platforms – 100% footprint National level Active access demand All platforms – 100% footprint Distributed by site Active access demand All platforms – Restricted footprint Distributed by site Sections / Number and types of building removed Penetration by type of buildings Coverage % of each access technology by municipality Active access demand HEO – Restricted footprint Distributed by site Platform market share HEO: Urban: 60 % Rural: 100%

ACCESS CORE

Active CORE demand HEO – 100% footprint National level PTS total market statistics (#subscribers, traffic) HEO market share for core services Calibration (for 100% footprint) Platform  CORE actives ports + LLU = Access active lines

PTS – Operators meeting 21 September 2017

The HEO core network maps Telia’s network structure

Modelling approach

21

COMMON Edge CR Two seperate Network METRO "Common with redundacy"

• Telia has provided a file describing the

architecture of the core network, in 3+1 hierarchical levels:

• The IP level, the higher level, made of two

parallel network of a dozen of nodes, the red and the blue networks;

• The Edge level, the second level of the core

network, made of 139 edge routers.

• The Common level, the third level of the core

network, which connects all access nodes to the Edge nodes.

• The Metro level, a redundant level of the

core network, which provides additional links among the Common and/or Edge nodes, in

• rder to ensure the reliability of the network.

Structure of the core network

Network dimensioning

PTS – Operators meeting 21 September 2017

Core nodes are dimensioned based on downlink and uplink demand

Modelling approach

22

• Core nodes are either switches (performing conversion between the electrical

signals used by the service provider's equipment and the fiber optic signals used by the passive optical network) or routers (forwarding data packets between computer networks).

• These nodes are dimensioned for each site according to downlink demand

(number of active end-users or number of ports of daughter sites) and uplink demand ( number of ports needed to link parent core sites or inter-IP sites)

• In

contrast with HY model where it was one-size-fits-all for each node type -

Network dimensioning

Rack Card 10G Card 10G Card 10G Card 10G Card 10G

SFP SFP SFP SFP SFP SFP SFP

Edge A Edge B Core IP router Core IP router (twin) Downlink traffic Core IP router Core IP router

SFP SFP SFP SFP

Uplink traffic Blue node

Example - Core routers’ configuration

PTS – Operators meeting 21 September 2017

Investment are depreciated using tilted annuity

Modelling approach

• The costs are depreciated using a tilted annuity after calculating a depreciation

factor per asset using the following formula:

𝐸𝑓𝑞𝑠𝑓𝑑𝑗𝑏𝑢𝑗𝑝𝑜 𝑔𝑏𝑑𝑢𝑝𝑠 = (𝜕 − 𝑞) × (1 + 𝑞)𝑢 1 − 1 + 𝑞 1 + 𝜕

𝑜

• Where:

– 𝜕 is the nominal pre-tax WACC of 6.6% and 𝑞 the price trend for the asset – 𝑜 the lifetime of the asset – (1 + 𝑞)𝑢 the index for deriving the current price of the asset

23

Network Costing

PTS – Operators meeting 21 September 2017

Costing and pricing of the network

Modelling approach

24

Network Costing Network cost allocation

PTS – Operators meeting 21 September 2017

Interconnection services cost is calculated following the Pure LRIC cost standard

Modelling approach

25

• Similarly to the HY model, the Fixed Termination services, as well as the other voice services

(retail, origination, transit) use two specific assets:

• The TDM (Time-division multiplexing) Gateways :
• The IMS (IP Multimedia Subsystem)
• The costs allocated to the termination services are the incremental costs (or “Pure LRIC”) of

TDM and IMS associated with offering termination when already providing retail voice, voice

• rigination and transit, following 3 steps:
• Dimensioning according to the total voice traffic (retail voice and origination, wholesale termination and transit);
• Dimensioning according to the total voice traffic excluding the voice termination traffic;
• The incremental inventory for the termination is assessed as the difference between the two latter.
• The voice service demand is assessed as follows:
• The future number of voice subscribers is based on the 2013-2016 trends (geometric growth rate) for 2017-2019

and considered stable from 2020.

• The traffic for the HEO is assessed based on the total market voice traffic (as published on PTS website)

multiplied by the HEO market share for voice services.

• The market numbers of minutes are published on PTS website until year 2016. Forecasts for 2016 and further are

extrapolated according to the geometric growth rate for each individual service between 2013 and 2016. The traffic is considered stable from 2020. NB: No PSTN network, all the voice traffic is handled by the IP network. Network cost allocation

PTS – Operators meeting 21 September 2017

Agenda

26

• 1. Context
• 2. Main modelling assumptions
• 3. Modelling approach
• 4. Model implementation and usage

PTS – Operators meeting 21 September 2017

Interaction between the different cost models

Model implementation and usage

27

Geomarketing database Access model (MS Access) Access model (Excel) Core model Colocation model Consolidation model

Unit costs Unit costs Unit costs Inventory Road network Core links Costs of core infra

Demand model

Lines passed Active demand Active demand Common costs and wholesale uplifts

PTS – Operators meeting 21 September 2017

Overview of the access model (MS Access)

Model implementation and usage

28

PTS – Operators meeting 21 September 2017

Overview of the access model (MS Excel)

Model implementation and usage

29

Unit cost

• f assets

Import from ACCESS Import from Demand Inventory Investment Dashboard Services Export to Core model Routing matrix per services Annual Capex Opex calculation Demand Export to consolidat ion Fibre Export to consolidat ion Copper

PTS – Operators meeting 21 September 2017

Overview of the Core model

Model implementation and usage

30

Demand per node Design rules Unit costs Core demand Total traffic OPEX and accommodation Investment Import from the ACCESS model Dashboard Services Total costs Traffic Routing tables Line-driven assets Traffic-driven assets TDM-IMS Export to consolidation Import from demand

PTS – Operators meeting 21 September 2017

Overview of the Colocation model

Model implementation and usage

31

Unit costs

Equipment, wages, accommodation

Dimension

Size of cables and racks

Resource

Hours of staff for installing services

Costs of services

CAPEX and OPEX

Cost summary

CAPEX, OPEX and demand

Dashboard

WACC, price trends

Annual cost

CAPEX, annual cost, demand

Services

Unit costs per service

Product list

And demand for the services

PTS – Operators meeting 21 September 2017

Overview of the Consolidation model

Model implementation and usage

32

Import from Access - Copper Import from Access - Fibre Import from Core Dashboard Results Import from Colocatio Upift Calculation Output Services