Explanatory Memorandum MTR version track changes RTR and ILR
Transcription
Explanatory Memorandum MTR version track changes RTR and ILR
2014 Explanatory Memorandum Regulatory project regarding The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Consultation publique nationale Du 17 novembre au 17 décembre 2014 Version publique Error! Reference source not found. Error! Reference source not found. The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table of Contents 0 Introduction .................................................................................................................................... 5 1 Mobile Termination Rate Determination Process .......................................................................... 6 2 Legal context ................................................................................................................................... 8 3 Results and Conclusion ................................................................................................................... 9 4 Cost accounting rules .................................................................................................................... 11 5 4.1 Recommendation of the European Commission .................................................................. 11 4.2 The adequacy of pure LRIC ................................................................................................... 12 Determination of the hypothetical efficient operator.................................................................. 13 5.1 Determination of the demand .............................................................................................. 13 5.2 Network design ..................................................................................................................... 28 5.3 Economic parameters ........................................................................................................... 32 6 Determination of the price cap ..................................................................................................... 39 7 Sensitivity analysis ........................................................................................................................ 44 8 Appendix A .................................................................................................................................... 46 9 8.1 Population density, zones and cell hubs ............................................................................... 46 8.2 Input Parameters .................................................................................................................. 48 Appendix B – List of Abbreviations ............................................................................................... 52 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Figure 3-1: Determination of the costs given different scenarios [source: ILR, cost model, 2014] ....... 9 Figure 4-1: illustration of the pure LRIC concept .................................................................................. 12 Figure 5-1: Overview of the mobile bottom-up model......................................................................... 13 Figure 5-2: Demand determining factors [Source: ILR (2014)] ............................................................. 13 Figure 5-3: Illustration of the increase in passengers at the Luxembourgish airport (Source: Statec 2012) ..................................................................................................................................................... 15 Figure 5-4: Demand modelling process ................................................................................................ 15 Figure 5-5: Differentiation between types of voice traffic [source: ILR (2014)] ................................... 17 Figure 5-6: Hourly distribution of voice, data and total volume for 2013 [source: ILR, operators’ data (2014)] ................................................................................................................................................... 24 Figure 5-7: Total annual network costs depending on the number of controller locations [source: ILR, mobile cost model (2014)] .................................................................................................................... 29 Figure 5-8: Total annual network costs depending on the number of core network locations [source: ILR, mobile cost model (2014)] ............................................................................................................. 30 Figure 5-9: Total annual network costs depending on the minimum distance between controller locations [source: ILR, mobile cost model (2014)]................................................................................ 31 Figure 5-10 : Economic parameters [Source: ILR (2014)] ..................................................................... 32 Figure 5-11 : Distribution of the wholesale commercial cost per minute of fixed termination per operator [source : ILR, operators, 2014]............................................................................................... 38 Figure 5-12 : Distribution without operator « E » [source : ILR, operators, 2014] ............................... 38 Figure 5-13: Additional wholesale comercial cost analysis [source : ILR, 2014]................................... 38 Figure 6-1 : Determination of network equipment volume [source: ILR, 2014] .................................. 39 Figure 6-2: Determination of total annual network costs [source: ILR, 2014] ..................................... 41 Figure 6-3: Determination of the pure LRIC MTR and the price cap [source: ILR, 2014] ..................... 42 Figure 8-1 : Zones and cell hubs of the hypothetical efficient mobile operator in Luxembourg ......... 46 Figure 8-2 : Population density 2014 .................................................................................................... 47 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 3-1: Price cap [source: ILR, cost model, 2014] ............................................................................ 10 Table 5-1: Thresholds for the definition of geo-types (population/km2) [Source: ILR (2014)] ............. 14 Table 5-2: Number of users per technology and geo-type [Source: ILR (2014)] .................................. 14 Table 5-3: Aggregated technical voice volume for 2013 [Source: ILR, operators’ data (2014)] ........... 18 Table 5-4: Weighted average of the share of unbilled traffic [source: ILR, operators’ data (2014)] ... 18 Table 5-5: Aggregated voice volumes and traffic shares for 2013 [Source: ILR, operators’ data (2014)] .............................................................................................................................................................. 19 Table 5-6: Aggregated annual voice volumes distribution per hour [Source: ILR, operators’ data (2014)] ................................................................................................................................................... 20 Table 5-7: Busy Hour Voice Factor and Daily Share of Voice in Network Busy Hour [Source: ILR (2014)] ................................................................................................................................................... 20 Table 5-8: Aggregated data volumes and traffic shares for 2013 [Sources: ILR, operators’ data (2014)] .............................................................................................................................................................. 21 Table 5-9: Distribution of the aggregated annual data volumes [Sources: ILR, operators’ data (2014)] .............................................................................................................................................................. 22 Table 5-10: The distribution of the aggregated total volume [Sources: ILR, operators’ data (2014)] . 23 Table 5-11: Traffic values for voice and data services for 2013 [Sources: ILR, operators’ data (2014)] .............................................................................................................................................................. 25 Table 5-12: Traffic values for SMS services for 2013 [Sources: ILR, operators’ data (2014)] ............... 26 Table 5-13: Input values into the bottom-up cost model for different services [source: ILR, mobile cost model (2014)] ................................................................................................................................ 26 Table 5-14 Frequency spectrum [source: ILR, mobile cost model (2014)] ........................................... 28 Table 5-15: Expected price change, expected growth rate and economic lifetime of network assets [source: ILR, mobile cost model, 2014]................................................................................................. 34 Table 5-16: Expected price change, expected growth rate and economic lifetime of network support assets [source: ILR, mobile cost model, 2014] ...................................................................................... 35 Table 5-17: OPEX mark-up [source: ILR, mobile cost model, 2014] ..................................................... 35 Table 5-18: OPEX mark-up on direct investment [source: ILR, mobile cost model, 2014] ................... 36 Table 6-1: Values of the inputs and outputs outlined in Figure 6-1 [source: ILR, cost model, 2014]... 40 Table 6-2 : Values corresponding to the inputs and outputs outlined in Figure 6-2 [source: ILR, cost model, 2014] ......................................................................................................................................... 42 Table 6-3: Values of the inputs and outputs outlined in Figure 6-3 [source: ILR, cost model, 2014]... 42 Table 8-1 Gross Replacement Costs Considered as Input in the model ............................................... 48 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 0 Introduction (1) The provision of mobile termination represents an essential service for all active mobile operators. Therefore it is necessary to set a price cap for this service, so that competition barriers in the sense of exceeding price settings are prohibited. (2) In January, the Institut Luxembourgeois de Régulation (ILR) concluded that on the market for voice call termination on individual mobile networks (market 7/2007) the following operators have significant market power on their own networks: EPT (Entreprise des Postes et Télécommunications), Orange Communications Luxembourg S.A. and Tango S.A.. Furthermore, ILR launched in September 2014 a public hearing on his proposal to designate the new entrant JOIN Experience S.A. as an operator with significant market power. (3) During this market analysis, a price control obligation based on cost orientation is put on all operators having significant market power. Therefore, ILR has elaborated a cost model that calculates the costs generated in relation with the provisioning of mobile termination on a network of a hypothetical efficient operator in Luxembourg. Based on this cost model, ILR determines the price cap which has to be respected by all the operators with significant market power on the mobile termination market. (4) This document represents the explanatory memorandum detailing how the ILR determines the price cap for the mobile termination rates (MTR) in Luxembourg : - In Chapters 1 and 2, the background and the process of determining a price cap is elaborated as well as the legal context this work and the cost model rely on. - Chapter 3 provides an overview of the resulting price cap, consisting of calculated costs as well as additional commercial costs. - Chapter 4 outlines the cost accounting rules as recommended by the European Commission. - The hypothetical efficient operator, which represents the starting point for being able to conduct the calculations using the cost model, and his properties are determined in Chapter 5. - Chapter 6 outlines the workflow of the calculations computed by the cost model in order to illustrate the different inputs and outputs of the model. - Finally, the last Chapter 7 provides a conclusion of the computations and the resulting price cap. 5/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 1 Mobile Termination Rate Determination Process (5) During the first quarter of 2013, ILR conducted a “Bill and Keep” study in order to analyse the effects of removing completely the billing from the wholesale market of mobile termination. The study let conclude that due to the important amount of international traffic on the Luxembourg market, a Bill and Keep approach could not be considered as an appropriate solution1. Therefore, a cost oriented method needs to be applied. Thus ILR has defined in the market analysis2 according to the MTR recommendation3 to use a bottom-up (BU) pure long run incremental costs (pure LRIC) approach. (6) The bottom-up approach starts by evaluating the demand level of the market before defining an efficient network able to satisfy this forecasted demand level. The costs corresponding to the modelling of this network can thus be determined. Using this approach, a hypothetical efficient operator is modelled by creating a modern and efficient network. This reflects the principal of economic efficiency and avoids that access seekers would need to pay for potential inefficiencies remaining in the network of an existent operator. (7) The pure long run incremental costs of the mobile termination corresponds to the additional costs associated with the production of termination with respect to the existing costs without providing termination. The LRIC method is forward looking and considers all costs as variables. This means, the method does not take into consideration any historical costs but calculates the costs that would be generated by an operator deciding today to deploy a new network for satisfying the forecasted demand. (8) Before being able to determine the pure LRIC costs related to mobile termination, the ILR needs to define the characteristics of the hypothetical efficient operator (HEO) in Luxembourg. It is important to note that the HEO is not an average operator based on eventual existing market players. By not taking the average of the market players, the integration of potential inefficiencies of the existing operators is avoided. (9) Therefore, the demand is defined based on demographical and geographical input data. Data provided by the government statistics service “STATEC” and by the government cadastral and topology service “ACT” are considered in order to derive the subscriber distribution and penetration rate. Furthermore, ILR takes into account the data and information provided by the mobile operators in relation to the technical specifications of the network. More details for the determination of the hypothetical efficient operator are described in Chapter 5. (10) The price cap is determined by ILR by calibrating the model based on the HEO’s characteristics and validated by sensitivity analyses on relevant model input data. Analyse du « Bill and Keep » of 6th March 2013 http://www.ilr.public.lu/communications_electroniques/avis_consultations/contrib_070313/Rapport_Bill_Keep.pdf. 1 Analyse du marché de la terminaison d’appel vocal sur réseaux mobiles individuels http://www.ilr.public.lu/communications_electroniques/avis_consultations/conspub_120713_m7/Analyse_du_marche_7.pdf 2 Commission Recommendation of 2009 on Regulatory Treatment of Fixed and Mobile Termination Rates in the EU (2009/396/EC) 3 6/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (11) The design and implementation details of the BU pure LRIC cost model are specified in the reference document4 for which the national call for inputs was held from 21st November 2013 to 21st January 2014 and thus validated. (12) The regulation concerning the determination of the mobile termination rate (Market 7/2007) will go through public hearing from November 17th until December 17th 2014. 4 Development of a Bottom-Up Mobile Network and Cost Model for the Determination of the Cost of Terminating Calls in Mobile Networks. http://www.ilr.public.lu/communications_electroniques/encadrement_tarifaire/ILR_MTR_ReferenceDocument_20140320.pdf 7/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 2 Legal context (13) The analysis of the relevant market for voice call termination on individual mobile networks (market 7/2007)5 and the corresponding regulation 14/172/ILR6 determine that the following market players have significant market power (SMP): Entreprise des postes et télécommunications (EPT), Orange Communications and Tango. An additional market analysis7 proposes that the new entrant JOIN Experience also disposes of significant market power in market 7/2007. The former regulation 06/92/ILR8 is repealed by the regulation 14/172/ILR. (14) As a result and according to article 7 of the regulation 14/172/ILR, a price control obligation is imposed on SMP operators on market 7/2007. ILR sets a price cap based on a bottom-up pure LRIC method determined through the definition of a hypothetical efficient operator in Luxembourg as well as the implementation of a cost model. The concerned operators are free to choose their pricing as long the price cap imposed by ILR is respected on an annual basis. 5 COMMISSION RECOMMENDATION of 17 December 2007 on relevant product and service markets within the electronic communications sector susceptible to ex ante regulation in accordance with Directive 2002/21/EC of the European Parliament and of the Council on a common regulatory framework for electronic communications networks and services 6 Règlement 14/172/ILR du 6 janvier 2014 portant sur la définition des marchés pertinents de la terminaison d’appel vocal sur réseaux mobiles individuels (Marché 7), l’identification des opérateurs puissants sur ces marchés et les obligations imposées à ce titre 7 http://www.ilr.public.lu/communications_electroniques/avis_consultations/conspub140915/adm_M7_complement_non_confidentielle.pdf 8 Décision 06/92/ILR du 2 mai 2006 concernant le marché de gros de la terminaison d'appel vocal sur les réseaux mobiles individuels (marché 16) 8/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 3 (15) Results and Conclusion In order to determine the costs for the provision of mobile call termination, ILR considers different scenarios for the demand of voice services and of data services. The voice traffic is varied from 90% to 110% while the data traffic is sampled between 100% and 135% based on the defined traffic in Chapter 5.1.4. The data points for the related costs are shown in the Figure 3-1. Figure 3-1: Determination of the costs given different scenarios [source: ILR, cost model, 2014] (16) The ILR considers that the costs of the hypothetical efficient operator for the provision of mobile call termination are 0.8702 €cents/minute, which represent the highest result of the different scenarios. The maximum was selected in order to avoid that an operator has to recover the costs for the termination service from revenues of other services (i.e. cross-subsidising). (17) The maximum value of 0.8702 €cents/minute corresponds to the calculated costs for 115% of data traffic and 90% of voice traffic. The minimum value of 0.0771 €cents/minute results for 132% data traffic and 110% of voice traffic. A total of 756 data points are calculated and evaluated. (18) In order to assure that operators will be able to cover the incremental costs for providing the termination service, it is appropriate to select a price that represents the upper limit of the calculated costs. Due to the pure LRIC approach taken, a lower price would not necessarily guarantee to cover the generated costs and thus the operators would need to cross-subsidise using other provided services. Therefore, the ILR considers that the calculated costs of 0.8702 €cents/minute are appropriate to determine the price cap. 9/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (19) These costs are adjusted with the additional wholesale commercial costs (see Chapter 5.3.8), equalling 0.1013 €cents/minute, in order to obtain the price cap (Table 3-1). Table 3-1: Price cap [source: ILR, cost model, 2014] Service Price cap [€cts/min] Voice call termination on individual mobile networks (market 7/2007) (20) 0.97 This price cap will be applicable from 2014 to 2016, since no major changes are expected during the considered period. The details concerning the determination of this price cap as well as the information collected and used are outlined in the following chapters. 10/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 4 Cost accounting rules 4.1 Recommendation of the European Commission (21) On May 7th 2009 the European Commission (EC) published a „Recommendation on the Regulatory Treatment of Fixed and Mobile Termination Rates in the EU“3. According to this recommendation national regulatory authorities shall impose symmetric mobile termination rates based on pure LRIC costs of an efficient operator. The following essential basics of this method are part of the recommendation: - Cost accounting should be based on the method of long-run incremental costs (LRIC) of an efficient operator. Termination fees are therefore equal for all operators. National regulatory authorities have to calculate the difference between long-run total costs of an operator offering the whole range of services and long-run total costs of an operator without termination. - EC recommends to base the valuation of efficient costs on current costs, i.e. gross replacement costs and not historical costs. - EC recommends to use a bottom-up model (BU-LRIC). National regulatory authorities can compare the results of a bottom-up model with a top-down model, which is based on certified accounts, in order to check and improve the resilience of the results and to be able to make the necessary adjustments. - The cost accounting model shall be based on efficient technologies which are available during the relevant period. The core network can be assumed to be NGN-based. The mobile access network should be based on 2G and 3G. - EC recommends to use economic depreciation for all assets as far as possible. - The decision on the proper size of the efficient operator shall refer to a market share higher than or equal to 20% of the retail market. If a national regulatory authority can verify that the market in their country demands a deviating efficient size, they can digress from the recommendation. - Any deviation from these principles, has to be justified by objective cost differences which are not under the influence of the operator in question. The recommendation mentions two reasons for deviations: Cost differences which are due to inhomogeneous allocation of spectra, as for as these cost differences do not diminish over time through the application of market based award procedures (frequency auctions). Or, for a maximum period of four years a new entrant (with an own network) can be allowed higher termination fees, if it is possible to prove that a new entrant to the mobile market, who is smaller than the minimum optimum size, has higher unit costs than the hypothetical operator. (22) In the course of calculation of pure LRIC costs for termination, first of all the total costs (TCS) for all services (S) of an operator have to be calculated based on a bottom-up model and following the above mentioned rules. In a second step total costs for all services excluding termination (T) have 11/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) to be calculated (TCS-T). The deviation between those two calculations gives the absolute pure LRIC costs for the service termination (ICT). ICT = TCS - TCS-T (23) The average pure LRIC costs per minute can be determined through the division of the absolute incremental costs by the termination minutes (MoUT). AICT = ICT / MoUT Figure 4-1: illustration of the pure LRIC concept Total Cost TCS ICT TCS-T MoUT Trafficrelated cost Non-trafficrelated cost Increment 1 (e.g. Data) Increment 2 (e.g. SMS) … Last Increment (i.e. Termination) Output 4.2 The adequacy of pure LRIC (24) In the previous market analysis, termination fees were set on the basis of an international benchmark. (25) The actual cost accounting approach recommended by the EC (i.e. pure LRIC based on a bottomup model) constitutes an approximation to the long-run marginal costs. The recommendation aims explicitly at traffic-related and therefore long-run variable costs. Traffic-related costs are caused by capacity enhancements, which are necessary to manage the increase in traffic. Non trafficrelated (i.e. long run fix common and joint) costs are explicitly excluded, unless they can be directly attributed to the service increment in termination. 12/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5 (26) Determination of the hypothetical efficient operator This section describes the details of the hypothetical efficient operator (HEO) in Luxembourg, for which the termination tariffs are determined. The characteristics of the HEO are based on an efficient network as well as on efficient technologies. Therefore, this section is structured according to the following modelling blocks: - Determination of the demand; - Network design; - Economic factors. (27) As shown in Figure 5-1, these blocks are reflected by the cost model and allow the calculation of the total network costs, and hence the pure LRIC costs caused by the termination service provided by the HEO. Details regarding the mobile cost model are described in the reference document4. Figure 5-1: Overview of the mobile bottom-up model 5.1 Determination of the demand (28) In order to define the network properties and its dimension, the demand that the HEO needs to satisfy has to be determined first. Therefore, fundamental settings such as demographic and geographic data, market share, service categories and technology mix are elaborated in this section. The inputs that are used for determining the demand are shown in Figure 5-2. Figure 5-2: Demand determining factors [Source: ILR (2014)] 13/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.1.1 Demographic and geographic input data 5.1.1.1 Demographic input data (29) The amount of potential users is determined based on available demographic data for Luxembourg. To define the demand, not only the population is considered but also travellers within Luxembourg. Furthermore, different points of interest (Hotspots) are taken into account in order to define the population density per area. Two peak loads are calculated, one for day-time and one for nighttime. This is needed in order to take into account that residents will generate load at two places, e.g. at work during the day and at home in the evening, while travellers will only generate load at work. Finding the highest demand per geographical location is necessary for dimensioning the network. The complete process is described in Chapter 2 of the reference document4. (30) It is crucial for the model to capture the specificities of Luxembourg. One major point that needs to be considered for a correct modelling is the fact that on a daily basis approximately 160 000 commuters are crossing the borders towards their workplaces in Luxembourg9. (31) Based on the derived population density, different geo-types are elaborated. These geo-types are calculated using the maximum amount of users during the day as well as during the night. Table 5-1 shows the classification scheme of the geo-types as implemented in the cost model. Table 5-1: Thresholds for the definition of geo-types (population/km2) [Source: ILR (2014)] Geo-type Population density (32) Urban more than 602 Suburban 602 to 172 Rural less than 172 The market penetration, meaning the SIM cards in use in the market, is set to 108% which results in a population coverage of 870 000 . The amount of users per geo-type is shown in Table 5-2. Table 5-2: Number of users per technology and geo-type [Source: ILR (2014)] Geo-type Technology GSM/EDGE/UMTS/HSPA Urban Suburban Rural Total 424 651 228 445 216 904 870 000 (33) Following the assumption that additional mobile demand arises along the motorways and major railway lines, additional mobile base stations are modelled along these traffic arteries, ensuring the provision of this supplementary supply. (34) Besides the motorways and railways, specific hotspots are considered, e.g. the airport. Figure 5-3 illustrates the increase of passengers and thus the importance of the airport. Furthermore, during the call for input of the reference document, operators suggested to add the main bus station in Luxembourg upper city and some specific high schools in Luxembourg as hotspots. 9 http://www.statistiques.public.lu/ 14/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Figure 5-3: Illustration of the increase in passengers at the Luxembourgish airport (Source: Statec 2012) 5.1.1.2 (35) The signal propagation is, among others, dependent on the surroundings, for instance whether or not there are obstacles. To take this into consideration, the topography of Luxembourg needs to be integrated into the model4. Accordingly the network needs to be adapted to ensure a full coverage of the country and population. 5.1.1.3 (36) Topographic input data Network coverage of an efficient mobile operator In Luxembourg, a complete population coverage (pop-coverage) for mobile voice services can be presumed. Thus, a pop-coverage for 2G (voice, EDGE) as well as for 3G (UMTS, HSPA) of 100% is modelled. On top of this full pop-coverage, a surface coverage of the whole country is considered. 5.1.2 Service categories and user types (37) The total demand in terms of traffic is needed for modelling the mobile network of the hypothetical efficient operator. The traffic is determined by multiplying the individual demand based on a representative (average) mobile user in the busy hour and the number of users of this operator. The busy hour is defined through the total traffic of all services. The number of users as well as the geographical distribution are determined by geographic and demographic data as well as market penetration. Figure 5-4: Demand modelling process (38) The following sections describe the service categories as well as the user types considered for demand modelling. 15/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.1.2.1 (39) Service categories The cost model allows for a differentiation of service categories. Depending on the category, the parameters have different characteristics which influence the network planning for the traffic generated by an average mobile user. The parameters for a given service are composed of the following properties: - Bandwidth required; - Quality of Service (QoS) requirements; - Direction of the route in the network. (40) The model is able to emulate a network based on eight different service categories, as described in the reference document4. However, for the national circumstances it is considered to be appropriate to model the network for the HEO based on only four different service categories. The service categories that are not modelled are either not used at all in the market or so rarely that they do not provide any relevance to the demand modelling. The following service categories are considered: - Real time voice; - Best effort; - SMS; - Mobile broadband access. (41) The service categories describe the features of the generated traffic, without considering the frequency a service or service category is used. The degree of utilisation of a service or a service category will be defined by user types. 5.1.2.2 (42) User types Users can ask for different services in varying degrees. Therefore, the model allows the clustering of the users in different categories based on their utilisation behaviour. The following three user types can be modelled4: - Business user; - Premium user; - Standard user (customer). (43) Despite these modelling possibilities, ILR took only one category into account (i.e “standard user”). This choice relies on the lack of information made available by the Luxembourgish mobile operators. (44) The utilisation degree (BH Erlang) per service category is captured during the busy hour for the given user type. As the busy hour is determined as a function of the total traffic, the traffic of all service categories has to be considered. This total traffic forms the basis for dimensioning the whole network. Some network elements are only used by one service category (e.g. MSC-S, MGW) 16/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) or/and are only used for voice traffic. To dimension these voice-specific network elements, the traffic per voice category and the voice specific busy hour is taken into account. To get the busy hour Erlang value for voice traffic, the busy hour Erlang value for the total traffic is adapted by using a conversion factor10. 5.1.3 Market share of an efficient mobile operator (45) Following the Commission recommendation, mobile termination rates3 are to be defined for operators with an efficient scale of market share. Therefore, the market share considered for this cost determination must be higher than or equal to 20% of the retail market. In the Luxembourg market three operators have their own networks. Hence, ILR considers that a market share of 33,33% reflects a realistic situation. The exact value of 1/3 is implemented in the cost model. (46) The market share is used in the model to derive the amount of subscribers based on the total amount of users in each area. 5.1.4 Determination of traffic volumes (47) Determining the traffic volumes is a major part of the dimensioning process of the demand to be covered by the modelled network. Therefore, real volumes are taken into account. The market players were asked to provide their traffic volumes for 2013 with a forecast for 2014 and 2015. (48) The exact determination of the different traffic volumes is described in the following sections. 5.1.4.1 (49) Determination of voice traffic The data used for the determination of voice traffic forms the annual traffic volumes of voice services (voice volumes measured in technical minutes), consisting of traffic volumes between own customers (on-net), origination to other networks (off-net-outgoing) as well as termination from other networks (off-net-incoming). The different voice traffic volumes are illustrated in Figure 5-5. Figure 5-5: Differentiation between types of voice traffic [source: ILR (2014)] 10 cf Formula 10. 17/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Formula 1 𝑇𝑒𝑐ℎ𝑛𝑖𝑐𝑎𝑙𝑉𝑜𝑙𝑢𝑚𝑒 [𝑚𝑖𝑛] = 𝑇𝑒𝑐ℎ𝑛𝑖𝑐𝑎𝑙𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑛−𝑛𝑒𝑡 + 𝑇𝑒𝑐ℎ𝑛𝑖𝑐𝑎𝑙𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑜𝑢𝑡𝑔𝑜𝑖𝑛𝑔 + 𝑇𝑒𝑐ℎ𝑛𝑖𝑐𝑎𝑙𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑖𝑛𝑔 (50) Based on stakeholders’ data, the aggregated volume of voice traffic for 2013 is shown in Table 5-3. Table 5-3: Aggregated technical voice volume for 2013 [Source: ILR, operators’ data (2014)] Voice – technical minutes (in 1 000 minutes) On-net Origination Termination On-net + origination On-net + termination On-net + origination + termination (51) 2013 367 673 578 035 511 369 945 708 879 042 1 457 077 To take into account the share of unbilled traffic, the voice volume is marked-up with the following formula: Formula 2 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒 [𝑚𝑖𝑛] = (52) 𝑇𝑒𝑐ℎ𝑛𝑖𝑐𝑎𝑙𝑉𝑜𝑙𝑢𝑚𝑒 (1 − 𝑆ℎ𝑎𝑟𝑒 𝑂𝑓 𝑈𝑛𝑏𝑖𝑙𝑙𝑒𝑑 𝑇𝑟𝑎𝑓𝑓𝑖𝑐) Normally a share of unbilled traffic provided by the operators resides between 1% and 4%. This value is included in Formula 2 to incrementally apply on the respective technical volume of all operators. For the Luxemburgish context, a weighted average of the share of unbilled traffic indicated by the operators is used for the pure LRIC calculations as shown in Table 5-4. Table 5-4: Weighted average of the share of unbilled traffic [source: ILR, operators’ data (2014)] Share of unbilled traffic weighted over all operators 2.97% Note: This value is also used for the pure LRIC calculations after the network dimensioning in the following way: The relevant networkdimensioning minutes are reduced by this factor by the application of Formula 2. (53) Furthermore, the on-net volume is considered twice (see Formula 3) in order to reflect the double utilisation of network resources for a communication between two end-customers on the same network. Formula 3 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑛−𝑛𝑒𝑡 ×2 = 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑛−𝑛𝑒𝑡 × 2 (54) Based on these calculated voice volumes, the respective portions of the different traffic types are elaborated as follows: Formula 4 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 [𝑚𝑖𝑛] = 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑛−𝑛𝑒𝑡 ×2 + 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑜𝑢𝑡𝑔𝑜𝑖𝑛𝑔 + 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑖𝑛𝑔 18/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Formula 5 𝑆ℎ𝑎𝑟𝑒 𝑂𝑓 𝑇𝑟𝑎𝑓𝑓𝑖𝑐𝑜𝑛−𝑛𝑒𝑡 = 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑛−𝑛𝑒𝑡 ×2 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 Formula 6 𝑆ℎ𝑎𝑟𝑒 𝑂𝑓 𝑇𝑟𝑎𝑓𝑓𝑖𝑐𝑜𝑓𝑓−𝑛𝑒𝑡 𝑜𝑢𝑡𝑔𝑜𝑖𝑛𝑔 = 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑜𝑢𝑡𝑔𝑜𝑖𝑛𝑔 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 𝑆ℎ𝑎𝑟𝑒 𝑂𝑓 𝑇𝑟𝑎𝑓𝑓𝑖𝑐𝑜𝑓𝑓−𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑖𝑛𝑔 = 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑓𝑓−𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑖𝑛𝑔 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 Formula 7 (55) The aggregated voice volumes for all operators calculated for 2013 and the respective traffic portions are illustrated in Table 5-5. Table 5-5: Aggregated voice volumes and traffic shares for 2013 [Source: ILR, operators’ data (2014)] Voice (in 1 000 minutes) On-net (x2) Origination Termination Portion of on-net traffic Portion of origination traffic Portion of termination traffic (56) 2013 757 381 596 607 526 309 40.28% 31.73% 27.99% The annual voice volume as well as the Busy Hour (BH) of an operator are determined using the aggregated traffic distribution of all operators per hour during one day. This distribution is the average of the most traffic intense 30 days of one year. The determination of this volume and the Busy Hour looks as follows: Formula 8 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑖=[00ℎ00−24ℎ00] [𝑚𝑖𝑛] = 𝐷𝑢𝑟𝑖𝑛𝑔𝑇ℎ𝑒𝐷𝑎𝑦 𝑉𝑜𝑙𝑢𝑚𝑒 𝑆ℎ𝑎𝑟𝑒𝑖 × 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 Formula 9 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑉𝑜𝑖𝑐𝑒𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 [𝑚𝑖𝑛] = 𝑚𝑎𝑥(𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑖=[00ℎ00−24ℎ00] ) (57) Table 5-6 illustrates the traffic distribution of the aggregated voice volume per hour of all operators in Luxembourg. The table shows that the Voice Busy Hour is between 17h00 and 18h00 while the overall Network Busy Hour is between 19h00 and 20h00. 19/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 5-6: Aggregated annual voice volumes distribution per hour [Source: ILR, operators’ data (2014)] Aggregated annual voice volumes per hour (in 1 000 minutes) 00h00 – 01h00 01h00 – 02h00 02h00 – 03h00 03h00 – 04h00 04h00 – 05h00 05h00 – 06h00 06h00 – 07h00 07h00 – 08h00 08h00 – 09h00 09h00 – 10h00 10h00 – 11h00 11h00 – 12h00 12h00 – 13h00 13h00 – 14h00 14h00 – 15h00 15h00 – 16h00 16h00 – 17h00 Voice Busy Hour 17h00 – 18h00 Network Busy Hour 18h00 – 19h00 19h00 – 20h00 20h00 – 21h00 21h00 – 22h00 22h00 – 23h00 23h00 – 24h00 Total (58) 2013 23 272 10 065 5 890 3 830 2 676 3 368 8 390 31 164 84 875 113 236 130 745 132 535 119 704 115 995 128 044 127 631 137 654 149 011 137 501 128 164 105 362 90 462 59 743 30 980 1 880 298 The network is dimensioned based on the overall traffic of the busy hour for all service types, not only on voice traffic. Therefore a Busy Hour Voice factor is defined: Formula 10 𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 𝑉𝑜𝑖𝑐𝑒𝐹𝑎𝑐𝑡𝑜𝑟 = (59) 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑉𝑜𝑖𝑐𝑒𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 The portion of the voice volume in the network Busy Hour is put in relation to the total voice volume of one day. Formula 11 𝐷𝑎𝑖𝑙𝑦𝑆ℎ𝑎𝑟𝑒𝑂𝑓𝑉𝑜𝑖𝑐𝑒𝐼𝑛𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 = (60) 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 𝑉𝑜𝑖𝑐𝑒𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑡𝑜𝑡𝑎𝑙 Using the Busy Hour for the aggregated traffic of all operators (defined in Section 5.1.4.3), the Busy Hour voice factor as well as the Daily Share of voice in the network Busy Hour can be calculated. The results are illustrated in Table 5-7. Table 5-7: Busy Hour Voice Factor and Daily Share of Voice in Network Busy Hour [Source: ILR (2014)] Busy Hour Voice Factor (bhvf) Daily Share of Voice in Network Busy Hour 1.08 7.31% 20/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (61) A Busy Hour voice factor of 1.08 means that the voice volume in the Voice Busy Hour is 1.08-times the overall voice volume in the Network Busy Hour. The Daily Share of voice in the Network Busy Hour represents the portion of voice traffic in the Network Busy Hour in relation to the total voice day traffic. 5.1.4.2 Determination of data traffic In order to determine the data traffic, annual traffic volumes of data services (volume in GiB11) were collected per operator. These traffic volumes are separated in those that go from the network to the subscriber (downlink) and in those that go from the subscriber to the network (uplink): (62) Formula 12 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 𝑖=[00ℎ00,24ℎ00] [𝐺𝑖𝐵] = 𝑉𝑜𝑙𝑢𝑚𝑒𝑑𝑜𝑤𝑛𝑙𝑖𝑛𝑘 + 𝑉𝑜𝑙𝑢𝑚𝑒𝑢𝑝𝑙𝑖𝑛𝑘 (63) The aggregated data volumes of all operators for 2013 as well as the corresponding traffic shares are outlined in the following table. Table 5-8: Aggregated data volumes and traffic shares for 2013 [Sources: ILR, operators’ data (2014)] Data Downlink data Uplink data Share of downlink data Share of uplink data (64) 2013 4 247 588 GiB 472 947 GiB 89.98% 10.02% The annual volume as well as the Busy Hour (BH) of an operator are determined using the aggregated traffic distribution of all operators per hour during one day. This distribution is the average of the most traffic intense 30 days of one year. The determination of this volume and the Busy Hour looks as follows: Formula 13 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑠𝑖=[00ℎ00,24ℎ00] [𝐺𝑖𝐵] = 𝐷𝑎𝑖𝑙𝑦𝑆ℎ𝑎𝑟𝑒𝑂𝑓𝐷𝑎𝑡𝑎𝑖 × 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 Formula 14 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝐷𝑎𝑡𝑎𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 [𝐺𝑖𝐵] = 𝑚𝑎𝑥(𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑖=[00ℎ00,24ℎ00] ) (65) The calculated distribution for the aggregated annual data volume is shown in Table 5-9, which indicates that the Data Busy Hour is between 21h00 and 22h00. Using the Unit GB (Gibga Byte) for data volumes is not always clear. The prefix G (Giga) can be considered as a decimal prefix 10^9 (Système international d’unités – SI prefix). This would mean that 1GB = 1.000.000.000 Byte. Nevertheless, the prefix G can also be considered as a binary prefix 1024, which would result in 1GB = 1.073.741.824 Byte. To differentiate between the two prefixes, the IEC (International Electrotechnical Commission) introduced the binary prefix GiB (Gibibyte), so 1 GiB = 1.073.741.824 Byte. 11 21/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 5-9: Distribution of the aggregated annual data volumes [Sources: ILR, operators’ data (2014)] Aggregated annual data volumes per hour [in GiB] 00h00 – 01h00 01h00 – 02h00 02h00 – 03h00 03h00 – 04h00 04h00 – 05h00 05h00 – 06h00 06h00 – 07h00 07h00 – 08h00 08h00 – 09h00 09h00 – 10h00 10h00 – 11h00 11h00 – 12h00 12h00 – 13h00 13h00 – 14h00 14h00 – 15h00 15h00 – 16h00 16h00 – 17h00 17h00 – 18h00 Network Busy Hour 18h00 – 19h00 19h00 – 20h00 20h00 – 21h00 Data Busy Hour 21h00 – 22h00 22h00 – 23h00 23h00 – 24h00 Total (66) 2013 174 809 129 987 89 293 60 763 47 605 47 597 71 161 134 248 166 193 204 473 204 681 227 563 249 518 272 799 254 356 254 809 264 312 270 209 276 492 268 321 268 336 281 172 265 478 236 359 4 720 535 The share of data traffic of 5.86% in the Network Busy Hour can be deduced with respect to the overall total data traffic per day in relation to the voice traffic. 5.1.4.3 Determination of total traffic in the network (67) The network dimensioning is based on the total traffic of the Busy Hour of all services. In order to define the Busy Hour of the overall traffic, the traffic for voice and data needs to be combined. The network is an all-IP NGN network and therefore all the services are transported over the same IP based network. For this reason it is necessary to convert the voice volume, expressed in minutes, to an equivalent of data volume. (68) The conversion of the voice volume into data volume is related to the voice coding. In mobile networks, the voice coding is generally done using the Adaptive Multi-Rate (AMR) Narrowband Codec with a bit rate of 12.2 kbps per communication way (mouth to ear). As voice is a bidirectional service, the double bit rate (24.4 kbps) is used. The annual voice volume distribution over one day is calculated as follows: Formula 15 𝑉𝑜𝑖𝑐𝑒 𝑉𝑜𝑙𝑢𝑚𝑒𝑖=[00ℎ00,24ℎ00] [𝐺𝑖𝐵] 𝑉𝑜𝑖𝑐𝑒 𝑉𝑜𝑙𝑢𝑚𝑒𝑖=[00ℎ00,24ℎ00] [𝑚𝑖𝑛] × 60 [𝑠/𝑚𝑖𝑛] × (2 × 12.2 × 1000 [𝑏𝑝𝑠]) = 8 [𝑏𝑖𝑡⁄𝐵𝑦𝑡𝑒] × 1073741824[𝐵𝑦𝑡𝑒/𝐺𝑖𝐵] 22/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (69) Using this calculation, the voice volume can now be put together with the uplink and the downlink of the data volume as follows: Formula 16 𝑇𝑜𝑡𝑎𝑙 𝑉𝑜𝑙𝑢𝑚𝑒𝑖=[00ℎ00,24ℎ00] [𝐺𝑖𝐵] = 𝑉𝑜𝑖𝑐𝑒 𝑉𝑜𝑙𝑢𝑚𝑒𝑖 [𝐺𝑖𝐵] + 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑖 [𝐺𝑖𝐵] (70) Accordingly, the Network Busy Hour can be defined as shown in Formula 17. The corresponding results are shown in Table 5-10. Formula 17 𝑇𝑜𝑡𝑎𝑙 𝑉𝑜𝑙𝑢𝑚𝑒𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 [𝐺𝑖𝐵] = 𝑚𝑎𝑥(𝑇𝑜𝑡𝑎𝑙 𝑉𝑜𝑙𝑢𝑚𝑒𝑖=[00ℎ00,24ℎ00] ) Table 5-10: The distribution of the aggregated total volume [Sources: ILR, operators’ data (2014)] Aggregated annual voice and data volumes per hour (in GiB) 00h00 – 01h00 01h00 – 02h00 02h00 – 03h00 03h00 – 04h00 04h00 – 05h00 05h00 – 06h00 06h00 – 07h00 07h00 – 08h00 08h00 – 09h00 09h00 – 10h00 10h00 – 11h00 11h00 – 12h00 12h00 – 13h00 13h00 – 14h00 14h00 – 15h00 15h00 – 16h00 16h00 – 17h00 17h00 – 18h00 Network Busy Hour 18h00 – 19h00 19h00 – 20h00 20h00 – 21h00 21h00 – 22h00 22h00 – 23h00 23h00 – 24h00 Total 23/54 2013 178 775 131 703 90 296 61 416 48 061 48 171 72 591 139 559 180 659 223 772 226 964 250 151 269 920 292 569 276 179 276 561 287 773 295 605 299 926 290 164 286 293 296 590 275 660 241 639 5 040 998 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (71) The busy hour of the total traffic of all services is between 18h00 and 19h00. Figure 5-6 shows the graphical distribution of the voice, data and total volume. It can be observed that the voice traffic is far less important compared to the data traffic. Nevertheless, the voice traffic has enough influence to shift the Network Busy Hour away from the Data Busy Hour towards the Voice Busy Hour. Figure 5-6: Hourly distribution of voice, data and total volume for 2013 [source: ILR, operators’ data (2014)] 5.1.4.4 (72) Setting of traffic for different service categories In Section 5.1.2.2 it is explained that per service category a different utilisation degree (BH Erlang) is captured. In order to calculate the traffic values (in Erlang) per service category, the Network Busy Hour is used as follows: Formula 18 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒𝑉𝑜𝑖𝑐𝑒 [𝐸𝑟𝑙] = 𝑉𝑜𝑖𝑐𝑒 𝑉𝑜𝑙𝑢𝑚𝑒𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 [𝑚𝑖𝑛] 60 [𝑚𝑖𝑛] × #𝑈𝑠𝑒𝑟𝑠 × #𝐷𝑎𝑦𝑠 Formula 19 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒𝐷𝑜𝑤𝑛𝑙𝑖𝑛𝑘𝐷𝑎𝑡𝑎 [𝑏𝑝𝑠] 𝐷𝑎𝑡𝑎 𝑉𝑜𝑙𝑢𝑚𝑒𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 [𝐺𝑖𝐵] × 1073741824 [𝐵𝑦𝑡𝑒⁄𝐺𝑖𝐵] × 8[𝑏𝑖𝑡/𝐵𝑦𝑡𝑒] = 3600 [𝑠/ℎ] × #𝑈𝑠𝑒𝑟𝑠 × #𝐷𝑎𝑦𝑠 × 𝑆ℎ𝑎𝑟𝑒𝑂𝑓𝐷𝑎𝑡𝑎𝐷𝑜𝑤𝑛𝑙𝑖𝑛𝑘 24/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (73) For network dimensioning purposes, only the data service with a high utilisation degree is relevant. Therefore only the downlink data traffic is calculated. (74) In order to have a valid input for the model, the traffic value that is represented in bps needs to be converted into Erlang as follows: Formula 20 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒𝑠𝑒𝑟𝑣𝑖𝑐𝑒 [𝐸𝑟𝑙] = (75) 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒𝑠𝑒𝑟𝑣𝑖𝑐𝑒 [𝑏𝑝𝑠] 𝑏𝑖𝑡 𝑟𝑎𝑡𝑒 𝑠𝑒𝑟𝑣𝑖𝑐𝑒 [𝑏𝑝𝑠] The traffic value (bps) of a voice service can be converted into the Voice-Equivalent-Erlang by using the voice bit rate of 12.2 kbps as follows: Formula 21 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒 [𝑉𝑜𝑖𝑐𝑒𝐸𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡𝐸𝑟𝑙] = (76) 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒 [𝑏𝑝𝑠] 12.2 × 1000 [𝑏𝑝𝑠] Table 5-11 shows the resulting traffic values regarding the aggregated traffic volume of the operators, the total amount of users and the amount of relevant days of a year. Table 5-11: Traffic values for voice and data services for 2013 [Sources: ILR, operators’ data (2014)] Traffic values per user in the network busy hour Number of users Number of days per year Voice [Erl] Voice [kbps] Data per downlink [kbps] Voice and data downlink [kbps] (77) 2013 870 000 250 0.010536 0.128545 2.729362 2.857907 The modelled service categories are introduced in Section 5.1.2.1, they include the SMS message service category. Therefore, also a traffic value for these messages needs to be determined, which is done as follows: Formula 22 𝑂𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑇𝑖𝑚𝑒𝑚𝑒𝑠𝑠𝑎𝑔𝑒 [𝑚𝑖𝑛] = 𝐿𝑒𝑛𝑔𝑡ℎ𝑚𝑒𝑠𝑠𝑎𝑔𝑒 [𝐵𝑦𝑡𝑒] × 8[𝑏𝑖𝑡/𝐵𝑦𝑡𝑒] 1 × 𝑏𝑖𝑡 𝑟𝑎𝑡𝑒𝑚𝑒𝑠𝑠𝑎𝑔𝑒 [𝑏𝑝𝑠] 60 [𝑠/𝑚𝑖𝑛] Formula 23 #𝑀𝑒𝑠𝑠𝑎𝑔𝑒𝑠𝑃𝑒𝑟𝑈𝑠𝑒𝑟𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 = (78) #𝑀𝑒𝑠𝑠𝑎𝑔𝑒𝑠𝑌𝑒𝑎𝑟 × 𝑆ℎ𝑎𝑟𝑒𝑂𝑓𝑇𝑟𝑎𝑓𝑓𝑖𝑐𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 #𝑈𝑠𝑒𝑟𝑠 × #𝐷𝑎𝑦𝑠 To define the traffic portion during the Busy Hour, the Daily Share of Voice in Network Busy Hour is considered as follows: 25/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Formula 24 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑉𝑎𝑙𝑢𝑒𝑚𝑒𝑠𝑠𝑎𝑔𝑒 [𝐸𝑟𝑙𝑎𝑛𝑔] = 2 × #𝑀𝑒𝑠𝑠𝑎𝑔𝑒𝑠𝑃𝑒𝑟𝑈𝑠𝑒𝑟𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝐵𝑢𝑠𝑦𝐻𝑜𝑢𝑟 × 𝑂𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑡𝑖𝑚𝑒𝑚𝑒𝑠𝑠𝑎𝑔𝑒 1 × 60 [𝑚𝑖𝑛/ℎ] (79) Using the double amount of messages per user is analogical with the on-net voice traffic and is necessary to take into account the double occupation of network resources for the transmission of a message between two end-users. (80) The following traffic values for SMS result from these calculations: Table 5-12: Traffic values for SMS services for 2013 [Sources: ILR, operators’ data (2014)] SMS Number of SMS per year Occupied time of a SMS [min] Number of sent SMS per user in the network busy hour SMS traffic value [Erl] 2013 915,730,000 0.001 0.308 0.000014 (81) The traffic distribution of the best-effort and mobile broadband services are based on the contributions of the operators as well as on international benchmarks. (82) In order to represent undelivered or dropped traffic, a blocking factor is integrated into the model. The amount of traffic provided by the operators represent the delivered traffic. Therefore this traffic is increased by the blocking factor. This is necessary in order to correctly dimension the network. The blocking factor, by which the traffic value is multiplied, depends on the service category and is shown in Table 5-13. Besides the blocking factor, this table also summarises the inputs for the service categories for the model. Table 5-13: Input values into the bottom-up cost model for different services [source: ILR, mobile cost model (2014)] Service name Traffic value downlink [kbps] Traffic value in Erlang Blocking factor Network traffic in Erlangs of a user in the network BH Average data rate required for the service in the fixed network for uplink [kbps] Average data rate required for the service in the fixed network for downlink [kbps] Average data packet length in bytes in uplink Average data packet length in bytes in downlink Average service session duration in minutes Voice Best effort SMS Mobile broadband 0.128545 0.010536 1.017 0.010716 0.191055 0.002388 1.01 0.002412 0.000137 0.000014 1.01 0.000014 2.538307 0.009401 1.02 0.009589 12.2 20 9.6 30 12.2 80 9.6 270 16 200 100 200 16 200 100 1150 3 3 0.001 5 Source: ILR, operators’ data (2014) 26/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.1.5 Quality and Security (83) A major aspect for dimensioning the network is the guarantee of network security and quality. The network needs to guarantee a specific quality for delivering the various services. Especially voice and real-time services require a given level of quality of service to remain fully functional. On the other hand, a network also needs to be able to adapt to unforeseen incidents, which means that a network needs to be redundant so that even when facing a failure, the services can still be delivered. (84) Respecting the quality and security requirements implies that some equipment will be required twice, even though only one is needed to satisfy the demand. 27/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.2 Network design (85) The model provides the ability to define the network design based on numerous parameters as shown in the Reference document4. Two different types of parameters can be distinguished. On the one hand, parameters that are accessible for optimisation, such as the number of sites per network layer. On the other hand, parameters which are not accessible for optimisation. These parameters are e.g. parameters that have to be set from a regulatory point of view or with which the degree of redundancy or resilience against failure of network components is determined (e.g. doubled connections, protection level). The total network costs are (as expected) lower, the lower the redundancy or the selected security is. The corresponding parameters are thus set in the model corresponding to an acceptable business practice level regarding redundancy and protection. (86) The determination of the parameters is based on regulatory decisions, input from Luxembourg operators and past experience4. This forms the starting point for the network optimisation process. (87) Optimising and determining the effective network is carried out as follows: - (88) First, the required frequency spectrum and the necessary 2G as well as 3G coverage is determined from a regulatory point of view. Then, the number of controller locations and the optimal number of core network locations are elaborated; With these nodes, the minimum distances between the controller/core network locations are optimised; All the previous results are iteratively verified in order to consider possible changes resulting from the optimisation process. In the next sections the results of these steps are shown in detail. 5.2.1 Frequency spectrum (89) The Luxembourg mobile operators use allocated frequencies in the 900 MHz, 1800 MHz and 2100 MHz frequency band. According to the regulatory decision to apply a market share of 33% (see Section 5.1.3) on the HEO, the HEO is modelled with 33% of the total spectrum. The Table 5-14 gives an overview of the total available spectrum as well as the spectrum of the hypothetical 33% operator. Table 5-14 Frequency spectrum [source: ILR, mobile cost model (2014)] Overall spectrum 900 MHz band 33,8 MHz 1800 MHz band 14,2 MHz 2100 MHz band 35 MHz 11,2 MHz 4,6 MHz 10 MHz (either downlink or uplink) 33% operator*) *) The values are rounded since these frequency bands are used for GSM and UMTS and therefore only 0,2 and 5 MHz blocks can be utilised respectively. 5.2.2 Radio access network (90) According to Section 5.1.1.3, the modelled hypothetical operator provides full area and full population coverage for both 2G and 3G services. (91) The optimised mobile network of the hypothetical operator has approx. 570 sites with 2G and/or 3G base stations, approx. 440 BTSs, approx. 330 Node Bs, approx. 1110 TRXs, approx. 1570 carriers 28/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) and 31 cell hubs. According to the data provided by the Luxembourg operators, their networks have between 350 and 600 sites with base stations, 350 to 800 base stations (BTSs and Node Bs), up to 2800 TRXs and up to 1500 carriers. Therefore, the modelled network of the hypothetical operator approximates the real mobile networks in Luxembourg quite well. 5.2.3 Controller locations (92) For these comparative static model calculations, the number of controller locations (i.e. locations with BSC and RNC nodes) are varied12 between 2 and 7. The results are shown in the following figure. Figure 5-7: Total annual network costs depending on the number of controller locations13 [source: ILR, mobile cost model (2014)] (93) The minimum of the total network costs is achieved with two controller locations, which is also in line with the amount of controller locations deployed by some Luxembourg operators due to redundancy reasons. Hence for the network of the hypothetical operator the number of controller locations has been set to two. (94) The differences in the overall network costs for up to seven locations and between the cost in case of two locations is up to 7,06%. 12 Less than two controller locations are not considered as a viable option because of a negative impact for 100% of all network services in case of such a single point of failure. This holds true for the core network locations investigated later on too. 13 This is a cumulative diagram representation of the costs. The values are stacked in the order BTS/Node B/HSPA, MSC call server, media gateway, LER etc. finally up to PCU, hub controller links, controller-core links and BSC/RNC. 29/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.2.4 Core network locations (95) In this section, the analysis regarding the optimal number of core network locations are presented. The following figure shows the dependence of the total annual network costs and the number of core network locations. Figure 5-8: Total annual network costs depending on the number of core network locations14 [source: ILR, mobile cost model (2014)] (96) From the above figure it can be seen that the minimum of the total annual network costs is determined with two core network locations. The differences in the total network costs for the range of up to seven core network locations and the costs in case of two core network locations are up to 13,49%. Note that an increase of the core network locations involves an increase of the number of controller locations, because the core network locations are always co-located with the controller locations15. 14 In order to give a better resolution of the view, the presentation of the Y-axis starts at € 19 million. The costs up to € 19 million are entirely attributable to base station costs (BTS/Node B/HSPA), which are independent of the number of core network locations. See also footnoteError! Bookmark not defined.. 15 This is necessary because the core network locations are selected from the set of controller locations by the model and thus cannot be greater than the number of controller locations. 30/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (97) The control layer nodes MSC-S, HLR, SMSC and IN located on core network locations have enough capacity to compensate a complete failure (100%) of a core network location. (98) The model implements redundant connections between the controller sites and the core network locations. Hence, the complete traffic is doubled and the MGWs and their interconnection interfaces are dimensioned for the voice busy hour in a redundant manner. Similarly, the capacities of SGSN, GGSN and LER nodes and switches are doubled due to these doubled connections. According to this redundancy approach, the network of the HEO includes two core network locations. 5.2.5 Minimum distances (99) The final fine adjustment on the hypothetical efficient operator’s mobile network relates to the determination of the minimum distances between the controller/core network locations. The minimum distances between the controller locations were varied between 0 km and 59,316 km, with the following results. Figure 5-9: Total annual network costs depending on the minimum distance between controller locations [source: ILR, mobile cost model (2014)] 16 For a minimum distance of more than 59,3 km, the model doesn’t find a solution for the specified number of controller locations. 31/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (100) The total network costs’ minimum is determined when the controller locations are located at a distance of 16,4 km. This results in two controllers located in Luxembourg City (Gare) and in Differdange. The differences in the total network costs for all minimum distances and the mentioned minimum distance are below 1,34%. In case of excluding the peaks caused mainly by an additional RNC for distances of 23,8 km and those below 16,4 km, this difference is below 0,07%. (101) Hence, the minimum distance between controller locations of the HEO is fixed to 16,4 km. (102) Because the number of the core network locations is the same as the number of the controller locations, these locations are co-located. Hence the distance between the core network locations is the same as for the controller locations, so the core network locations are located in Luxembourg City (Gare) and in Differdange. (103) A 100% protected physical ring/link connects the two controller/core network sites. By means of this ring, a controller location is connected using doubled link connections with 100% protection to the other core network location. 5.2.6 Interconnection locations for voice services (104) The number of interconnection locations is set to two and these are collocated with the selected core network locations. More than two interconnection sites are, due to merely two core network locations, not efficient, whereas only one site represents from a redundancy and resilience point of view a single point of failure, which is not recommended. 5.3 Economic parameters (105) The determination of the mobile termination costs requires as input the economic factors outlined in Figure 5-10. Figure 5-10 : Economic parameters [Source: ILR (2014)] 32/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.3.1 Gross replacement cost (106) For the determination of the gross replacement cost of the network (i.e. GRC), ILR has opted for a valuation method using the current prices of the assets and as such the resulting values represent those of a new network (i.e. modern equivalent assets). In other words, the adopted conceptual approach considers that the cost of the network equals those incurred by a new entrant on the market. (107) The total GRC of the hypothetical efficient operator’s network equals 108.62 mio €. The investment related inputs used by the ILR in its model are based on operators’ inputs as well as on international benchmarks. 5.3.2 Depreciation method (108) The annual amortisation values of the network assets cover both the depreciation of and the interest on the capital resources employed. The ILR has opted for the annuity approach [i.e. a total annualised CAPEX of 18 894 035 €]. (109) For the determination of the economic depreciation, the ILR considers also the variation of the output (Δg) as well as the evolution of the assets’ prices (Δp) in order to compensate for the asset’s value loss over its economic lifetime. 5.3.3 Expected price change (Δp) (110) The ILR considers an estimated yearly change in the prices of the specific assets (i.e. Δp) for the determination of the GRC. Thus, e.g. a negative price change represents a positive technological change in the domain of network assets as well as network support assets. The expected price changes of network assets taken into account by the ILR are shown in Table 5-15. The expected price changes of network support assets are illustrated in Table 5-16. 5.3.4 Expected growth rate (Δg) (111) As referred supra, the ILR considers the estimated average variation of the utilisation rate of the installation during its economic lifetime (i.e. Δg). This expected growth rate is used in order to be able to react to changes on the demand side with regards to the depreciation as well as to allow a better approximation between the demand and the depreciation of the investment. (112) The mobile market in Luxembourg is characterised by increasing capacities in voice traffic and strongly increasing capacities in data traffic. The ILR therefore considers in its bottom-up cost model: - Δg = 0 for all active components used by voice and data as well as those which have a short lifetime and can therefore be adapted relatively quickly to changes; - Δg > 0 for passive elements used by voice and data. The considered expected growth rates of the network assets, computed according to the historical usage pattern for voice and data, are shown in Table 5-15. The expected growth rates of the network support assets are represented in Table 5-16. (113) 33/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.3.5 Economic lifetime (114) The corresponding economic lifetimes of the different network assets considered for the determination of the annualised CAPEX are illustrated in Table 5-15. The corresponding economic lifetimes of network support assets used by the ILR are represented in the Table 5-16. Table 5-15: Expected price change, expected growth rate and economic lifetime of network assets [source: ILR, mobile cost model, 2014] Network assets Expected price change Expected growth rate Economic lifetime (years) BTS / NodeB / HSPA Sites 1.10% 12.41% 18 Equipment -3.67% 0% 8 TRX / Carrier -3.78% 0% 8 BSC site 1.10% 12.41% 18 RNC site 1.10% 12.41% 18 BSC hardware -1.50% 0% 8 BSC software -4.75% 0% 5 RNC hardware -1.50% 0% 8 RNC software -4.75% 0% 5 BSC ports -1.50% 0% 8 RNC ports -1.50% 0% 8 PCU BSC -1.50% 0% 8 MSC call server -3.20% 0% 8 MGW Media Gateway Ports -3.20% -1.50% 0% 0% 8 8 Core sites 1.10% 12.41% 18 BSC/RNC Other core location equipment HLR -1.50% 0% 8 AUC -1.50% 0% 8 EIR -1.50% 0% 6 LER / LSR -1.50% 0% 8 SMSC -1.50% 0% 8 SGSN -1.50% 0% 8 GGSN -1.50% 0% 8 IC interface -1.50% 0% 8 Network management system -1.50% 0% 8 IN -1.50% 0% 6 Aggregation systems Aggregation systems ports -0.90% -0.90% 0% 0% 8 8 Radio links -1.80% 0% 8 Other 34/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 5-16: Expected price change, expected growth rate and economic lifetime of network support assets [source: ILR, mobile cost model, 2014] Network support assets Expected price change Expected growth rate Economic lifetime (years) Vehicles 2.00% 0% 5 Office equipment Workshop Equipment 2.00% 2.00% 0% 0% 6 5 IT / general purpose computer -4.75% 0% 5 Network management Land and buildings 2.00% 1.10% 0% 0% 6 28 5.3.6 Weighted Average Cost of Capital (WACC) (115) According to articles 28 (1) c) and 33 (2) of the “Loi du 27 février 2011 sur les réseaux et les services de communications électroniques“, the ILR, when setting price caps of regulated wholesale products, should consider the investments realised by the SMP operators and allow them a reasonable remuneration of the capital employed, which should reflect the associated investment risk. (116) The cost of capital equals the weighted average cost of capital (WACC). The ILR considers a pre-tax nominal WACC of 12% for the mobile network activities under review which corresponds to the highest WACC provided by Luxembourgish operators. 5.3.7 Operational expenditure (OPEX) (117) The considered operational expenditure of the hypothetical efficient operator is equal to the expenses required for the functioning/operation of the network. (118) ILR determines the OPEX as a percentage mark-up on the GRC of the relevant facilities. The values of these percentage mark-ups are determined on the basis of values provided by the Luxembourgish operators as well as the WIK database (Table 5-17 and Table 5-18). Table 5-17: OPEX mark-up [source: ILR, mobile cost model, 2014] Network support assets OPEX mark-up Motor vehicles 11% Office equipment 11% Workshop Equipment 11% IT / general purpose computer 11% Network management 30% Land and buildings 5% 35/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 5-18: OPEX mark-up on direct investment [source: ILR, mobile cost model, 2014] Network segment OPEX mark-up on direct investment BTS / NodeB / HSPA BTS (2G) 14.75% NodeB (3G) 14.08% HSPA 14.08% 2G/3G 14.41% 2G/HSPA 14.08% 3G/HSPA 14.08% 2G/3G/HSPA 14.08% BSC / RNC BSC/PCU 14.41% RNC 14.75% MSC call server, MGW and LER MSC call server 12.46% Media Gateway 13.13% LER / LSR 13.13% Core sites 11.46% Other core location equipment HLR 17.95% AUC 17.91% EIR 17.91% SMSC 30.98% SGSN 16.52% GGSN 13.36% Others (119) IC interface 10.63% Network management system 9.97% IN 9.97% Aggregation systems 10.13% Radio links 9.95% The annual OPEX of the considered hypothetical efficient operator’s network equals 15 416 538€. 36/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 5.3.8 Additional wholesale commercial costs (CCAdd) (120) The costs of the service of termination is not only consisting of costs generated by the operator’s network, but also of an additional wholesale commercial cost (e.g. billing). If the service of termination is neither offered nor billed, these additional wholesale commercial costs will not occur. Given that these costs are not included in the results of the model, ILR considers it appropriate to allocate these costs to these results. (121) For the determination of the additional wholesale commercial cost (CCAdd) of the hypothetical efficient operator, ILR inquired for values among the Luxembourgish operators. The three operators offering the service of mobile call termination in Luxembourg are also offering the service of fixed call termination. (122) This additional wholesale commercial cost represents the sale of termination provision. The task of sale in itself is independent on whether the termination that is sold is provided over a fixed or a mobile network. Therefore the ILR considers that the related cost for that task is the same for both termination types. Hence, the ILR uses the same statistical analysis as is done with respect to the price cap determination for termination on fixed networks17. This analysis is more representative as more data points (number of operators providing the given termination) are taken into account than if a separate analysis would solely be done for three mobile network operators. (123) In order to determine this CCAdd, the operators that provide termination on fixed or mobile networks were asked to contribute their wholesale commercial cost related to that task. To be able to compare the different values, the costs per operator are normalised by dividing the cost with the termination volume of each operator. The resulting outcome is represented in Figure 5-11. The figure shows that one operator has a very divergent value with respect to the other operators. This data point is therefore considered as an outlier and removed from the data sample. Figure 5-12 shows a more detailed overview of the remaining operators. Nevertheless, the figure shows that this time two data points form outliers. Again, these data points are removed from the data sample. By analysing why these operators represent outliers, it is noticed that those three operators have a very limited termination volume. Having a limited termination volume is not comparable to the volume of the HEO and therefore not considered as being relevant for this analysis. 17 Price cap determination for the provision of termination on fixed networks: Exposé des motifs - Projet de règlement portant Fixation des plafonds tarifaires pour les prestations de départ d’appel sur le réseau téléphonique public en position déterminée (Marché 2/2007) et de la terminaison d’appel sur divers réseaux téléphoniques publics individuels en position déterminée (Marché 3/2007). 37/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Figure 5-11 : Distribution of the wholesale commercial cost per minute of fixed termination per operator [source : ILR, operators, 2014] (124) Figure 5-12 : Distribution without operator « E » [source : ILR, operators, 2014] Figure 5-13 displays the final data set considered for the calculation of the CCAdd. The mean value of all remaining operators is taken. In order to represent the fact that more of the data points is underneath the mean value, the variance is subtracted from the mean value. The resulting CCAdd that is considered for the HEO is 0.1013 €cents/minute. Figure 5-13: Additional wholesale comercial cost analysis [source : ILR, 2014] 38/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 6 Determination of the price cap (125) In this section the process for determining the mobile termination rates (MTR) based on a pure LRIC method for a hypothetical efficient operator using a cost model is explained. Therefore the corresponding inputs needed by the cost model as well as the generated outputs are identified in the following paragraphs. (126) The figures regarding the various steps contain the different inputs [green background] that lead to intermediate results [orange background] and final results [orange background, red border]. The corresponding values are also illustrated later in this section or a reference is made to the corresponding appendix. (127) As already mentioned, ILR uses a BU pure LRIC approach for determining the costs of mobile call termination in order to set a price cap on the service of voice call termination on individual mobile networks (market 7/2007) [Output_10: 0.97 €cts/min.] (128) The process is divided into five steps: 1. 2. 3. 4. 5. (129) Determination of the HEO’s number of SIM cards; Determination of the quantity of the network elements; Determination of the total annual network costs; Determination of the pure LRIC cost; Cost increased by the additional wholesale commercial costs. The first two steps, i.e. determination of the HEO’s number of SIM cards and the determination of the quantity of the network elements, are outlined in Figure 6-1. Figure 6-1 : Determination of network equipment volume [source: ILR, 2014] 39/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 6-1: Values of the inputs and outputs outlined in Figure 6-1 [source: ILR, cost model, 2014] Input_01 Demographic and geographic input data (Described in chapter 5.1.1) Input_02 1.08087 Input_03 1/3 Input_04 GSM/EDGE/UMTS/HSPA (see Table 5-14 Frequency spectrum) Input_05 Table 5-13: Input values into the bottom-up cost model for different services Input_06 Technical parameters are specified in the reference document4 Output_01 # of SIM cards 870 000 Output_02 # of SIM cards 290 000 Output_03 Unit# With termination Without termination BTS / NodeB / HSPA Sites # 571 551 Equipment # 1417 1371 TRX / Carrier # 2895 2767 BSC/RNC BSC site # 2 2 RNC site # 2 2 BSC hardware # 2 2 BSC software # 2 2 RNC hardware # 12 11 RNC software # 12 11 BSC ports # 880 804 RNC ports # 1146 1136 PCU BSC # 2 2 MSC call server # 2 2 MGW Media Gateway # 2 2 Type 1Ports # 4 4 76 48 2 2 E1 ports, for interconnection at MGW Core sites # Other equipment HLR # 2 2 LER / LSR # 2 2 SMSC # 2 2 SGSN # 2 2 GGSN # 2 2 IC interface # 2 2 Network management system # 1 1 IN # 2 2 Aggregation systems Aggregation systems ports # # 41 1619 41 1556 Radio links # 568 548 40/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) (130) Figure 6-1 shows that the number of total SIM cards in Luxembourg [870 000] is determined by considering the demographic data and the average penetration rate. Then the number of the HEO’s SIM cards [290 000] is determined by considering the total SIM cards in Luxembourg multiplied with the market share. (131) The information of the HEO’s SIM cards, the technical parameters, the frequencies the HEO has obtained, the technology [i.e. GSM/EDGE, UMTS/HSPA] to be applied as well as the average BH traffic per user are considered for dimensioning the HEO’s network. This needs to be done in order to define the type and quantity of elements required to satisfy the demand of the network busy hour and to guarantee a 100% coverage of Luxembourg. This is done once for a network with termination and for a network without termination, which is needed for the next step. An overview of the different input values and the generated outputs for these two steps are illustrated in Table 6-1. (132) In the next step, the total annual network costs are determined, as illustrated in the Figure 6-2. Figure 6-2: Determination of total annual network costs [source: ILR, 2014] 41/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Table 6-2 : Values corresponding to the inputs and outputs outlined in Figure 6-2 [source: ILR, cost model, 2014] Input_07 Table 8-1 Gross Replacement Costs Considered as Input in the model Input_08 Table 5-15: Expected price change, expected growth rate and economic lifetime of network assets [source: ILR, mobile cost model, 2014], Table 5-16: Expected price change, expected growth rate and economic lifetime of network support assets [source: ILR, mobile cost model, 2014] Table 5-17: OPEX mark-up [source: ILR, mobile cost model, 2014], Table 5-18: OPEX mark-up on direct investment [source: ILR, mobile cost model, 2014] 12% Input_09 Input_10 Unit With termination Without termination Output_03 # 8 924 8 506 Output_05 € 112 941 110 108 620 177 Output_06 €/year 16 049 088 15 416 538 Output_07 €/year 19 594 717 18 894 035 Output_08 €/year 35 643 806 34 310 574 (133) In this step, the annualised CAPEX of the network equipment [Output_07] is calculated considering the sum of unit gross replacement costs [Output_05] as well as economic parameters [Output_04]. (134) The total annual network costs [Output_08] are determined by the sum of the annualised CAPEX [Output_07] and the total operational expenditure [Output_06]. (135) According to the BU pure LRIC approach, ILR determines the total annual network costs with total volume of termination [Output_08A] and without voice termination [Output_08B]. Thus, the Output_08A and Output_08B are calculated with and without mobile call termination, as outlined in the Figure 6-2. (136) The Figure 6-3 outlines the step for the determination process of the BU pure LRIC cost for call termination [Output_09]. Figure 6-3: Determination of the pure LRIC MTR and the price cap [source: ILR, 2014] Table 6-3: Values of the inputs and outputs outlined in Figure 6-3 [source: ILR, cost model, 2014] 42/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Unit Output_08A Output_08B €/year 35 643 806 €/year 34 310 574 Input_11 Minutes 153 210 491 Output_09 €cts/min 0.8702 Input_12 €cts/min 0.1013 Output_10 €cts/min 0.9715 (137) The difference between Output_08A and Output_08B refers to the costs effectively incurred for mobile call termination. These costs are divided by annual termination minutes [Input_11] in order to get the incremental cost per minute [Output_09]. (138) The price cap [Output_10] determined by ILR consists of the BU pure LRIC cost for mobile call termination [Output_09] increased by additional wholesale commercial costs directly related to the provision of mobile call termination [Input_14] (see Chapter 5.3.8). 43/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 7 (139) Sensitivity analysis Pure LRIC costs for termination are calculated based on a hypothetical mobile operator with a market share of 33% as described in chapter 5. The network with optimum costs for this operator was determined in a bottom-up cost accounting model. The traffic demand for services included in the calculation was estimated with 100%. This equals 33% of total market demand from 2013. Due to the design of the bottom-up cost accounting model, different demand situations lead to different quantity structures for network dimensioning. This again leads to different values for pure LRIC costs of termination. Therefore calculations were not based on a pure static view of 100% demand for all services, but on various demand settings. The services referred to in chapter 5.1.4, are treated as follows: Demand for voice service has risen by 5% during the last year, but there might equally be a decrease in voice service demand in the future. Therefore range of demand was set with 90% to 110%. Demand for broadband services (best effort and mobile broadband) is expected to increase further in the future. Last year the increase was approximately 30%. So demand for broadband services was set with 100% to 135% (simultaneously for both data services). Table 7-1: change of demand per service service VOICE BEST_EFFORT SMS MMS MOBILE_BROADBAND (140) change of demand 90% to 110% 100% to 135% no change no change 100% to 135% In the course of this variation, demand for voice was changed in steps of 1% and demand for best effort and mobile broadband were changed jointly in steps of 1% in a combinatorial way. This results in 756 possible combinations. The following picture shows the sample space of these calculations. 44/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Figure 7-1: Pure LRIC range [source: ILR, mobile cost model, 2014] (141) The pure LRIC value for mobile termination varies between 0.0771 €cents/minute and 0.8702 €cents/minute according to the calculated scenarios. (142) As clarified in chapter 4, pure LRIC costs represent the absolute minimum for the determination of termination fees. In order to avoid that the operator has to recover the costs for the termination service from revenues of other services (i.e. cross-subsidising), termination fees have to be determined at the upper limit of the calculated range. If the arithmetic mean or any other statistical method (median, x%-quantile etc.) were used, the terminating operator would not even recover incremental costs with a probability of 50% (or 1 minus x% at an x%-quantile). To determine a termination fee lower than costs seems inadequate. 45/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 8 Appendix A 8.1 Population density, zones and cell hubs (143) For the radio access network modelling process, the national territory of Luxembourg is divided into zones. Within each of these zones, homogeneous conditions are assumed to apply so that a corresponding cell deployment can be performed. In addition, the resulting zones form the first level of aggregation regarding the whole network, over which the incoming and the outgoing traffic from the base stations are carried. The process of determining the zones is described in the reference document4. The aggregation network connects the base stations with the controllers. With regard to the physical network, it is useful to divide the aggregation network into two separate parts: (a) connections from the individual cell sites of a zone to a central location, cell hub, which represents the first concentration point of the mobile radio network, and (b) the connections of the cell hubs to a corresponding controller location (BSC in 2G and RNC in 3G). In addition, controller locations are considered as part of the aggregation network in the dimensioning process. Figure 8-1 : Zones and cell hubs of the hypothetical efficient mobile operator in Luxembourg Source: ILR (2014) 46/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Figure 8-2 : Population density 2014 Source: STATEC (2014) 47/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 8.2 Input Parameters Table 8-1 Gross Replacement Costs Considered as Input in the model BTS / NodeB Average investment for Macrocell Construction per GSM/EDGE or UMTS/HSPA site 102 000 € Hybrid site upgrade 46 200 € Average investment for Microcell Construction per GSM/EDGE or UMTS/HSPA site 70 733 € Hybrid site upgrade 35 000 € Average investment for Picocell Construction per GSM/EDGE or UMTS/HSPA site 59 330 € Hybrid site upgrade 30 500 € GSM/EDGE equipment: Average investment per unit Macrocell 1-sector BTS equipment 14 127 € Macrocell 2-sector BTS equipment 18 670 € Macrocell 3-sector BTS equipment 24 220 € Microcell 1-sector BTS equipment 12 357 € Microcell 2-sector BTS equipment 16 900 € Microcell 3-sector BTS equipment 22 450 € Picocell 1-sector BTS equipment 9 807 € Picocell 2-sector BTS equipment 14 350 € Picocell 3-sector BTS equipment 19 900 € Overlay of BTS equipment (2-Band) 17 650 € Overlay of BTS equipment (3-Band) 20 150 € UMTS/HSPA equipment: Average investment per unit Macrocell NodeB equipment without carriers 15 285 € Microcell NodeB equipment without carriers 15 465 € Picocell NodeB equipment without carriers 15 630 € Macrocell sector 6 117 € Microcell sector 5 117 € Picocell sector 4 451 € Macrocell NodeB equipment without carriers GSM/EDGE/UMTS/HSPA equipment: Macrocell 1-sector BTS equipment 23 530 € Macrocell 2-sector BTS equipment 27 164 € Macrocell 3-sector BTS equipment 31 604 € Microcell 1-sector BTS equipment 22 114 € Microcell 2-sector BTS equipment 25 748 € Microcell 3-sector BTS equipment 30 188 € Picocell 1-sector BTS equipment 20 074 € Picocell 2-sector BTS equipment 23 708 € 48/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Picocell 3-sector BTS equipment 28 148 € Microcell NodeB equipment without carriers GSM/EDGE/UMTS/HSPA equipment: Macrocell 1-sector BTS equipment 23 674 € Macrocell 2-sector BTS equipment 27 308 € Macrocell 3-sector BTS equipment 31 748 € Microcell 1-sector BTS equipment 22 258 € Microcell 2-sector BTS equipment 25 892 € Microcell 3-sector BTS equipment 30 332 € Picocell 1-sector BTS equipment 20 218 € Picocell 2-sector BTS equipment 23 852 € Picocell 3-sector BTS equipment 28 292 € Picocell NodeB equipment without carriers GSM/EDGE/UMTS/HSPA equipment: Macrocell 1-sector BTS equipment 23 806 € Macrocell 2-sector BTS equipment 27 440 € Macrocell 3-sector BTS equipment 31 880 € Microcell 1-sector BTS equipment 22 390 € Microcell 2-sector BTS equipment 26 024 € Microcell 3-sector BTS equipment 30 464 € Picocell 1-sector BTS equipment 20 350 € Picocell 2-sector BTS equipment 23 984 € Picocell 3-sector BTS equipment 28 424 € TRX and carrier: Average investment for TRX: 2 941 € Av. inv. for UMTS/HSPA carrier (driven by native UMTS services): 3 744 € License cost: License cost per year (800 MHz), in 2.5 MHz blocks: 93 750 € License cost per year (900 MHz), in 2.5 MHz blocks: 93 750 € License cost per year (1800 MHz), in 2.5 MHz blocks: 93 750 € License cost per year (2100 MHz), in 5 MHz blocks: 120 000 € License cost per year (2600 MHz), in 5 MHz blocks: 120 000 € Administrative licences cost per year: -€ BSC / RNC Average investment for site construction (BSC), in €: 95 350 € Average investment for site construction (RNC), in €: 95 350 € Hardware investment per BSC unit of type: 190 000 € Software investment per BSC unit of type: 170 000 € Hardware investment per RNC unit of type: 360 000 € Software investment per RNC unit of type: 680 000 € 49/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) BSC/RNC Ports 1 2 Average investment per E1 port installed at the BSC: 1 780 € 1 780 € Average investment per type 1 port installed at the RNC i (i=1 to 2): 950 € 1 800 € Average investment per type 2 port installed at the RNC I (i=1 to 2): 3 200 € 4 100 € Average investment per type 3 port installed at the RNC i (i=1 to 2): 3 300 € 4 100 € Average investment per type 4 port installed at the RNC i (i=1 to 2): 3 300 € 4 100 € PCU (only for BSC): 1 Average investment per packet control unit installed at BSC unit of type i: (i=1 to 2) 12 000 € MSC call server Average investment per core site: 327 267 € Average investment for MSC call server unit: 581 112 € MediaGateway Investment in material and installation per MediaGateway: 466 667 € Investment per type 1 port facing LER at Media Gateway: 648 € Investment per type 2 port facing LER at Media Gateway: 9 397 € Investment per E1 port, facing interconnection at Media Gateway: 763 € Investment per Ethernet based type 1 port, facing interconnection: 648 € Investment per Ethernet based type 2 port, facing interconnection: 9 397 € Label Edge Router Investment in material and installation per LER unit of type i: 211 667 € Average investment per type 1 port: 1 233 € Average investment per type 2 port: 7 333 € Average investment per type 3 port: 21 744 € Average investment per type 4 port: 41 960 € Label Switch Router Investment in material and installation per LSR unit of type i: 215 000 € Average investment per type 1 port: 1 233 € Average investment per type 2 port: 7 333 € Average investment per type 3 port: 21 744 € Average investment per type 4 port: 41 960 € HLR Investment in material and installation per 2G HLR functionality: 3 091 266 € Investment in material and installation per 3G HLR functionality: 3 091 266 € Investment in material and installation per Hybrid HLR functionality: 1 031 179 € Authentication centre included in the HLR EIR included in the HLR SMSC Investment in material and installation per SMSC unit: 295 316 € SGSN 50/54 2 13 500 € The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) Investment in material and installation, per SGSN unit: 731 157 € Average investment per type 1 port: 4 600 € Average investment per type 2 port: 5 100 € GGSN Investment in material and installation per GGSN unit: 474 443 € Average investment per type 1 port: 1 400 € Average investment per type 2 port: 6 800 € IC Interface Investment in interconnection interface: 575 000 € Network management system Investment in Network management system: 306 570 € Intelligent network (IN) Investment in Intelligent network: 1 988 870 € Aggregation systems 1 2 Investment in Aggregation systems of type i (i=1 to 2) 22 145 € 38 584 € Average investment per type 1 port: 230 € 280 € Average investment per type 2 port: 850 € 1 050 € Average investment per type 3 port: 5 550 € 6 300 € Average investment per type 4 port: 5 550 € 6 300 € Radio links (i=1 to 2) 1 2 Licence charge per radio link (one off investment): -€ Annual price per 25 MHz (frequency cost for radio links): 429 € Investment per radio link of type i (cell site - hub) 8 400 € 8 900 € Investment per radio link of type i (hub - controller) 8 550 € 9 800 € Investment per radio link of type i (controller - core) 6 100 € 6 100 € Investment per repeater for radio link of type i 9 050 € 10 050 € Leased lines (i=1 to 2) 1 2 Annual charge for the provision of a local link of type i 2 820 € 5 400 € Annual charge for the provision of a regional link of type i 4 008 € 5 400 € Annual charge for the provision of a long distance link of type i 5 412 € 5 400 € Hub - controller links: 1€ 2€ Annual charge for the provision of a local link of type i 2 820 € 5 400 € Annual charge for the provision of a regional link of type i 4 008 € 5 400 € Annual charge for the provision of a long distance link of type i 5 412 € 5 400 € Controller - core links (i=1 to 2) 1 2 Annual price per Km for link of type i (local) - € - € Annual charge for the provision of a local link of type i 2 820 € 5 400 € Annual price per Km for link of type i (regional) - € - € Annual charge for the provision of a regional link of type i 4 008 € 5 400 € Annual price per Km for link of type i (long distance) - € - € Annual charge for the provision of a long distance link of type i 5 412 € 5 400 € 51/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) 9 Appendix B – List of Abbreviations 2G 2nd Generation of mobile network technology, collective name for GSM, GPRS and EDGE 3G 3rd Generation of mobile network technology, collective name for UMTS und HSPA 3GPP 3rd Generation Partnership Project AMR Adaptive Multi Rate AN Access Network ATM Asynchronous Transfer Mode AuC Authentication Centre BN Backhaul Network bps bits per second, abbreviation of the standard bit rate unit bit/s BS Base Station BSC Base Station Controller BSS Base Station System BTS Base Transceiver Station CDR Call Data Record CN Core Network CS Circuit Switched DL Downlink E1 Primary Rate Interface with a gross data rate of 2048 kbit/s, subdivided into 32 channels of 64 kbit/s EDGE Enhanced Data Rates for GSM Evolution EIR Equipment Identity Register ETSI European Telecommunication Standards Institute G-MSC Gateway-MSC GB (also GByte) Gigabyte, equals 10^9 or 1.000.000.000 Byte GGSN Gateway GPRS Support Node GiB Gibibyte, equals 1024^3 or 1.073.741.824 Byte GMSK Gaussian Minimum Shift Keying GPRS General Packet Radio Service GSM Global System for Mobile Communication HLR Home Location Register 52/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) HSDPA High Speed Downlink Packet Access HSPA High Speed Packet Access, collective name for HSDPA and HSUPA HSUPA High Speed Uplink Packet Access IC Interconnection IM Instant Messaging IMS IP Multimedia Subsystem IN Intelligent Network IP Internet Protocol ISDN Integrated Services Digital Network Iu-Interface Interface in UMTS between RNC and core network kbit/s Kilobit per Second LER Label Edge Router LSR Label Switch Router MGW Media Gateway MMS Multimedia Messaging Service MSC Mobile Switching Center MSC-S MSC (Call) Server MTR Mobile Termination Rate NGN Next Generation Network NM Network Management Node B Base station in UMTS PCU Packet Control Unit PSTN Public Switched Telephone Network QoS Quality of Service RNC Radio Network Controller SGSN Serving GPRS Support Node SMS Short Message Service SMSC Short Message Service Center TDM Time Division Multiplex TRX Transceiver, from “transmitter” and “receiver” UL Uplink 53/54 The determination of the price cap for the provisioning of voice call termination on individual mobile networks (market 7/2007) UMTS Universal Mobile Telecommunications System UTRAN Universal Terrestrial Radio Access Network VLR Visitor Location Register WACC Weighted Average Cost of Capital WLAN Wireless Local Area Network 54/54