This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts

for publication in the ICC 2008 proceedings.

Infrastructure Sharing and Shared Operations for Mobile Network Operators

From a Deployment and Operations View
Dr. Thomas Frisanco, Member, IEEE, Dr. Paul Tafertshofer, Pierre Lurin
Nokia Siemens Networks Munich, Germany frisanco@ieee.org

Rachel Ang
Ernst & Young Bucharest, Romania rachel.ang@ro.ey.com

AbstractThe traditional and still prevailing mobile network operator (MNO) business model is based on the carriers full ownership of the physical network assets. However, rapid and complex technology migration, regulatory requirements, and increasing capital expenditures on one side and competitive environments, saturated markets, and pressure on margins on the other side advocate a new paradigm: the focus on critical success factors and key assets. Simultaneously, telecommunications equipment is commoditized. These trends are paving the way for the sharing of network infrastructure in the core and radio access networks among multiple operators. Challenges arise with regard to technical solutions to enable such business models in a multi-vendor landscape, but also in the context of the principal-agent-problem accompanying the re-allocation of assets and operational duties. This paper investigates the current technological, regulatory, and business landscape from the perspective of sharing network resources, and proposes several different approaches and technical solutions for network sharing. We introduce a model for estimating savings on capital and operating expenses, and present the results of our simulations for the various scenarios. Finally, we assess the benefits of Managed Services for the shared network case, a potentially highly attractive model to overcome some of the challenges posed by infrastructure sharing. KeywordsInfrastructure sharing; network sharing; RAN sharing; network management; managed services; outsourcing.

3.5G features like high-speed packet access (HSPA), and finally the introduction of 4th generation (4G) technologies, is becoming increasingly rapid and complex. Regulatory requirements might mandate the coverage of areas that is not attractive from a business perspective. At the same time, markets are highly competitive and increasingly saturated, so that operators have to be conscientious of costs in order to keep their profitability margins. Operators need to focus on activities that are really differentiating themselves in the marketplace, and the mere provisioning of coverage and capacity is gradually seen less of such a key success factor. Specialized providers may be more competitive for running a particular part of the business, e.g. the network, leading to a vertical disaggregating of the value chain, shown in fig. 1. Operations outsourcing and outtasking are typical examples, and so is the NetCo business model, where the operating party is also owner of the network assets. Considering now not just one, but multiple operators in a market, the sharing of certain, non-strategic platforms and assets, operating them together, or having them operated by a 3rd party, become viable options, and horizontal partnerships emerge, see fig. 2. In this context, infrastructure sharing is an important topic that will be investigated in detail in this paper, describing technical concepts as well as modeling the financial impacts for the operators business case. Infrastructure sharing is considered both in growth or new roll-out (e.g. new technology or additional coverage) and in consolidation (e.g. phase out of old technology, re-location) scenarios. At present, publication coverage of the topic is still limited. Although both academia and industry contribute relevant research, most papers focus on one or few single aspects of infrastructure sharing. E.g., [7] proposes a technical resource sharing framework tailored for the MNOMVNO-context with an emphasis on service level agreements (SLA). [8] is a technical analysis of options aimed specifically at shared rural 3G roll-outs without indepth consideration of operational issues or consolidation

I. INTRODUCTION Current incumbent mobile network operators are characterized by a high degree of vertical integration: The MNO plans the network architecture and topology, acquires (buys or leases) and develops (in terms of civil engineering) the sites needed for rolling out the network, oversees the network implementation by suppliers and subcontractors, operates and maintains the network, creates, markets and provides services to its end users, and manages the customer relationship. This traditional model is shown in fig. 1. However, technology migration, such as the introduction of 3rd generation (3G) wireless technologies on top of 2nd generation (2G) networks, the further implementation of

978-1-4244-2075-9/08/$25.00 2008 IEEE

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

scenarios. [9] puts regulatory issues at the center of a feasibility study of infrastructure sharing and in that respect addresses similar issues as [4], a legal opinion paper that delivers an analysis from a regulatory, legal, and public policy point of view, while neglecting any technological issues. In general, strategic issues are only covered as far as competitive (and anti-trust) aspects are concerned, and statements regarding economic impact are purely qualitative. [10] is a market and policy based discussion without detailed technical scenarios. Contributions from industry typically focus on the description of off-the shelf technical solutions, but fail to study the operators processes and to quantify economic implications (e.g. [1]). Also, given the vendor perspective, most of the attention goes to fixed assets rather than operations considerations compare [3], which addresses investment, coverage and time-to-market issues for new rollouts. [2] proposes a decision matrix and argument checklists for 3G greenfield scenarios for the different morphologies (urban-suburban-rural) and phases (coverage- vs. capacitydriven), but discusses only a subset of the available sharing options. Typically, greenfield scenarios and new roll-outs are emphasized, while there is little knowledge about network consolidation.

II. THE OPERATOR LIFECYCLE Infrastructure sharing can be initiated at various stages of the operators network roll-out and technology refresh cycles. Figs. 3 to 5 show the three principal scenarios in a two-operator environment: greenfield: both operators build their networks from the scratch (e.g. two completely new carriers, or 2G incumbents rolling out 3G networks); buy-in: one operator already has footprint and acts as a host to the newcomer; for the newcomer, this scenario is equivalent to a greenfield situation; consolidation: the operators consolidate and merge their exiting networks according to their chosen sharing strategy.

Customer Interaction

Carrier Functions

Business Support Products & Services NW Operate & Maintain NW Build

Traditional Vertical Telco

SalesCo

ServCo

NetCo

Figure 3. Infrastructure sharing in greenfield operator lifecycle Figure 1. Disaggregating of the telecoms value chain

Horizontal Partnerships

SalesCo 1 ServCo 1 NetCo 1

SalesCo 2 ServCo 2 NetCo 2

Vertical Partnerships

Managed Service Provider

Figure 2. New telecoms business ecosystem

Figure 4. Infrastructure sharing in buy-in operator lifecycle

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

Alternatively, depth of sharing is used as a synonym for the technology domain, and reach of sharing represents the geographic domain. These two dimensions recur throughout the literature. E.g., [11] uses them, plus the criteria of which network generations are shared, and how many operators are involved (fig. 7). Since we performed our quantitative analysis for a two-operator environment and assumed both 2G and 3G technologies to be included within the scope, we consider these two dimensions to be fixed and not subject of our study. The three dimensions are interrelated, since the choice of a certain option regarding one dimension will limit the degrees of freedom for reasonable choices for the other dimensions. The decision on business model and geographic model largely depend on the involved operators relative conditions, installed bases, and future roll-out plans. Incumbents with similar roll-out cycle would probably prefer mutual service provision agreements, or establish a joint venture to run the shared network. If both incumbent and new entrant (greenfield) operators are involved, unilateral service provisioning would be an appropriate choice. If operators want to focus on service development and sales, the delegation of the network provisioning to a 3rd party network provider (which holds the assets and operates them) would be an interesting alternative. Operations outsourcing and outtasking are options that can deliver cost reductions to operators in any constellation: standalone, unilateral and mutual service provisioning agreements, and joint ventures. For the collaboration schemes mentioned, however, outsourcing becomes especially interesting, because on one side the outsourcing provider can achieve higher synergies out of the alignment of the services for the combined scope, and on the other side this external partner can facilitate the sharing process, provide neutral governance models (avoiding the principalagent-problem), and guarantee the confidentiality of each operators data, such as customer-specific traffic data and service-specific configuration settings.

Figure 5. Infrastructure sharing in consolidation operator lifecycle

In cases with prevailing greenfield portions, the main rationale is to allow a larger network footprint than standalone, or allow roll-out of a network at all, especially in situations where the actual utilization and revenue-generation of certain assets is low. These scenarios generally offer higher, and shorter-term, investment and operations savings potentials, because less network elements are needed from the start, aligned planning and implementation ex-ante provides higher efficiency and lower complexity, and operations of the shared network imply less effort compared to the sum of the parts. However, when consolidation is concerned, although the number of network elements decreases, initial cash outlay occurs for planning and implementing the relocation/colocation of sites, especially installation & commissioning costs; furthermore, some cost reductions related to sites and staffing will not we exploited immediately due to the longerterm nature of some contracts. From an accounting point of view, the decommisioning of equipment usually means unanticipated write-offs, or asset sales below book value, if such assets are handed down to the secondary market. Recurring operational savings later have to compensate the initial one-time expenses, and the transaction will only be economically attractive from a mid-term perspective (an aspect also highlighted in [11]). III. TAXONOMY OF INFRASTRUCTURE SHARING

Business Model
Unilateral Service provisioning Mutual Service Provisioning Joint Venture 3rd Party Network Provider

For the purposes of our study, we characterize infrastructure sharing transactions by three dimensions, see fig. 6: the business model, describing the parties involved, and their contractual relationship; the geographic model, describing operators physical footprint; each

Technology Model
Site Sharing Access Transmission Sharing Active RAN Sharing 3G Multi-Operator Core Network Roaming Based Sharing

Geographic Model
Full Split Common Shared Region Unilateral Shared Region Full Sharing

the technology model, describing the technical solution used.

Figure 6. The three dimensions of infrastructure sharing

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

Depth of sharing

Full sharing Active Nationwide infrastructure Passive Rural and infrastructure selected urban Sites Rural

Reach of sharing

3G 2G Full

2 3 4

Extent of sharing

Number of sharing parties

Figure 7. Alternative taxonomy of infrastructure sharing (Oliver Wyman)

D. Common Shared Region Operators of similar scale will establish a common shared region, if they both want to be physically present in an area (i.e. expanding into that area in the growth case, refraining from moving out from that area in the consolidation case), but still want to share infrastructure, or at least sites, in order to reduce capital and operating expenditures. Since no roaming is required and new technical features have recently been added by the infrastructure vendors, the subscriber will not necessarily notice even an infrastructure sharing (as is the case with roaming), because solutions are available that permit both operators to use their individual network identifiers. E. Full Sharing With full sharing, operators combine either all sites, or their entire radio or even core networks (depending on the technical solution). In the extreme case , they retain separate only a part of the core network, especially the portion attached to subscriber ownership, such as home location register (HLR), authentication and billing system. A geographical full sharing implementation is of course always more efficient than a partially sharing implementation of the same technical approach. For a roaming-based solution, the only difference between full split and full sharing is the regional selection criterion for the first (i.e. one operator rolls out or concentrates in one area), while the second means a case-by-case decision on the roll-out or phase-out without regional decision criterion. In a growth environment, full sharing mandates an optimal joint network planning ex-ante; in a consolidation environment, operating costs are reduced by concentrating sites (relocation costs apply, though), and by retiring equipment that is no longer needed under a capacity point of view.

IV. GEOGRAPHIC DIMENSION OF NETWORK SHARING All approaches described in this paper can be applied to an unlimited number of operators doing business in the same geography, however, for the sake of simplicity, we will limit our illustrations to a two-operator case. A. Base case Standalone As depicted in fig. 8, in the standalone case each operator provides full service coverage for the complete geography (e.g. the whole country) by operating its own network. Departing from this structure, various options are available, shown in fig. 9.
geography to be covered

Service coverage for Operator A customers Service coverage for Operator B customers

Operator A network Operator B network

Figure 8. Non-sharing base case

Full Split

Op A

Op B

B. Full Split In the full split case, the operators cover disjoint, complementary areas. This approach is interesting for operators of comparable strength that want to enter a mutual service (roaming) agreement. In a growth scenario, it allows extended coverage or introduction of new technology at lowest combined cost; in a consolidation scenario, it requires discretionary phase-out coordinated between the operators, but no relocation of equipment. C. Unilateral Shared Region The unilateral sharing is a model particularly aimed at combining incumbents and new entrants roll-out requirements, because it allows the operator holding a large installed base to leverage it to generate additional volume and revenues, while relieving the greenfield operator from the burden of investing in an own full-coverage infrastructure that may be incommensurate compared to its small subscriber number. Again, roaming would be the corresponding technical solution.

Unilateral Shared Region

Op A Op B Op A Op B

Op A

Common Shared Region

Op A+B

Full Sharing

Op A + B

Figure 9. Geographic sharing options

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

V. TECHNICAL APPROACHES The technical approaches, illustrated in fig. 11 and discussed in more detail later, can largely be allocated to three clusters: passive radio access network (RAN) sharing; active RAN sharing; roaming-based sharing.

Level of Control No Sharing

Common Region

Passive RAN Full Sharing Sharing

Common Region

The technical solution must closely match with the geographic dimension, as we already noticed in the section above. Tab. I shows an argument checklist that summarizes strategic issues that should be considered. While infrastructure sharing is by definition the most cost-efficient design principle for any new roll-out and the recurring-costoptimizing approach for consolidation (however, some onetime costs arise from potential relocation), its greatest benefits can be experienced in the coverage-driven domain, i.e. areas where the number of network elements is driven by coverage needs (rural areas), as opposed to the capacitydriven domain (hotspots, urban and suburban morphology), where the number of users dictates the network dimensioning. Apart from the addressable market resident in the rural areas, coverage requirements might be set by the regulator, or demanded by highly mobile customers originating in covered areas, but traveling to or through remote areas. At the same time, the cost savings potentials from infrastructure sharing are earned through sacrificing some of the control that the standalone operator has over its network, as depicted in fig. 10. This is why, considering both the appeal of sharing to the operators, and their strategic interests, the stronger forms of sharing are usually recommended for coverage-driven roll-outs in rural areas that have limited business potential, and where differentiation (which requires autonomy) is less important. Also other than coverage-requirement-related regulatory policies need to be taken into consideration, especially regarding frequency pooling ([4]): If permitted, operators can share even the same frequency carrier on one base station; if forbidden, multiple carrier units need to be deployed. Tab. II gives some examples from EU economies.
TABLE I. TECHNICAL APPROACH SELECTION SCOREBOARD

Active RAN Full Sharing Sharing

Unilateral Shared Region Full Sharing Full Split

Roaming-Based Sharing

Potential Cost Savings

Figure 10. Trade-off between network control and potential cost savings

TABLE II.

EXAMPLES OF REGULATORY FRAMEWORK IN EU

COUNTRIES
Passive RAN Sharing Active RAN Sharing admissible admissible admissible admissible admissible Geographical Split no comment admissible admissible no comment partly admissible Frequency Pooling / Trading not admissible not admissible not admissible together with core network sharing not admissible

Austria Germany UK Sweden Switzerland

admissible admissible admissible admissible admissible

Degree of sharing Degree of Network Control

Pure Site Sharing
Service Platfor ms HLR Servi ce Platforms HLR Ser vice Platforms HLR MSC/ SGSN MSC/ SGSN BSC/ RNC MSC/ SGSN

Site & Access Sharing

Service Platforms HLR

MORAN
Service Platforms HLR HLR

MOCN
Ser vice Platforms HLR HLR

Shared RAN w/ Gateway Core

Service Platforms HLR HLR

Full Network Sharing

Service Platfor ms HLR

Service Platforms

Ser vice Pl atforms

Service Platforms HLR

MSC/ SGSN

MSC/ SGSN

MSC/ SGSN MSC/ SGSN

MSC/ SGSN

MSC/ SGSN MSC/ SGSN MSC/VLR/ SGSN

BSC/ RNC

BSC/ RNC

BSC/ RNC

GMSC/VLR /SGSN BSC/ RNC BSC/ RNC BSC/ RNC

BSC/ RNC

BTS/ NodeB BTS/ NodeB BTS/ NodeB

BTS/ NodeB

BTS/ NodeB

BTS/ NodeB

BTS/ NodeB

BTS/ NodeB

Dedicated frequencies

Shared frequency

Technical approach for infrastructure sharing Passive RAN Areas with high business volume Heavy competition between operators Service and performance differentiation needed Full control of own network assets Base method for network consolidation Active RAN Areas with moderate business potential Competition between operators Partial control of network assets Roaming-based Areas with low business potential Possibly regulatory coverage requirements

Passive RAN Sharing

Active RAN Sharing

Roaming Based Sharing

Figure 11. Technical approaches for infrastructure sharing

A. Passive RAN Sharing The Traditional Concepts: Site sharing or co-location Exploiting opportunities for sharing the radio sites, i.e. the locations where the RAN components base transceiver stations (BTS) for 2G and Node B for 3G networks are installed, has become popular since around year 2000. Operators can directly enter an agreement to share sites, but more commonly there is an enabling 3rd party involved; in fact, providing towers to telecommunications operators has

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

become a standalone business in many markets, run by socalled tower companies such as: American Tower, Crown Castle International, Global Signal, SBA Communications, Spectra Site, or AAT Communications, just to name a few. While these established a footprint in rather mature markets like the USA or the UK, the phenomenon can also be observed in emerging markets, like India. Regulators encourage the sharing of sites, because it means that in total less sites will be needed, which is desirable considering environmental and aesthetic concerns that are present in the population. Traditional site sharing, or co-location, usually comprises the shared use of the site itself, the mast, shelters and cabinets, the power supply including backup batteries, air conditioning, and diesel generators, if present; depending on the frequency spectra used, antennas may also be shared. Both capital (CAPEX) and operating (OPEX) expenditures are reduced significantly by sharing these among multiple tenants. Site acquisition costs and expenses for civil works (erecting masts etc.) account for up to 40% of the initial investment into fixed assets (CAPEX). Besides the costs of site acquisition, the process of acquiring (buying or leasing) sites and getting all necessary permits and clearance can be very lengthy and time-consuming. Within recurring costs, site-related costs (site maintenance, site rental) typically make up 5-20% of network OPEX, with the bigger number applying for sites that are leased, not owned. The sharing of some electrical equipment, such as air conditioning, further makes power consumptions an addressable cost item, which represents roughly 3% of network OPEX. B. Passive RAN Sharing Access Transmission Sharing In addition to what was said for site sharing, access transmission sharing also includes sharing the transmission network between BTS and base station controller (BSC) for 2G and between Node B and radio network controller (RNC) for 3G networks. The transmission network can be implemented as leased lines (LL) or microwave (MW) links. Lines leasing will typically account for 10% of network OPEX, MW frequency fees for 3%, which become addressable through access RAN transmission sharing. Further savings result from less field services and network operation center (NOC) efforts as well as less spares and logistics and technical assistance center level 2 (TAC2) costs due to the lower number of network elements (LL and MW links). C. Active RAN Sharing Multi-Operator RAN (MORAN) Additional CAPEX and OPEX savings can be realized by also sharing the active RAN infrastructure, i.e. BTS and BSC in 2G or Node B and RNC in 3G networks. MORAN (see [6]) is a technical solution where operators maintain a maximum level of independent control over their traffic quality and capacity, e.g. each operator maintains its own set of cell-level parameters, only site-level parameters are shared. In principle, multiple virtual radio access network instances are implemented by splitting the BTS, BSC, Node

B, and RNC into logically independent units being realized by a single physical instance. These virtual radio access networks are then connected to the respective operator core network mobile switching center (MSC) and serving GPRS support node (SGSN) for circuit and packet switched traffic, respectively. Operators continue to use the dedicated frequency ranges that they were awarded by the licensing bodies, and broadcast their own individual network identifiers such that they maintain full independence in their roaming agreements and the sharing is not visible to their subscribers. With MORAN, all previously mentioned cost items are again addressable, but larger savings are obtained in various categories, like electrical power, and maintenance, because again the number of elements is reduced. Additionally, network planning and optimization can be shared. D. Active RAN Sharing 3G Multi-Operator Core Network (3GPP MOCN) MOCN is another active RAN sharing solution which has been defined in 3GPP Rel. 6 for 3G networks (see [5]), where Node B and RNC are shared among multiple operators and frequencies are pooled. Addressable cost items are identical to MORAN, but while frequency pooling results in further marginal savings of equipment investment and equipment-related costs operations (FS and NOC), spares and logistics, and electricity due to a lower number of carrier units in extremely low-traffic areas, operators have to give up their independent control on traffic quality and capacity to a large extent. Subscribers using pre 3GPP Rel. 6 mobile terminals may realize that the network is shared. Under regulatory aspects, 3GPP MOCNs feature of frequency pooling may exclude the MOCN solution from being used in certain markets. E. Roaming-Based Sharing Shared RAN with Gateway Core, and Full Network Sharing From the beginning of second-generation (2G), digital mobile telephony, roaming has always been employed as a means of virtually extending the geographic coverage of an operator by allowing its subscribers to use an other operators network. International roaming is the natural solution to serve ones customers abroad, where the operator has no license and no business. Roaming is also used on a domestic basis, as national roaming, typically to grant to a new entrant or greenfield operator nationwide coverage right from the start, when the operator rolls out its network initially in the urban and suburban areas and is not yet present in the rural areas. Incumbent operators are often forced into such a temporary national roaming agreement with the new entrant by the regulator. But the traditional MNO business model always foresees that every operator achieves its own full coverage after a certain introductory phase. Roaming-based options in the context of network sharing, instead, mean that one operator relies on another

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

operators coverage for a certain, defined footprint on a permanent basis. As already mentioned, such dependence can be either unilateral or mutual, regionally split or for the network as a whole. If operators decide to retain dedicated independent core networks or only share the radio access network in a certain region, the shared RAN with gateway core solution can be deployed. Similar from a point of view of addressable cost items, compared to the active RAN sharing solutions outlined above, this approach, however, does not require specific features in the RAN equipment, as the sharing is fully implemented by roaming features that need to be implemented in the core network. The shared RAN is connected to the core networks of the sharing partners via a so-called gateway core consisting of MSC, SGSN, and visitor location register (VLR). In this solution, either frequencies are pooled, or only the frequency spectrum of one of the participating operators is used, such that there is no independent control of the traffic quality and capacity for the operators. If only one spectrum is used, capacity is substantially reduced; the pooling of frequencies is again subject to restrictive regulatory policies. Unless mobile terminals are equipped with specially configured SIM cards, the network sharing is visible to the subscribers, because only a single common network identifier is broadcasted in the shared region. In the full sharing case, the operators only retain that portion of the core network separate which also an MVNO (mobile virtual network operator) would own, i.e. home location register (HLR), authentication and billing system. In this case, all network OPEX are shared. VI. NETWORK OPERATIONS

the results are given in a relative format (cost savings in percent); the results reflect a static, medium term state based on the set time period to be analyzed; for a dynamic analysis, an interpolation is done between beginning and ending value; OPEX savings are sustainable and recurring (repeating every year); CAPEX savings are cumulative.

In order to be generic, the model states a status-quo basis for current network OPEX and currently deployed network CAPEX (both including breakdown in the various categories that are affected by network sharing) on a common-value basis, i.e. in units that add up to 100. Both OPEX and CAPEX are split into a 2G and a 3G portion. Next, the network growth is modeled by measuring the planned new roll-out against the existing base and calculating incremental OPEX and CAPEX. Instead of a simple linear extrapolation, we also considered potential changes in the share of leased vs. owned sites and the microwave share of the transmission network which would make the future roll-out not directly comparable to the legacy base, and we modified the incremental values accordingly. By summing existing and incremental values, we get the future predicted values for OPEX and CAPEX. This calculation is first done without considering network sharing, to build a base case serving as benchmark for the other cases. Next, the simulation is repeated for all possible scenarios described above. Key drivers are the degrees of network overlap between the operators roll-outs and the degrees of desired sharing (both for 2G and 3G, for growth and consolidation areas) and the share of existing leased lines and MW links bearing excess spare (i.e. they are shareable). Additionally, modified simulations are down by combining all previous scenarios with operations outsourcing, either as total operations or field services only. B. Assumptions We modeled a two-incumbent-operator case and assumed that for each sharing strategy scenario, the same scenario will be applied to both 2G and 3G networks. We further assumed an operator with currently 80% 2G and 20% 3G share, where 5% of the 2G and 80% of the 3G network are still to be rolled out. 85% of the 2G growth area, 90% of the 3G growth area, and 95% of the 2G and 3G existing coverage are overlapping. There is an 85% share of leased sites in new roll-out vs. 65% in legacy base, and a 70% MW share in new roll-out vs. 50% in legacy base; 50% of existing leased lines and 40% of existing MW links have spare capacity. Regarding network deployment, we assumed that fullturnkey installation & commissioning services are provided by the equipment supplier.

The sharing of active equipment poses additional challenges to the participating operators, because it involves a far higher, and mission-critical degree of operations and maintenance aimed at shared resources compared to passive sharing. When only sites are shared, all tenants could in principle employ their own field service force, even if awarding outsourcing contracts to one common provider of managed services would lead to significant further cost cuts. But the maintenance of shared active equipment can only be reasonably done by one party in charge, which then would owe fiduciary duties to the partner. The potential principalagent problem can only be fully excluded if the operating body equally reports to all partners, meaning the carving-out of the relevant resources, e.g. by setting up a joint venture. In this case, outsourcing becomes a very attractive alternative. We include an evaluation of the financial impacts of outsourcing in the later section. VII. ECONOMIC CONSIDERATIONS A. Financial Simulation Model We employed a spreadsheet-based financial model to evaluate the impacts of various network sharing options on the operators business case. The characteristics of this model are:

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings.

C. Simulation Results Tab. III shows the output from our financial model. We could prove our assumptions regarding the increasing savings potential when moving from a passive sharing scheme towards active sharing and roaming solutions. Already in the two-operator scenario, the financial impact is large: Savings in the range of 20-40% of both OPEX and CAPEX could be obtained if operators were willing to move beyond classical site sharing. We could prove our assumptions regarding the increasing savings potential when moving from a passive sharing scheme towards active sharing and roaming solutions. The improvement in economic impact from MORAN to 3GPP MOCN and shared RAN with gateway core schemes appears to be rather small, both for OPEX and CAPEX, while many degrees of freedom have to be given up by the operators. Where strategically permissible, however, a full sharing approach could yield significantly higher benefits. The consideration of outsourcing, especially with total operations scope, increases the applicable OPEX savings in the respective scenario substantially. This is due to the fact that only for site-related costs, such as site rental, fees for leased lines and microwave links, and site infrastructure management, immediate savings out of sharing can be exploited by the operators on their own, without changes in processes; the operations-related items field services, maintenance, technical support, spares repair and logistics, network operations center, ongoing planning and optimization can be handled much more efficiently by one single party, such as the outsourcing service provider; if done by both operators, double teams will be present, and efforts could eventually even increase due to coordination costs. For similar reason, when considering the CAPEX portion, equipment should be supplied along with fullturnkey services by one supplier (as assumed in our calculations) in order to exploit synergies not only regarding the equipment itself, but also regarding planning and installation & commissioning.
TABLE III. Infrastructure sharing approach Standalone Site sharing / co-location Access transmission sharing MORAN MORAN & 3GPP MOCN Shared RAN w/ gateway core Full network sharing FINANCIAL MODEL OUTPUT Network OPEX No FS TO outoutoutsource source source n.a. -6% -11% -8% -15% -23% -24% -25% -35% -14% -20% -27% -28% -29% -38% -18% -24% -32% -33% -34% -42% New network CAPEX n.a. -18% -20% -33% -34% -34% -41%

VIII. CONCLUSIONS We can state that infrastructure sharing for telecommunications operators is one of the emerging hot topics on technologists as well as senior managements agenda. Infrastructure sharing proves particularly beneficial in greenfield and strong growth scenarios, situations that are typically present in mature markets where technology migration is an issue, or in emering/developing markets. Operators interested in infrastructure sharing should examine the various technical solutions that are readily available, taking into account the strategic, financial, technical, and regulatory frame conditions; especially, diverse approaches should be considered for new and old technology networks, and for the various geographic domains (urban, suburban, rural). In the roll-out phase, the provisioning of full-turnkey installation & commissioning (I&C) services generates additional benefits to the operator Regarding ongoing costs, maximum synergies can only be achieved if operation-related items are addressed in addition to the site-related items, either through reorganising processes and workforce, or through outsourcing/out-tasking. Due to increased complexity and alignment requirements and due to economies of scale, outsourcing should be considered as a serious option. An outsourcing service provider, or even network provider, can provide operator- neutral governance models, guarantee the confidentiality of operators data, and avoid the principal-agent problem.

REFERENCES
T. Leibner, Network and Infrastructure Sharing in 2G networks, Siemens, 2004. [2] White-paper: 3G Infrastructure Sharing. Siemens, 2001. [3] Shared Networks: An operator alternative to reduce initial investments, increase coverage and reduce Time To Market for WCDMA by sharing infrastructure. Stockholm: Ericsson, 2001. [4] E. Lichtenberger, E.-O. Ruhle, M. Uhlirs, Infrastructure-Sharing bei Mobilfunknetzen der 3. Generation (UMTS), Journal fr Rechtspolitik, Heft 2. Vienna: Springer-Verlag, 2003, pp. 79-86. [5] 3GPP TS 23.251 V2.0.0 (2004-06): Network Sharing; Architecture and Functional Description, Rel. 6. 3rd Generation Partnership Project, 2004. [6] Press release (23.05.2001): Nokia launches Multi-Operator Radio Access Network for controlled 3G network sharing. Nokia, 2001. [7] J. Hultell, K. Johansson, J. Markendahl, Business models and resource management for shared wireless networks, IEEE Vehicular Technology Conference, 2004. [8] J. A. Village, K. P. Worrall, D. I. Crawford, 3G shared infrastructure, Third Conference on 3G Mobile Communication Technologies, 2002, pp. 10-16. [9] C. Beckman, G. Smith, Shared networks: making wireless communication affordable, IEEE Wireless Communications, Vol. 12, Apr. 2005, pp. 78-85. [10] H. de Vlaam, C. F. Maitland, Competitive Mobile Access in Europe: Comparing Market and Policy Perspectives, Communications & Strategies, No. 50, pp. 69-94. [11] White-paper, The Rise of Network Sharing Risks and Rewards for Telecom Operatoirs, Oliver Wyman, 2007. [1]