Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Published by Analysys Mason Limited St Giles Court 24 Castle Street Cambridge CB3 0AJ UK Tel: +44 (0)1223 460600 Fax: +44 (0)1223 452800 Email: research@analysysmason.com www.analysysmason.com/research Registered in England No. 5177472 Analysys Mason Limited 2009 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher. ISBN 1 905495 86 2
Contents
About Analysys Mason 0 1
1.1 1.2 1.3
vi vii 1
1 2 4
2
2.1 2.2 2.3 2.4
5
5 8 11 14
3
3.1 3.2 3.3 3.4
18
18 19 20 21
4
4.1 4.2 4.3
22
22 23 24
5
5.1 5.2
25
25 26
Contents
iv
5.3 5.4
26 27 30 31 32
Key to acronyms Research from Analysys Mason Consulting from Analysys Mason
List of Figures and Tables Figure 0.1: Table 1.1: Table 1.2: Figure 1.1: Figure 1.2: Figure 2.1: Figure 2.2: Figure 2.3: Table 2.1: Table 2.2: Figure 2.4: Figure 3.1: Figure 4.1: Figure 5.1: Figure 5.2: Cornerstones of an operators strategy for deploying a data-optimised network vii Factors that are driving the increase in wireless network data traffic 1 Key forecasts for mobile traffic in developed and developing regions 3 Global wireless network traffic growth, 20082015 3 The cost of data traffic using legacy networks, and revenue 4 Typical mobile network costs per subscriber per cellular sector 5 LTE can deliver profit by significantly reducing opex and capex 7 All technology roadmaps can lead to LTE 9 SWOT analysis of migration from HSPA directly to LTE 12 SWOT analysis of migration from HSPA to HSPA+ (Release 7) 13 Network traffic generated by an example mobile service mix, split between indoor and outdoor usage, 20082012 15 Timeline for 25002690MHz spectrum auctions 18 Expected LTE device availability 22 LTE by geography/operator 28 Deployment strategy guidelines 29
Authors Helen Karapandi (Analyst) joined Analysys Mason in April 2007. Her research interests include wireless access, mobile broadband and mobile payment. Her most recent work includes several reports on the subjects of mobile payment, near field communications, WiMAX access technology and networks, radio access network evolution, mobile broadband and service bundling. Prior to joining Analysys Mason, Helen was Business Development Manager for an international asset management company, where she was responsible for Europe, the Middle East and Africa (EMEA). She has also held positions with Cambridge Investment Research (an independent, Cambridge-based strategy consultancy) and eBay AG. Helen has an MA in Modern Languages from the University of Cambridge. Terry Norman leads Analysys Masons Wireless Networks research and analysis programme, which focuses on the development of wireless access technologies and their commercial applications. Terry has more than 30 years experience in the radio communications industry in radio planning, access network design, operations and latterly ten years consulting. He has authored numerous reports on the
Contents
subject of wireless access technologies. He has been quoted in the broadsheet press and regularly contributes to media analysis. Terry holds a doctorate in radio propagation, and is a Chartered Physicist and a member of the Institute of Physics and the Institute of Technology.
Acknowledgements The authors would like to thank all those who assisted in the preparation of this report: Anthony Berkeley (Director, LTE Strategy, Alcatel-Lucent), Dan Warren (Director of Technology, GSMA), Mats Blumenberg (LTE Product Line Management, Huawei), Stephane Daeuble (Senior Manager, Global Marketing, Motorola), Manash Goswami (Director of Business Development, Platforms and Mobile Devices, Motorola), Sami Jockinen (Senior Manager, Wireless Access Technology Marketing, Nokia), Peter Carson (Senior Director, Product Management, Qualcomm), Herv Dubreil (Strategic Advisor, CTO Office, Orange Group), Frdric Gastaldo (Head of Development, IT Services, Swisscom), Rico Schwendener (Head, Network Innovation Centre, Swisscom) and Jaime Lluch (Radio Access Technology Manager, Telefnica S.A.). The authors would also like to thank Analysys Mason colleagues Matt Hatton and Andrew Parkin-White for their input, and Sarah Peake and Claire Varley for editorial assistance.
Disclaimer Figures and projections contained in this report are based on publicly available information only and are produced by the Research Division of Analysys Mason Limited independently of any client-specific work within Analysys Mason Limited. The opinions expressed are those of the stated authors only. Analysys Mason Limited recognises that many terms appearing in this report are proprietary; all such trademarks are acknowledged and every effort has been made to indicate them by the normal UK publishing practice of capitalisation. However, the presence of a term, in whatever form, does not affect its legal status as a trademark. Analysys Mason Limited maintains that all reasonable care and skill have been used in the compilation of this publication. However, Analysys Mason Limited shall not be under any liability for loss or damage (including consequential loss) whatsoever or howsoever arising as a result of the use of this publication by the customer, his servants, agents or any third party.
Contents
Industry Strategy Global Growth Markets Fixed Broadband Mobile Broadband Voice Consumer Content and Applications Telecoms Software Fixed Networks Wireless Networks Core Data and Forecasts
0 Summary
LTE is the only technology that can meet the requirements for the anticipated dramatic growth in wireless data traffic
Wireless data traffic is forecast to increase tenfold in developed countries between 2008 and 2015, and sevenfold in emerging markets over the same period. It is clear that mobile networks will reach their limits in terms of capacity and throughput within a few years. At the same time, revenue per megabyte continues to fall, driven down by increasing competition and the introduction of flat-rate pricing. Therefore, the decision to deploy a data-optimised radio network appears inevitable for most operators. There is, however, a complex interplay of choices, which operators will need to understand and sequence if they are to achieve an optimal position to deal with data demand and pressures on their revenue. In this report, we evaluate the four main factors (besides the individual operators business objectives) that influence mobile operators strategies, timing and costs of deploying a dataoptimised RAN architecture, as shown in Figure 0.1.
Figure 0.1: Cornerstones of an operators strategy for deploying a data-optimised network [Source: Analysys Mason, 2009]
Traffic growth
Technology mix
Device availability
Spectrum availability
A summary of our key findings in these four areas is as follows. Traffic growth. Total wireless network traffic generated from voice and data services will increase dramatically by 2015. Traffic per cellular user per month in developed regions will
0: Summary
viii
rise from an average of 56MB in 2008 to 455MB. Over the same period, the combination of a substantial rise in wireless network traffic per user and a relatively modest increase in ARPU will result in a significant decline in revenue per megabyte: this will fall from USD0.86 to USD0.12 in developed regions and USD0.57 to USD0.15 in emerging markets over the same period. The costs of the existing network will soon begin to erode profit margins and will eventually outstrip revenue. As a consequence, the operator must:
deploy a data-optimised network architecture reduce network carriage cost per megabyte ensure a level of indoor service that is comparable to alternatives, like Wi-Fi.
Device availability. The introduction of new and improved wireless devices (most notably USB modems and datacards) enabling a greater range of services has been one of the major drivers of mobile data usage and will be a key enabler of LTE. Major handset vendors operate under the assumption that LTE access infrastructure will be available from late 2009, with commercial services from 2010. Devices will be introduced in phases during 20092010. Vendors expect it to take 12 years thereafter for devices to become widely available. Technology mix. Compared with standard UMTS, LTE offers significantly increased capacity for a given bandwidth (spectral efficiency), and other important capex and opex savings. Therefore, operators will deploy LTE at some point to meet the rising demand for data and falling revenue (per megabyte). However, as operators evolve their networks towards LTE, they may wish to retain some access technologies and replace others, resulting in a technology mix. This mix depends upon the operators legacy infrastructure and business objectives:
LTE will not be deployed as a standalone solution. Instead, it will be used in conjunction with GSM, UMTS and a separate indoor solution. It is likely to be deployed as a capacity overlay on existing UMTS and GSM services, and a dedicated indoor coverage solution is needed in order to meet the demand for data from the home or office. For example, Telefnica plans to deploy femtocells, while Orange has opted for a UMA/Wi-Fi solution. Many operators will retain GSM900/1800 to support circuit-switched voice and legacy services, and to protect incoming revenue from roaming services for several years. The earliest GSM switch-off is likely to occur in 2015, the majority of GSM networks will remain until 2020. Operators are currently divided on HSPA+. Some operators, like T-Mobile, (typically those that have already deployed HSDPA7.2/HSUPA across a large proportion of their networks) will carry out software upgrades to HSPA and HSPA+ without deploying MIMO. Others, like Orange, plan to deploy HSPA+ as soon as the technology becomes available in order to provide capacity relief in high-density areas.
0: Summary
ix
WiMAX is expected to have limited impact on LTE deployments, although some global operators with footprints across both developed and emerging markets may choose to deploy WiMAX opportunistically in areas where spectrum is constrained.
Operators do still have scope to maximise their existing network assets, but this may be a short-lived opportunity in view of the exponential growth in mobile data in many markets. Spectrum availability. The success of wireless solutions relies on the availability of spectrum and bandwidths, and the timing of auctions. The ability to take advantage of new spectrum allocations and the opportunity to potentially refarm existing GSM spectrum are two important developments that will shape LTE deployment strategies. Depending on the country, operators have a choice of spectrum, including: 700800MHz, 900MHz and possibly 1800MHz, as well as new spectrum in 2.52.6GHz. The choices that operators make will determine range, capacity and building penetration, which, in turn, will affect cell count, and therefore opex and capex. It is crucial that operators understand what spectrum is going to be available and when. Collectively, technology, devices, spectrum and traffic growth, along with the operators business objectives, are the cornerstones of operators strategies for deploying dataoptimised networks. In developed markets, operators networks will typically reach full capacity within the next 23 years. We expect to see widespread deployment of LTE by 2014, with early adopters deploying by 2012. In emerging markets, operators typically have less-evolved legacy infrastructure. Data demand is not as great as in many developed markets, so we expect to see the first deployments of LTE nearer to 2015. In both developed and developing regions, we expect that many operators will evolve through the various access technologies (GSM, GPRS, EDGE, W-CDMA, HSDPA, HSUPA). We believe that there are four basic operator LTE deployment strategies, based on four operator types: the operator with both GSM and UMTS access networks; the UMTS-only operator, the GSM-only and the CDMA-only. There will be anomalous situations where early adopters in emerging markets leapfrog from GSM directly to LTE. We expect this to occur in certain South-East Asian markets. The same will also be true for a number of CDMA operators, including Verizon, which is expected to be one of the first operators to deploy LTE. The time is ripe for operators to develop a coherent network evolution strategy, based on the four cornerstone drivers (traffic growth, device availability, technology mix and spectrum availability) and to design a phased deployment that stays one step ahead of rising demand.
0: Summary
1.1 Network evolution will be largely driven by consumer demand for services delivered over high-performance networks
A number of factors are driving wireless data, as shown in Table 1.1.
Table 1.1: Factors that are driving the increase in wireless network data traffic [Source: Analysys Mason, 2009]
Type Network Factor Widespread deployment of 3G technologies that are supporting greater usage and range of services Improved mobile devices enable a greater range of data services Increasing indoor usage of mobile devices Increasing mobile penetration Description 3G enhancements (including HSPA and broadband EV-DO) offer faster throughput, greater capacity, better quality of service and lower costs than previous generations of mobile technology Plug-and-play wireless USB modems and advanced smartphones (with improved displays, user interfaces, processing power, memory and batteries) have become more widely available Heavy indoor usage of traffic-intensive data services, such as mobile broadband services and mobile TV, will drive the demand for increased capacity Mobile penetration has reached near-saturation in developing regions, but there are opportunities for users to adopt secondary devices, such as USB modems
Devices
Type Services/content
Description The amount of music and video traffic has grown significantly in the last couple of years as a result of popularity of YouTube, Napster and iTunes, the increased availability of streamed radio content and the burgeoning number of professional and amateur podcasts. Music and video streaming and downloads are some of the most intensive applications available on the Internet. It was reported in the Telegraph in May 2008 that YouTube consumed as much capacity in 2007 as the whole Internet did in 2000.1 Videos, which account for about 99% of all bytes transferred, are growing in terms of file size and length: 90% of videos last longer than three minutes, up from under one minute in 1997 The mobile industry has shifted its focus from mobile Web (WAP) to full-screen display of standard Web pages. Web pages are also becoming more complex, with increasing numbers of multimedia-rich pages. The average Web page size has more than tripled in the past five years2 Reduced prices and unlimited tariffs improve affordability (particularly for traffic-intensive services)
1.2 Wireless network traffic from voice and data services will increase dramatically by 2015
The total volume of wireless network traffic generated from voice and data services in 2015 will be nearly ten times that in 2008 in developed regions, as shown in Table 1.2. Of this total, data will account for 94% by 2015 and almost three quarters (74%) will be generated indoors. Wireless network traffic in developing regions will follow similar trends, but will lag behind and will not increase to the same extent in the period to 2015.
1 2
http://www.telegraph.co.uk/news/uknews/1584230/Web-could-collapse-as-video-demand-soars.html. Website Optimization, LLC (Ann Arbor, MI, 2008), Average Web Page Size Triples Since 2003. Available at http://www.websiteoptimization.com/speed/tweak/average-web-page/.
Table 1.2: Key forecasts for mobile traffic in developed and developing regions [Source: Analysys Mason, 2009]
Traffic metric Total traffic per month Traffic per mobile user per month Percentage of data in total traffic Percentage of total traffic generated indoors Revenue per megabyte Developed regions 2008 57PB 56MB 49% 54% 2015 557PB 455MB 94% 74% Developing regions 2008 50PB 22MB 7% 34% 2015 307PB 83MB 79% 62%
USD0.86
USD0.12
USD0.57
USD0.15
These forecasts assume a Base case scenario, as shown in Figure 1.1. In the Downside scenario, traffic per customer will be four times higher in 2015 than it was in 2008, while in the Upside scenario, traffic per customer will be 22 times higher. The variability of these forecasts implies that operators must carefully predict the demand for traffic from their customers and build a degree of flexibility into their plans.
Figure 1.1: Global wireless network traffic growth, 20082015 [Source: Analysys Mason, 2008]3 3000 Total wireless network traffic (PB per month) 2500 2000 1500 1000 500 0 2008 Base case Downside Upside
2009
2010
2011
2012
2013
2014
2015
For more information, see Brydon, A. and Heath, M., Wireless network traffic 20082015: forecasts and analysis, Analysys Mason (Cambridge, 2009).
Figure 1.2: The cost of data traffic using legacy networks, and revenue [Source: Analysys Mason, 2009]
Volume of network traffic Cost of existing network
Network costs outstrip revenue
Revenue
Voice era
Data era
Time
In developed countries, mobile broadband revenue is overtaking that of voice (developing regions will lag behind due to a greater initial reliance on circuit-switched voice). Mobile broadband revenue in Europe is expected to grow from USD9.07 billion in 2008 to USD33.91 billion in 2014, and is a key element of operators business models.4 In order to support usage-intensive applications profitably and maintain an optimum cost structure, network operators will need to develop their network architecture with the aim of reducing the network carriage cost per megabyte. Together, these conditions point to the need for: a data-optimised network architecture, a network carriage cost per megabyte that realises a return on investment and a level of indoor service that is comparable to outdoor coverage. The introduction of network enhancements, such as HSPA, HSPA+ and LTE possibly in combination will address these needs, but operators are facing complex technology choices.
For more information, see Hatton, M., Mobile broadband in Europe: forecasts and analysis 20092014, Analysys Mason (Cambridge, 2009).
the advantages of LTE other technology options CDMA, TD-SCDMA and WiMAX the implications of existing assets and legacy services for future deployment plans the provision of indoor coverage.
offers significant performance improvements has the support of the mobile industry can offer significant opex and capex savings offers improved use of spectrum.
LTE offers significant performance improvements LTE offers reduced latency, improved quality of service and the ability to operate with advanced antenna systems, which can improve spectral efficiency.5 For example, LTE with 22 MIMO is approximately 2.5 times more spectrally efficient than HSPA.6
LTE has strong industry support Radio access standardisation of LTE was finalised at the end of 2007 and the remaining specifications are expected to be completed by the end of March 2009. Already 85% of the worlds mobile subscribers are on GSM/UMTS-based networks and the number is expected to increase. Both base-station and handset vendors are taking advantage of this huge installed base of subscribers. Existing 3G infrastructure and handset vendors will offer LTE-compatible technology and devices, creating a multi-vendor market with interoperable products and potentially cost-effective upgrades from existing equipment, which will reduce costs and increase choice through competition. Base-station vendors are designing LTE base stations that use existing base-station sites to provide the same coverage as GSM and UMTS services. Base-station vendors claim that they will attempt to enable MNOs to re-use existing assets wherever possible, minimising the requirement for investment. Consequently, a number of major operators have confirmed that they have a long-term strategy to move to LTE, building confidence throughout the industry that the value chain of base-station and device manufacturers and vendors will be able to create economies of scale and support a variety of devices and services. Product development for LTE is in progress and the first LTE systems will go to market at the end of 2009.
Spectral efficiency is a measure of the ability of a radio bandwidth to carry information, and is measured in bits per hertz. MIMO multiple antenna technology improves the range and capacity of a communications link for a given bandwidth.
LTE can offer significant opex and capex savings Compared with legacy networks, LTE improves spectral efficiency and requires fewer nodes to deliver the same network capacity, or a greater capacity can be achieved for the same number of nodes. Furthermore, LTE employs a flat-RAN architecture, which also reduces the number of network nodes. Jointly, the benefits of improved spectral efficiency and a flat-RAN architecture reduce network carriage costs and create a cost-growth curve that tends to track revenue rather than demand, allowing the operator to maintain a healthy profit margin, as shown in Figure 2.2.
Figure 2.2: LTE can deliver profit by significantly reducing opex and capex [Source: Analysys Mason, 2009]
Volume of network traffic Cost of legacy network
Legacy network costs outstrip revenue Revenue
Voice era
Data era
Compared with previous access technologies, like GSM or UMTS, the smaller size of the LTE access node brings other operational gains, including easier and quicker siting of the node, lower power consumption and a much smaller space requirement. Deploying LTE networks could allow operators to switch off their GSM networks, which would offer further savings. However, this is one of the more contentious benefits. Switching off GSM would simplify network management, with the corresponding operational cost savings and the earlier an operator can decommission its GSM network, the greater such gains are likely to be. At the same time, there would be costs associated with:
migrating the GSM customer base and other legacy services the loss of GSM roaming and the attendant revenue stream the loss of revenue from GSM circuit-switched voice.
Many operators have tried RAN sharing as a way of reducing costs. When RAN sharing fails, it is often because operators are not able to agree how to apportion value to existing assets. Deploying a new (LTE) network offers operators the chance to plan a shared access network without being concerned about how to value existing assets. Sharing an LTE RAN will help operators to justify the investment. Indeed, network sharing may be the only way that operators can afford to invest in LTE.
LTE offers improved use of spectrum The choice of spectrum is an important one for operators as it directly affects capex and opex. LTE offers a greater degree of spectrum flexibility compared with earlier generations of mobile technology. It is available in a wider number of carrier frequencies and various bandwidths are available from 1.25MHz to 20MHz. In comparison, UMTS has a fixed, and therefore relatively inflexible, bandwidth requirement of 5MHz. LTE also supports both paired and unpaired radio spectrum, using its FDD and TDD modes. LTE is deployable in a range of spectrum bandwidths (from 1.2520MHz), allowing MNOs to use current spectrum allocations, and has also been designed to use the IMT-2000 extension band, which we discuss in more detail in Section 4.1. In later releases of the standard, LTE will be able to add together blocks of spectrum, or concatenate spectrum, for different parts of the radio spectrum to create one large bandwidth. This will be a major step forward in improving capacity, spectral efficiency and harmonising spectrum usage worldwide.
2.2 There are other technology choices, including CDMA, TDSCDMA and WiMAX, but most operators will migrate to LTE
LTE is the natural evolutionary successor to HSPA (and HSPA+), and it could also offer a smooth migration from all other the major systems, including CDMA, TD-SCDMA and WiMAX, as shown in Figure 2.3.
Figure 2.3: All technology roadmaps can lead to LTE [Source: Analysys Mason, 2009]
GSM
GPRS
EDGE
EDGE Evolution TD-SCDMA Evolutions HSPA HSPA+ LTE Rel. 8/9 FDD/TDD LTE Rel.10 FDD/TDD
TD-SCDMA
W-CDMA
CDMA 1
UMB
WiMAX 16d
However, LTE is not the only solution for a data-optimised network. It is important to consider the role of other technologies, including CDMA and TD-SCDMA, and particularly WiMAX. WiMAX 16e was initially presented as competing with LTE, however deployments have shown that this not the case. WiMAX 16e uses 2.3GHz, 2.5GHz and 3.5GHz licensed spectrum and 5.8GHz unlicensed, and the fact that WiMAX 16e is TDD has allowed it to find a number of applications in parts of the world where spectrum is an issue. Moving into 4G, the specification for the next generation of WiMAX IEEE 802.16m is very similar to LTE in terms of frequency, spectral efficiency, latency and advanced antenna systems. As a consequence, WiMAX 16m will compete with LTE for the same customer base. LTE will have the advantage of being able to draw on a significant subscriber base, but as there is no single service that demands LTE, it is plausible that there will be strong competition from WiMAX based on price. The outcome of this competition will depend on the willingness of members of the LTE and WiMAX value chains to subsidise price in order to see their technology succeed. Because LTE will evolve from the vast base of GSM/UMTS networks, LTE will have major commercial advantages over WiMAX. LTE will provide performance similar to that of WiMAX 16m, so there is no justification (from a performance perspective) for UMTS operators to change to WiMAX. Discussion with leading vendors and operators reveals that most MNOs will upgrade their networks to LTE rather than WiMAX. We predict that UMTS technologies will dominate wireless broadband services, and will have possibly twenty times as many users as WiMAX by the end of 2015. It is likely that most WiMAX deployments will be in developing regions, where it is already finding ready markets. We believe that 92% of the 98 million WiMAX customers at the end of 2015 will be in developing regions. While the majority of MNOs will follow an LTE roadmap,
10
WiMAX will be one of the technologies available to UMTS operators that might want to take advantage of opportunities in developing countries and/or countries where ADSL coverage at broadband speeds is poor. Telefnica is one such example.
Telefnica is working with WiMAX in selected markets in order to understand its potential in specific scenarios
In Latin America, lack of spectrum is a big issue. With the exception of Brazil, Telefnica does not have dedicated 3G spectrum in any of the Latin American markets in which it operates. Furthermore, in most of these markets, Telefnica has spectrum allocations of 212MHz or 5MHz at 850MHz, and 10MHz or 20MHz at 1900MHz, which is half the amount of spectrum it has in European markets. According to Telefnica, We have to make the most of the spectrum we have to deliver the capacity that our clients demand. We are exploring options that we have with our 2.5GHz and 3.5GHz assets. WiMAX continues to be an option in very specific market scenarios, but we have clearly defined a development path that focuses on 3GPP, not WiMAX.
Similarly, Orange has launched WiMAX in Africa, taking advantage of the availability of WiMAX licences where 3G licences are not yet available.
CDMA, a narrowband technology similar to W-CDMA, has been widely deployed globally, notably in the USA and parts of Asia. CDMA was pioneered by Qualcomm, but Qualcomm shifted its interest to LTE in November 2008, essentially closing CDMAs evolution path. TD-SCDMA is a proprietary technology based on TDD W-CDMA. It was devised in China to compete with Western technologies and to reduce the cost of intellectual property rights being paid to Western companies. While there is limited global interest in TD-SCDMA, it will have a significant impact in China, where China Mobile has adopted it as the 3G
11
evolution technology.7 It remains to be seen whether TD-SCDMA will evolve or join the LTE path. It is part of 3GPP specifications from Release 4 onwards.
2.3 Existing assets and legacy services will influence future deployment plans
One of the key questions facing operators is how much they will be able to maximise the use of existing network assets. By the end of 2008, 3.5 billion mobile subscribers were using networks based on the GSM family of standards, of which only 350 million subscribers were on 3G UMTS. In theory, operators could migrate a substantial number of subscribers to UMTS before moving on to LTE. However, this ignores local regulatory constraints, availability of suitable spectrum, local legacy services and the relatively limited roaming capability of UMTS. 3G spectrum has not yet been fully allocated in many countries and UMTS could be deployed more widely, but operators will need to weigh the cost of upgrading older technology against the cost of deploying a new technology with a longer lifespan ahead of it. In countries where UMTS is deployed, many operators feel that their existing assets are not yet fully utilised. For example, in November 2008, Vittorio Colao, CEO of Vodafone, commented that Vodafones European networks were running at only 32% of their capacity.8 Operators must carefully weigh up the longer-term improved profit of LTE networks compared with the investment that is lost if operators do not maximise their UMTS networks. For this reason, many operators will consider using LTE as a capacity overlay where required.
What are the reasons for deploying HSPA+? In developed countries, many operators with UMTS HSPA networks have to decide whether to upgrade to HSPA+ or whether to move directly to LTE. A number of operators, including Swisscom, Telefnica and T-Mobile, will carry out software upgrades to HSPA and HSPA+ without deploying MIMO. Typically, these operators have already deployed HSPA across a large proportion of their networks, offering HSDPA at 7.2Mbit/s in major cities.
The Chinese telecoms regulator issued China Mobile with a technology-specific licence for TD-SCDMA in January 2009. It is expected that China Mobile will begin trials in 2009 and will launch the technology commercially in 2010. The Financial Times: http://www.ft.com/cms/s/0/18d0ebb0-cf85-11dd-abf9-000077b07658.html.
12
There are a number of arguments in favour of migrating straight to LTE from HSPA, as shown in Table 2.1.
Table 2.1: SWOT analysis of migration from HSPA directly to LTE [Source: Analysys Mason, 2009]
Strengths Opportunities
Improved spectral efficiency at 20MHz Offers speeds of up to 144Mbit/s downlink and 50Mbit/s+ uplink speeds in <20MHz spectrum MIMO evolution path will be more likely for LTE Significantly more investment required in infrastructure, because LTE is based on OFDMA, a completely new modulation scheme (whereas HSPA+ is based on W-CDMA). LTE requires a new set of radio access and core infrastructure components not only are the radio access algorithms different from legacy networks, but also the signalling and control protocols from the access to the core are significantly different Backward compatibility handover from LTE to HSPA+ is complex, compared with handover from HSPA+ to HSPA LTE chips will be expensive initially, because they will require a new semiconductor ecosystem to support them Significant disruption to existing services and redirection of resources
Competitive differentiation Wider choice of carriers and bandwidths. LTE TDD will be available Can re-use GSM sites Future proofing against capacity demand
Weaknesses
Threats
If LTE operators are not able to secure 20MHz spectrum, they will not be able to realise the advantages of LTE in terms of speed May cause operators to overextend themselves financially Anticipated data demand may not be realised
Other Tier 1 operators, including Orange, have conducted public trials of HSPA+ with hardware MIMO: clearly there is a feeling that HSPA+ still has growth potential, as shown in Table 2.2. This is particularly true where operators can benefit from using HSPA+ as a high-capacity overlay when demand is great.
13
Table 2.2: SWOT analysis of migration from HSPA to HSPA+ (Release 7) [Source: Analysys Mason, 2009]
Strengths Opportunities
HSPA+ uses the existing modulation scheme (W-CDMA), whereas LTE requires a new set of radio access and core infrastructure components Spectral efficiency of HSPA+ (when MIMO is included) is close to that of LTE in 5MHz spectrum Backward compatibility handover from HSPA+ to HSPA is relatively simple (the connectivity between RNC and the core (SGSN and GGSN) remains as before), compared with handover from LTE to HSPA+ Handsets and terminals are widely available HSPA+ USB modems are likely to be cheaper than LTE equipment, which will require a new semiconductor ecosystem
If operators are unable to secure 20MHz spectrum, they will not be able to realise the advantages of LTE in terms of speed Reduces capex Less disruptive upgrade path Allows investment to match demand smooth upgrade path
Weaknesses
Threats
HSPA faces serious coverage issues, as increased data usage causes cell shrinkage. To counter this, network operators would have to split cells to handle the load The backhaul networks (being dimensioned for voice and low data rates) are not ready to support mass adoption of mobile data services. In addition to investment in MIMO equipment, mobile networks would need upgrades to the backhaul and the core network, which will be costly Requires another site visit to upgrade to LTE
Competition from operators that have migrated directly to LTE Running out of capacity
Orange plans to deploy HSPA+ as soon as the technology becomes available, which is likely to be in mid-2009.
Orange plans to upgrade to HSPA+ (Release 7) as soon as the technology becomes available
Orange plans to upgrade its network gradually, focusing on high-density areas, as and when capacity is required: HSPA+ allows you to select the options you want to include in your network and, as such, you can have focused deployment in areas where there are issues with loading or performance. Secondly, it is unlikely that we will see a complete deployment of LTE before 2011 or 2012. If you want to improve the user experience and cater for the growing demand for dongles and new terminals like the iPhone, you need to upgrade.
14
It is, however, clear that Orange is focusing on LTE in the long term. It is currently in discussion with LTE equipment vendors and aims to begin trials in 2009, with its first deployments envisaged for 2010. Oranges initial LTE deployments will focus on dense, urban areas where the demand for network capacity is high. From there it will extend coverage opportunistically. Oranges strategy will also vary between countries, depending on regulatory constraints.
As we will see in Section 2.4, the question of whether to upgrade the macrocell network to HSPA+ or LTE is closely linked to indoor coverage, which is a major consideration in operator access network deployment strategies.
2.4 LTE will not be deployed as a standalone layer, but as a complement to a dedicated indoor coverage mechanism
Data traffic has grown exponentially in the past 12 months, and an increasing proportion of data usage is occurring indoors. We predict that more than 86% of data traffic will originate indoors in developed regions and more than 60% in developing regions by 2015. To be successful, operators will need to fulfil subscribers expectations in terms of coverage. Operators will need to improve the 3G user experience in indoor environments to cater for this growing usage trend. Direct indoor coverage using femtocells or Wi-Fi is one way of achieving this. The widespread use of a direct indoor coverage solution will affect operators plans to deploy LTE by offloading the demand from the macrocell network. Figure 2.4 shows the growth in traffic from 2008 to 2012 for a mobile operator with 10 million subscribers. Demand will outstrip HSPA capacity in 20102011 and HSPA+ capacity by 2012. However, if indoor traffic is offloaded to a direct indoor solution network, HSPA and HSPA+ will offer sufficient capacity for 2012 and beyond.
15
Figure 2.4: Network traffic generated by an example mobile service mix, split between indoor and outdoor usage, 20082012 [Source: Analysys Mason, 2009]
10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 2008 2009 Outdoor Indoor 2010 2011 2012
HSPA+ (10MHz) HSPA (10MHz) LTE (10MHz) LTE (20MHz)
Network capacities
Femtocells Femtocells are small UMTS base stations, which users can install in their homes and connect to mobile operators switching networks through the home ADSL line. In contrast, femtocells for businesses are usually installed and integrated into the existing IT network of the business by a trained installation team provided by the operator. Backhaul is via the existing external connectivity. While the femtocell solution is technically elegant, it is not straightforward to implement and has at least three major hurdles to overcome.
The business case is challenging. The success of the solution depends on the use of an ADSL backhaul, which will disadvantage the mobile operator, unless its parent company also owns the backhaul. Operators will have to learn to manage millions of femtocells, as opposed to tens of thousands of base stations, which was their previous experience.
Much has been said about the technical challenges of a femtocell deployment associated with managing interference and handover, but these have largely been overcome. Mobile operators must carefully consider their position before relying on a femtocell solution to meet the increasing demand for capacity. A number of operators, including T-Mobile, TeliaSonera and Vodafone, are conducting femtocell trials. The hype
16
surrounding femtocells seems to have diminished and operators are doing more work behind the scenes to determine the validity of a femtocell solution.
Wi-Fi with ADSL Femtocells are not the only means of providing indoor coverage. Wi-Fi, combined with WiMAX or ADSL, is another serious contender. Wi-Fi routers are commonplace in both homes and businesses. These routers are self-installed devices, and have a number of advantages over the femtocell direct indoor solution. The business case for Wi-Fi is well proved and the solution is technically efficient. Furthermore, because Wi-Fi routers use unlicensed frequency bands, operators are not required to manage interference with an outdoor network. However, this ease of use also has its disadvantages. Wi-Fi routers do not offer a carrier class of service, which is particularly important for business deployments. Wi-Fi installations are not bound by the type of service level agreements that an enterprise
17
would expect. Finally, because of the nature of the Wi-Fi plug and play installation, enterprises tend to be casual about security and often do not implement a full link encryption.
the ITU allocation of a 3G expansion band in 2600MHz the refarming of GSM 900MHz (and possibly GSM 1800MHz) the availability of spectrum in the 700800MHz range, as a result of the switchover from analogue to digital broadcast TV the refarming of wireless cable 17002100MHz spectrum in the USA for Advanced Wireless Services.
July 2008
January 2009
2Q 3Q05 2009
2Q 2010
19
Unlike UMTS, LTE operates over a range of bandwidths, from as little as 1.25MHz, but with greater spectral efficiency at bandwidths above 10MHz, ideally 20MHz. The 2600MHz band offers good bandwidth, and most network operators have publicly declared an interest in deploying LTE in this band. However, compared with lower frequencies, the range of a 2600MHz cell is limited. This can be overcome, but requires a high density of cells, at greater expense. As a consequence, most operators are considering using the IMT-2000 expansion band as a capacity overlay where there is a greater demand for capacity and range is less of a concern, for example in urban areas.
3.2 The timing of GSM 900MHz (and GSM 1800MHz) spectrum refarming is critical
Operators are unable to speculate about whether they will use LTE in any refarmed spectrum: in most cases, it is not known when the spectrum will become available and, therefore, operators do not know how mature LTE may be at that time. Furthermore, they do not know what other spectrum may become available in the meantime. We believe that, where refarming is already permitted or regulatory approval is imminent, there may be a good business case for a GSM operator to deploy UMTS900, rather than LTE, in order to provide 3G rural coverage, increase 3G capacity, or both. Operators that are launching UMTS900 networks are expecting the greatest usage to come initially from mobile broadband services based on USB modems or datacards, and envisage that voice services will follow as subscribers replace their 2G handsets with phones supporting UMTS900/2100 this may take several years, especially in the current economic climate. If early switch-off of the GSM network is shown to offer significant cost benefits, operators will need to subsidise 3G handsets in order to encourage their subscribers to upgrade more rapidly, but the costs and benefits of doing so would require careful consideration.
Where refarming measures are delayed or take time to implement, it becomes increasingly likely that an operator will migrate to LTE900, rather than UMTS900. There are a number of benefits of deploying LTE900, rather than UMTS900.
20
UMTS deployments require a full 5MHz of spectrum, whereas LTE, which can be deployed in spectrum bands as small as 1.25MHz and up to 20MHz, allows for the allocation of spectrum to be more flexible and maximises the use of available spectrum. The coverage of GSM and LTE is similarly predictable, because like GSM, and unlike UMTS, the range of an LTE cell is unaffected by traffic on the cell. Therefore, GSM sites can readily be used to deploy LTE, yielding considerable capex savings.
If operators migrate to LTE900, then GSM900 is likely to be maintained for voice services, global roaming and certain M2M applications, possibly as late as 20182020. The future use of any 1800MHz spectrum that an operator holds is linked to the future of GSM900. Unlike UMTS900, for which the supply of network equipment and devices is rapidly increasing, there is relatively little vendor backing for UMTS1800 equipment or devices at present. As a result, we regard refarming to LTE at 1800MHz as a more probable outcome than deploying UMTS1800. Operators are therefore likely to retain GSM services at 1800MHz until refarming at that frequency is required for capacity reasons, and will then deploy the most spectrally efficient technology at their disposal at that time. How quickly migration to LTE1800 becomes feasible will depend on the availability of the LTE standards, network equipment and devices. Vendors are claiming that operators want to deploy LTE in the 1800MHz band. If this is true, then we expect network equipment and devices to be available within the next three years. In the short term (35 years), it is likely that LTE will not replace GSM1800, but will be deployed in addition to GSM1800.
21
3.4 The US regulator has made spectrum available for Advanced Wireless Services
In 20062007, the FCC auctioned spectrum in two segments (17101755MHz and 21102155MHz) for Advanced Wireless Services mobile broadband and other 3G services. The spectrum was previously allocated to MMDS or wireless cable. Most of the spectrum was awarded to T-Mobile USA in 2007. Advanced Wireless Services spectrum offers reasonable coverage, similar to GSM1800, and is offered in 10MHz and 20MHz bandwidth pairs, providing good capacity. Therefore, the spectrum could be used for coverage only, or for coverage and capacity. In summary, the availability of new spectrum allocations and the opportunity to refarm existing GSM spectrum are two key areas that will enable LTE deployments. In the future, operators should be aware of the various forthcoming spectrum auctions within their particularly market(s) and align them with their RAN evolution strategies.
Figure 4.1: Expected LTE device availability [Source: Analysys Mason, 2009]
USB modems and datacards Laptops with embedded modules
Handsets
2009 3Q02
3Q03
2010
3Q04
1Q05
3Q06
2012
USB modems or datacards will be the first devices to be introduced and are likely to be available in 2H 2009. These device forms are already common with UMTS. LTE will initially support data-intensive services for laptop users. Handsets will be available approximately one year after the first USB modems are available. Handsets will initially be for data only. Voice handsets will probably not be available until LTE VoIP is commercialised. This will first require the standardisation of
23
VCC, part of IMS service continuity, which is due in Release 10 of 3GPP standards, and not before 2010.
Laptops with embedded modules are likely to be made available 12 years after the introduction of USB modems. The design cycles are longer for embedded solutions. The embedded modules, particularly for LTE, require additional support for MIMO.
Clearly, the growth in demand for data from subscribers, which is the catalyst for LTE development, will also have an impact on the design of handsets and devices with embedded modules. The likely innovations are larger, higher-resolution handset screens on which to view video content, larger capacity for storing music tracks and dedicated keys to use solely for gaming. Long battery lives will be another key factor among the specifications. Within 23 years, the dominant device for mobile broadband access will be the laptop with an embedded module. In the longer term, the delivery of LTE services will not be limited to traditional devices, but will instead extend to a range of devices including cameras, cars, energy monitors, environmental sensors, health-monitoring devices and many other machines. As with any new technology, the retail cost of LTE devices is expected to come in at the top end of the market, targeting enterprise users and early adopters. We expect unit cost to fall as a result of economies of scale and competition. To this end, members of the value chain are working hard to ensure that take-up of LTE devices is strong. The Embedded Mobile Intiative, an industry group led by the GSMA, is aiming to foster collaboration between consumer electronics and industrial product manufacturers to encourage the adoption of embedded 3G modules, while developing common technical standards and platforms. Members of the initiative include operators and vendors, such as Huawei, KTF, Rogers Wireless, Smart, Softbank, Telecom Italia, Telefnica, Telstra and Vodafone. Qualcomm estimates that it will take a couple of years for LTE devices to become widely available. As unit costs fall, laptop manufacturers will increasingly fit broadband access modules as standard. This threatens to separate the device and access markets, and MNOs will lose control of an important element of the value proposition.
4.2 The first LTE-compatible devices are likely to support four frequency bands
The first devices (USB modems and datacards) in Europe and Asia Pacific are likely to support four frequency bands: 900/1800/2100 and 2600MHz, GSM and W-CDMA. Devices to support refarming of 900/1800MHz and possibly 2100MHz, will be available
24
from 2010. Commercial launch, depending on demand, will follow at a later date, following a period of interoperability testing. The 700MHz frequency band, auctioned recently in the USA, will pose a number of challenges for device manufacturers, including:
the need for a larger antenna due to the longer wavelength channel interference from adjacent frequency bands that are used for TV broadcasting image channel interference due to GPS in band harmonics.
The characteristics of the 700MHz band will constrain LTE devices, resulting in a more onerous design process, the cost of which has to be absorbed in the value chain.
4.3 Vendors have agreed a standardised royalty framework for LTE handsets
In April 2008, seven major vendors Alcatel-Lucent, Ericsson, NEC, NextWave Wireless, Nokia, Nokia Siemens Networks and Sony Ericsson announced that they had reached an agreement for a framework for licensing intellectual property rights relating to 3GPP LTE. The framework sets a single-digit percentage royalty or, in the case of notebook computers, a fee of under USD10 on equipment using LTE technology. The framework for royalties should allow operators to avoid the delays and costs incurred when disputes over intellectual property rights disrupted the development of 3G handsets. In summary, a number of key steps have already been taken to ensure that devices will be available in time for the launch of the first LTE networks. On-going co-operation between operators, application developers and handset manufacturers is essential if a reasonable choice of devices is to be available in time for future commercial launch dates. While vendors and operators are confident about device availability expecting them to be ready in sufficient numbers for LTE launches the success of LTE is particularly sensitive to the price and availability of devices. One would do well to remember the lessons of early deployments of UMTS, which were beset by the limited number of reasonably priced handsets. In the case of LTE, the range of frequencies and variety of device types may exacerbate the problem. If devices are not available or are available in insufficient quantities, this would represent a major barrier to the adoption of mobile broadband services over LTE.
direct indoor coverage with femtocells and/or Wi-Fi macrocell coverage using 900MHz macrocell coverage using 2100MHz and/or 2600MHz.
With direct indoor coverage using femtocells/Wi-Fi, the quality of coverage is likely to be far better than macrocell coverage. Wi-Fi has the advantage of being a reliable technology that can be deployed today with a viable business model. Femtocells are a natural progression for the mobile operator from a GSM/UMTS family background, but the business case and technical challenges may yet prevent the widespread deployment of femtocells. Both Wi-Fi and femtocell solutions depend on the ADSL backhaul. This may make the mobile operator unacceptably vulnerable to the fixed operator and may be a serious weakness that fixed operators could exploit, especially as our predictions are that mobile operators may have taken more than 20% of fixed operators broadband business in developed European markets by 2013. Of course, a converged operator, with both fixed and mobile networks, would not feel threatened by fixedmobile substitution. If it transpires that a profitable femtocell business case is not possible, and/or the operator does not wish to rely on ADSL backhaul to provide its mobile broadband offering, then the operator will need to provide indoor coverage with the macrocell network. Whether the operator decides to buy 700MHz spectrum when it becomes available or to replace 900MHz GSM with data-optimised access technology, such as LTE or HSPA+, depends on
26
when the spectrum becomes available, and how strong the demand for data is predicted to be. We believe that the operational costs of revisiting sites to make hardware upgrades mean that most operators will deploy LTE in 900MHz when it becomes available. It is likely that the 2G/3G operator will take advantage of incidental indoor coverage from 2100MHz, however the site density required to achieve good coverage indoors at 2600MHz will need to be weighed against the cost of acquiring new, low-frequency spectrum (900MHz).
27
28
CDMA operators, including Verizon Wireless, which is expected to be one of the first operators to deploy an LTE network. A number of operators have announced plans for LTE deployment, as shown in Figure 5.1.
T-Mobile Trials: now Launch: 2010/2011 NTT DoCoMo Trials: now Launch: 2010
6 8
Technology, devices, spectrum and demand, along with the operators business strategy, are the main factors that will determine the timing and strategy for LTE deployment for any particular operator. More than ever these will be different for each operator. Operators will need to consider their local choices carefully, in order to achieve an optimal position to deal with future data demand and demands on their revenue. We expect operators to differentiate themselves based on their understanding of the interplay of these factors and their subsequent technology choices and deployment strategies. Ultimately, there are as many pathways to a data-optimised RAN as there are operators. For all operators, finding their way through this maze of complex decisions can be frustrating, but there are a few simple guidelines they can follow, as shown in Figure 5.2.
29
Plan your technology mix and spectrum requirements 5 6 Phased Technology deployment mix Optimise a phased deployment with the following constraints: minimise network carriage costs maintain network service quality meet rising demand
Key to acronyms
2G 3G 4G 3GPP Second generation Third generation Fourth generation The 3rd Generation Partnership Project ADSL Asymmetrical digital subscriber line ARPU CDMA CEO CPE DSL EDGE EV-DO Average revenue per user Code-Division Multiple Access Chief executive officer Customer premises equipment Digital subscriber line Enhanced Data for GSM Evolution CDMA2000 1 Evolution Data Only EV-DV CDMA2000 1 Evolution Data and Voice FCC Federal Communications Commission FDD GB GGSN GHz GPRS GPS GSM Frequency division duplex Gigabyte Gateway GPRS support node Gigahertz General packet radio service Global positioning system Global System for Mobile Communications GSMA HD HSDPA GSM Association High definition High-Speed Downlink Packet Access HSPA HSPA+ HSUPA IMS IMT-2000 High-Speed Packet Access Evolved High-Speed Packet Access High-Speed Uplink Packet Access IP Multimedia Subsystem International Mobile Telecommunicationss-2000 IP kHz Internet protocol Kilohertz WAP Wi-Fi WiMAX USB VCC VoIP W-CDMA TV UMA UMB UMTS TDD TD-SCDMA Opex PB RAN RNC RTT SGSN SIM SWOT MNO OFDMA MMDS LTE MB Mbit/s MHz MIMO Long-term evolution Megabyte Megabits per second Megahertz Multiple-input multiple-output (a smart antenna system) Multichannel multipoint distribution system Mobile network operator Orthogonal Frequency Division Multiple Access Operational expenditure Petabyte Radio access network Radio network controller Radio transmission technology Serving GPRS support node Subscriber Identity Module Strengths, weaknesses, opportunities and threats Time-division duplex Time Division Synchronous Code Division Multiple Access Television Unlicensed Mobile Access Ultra Mobile Broadband Universal Mobile Telecommunications System Universal serial bus Voice call continuity Voice over IP Wideband Code Division Multiple Access Wireless Application Protocol Wireless Fidelity Worldwide Interoperability for Microwave Access (broadband wireless based on the IEEE 802.16 standard)
Key to acronyms
Telecoms Markets
Fixed Broadband
Voice
Telecoms Software
Service Assurance
Service Delivery Platform Strategies Network Equipment Manufacturer Strategies Wireless Networks Telecoms Software Market Shares Telecoms Software Forecasts
Telecoms Software Strategies Technology and Infrastructure Core Data and Forecasts Country Reports
Core Forecasts
For more information about our research programmes and custom research services, please visit www.analysysmason.com/research.
Mobile operators
Delivering value through strategy planning and implementation
Media companies
Helping to maximise revenue in converging markets
Regulators
Establishing and implementing policy frameworks in telecoms and media markets
Enterprise users
Helping to maximise returns from investment in ICT
Financial institutions
Supporting vendors, sellers and financiers of industry transactions
Fixed operators
Defining new strategies, optimising service portfolios and supporting policy development
For more than 20 years, Analysys Mason has been working with a range of global clients to resolve business-critical issues by applying its intellectual rigour, passion and expertise. For more information about our consulting services, please visit www.analysysmason.com/consulting.