Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Maravedis-Rethink:White Paper Sensification SCWS


Published on

This short paper looks at the requirement for densification; the scale of capacity that operators will need by 2018; and the critical enablers, such as self-optimizing network (SON) technology. The data points referenced come from surveys of over 75 mobile operators by Maravedis-Rethink’s RAN Research Service. Fuller findings will be published in June 2014 in a report entitled ‘Towards the Hyper-Dense Network: the shape of the HetNet 2013-2019’.

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

Maravedis-Rethink:White Paper Sensification SCWS

  1. 1. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 Summary Densification may be a horrible word, but it is an important one to almost any mobile operator which is fighting the twin battles of meeting rising data capacity demands; and doing that profitably. There are many ways to add capacity to a network, but they have to be cost-effective enough for a world where data ARPUs are stagnant or falling, and many added value revenues are going to over-the- top or cloud providers rather than mobile carriers. Massive capacity at low cost and power, and with the flexibility to target resources where they are needed, rather over-provision the whole network – these are the drivers behind densification, a wave which will affect all layers of the infrastructure, but particularly the small cells. This short paper looks at the requirement for densification; the scale of capacity that operators will need by 2018; and the critical enablers, such as self-optimizing network (SON) technology. The data points referenced come from surveys of over 75 mobile operators by Maravedis- Rethink’s RAN Research Service. Fuller findings will be published in June 2014 in a report entitled ‘Towards the Hyper-Dense Network: the shape of the HetNet 2013-2019’. Drivers of density Mobile operators are facing a period when users will consume up to 40 times more wireless data by 2019, requiring massive increases in network capacity; but in which ARPU will rise by only 2.5%. A network design which delivers substantial, and well-targeted, capacity, while slashing cost of delivery, is essential to any business model. Among the drivers for a new network are: Users will demand four times more gigabytes of data for the same cost by 2018  Mobile data traffic will rise 12 times to reach 10.5 exabytes by 2018.  Consumer and enterprise ARPU will rise by just 2.5%, and will fall sharply in voice/messaging. Towards the hyper-dense network: SON will be the key enabler Caroline Gabriel, May 2014
  2. 2. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014  New revenue streams such as broadband M2M will be enabled by changing networks and market conditions, but these will add only 11% to total revenue, on average.  Using conventional network architectures, by contrast, would require a 44% increase in capital spending in the years between 2012 and 2018, when most operators are targeting less than 5% a year.  Operating costs will rise too using current architectures, notably backhaul and power  A reduction in radiated power is essential – mobile usage is limited even more by battery life than by coverage. The role of small cells: No single solution will provide all the answers, and carriers will use a combination of tools including Wi-Fi offload, small cells, distributed and cloud-based RANs, and LTE-Advanced upgrades. Software intelligence will be critical and the period will see the complete reworking of the OSS (operations support system); convergence with the BSS and IT platforms; and the start of a significant uptake of software defined networking, particularly NFV (Network Functions Virtualization). Hallmarks of the new RAN will include:  increasing deconstruction of the base station to distribute processing effort efficiently and reduce cost. The extreme will be full Cloud-RAN, where all processing is virtualized in the cloud and stripped-down radio/antenna units remain at the cell sites.  densification - smaller and smaller cells, to maximize capacity by moving it closer to the user, often combined with more and more antennas in the macro layer (Massive MIMO).  multiple layers of base stations within each macrocell, often using different frequencies and air interfaces within one cell to deliver more capacity.  extensive offload of macro network data to Wi-Fi and other technologies.  flexible backhaul techniques.  new, sometimes unconventional spectrum bands and carrier aggregation.  increasing investment in software technologies to make networks more efficient and intelligent – from adaptive networking tools to fully fledged SDN (software defined networking) and NFV (Network Functions Virtualization). Within that broader landscape, a dense layer of small cells will be an important tool. Among the operators surveyed by Maravedis-Rethink, on average it was expected that public access small cells would deliver 14% of the additional capacity required in the first year of commercial deployment, and 18% of the targeted reduction in total cost of ownership (TCO). By year four of a deployment, those figures will rise to 24% and 23% respectively.
  3. 3. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 New technology % required capacity increase Y1 % required TCO decrease Y1 % required capacity increase Y4 % required TCO decrease Y4 LTE upgrade 17% 14% 4% 4% Wi-Fi offload 16% 20% 9% 13% Public access small cells 14% 18% 24% 23% Deconstructed RAN (RRH and distributed antennas) 11% 17% 13% 17% New or refarmed spectrum and carrier aggregation 17% 9% 16% 12% SuperMacro/LTE- A eg MIMO, CoMP 14% 10% 18% 15% Adaptive networking/SON 11% 12% 16% 16% Figure 1. Contributions of various technologies to capacity and TCO improvements in mobile networks, in year one and year four of a HetNet roll-out. Source: Maravedis-Rethink RAN Service survey (75 MNOs) In year 1, small cells are the fourth most important contributor to additional capacity, after new spectrum, the LTE or LTE-A upgrade, and WiFi offload, and equal with macro layer enhancements such as enhanced MIMO. In year 4, small cells become the most important contributor to capacity and TCO improvements. This is partly because some other tools are delivering diminishing returns, such as the air interface upgrade, but also because small cells will have become far denser, improving the cost:capacity ratio and supporting ‘hyper-dense’ capacity where this is required (in sports stadiums, meeting places or business parks, for instance). While first-wave LTE networks have generally been built out in the conventional way – macrocell overlays or modernizations in conventional mobile spectrum bands, with coverage the first goal – the next wave of LTE, LTE-Advanced and later ‘5G’ will be very different. Densification will see MNOs deploying large numbers of small cells, often in a separate layer from the macrocells, to deliver high capacity in a targeted way close to the users. In the HetNet, cells of many sizes, air interfaces and spectrum (including unlicensed/WiFi) are combined to form a seamless pool of flexible capacity. Because the cells are location aware, they can be used to deliver
  4. 4. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 very targeted services, which makes them valuable for supporting IoT, personalization and specialized MVNOs. Key benefits of small cells, such as lower cost of data delivery and location awareness, require density and operators are already talking about hyper-density as a way to deliver even greater benefits and ROI in future. ie the benefits scale at least linearly with the number of cells per square km, if not better in some scenarios. So while most operators have tended to deploy relatively high power public access small cells first, mainly for coverage, most acknowledge that the returns will be maximized by greater density, which will mean far cheaper, smaller and lower powered mini-base stations, as well as important changes to the way these cells are managed. Qualcomm, Airspan and Sprint demonstrated a ‘hyper-dense’ network at a Nascar race track, looking towards densities of 1,000 small cells per square kilometer, but we believe a more realistic definition of hyper-density would be about 400 cells per square kilometer. Densifying all layers Adding dense capacity will not just occur in small cells but across the whole multilayered HetNet. It is increasingly clear that the maximum benefits of small cells are achieved when they work in harmony with an enhanced macro layer. This may undergo significant changes of its own, falling into two broad categories. Figure 2. Numbers of MIMO antennas and small cells deployed 2013-2018. Source: Maravedis-Rethink RAN Service 2014 0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000 6,000,000 7,000,000 8,000,000 2013 2014 2015 2016 2017 2018 units MIMO Small cells
  5. 5. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 One is labeled ‘SuperMacro’ and relies on various ways to improve the cost and power efficiency of the macrosite, and its capacity, with MIMO antenna arrays a particularly important tool. Some vendors are talking about ‘massive MIMO’ technologies. The other focuses on deconstructing the RAN, by virtualizing the digital processing in shared platforms or in the cloud – the most extreme is Cloud-RAN, where the network processing is entirely centralized, leaving just a very low cost antenna/radio unit at the cell site. These cell sites may be almost as small in radius as those powered by an actual small cell, the main difference lying in the functionality. The small cell is a single autonomous unit with baseband integrated and will have greater local intelligence, in many cases, than a centralized RAN cell site, to support functions like data caching and presence-based services. It will often have its own connection to the cloud (many operators are looking to inhouse or outsourced cloud services to manage their rising number of small cells and support functions like SON). Operators are divided on whether to densify the macro layer first, before embarking on small cells, but eventually, nearly all agree that maximum efficiency will depend on strong interoperability between both layers, so that resources are shared and flexibly allocated where they are most needed. Efficient inter-layer operations can minimize interference and improve power consumption, and the coverage and full mobility of the macro layer can complement the targeted capacity of the small cells. TDD (unpaired) spectrum is particularly important in this scenario, since it allows the use of excess antennas to reduce inter-tier and intra-tier interference. While both layers will be progressively densified, the immaturity of ‘massive MIMO’ is currently even higher than that of hyperdense small cells, so we expect progress to be similarly gradual in the first 2-3 years. Figure 2 shows the increase in deployment of the two technologies, with a significantly higher curve for small cells, reaching a cumulative base of 7.5m by the end of the study period.
  6. 6. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 Critical enablers Figure 3 shows that in 2013-2014, most cells are still deployed in low or medium density scenarios (90% and 88% respectively) but from 2015 there is a shift towards higher density roll-outs. By 2020 40% are in hyperdense networks and 34% in high density. (Hyperdense = 400 cells/sq km or 80 per macro; Dense = 140; Moderate = 45; Non-dense = 10). Figure 3. Number of public access small cells deployed in various levels of density (definitions of density above). To achieve these projected levels, however, several key enablers still need to be in place, which is why progress is initially gradual. So far, there is little sign of commercial densification. Early public access small cell deployments have generally been characterized by at least one of the following features;  Small scale – 2-3 cells per macro in limited areas such as urban hotzones or business parks  Geared to filling coverage black spots, not to capacity  Pre-commercial  Carrier WiFi-only, with limited integration into the mobile core  A few larger projects have been heavily proprietary and hand-crafted for the carrier (eg SK Telecom) Only a few carriers can afford the hand-crafted approach, and the targeted benefits of small cells are only fully achievable with greater density, and a shift to high capacity rather than coverage. Public access small cells have been rolled out more slowly than expected to date, though carriers’ capacity and cost pressures will accelerate the pace from 2014. Reasons for the delayed activity, outside a few flagship markets like Japan, include (according to the operator survey): 0 5,000,000 10,000,000 15,000,000 20,000,000 25,000,000 30,000,000 2013 2014 2015 2016 2017 2018 2019 2020 noof smallcells No in hyperdense No in dense No in medium dense No in non-dense Total
  7. 7. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014  In many areas, coverage has been the first priority, requiring fewer cells than high capacity  The technology remains immature and key standards have not been agreed especially in areas like SON  Many operators are deterred by site acquisition and backhaul challenges and are looking for third party services to emerge to address these  Equipment prices need to come down further to justify dense deployment. At current levels, a new macro base station can sometimes be more cost effective than deploying four small cells or more in the same area, yet many operators see optimal benefits only when they can deploy densely  LTE and multimode small cells have been slower to get to market than expected and operators do not want to make investments in dense 3G only to have to upgrade quickly  There has been a shift of short term attention to WiFi offload to solve short term capacity problems, and some carriers are waiting for integrated WiFi to improve the small cell business case There are well known barriers to small cell deployments of any scale, notably backhaul and site issues, cost, scalability and availability of devices if new spectrum is used. With density, other barriers become important – the challenges of management and of BSS/OSS/analytics; the threat of chaotic capacity; and with the inclusion of homespots and WiFi, the reduction in cells which are fully under the operator’s control, with potential impact on quality of service. Therefore, the level of density indicated in Figure 3 will only happen if certain factors are achieved:  Full small cell SON capable of automating nearly all the functions.  MOCN for multi-operator support  Full WiFi/cellular integration at management/mobile core levels, and probably cell levels  Automated or streamlined processes for securing sites and backhaul Additional enablers will need to be in place to allow for hyper-density:  Support for fully dynamic provisioning and viral networks  Use of high frequency spectrum to support such large numbers of cells (3.5GHz, 60GHz etc).  Next gen SON and true plug and play  Next gen interference management, eICIC  Dynamic coordination  Cognitive radio
  8. 8. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014  Ad hoc viral networks and opportunistic cells  Multiflow  ‘Indoor out’, adding indoor LTE homespots (residential private access points with a portion of capacity left open for passers-by) The criticality of SON Some of the above enhancements are nearly available in commercial products and others will not appear until another generation of equipment has gone by. In our opinion, automation elements – in particular SON and zero touch provisioning/management – are the most critical to the dense network being affordable and efficient, with interference minimized. SON is one of the technologies which operators hope will make their mobile broadband investments commercially viable. By automating many planning and management tasks, it can speed time to market and save on the operating costs of manual processes. This is especially critical when carriers start deploying large numbers of small cells, in order to ensure they deliver targeted capacity, rather than radio chaos, and in challenging cost and time parameters. To read some discussions, it could be imagined that SON was already a widely deployed technology, but in fact it is in its infancy. There has been some activity, mainly in 3G – for which it was not originally designed – but while nearly all cellcos have a high level of interest, widespread commercial implementation for 4G or small cells remains weighted to 2015 and later. This is because there are many questions still to be answered, about which of the 30+ use cases will contribute to the operator’s business model, and which architecture will best support those cases. The uncertainties are even greater when it comes to small cells, and indeed, the wait for fully mature SON tools is one of the reasons that some carriers have delayed mass roll-out. Despite challenges of defining standards and use cases sufficiently quickly to meet market needs, there is high motivation for operators and their suppliers to deliver fully fledged SON tools. The impact on the business case, especially on operating costs, can be significant. Automated, closed-loop SON processes are increasingly seen as necessary because:  Networks are becoming more complex, with far more cells to manage, and this stretches the capabilities of existing manual processes and tools  SON is a key enabler of large-scale small cell networks, since unless these are automated, the cost of ownership is too high to make such large numbers of base stations viable.  As users consume more data without significantly increased ARPUs, the cost of running the network must be driven down, and automation contributes to this  In a data-driven mobile networks, patterns of traffic and user behavior are far more unpredictable than in a voice-led system, so the network needs to be able to adapt dynamically to high peak loads, without having to provision every cell for the maximum  When data volumes are high, there is far greater need to optimize the network to avoid outages or poor customer experience
  9. 9. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 This last point is the key to interest in SON. While automated processes are important to reducing capex and opex, and so improving the mobile data profit model, the operators are also competing fiercely based on quality of experience for users, especially premium ones, and this is an important driver for small cells with their location and presence awareness and ability to track subscribers’ experience. If a network is well optimized, and adapts to changes such as time of day or a base station fault, the user experience will be better. When SON tools are integrated with small cells and location awareness, they can even help carriers to target optimal QoE at specific cells, to support high value customers. Dynamic load balancing and such tasks can only effectively be done with automation, since changes must be made constantly, whereas manual optimization was typically a multi-month exercise, and tasks like cell positioning might be done once a day. SON will never automate all the tasks of optimization but early adopters claim significant results – a typical result is to increase data throughput by about 10% because networks are better optimized. That, in turn, can delay further capex investment in additional capacity, as well as improving user satisfaction and saving on the operating costs of manual processes. As indicated above, the area where SON will be most essential – though currently least mature – will be in small cells. Figure 4 indicates how the fortunes, and benefits, of SON will be closely tied to mass roll-outs of these tiny base stations. Fig 4 Number of newly deployed cells managed by SON 2013-2018. Source: Maravedis- Rethink RAN Service SON survey 2014 Deployment of new macro base stations will slow as operators rely more on software upgrades and small cells, though the percentage of macrocells managed by SON will rise throughout the period and reach about 75% in 2017-2018. By contrast, almost all small cells will be managed by SON as 0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000 6,000,000 7,000,000 8,000,000 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000 2013 2014 2015 2016 2017 2018 noofsmallcells noofsites LTE 3G LTE-A Small cells
  10. 10. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 soon as the carrier moves beyond small trials. Deployment of SON-controlled small cells will enjoy CAGR of 182% in 2013-2018, compared to 55% for LTE macrocells. Conclusion The best way for operators to address the challenges of dramatically increasing mobile data capacity, while reducing TCO and power consumption, will be to densify their networks, preferably in two interworking layers, macro and small cell. The small cell layer will support WiFi integration and increased ability to harness high frequency and TDD spectrum. Its benefits – coverage, capacity, power efficiency, improved customer experience and personalization – tend to increase, linearly or even more, with the increase in density, even if Qualcomm’s vision of 1,000 cells in a square kilometer may be unachievable for most. Denser deployment will justify small cell roll-out far better than the current hotspot pattern, but it will not happen until certain key barriers are overcome – notably in the cost of the cells and the ability to manage them efficiently. Among all the technologies which will help to enable the ‘hyper-dense’ network later this decade, the most critical will be advanced SON.
  11. 11. Maravedis-Rethink RAN Service Maravedis-Rethink Copyright February 2014 RAN Research Service This note is part of the RAN research service and is not available for sale individually. Led by Caroline Gabriel, the RAN service covers the whole wireless access and packet core industry with a special focus on LTE-related technologies, including small cells, EPC, carrer WiFi, evolved packet core, and other emerging trends. The service provides in-depth operator strategy tracking, as well as vendor profiles and SWOT analysis over the whole RAN ecosystem, from chips to infrastructure to management systems. Forecasts and full industry reports are produced twice a year and are complemented by 18 research notes per year on important topical developments, as well as one-to-one analyst support. About the author: Caroline Gabriel has spent 25 years analyzing the technology market, starting as a journalist and then moving into research and consulting. She was European content director at VNU, then Europe’s largest technology publisher and co-founded Rethink Technology Research in 2000 to focus on emerging wireless platforms and operator business models. After Rethink combined its research offerings with those of Maravedis, she became the Research Director for the whole venture. In addition, she leads the RAN practice and the Wireless Watch weekly research newsletter. To subscribe to the service, please contact Adlane Fellah at or call + 1 305 865 1006 All data contained in this research material is proprietary to Maravedis-Rethink . and may not be distributed in either original or reproduced form to anyone outside the client’s internal organization within five years of the research material date without prior permission of Maravedis-Rethink. The research material contained herein is for individual use of the purchasing Licensee and may not be distributed to any other person or entity by such Licensee including, without limitation, to persons with the same corporate or other entity as such Licensee, without the express written permission of the Licensor. Disclaimer: Maravedis-Rethink makes no warranties express or implied as to the results to be obtained from use of this research material and makes no warranties expressed or implied of merchantability or fitness for a particular purpose. Maravedis-Rethink shall have no liability to the recipient of this research material or to any third party for any indirect, incidental, special or consequential damages arising out of use of this research material. Maravedis-Rethink Return Policy Downloaded or sent research materials in any format are not refundable, nor credited under any circumstances. It is the sole responsibility of the buyer to verify through the Table of Contents and the Executive Summary that the research material fits the buyer’s information needs.