The Vault

Street Light Small Cells
White Paper / Sep 2014

A REVOLUTION IN MOBILE OPERATOR NETWORK ECONOMICS

Data demand is predicted to maintain its steep growth curve for the indefinite future. This growth isn’t being fueled simply by adoption, so there’s no peak in sight. Status quo network planning, such as splitting a macro cellular network or offloading additional traffic to Wi-Fi, isn’t going to scale enough to deal with the impending crunch.

Street light small cells, which are analyzed in this white paper, can reduce network costs, allow for dramatic increases in density and address morphologies from dense urban through suburban, representing almost 80% of wireless demand. The concept is viable: backhaul technology is already mature enough to support deployment, it permits blanket agreements with municipalities and power companies, and, crucially, it’s already been proven in a deployment constructed years ahead of its time.

This paper dives into mobile data demand predictions through 2020, the architecture and economics of a street light small cell network, the benefits of a hybrid solution leveraging macro cellular networks, and more.

We found some problems with the form:
To view this content please fill out the form below to tell us about you.
Privacy Policy
STREET LIGHT SMALL CELLS ? A REVOLUTION IN MOBILE OPERATOR NETWORK ECONOMICS How exponentially growing smartphone data traffic could lead to an unexpected architectural solution October 2014 Prepared by Signals Research Group Sponsored by InterDigital Signals Research Group recently completed a study of mass-deployed outdoor small cells. We call them street light small cells, even though they may not always be deployed on street lights. The concept demands that they be low cost, reasonably effective in offloading traffic from the macro network, and aesthetic ? blending into the urban landscape and barely noticeable to the average pedestrian. Finally, they must be deployed in large numbers to achieve (1) a decline in unit costs and (2) a steep reduction in the growth of macro cellular sites. The paper examines the economics of street light small cells in the context of other network technologies. It shows how the unit costs of a mass deployed street light small cell layer decline rapidly over time and compares those results to the economics of the macro cellular layer. As the sole authors of this paper, we stand fully behind the analyses and opinions that are presented in this paper. If you have questions feel free to contact J. Randolph Luening at . In addition to providing consulting services on wireless-related topics, Signals Research Group is the publisher of the Signals Ahead research newsletter (www.signalsresearch.com). www.signalsresearch.com STREET LIGHT SMALL CELLS ? A REVOLUTION IN MOBILE OPERATOR NETWORK ECONOMICS How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Page 3October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 1.0 Executive Summary Mobile operators have seen demand increase year after year for decades. However, the data growth that dominates the current era is qualitatively different than previous growth trajectories. It is not simply a matter of increasing adoption (a curve that eventually flattens) as in the voice era. Today?s data-oriented demand will grow at a steep compound annual rate for the indefinite future. Every mobile data forecast points in this direction. Operators currently have a limited toolkit: ?? Macro cellular cell splitting, which has resulted in a few percentage points of growth per year, is vastly inadequate. ?? Radio access technology improvements (the next release of LTE) and new spectrum, while both very helpful, will collectively absorb only a small percentage of the growth ?? Enterprise small cells/distributed antenna systems/public venue small cells, while all highly beneficial and generally cost-effective, provide relief only in very specific venues ?? Wi-Fi already carries most traffic from small devices. Can it reasonably offload a lot more? It is most effective in a small number of user-controlled venues (principally home and office). Operators need new tools if they are to scale their networks to carry the increased traffic. Fortu- nately, an economic cross-over is about to occur where another tool will become available and cost-effective. That tool is the mass-deployed street light small cell: ?? Street light small cells enable network cost reductions. A densely deployed layer of street light small cells can carry traffic at a lower cost than the macro cellular network. Street light small cells are expected to be 12% less expensive per GB ($1.02 vs. $1.16) than the macro cellular layer in 2020 with LTE traffic only, 36% less expensive per GB ($0.74 vs. $1.16) with LTE and Wi-Fi, and 57% less expensive per GB ($0.50 vs. $1.16) with LTE and Wi-Fi in a neutral host environment, where a single small cell layer carriers traffic for more than one operator. These are not theoretical numbers. They come from a detailed planning model reflecting carefully forecast demand for macro cellular LTE traffic and street-level (excluding residences and businesses) Wi-Fi traffic and historically achieved operator costs. ?? Street light small cells enable a dramatic increase in density. Mass deployed street light small cells deliver a dramatic increase in capacity relative to the macro cellular layer. There An economic cross-over is about to occur where mass- deployed street light small cells will become cost-effective. Key highlights from this study Mass deployed street light small cells have the ability to provide wide-area capacity at less than half the cost per GB of the macro cellular network. They can enable a dramatic increase in capacity and allow operators to support the anticipated 40% CAGR in data usage, while keeping macro cellular site growth to 5% to 6% per year through 2020. If deployed in the thousands per city, street light small cells can complement existing network elements, such as in-building small cells and macro cellular sites. However, they must be simple, inexpensive, self-configuring, wirelessly linked, nearly invisible, and ulti- mately deployed widely in densely populated areas. Page 4October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution is a 481 times increase in density (sites per square kilometer) between a 700 MHz suburban macro cellular site, designed for high quality in-building coverage, and a grid of street light small cells, covering a hexagonal area, with 300 meters of separation. Inexpensive small cells occupy less spectrum (2x20 MHz vs. 2x65 MHz) and experience greater adjacent-cell inter- ference. Despite this their usable capacity is still 75 times greater than that of a macro cell. In an urban environment the usable capacity is 2.5 to 9 times greater. In a dense urban envi- ronment macro cellular sites are closely spaced because the required margins for in-building coverage are high. As a result, the usable capacity from a street light small cell is closer to parity, ranging from 0.7 to 4 times that of a macro cellular site. ?? Street light small cells are deployable across a range of geographies. These include dense urban, urban, and suburban morphologies. These regions collectively represent 80% of wire- less demand and close to 80% of wireless network investment. A fully deployed LTE street light small cell layer is able to capture at least 37% of all 2G+3G+4G data traffic by 2018. The potential of street light small cells to slow macro cellular site growth is very important. ?? Backhaul technologies are sufficiently mature to enable the widespread deployment of inexpensive street light small cells - a self-organizing combination of fiber, coaxial, and wire- less links integrated into the small cell hardware. The 60 GHz V band, with several GHz of spectrum, is particularly appealing. ?? Blanket agreements with municipalities and utilities/power companies, combined with aesthetically pleasing and easy-to-install equipment, are key to the economics and public acceptance. ?? History shows that street light small cells do work ? One mass deployment, years ahead of its time, covering 20% of the US population, and other smaller deployments demonstrate the deployment processes and the legal/contractual frameworks that make mass deploy- ment possible. We will discuss how these points of reference establish a roadmap for the modern mobile operator. We?ve taken the many pieces of this technological and economic mosaic and asked ?If not now, when?? and ?If not street light small cells, then what?? How else will operators address the systematically increasing demand for data? Is there a way to deliver similarly geographically distributed capacity using some other technology? What is the likelihood that operators would meet their numerical deployment objectives using a conven- tional toolkit? Our analysis suggests that the likelihood of success using incremental technology improvements + newly auctioned spectrum + cell splitting is low because the expected growth in demand is so high (we?ll quantify it in a few pages). Subtle advantages of street light small cells include: ?? Street light small cells can leverage underutilized spectrum (2500 MHz and the new ASA bands) or reuse mainstream spectrum ?? Carrier aggregation, in particular dual connectivity, and Enhanced Inter-Cell Interference Coordination (eICIC) will enable creative mingling of outdoor small cell and macro cellular systems ? optimizing the utility of both ?? Street-level structures are plentiful. A study of street lights in San Francisco shows that actual street lights outnumber the required number of sites in a dense 300 meter street light small deployment by a factor of 7.6 ? making site selection a breeze Wireless backhaul technologies are sufficiently mature to enable the widespread deployment of inexpensive self-organizing street light small cells Street lights outnumber the required number of street light small cells by a factor of 7.6 to one. Page 5October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution ?? Self-organizing networks (SON) will enable operators to quickly deploy large numbers of small cells with a small fraction of the effort of deploying a traditionally engineered network ?? Inexpensive millimeter wave wireless backhaul will enable significant operating expense savings and will provide a high degree of resiliency ?? Self-configuring backhaul technologies will lower installation costs and allow network extensions for multiple small cells from a single leased connection In countries with limited spectrum available for LTE the need for outdoor small cell capacity (3G and/or 4G) may be even greater. This whitepaper discusses: ?? The exponentially increasing demand for mobile data through 2020 ?? The inability of ?status quo? network planning to meet that demand ?? The architecture of a mass-deployed street light small cell network ?? The essential enabling technologies in a mass-deployable street light small cell ?? The economics of mass-deployed street light small cell network ?? The rationale for a multi-layered network (macro cellular + in-building + public venue + street light small cell) and why one technology is not enough ?? The economics of a combined macro cellular/street light network and why operators should embrace such a hybrid solution ?? The economics of a neutral host that builds the small cell network and sells capacity to multiple operators and why this business model deserves consideration ?? The key operational issues (municipal right of way, data connectivity, power) and why none of these are unsolved problems Page 6October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Table of Contents 1.0 Executive Summary ?????????????????????????????????? 3 2.0 Introduction ????????????????????????????????????? 9 3.0 The Mobile Operator?s Journey ???????????????????????????? 11 3.1 Operating on Auto-Pilot ??????????????????????????????? 11 3.2 Logistical Feasibility ???????????????????????????????? 14 3.3 In-Building Solutions ????????????????????????????????15 3.4 Surgical Deployment of Small Cells ?????????????????????????15 4.0 A World with Street Light Small Cells ???????????????????????? 16 4.1 Small, Inexpensive, Invisible ???????????????????????????? 16 4.2 Layers of the Network ????????????????????????????????17 4.3 Elevation: Many Implications ??????????????????????????? 19 4.4 Metro Area Wi-Fi ????????????????????????????????? 19 4.5 Economics ????????????????????????????????????? 20 5.0 Deployment Decisions ??????????????????????????????? 22 5.1 Street Light Architecture ????????????????????????????? 22 5.2 LPA Power, Device Cost ?????????????????????????????? 23 5.3 Enabling Technologies ??????????????????????????????? 24 5.4 Why 60 GHz Millimeter Wave? ?????????????????????????? 25 5.5 Key Operating Expenses ?????????????????????????????? 26 5.6 Site Density ???????????????????????????????????? 27 5.7 Which Frequency Band? ?????????????????????????????? 28 5.8 Voice Traffic ???????????????????????????????????? 28 5.9 Deployment Geometry (Linear vs. Area) ?????????????????????? 28 5.10 Key Assumptions ????????????????????????????????? 29 5.11 Utilization and Economics ????????????????????????????? 30 6.0 Neutral Host ???????????????????????????????????? 34 6.1 Business Model ?????????????????????????????????? 34 6.2 Economics ????????????????????????????????????? 36 7.0 Technology and Incremental Revenue ??????????????????????? 37 7.1 Next Generation Wi-Fi (802.11ac) and WiGig (802.11ad) ??????????????? 37 7.2 Municipal Services ????????????????????????????????? 37 7.3 Alternative Architectures ????????????????????????????? 38 8.0 Conclusions ????????????????????????????????????? 39 Appendix I: Historical Networks ????????????????????????????? 41 Back to the Future: Metricom Ricochet ?????????????????????? 41 Google: Free Metro Wi-Fi, an Experiment ????????????????????? 46 Appendix II: Enabling Technologies ??????????????????????????? 50 Fiber and Coaxial Cables ?????????????????????????????? 50 Poles and Lights ?????????????????????????????????? 53 Financial Assumptions ??????????????????????????????? 57 Page 7October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Appendix III: Revolutionary Visions ??????????????????????????? 59 The Concept of ?1000x? ?????????????????????????????? 59 Inside Out ????????????????????????????????????? 59 Invasion of the DVRs ???????????????????????????????? 59 Appendix IV: Sensitivity Analysis ???????????????????????????? 62 LTE Only, Major Operator ????????????????????????????? 62 LTE + Wi-Fi, Major Operator ???????????????????????????? 64 LTE + Wi-Fi, Neutral Host ????????????????????????????? 65 Notes ?????????????????????????????????????????? 66 Page 8October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Index of Figures Figure 1. Data Demand per Device ?????????????????????????????? 11 Figure 2. Macro Sites Required with and without Street Light Small Cells??????????? 13 Figure 3. Street Light Small Cells Absorb Macro Cellular Traffic ????????????????18 Figure 4. Macro Cellular and Street Light Economics (LTE + Wi-Fi) ????????????? 20 Figure 5. Linear Street Light Deployment (Dense Urban and Urban) ????????????? 22 Figure 6. Area Street Light Deployment (Suburban) ?????????????????????23 Figure 7. Utilization by Morphology, LTE Only ??????????????????????? 30 Figure 8. Economics of Street Light Small Cells, LTE Only ?????????????????? 31 Figure 9. Economics of Street Light Small Cells, LTE + Wi-Fi ??????????????????32 Figure 10. Traffic by Network Layer, Including Wi-Fi ?????????????????????33 Figure 11. Economics of Street Light Small Cells, Neutral Host with LTE + Wi-Fi ??????? 36 Figure 12. Metricom Street Light Small Cell (Circa 2001) ?????????????????? 42 Figure 13. Street Light Small Cell Density ?????????????????????????? 44 Figure 14. Capital Investment per Square Mile ??????????????????????? 45 Figure 15. Coverage of Mountain View Wi-Fi Network ??????????????????? 46 Figure 16. Outdoor Usage vs. Residential Usage in 2014 ??????????????????? 47 Figure 17. Comcast Hybrid Fiber Coax Network, San Francisco ??????????????? 50 Figure 18. Typical Cable Wiring ????????????????????????????????51 Figure 19. Street Lights in San Francisco ???????????????????????????53 Figure 20. Mid-Town Manhattan ?????????????????????????????? 54 Figure 21. Neighborhood in the Bronx ????????????????????????????55 Figure 22. Major Operator, LTE Only, Sensitivities ????????????????????? 63 Figure 23. Major Operator, LTE+ Wi-Fi, Sensitivities ????????????????????? 64 Figure 24. Neutral Host, LTE + Wi-Fi, Sensitivities ?????????????????????? 65 Index of Tables Table 1. Spectrum Allocation of a Typical Tier 1 Operator ??????????????????12 Table 2. Distribution of LTE Traffic ?????????????????????????????? 17 Table 3. Street Light Small Cell Deployment Assumptions ????????????????? 29 Table 4. Share of Traffic for Major Operator and Neutral Host Scenarios ???????????35 Table 5. Metricom Street Light Small Cells by City ????????????????????? 43 Table 6. Pole Costs Reported to FCC ???????????????????????????? 56 Page 9October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 2.0 Introduction Smartphones have largely replaced feature phones in most developed countries. Smartphone displays ? measured in pixels ? resemble laptop displays and in many cases surpass them. Mobile devices represent a rapidly increasing percentage of internet page views. Mobile data is growing exponentially. Unlike the growth in voice that grew rapidly over a generation and then flattened as the market saturated, mobile data is likely to continue growing indefinitely. It is a function not of adoption but of silicon economics. As processing power, storage, and display capabilities increase, so will the desire to transmit greater amounts of data. Operators in the US have enjoyed a rare and refreshing season of relatively easy gains in capacity. LTE has been a tremendous success. The 700 MHz band has permitted rapid deployment in the United States of this next generation of technology. It happened quickly in the existing macro cellular footprint thanks to the excellent coverage of a new sub-GHz spectrum band. Other frequency bands have also become available. The flexible spectrum doctrine of the FCC gives operators the freedom to use spectrum in a nearly optimal fashion. In the next few years the sizable inventory of spectrum controlled by each major operator (~100 MHz in major cities) will increasingly transition to LTE. One could be lulled into believing that there is no capacity crunch, despite rapidly increasing levels of data usage. An appropriate analogy might be a tropical depression. It is barely noticeable to meteorologists as it is developing at sea. It quietly grows in strength then eventually hits land ? at which time its force is fully evident. The spectacular gains in network capacity that result from new technologies (e.g. LTE) and new spectrum are discontinuous, not recurring, events. Operators have enjoyed them, but they are not sustainable or ongoing. Macro cellular site growth today is modest: 5% to 6% per year for the two largest operators. Spectrum growth is also modest. We estimate 4% to 5% per year, from 2013 to 2020. Importantly, it arrives in unpredictable waves (new allocations followed by auctions). In-building solutions (DAS or small cell) are powerful, if you have a large office building or a large public venue. What about the vast portion of the population that lives and works in ?small? venues, ones that are unlikely to attract the attention of an in-building infrastructure team? What happens after operators have fully transitioned their networks to LTE and have filled newly purchased bands of spectrum and have deployed in-building solutions in all large venues? Some observers speculate that mobile operators will cease to be profitable, as data usage increases, but as revenues remain flat. Others suggest that data usage will flatten or that virtually all data usage will be offloaded to Wi-Fi, making mobile operators less relevant. Some believe that usage- based bundles will increase price awareness and moderate the growth of macro cellular traffic. Another possibility ? which we will explore today ? is that radical new architectures will emerge that will enable operator cost curves to continue to decline. These radical new architectures will provide the right amount of capacity in the right location at the right time in history. We will consider the concept of mass-deployed outdoor small cells. Let?s call them ?street light? small cells. They won?t necessarily be deployed only on street lights. The concept demands that they must be low cost, reasonably effective in offloading traffic from the macro cellular network, and aesthetic. They must blend into the landscape in such a way that they are barely noticed by the average consumer ? or else they will face resistance. Finally, they must be deployed in large Mobile data usage is likely to continue growing exponentially. In the next few years the sizable inventory of spectrum controlled by each major operator will transition to LTE. Macro cellular site growth is modest, 5% to 6% per year for the two largest operators. It is possible that radical new architectures will emerge that enable operator cost curves to continue to decline Page 10October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution numbers. Scale is essential for two reasons: (1) to achieve the desired price points and (2) to be effective in slowing the growth of the macro cellular network. Street light small cells are important as a concept because they represent one of the few archi- tectural revolutions that could be deployed in large geographic areas and could ? for five or ten years ? offset the explosive growth of mobile data, and do so in a way that is cost effective for the operator. We will briefly consider other complementary concepts to meet the projections of rapidly growing data traffic. The concepts include existing technologies, such as in-building small cells/ DAS systems, and the disruptive paradigms discussed in Appendix III. Our view of the future ? with or without streetlight small cells ? includes multiple layers of radio access technologies, many of which are venue-specific. The story we are about to tell is set in the United States. It will also play out in London, Dubai, Beijing, Caracas, and other large cities worldwide. The timing will be different and the precise factors driving change (spectrum, demand, and real estate constraints) will be local. We will start with this one geography because ? as the reader will soon discover ? there are lots of numbers and assumptions associated with the story. Page 11October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 3.0 The Mobile Operator?s Journey 3.1 Operating on Auto-Pilot Absent any architectural innovation, mobile operators in the US will experience an increasing need for capacity beginning in 2016. At roughly that point in time the windfall of capacity from LTE will have been realized and the networks leveraging the spectrum purchases of the last decade will be largely operational. Operators will have newly acquired spectrum (600 MHz and/or 1700 MHz) but, if history is an indicator, the process of clearing incumbents, specifying devices, and readying the band for operation may still be under way, especially in the 600 MHz band. If we carefully analyze consumer demand, drawing from the best industry sources1, we see the numbers in Figure 1. Not only is demand per device increasing but the number of devices is also increasing. A growing number of tablets will have mobile broadband connectivity. Many other devices: cameras, wear- able items, machines, and automobiles ? all unconventional ?subscribers? ? will connect to the mobile network. The portion of usage offloaded to Wi-Fi will increase slightly, but will be over- whelmed by the shear growth in demand. Our demand model (derived from sources noted above) shows a 40% compound annual growth in mobile data (excluding traffic offloaded to Wi-Fi) in the United States from 2013 to 2020. This is slower that the 60% CAGR anticipated on a world- wide basis but still extremely fast ? placing high expectations on every mobile operator. Many new categories of devices: cameras, wearable items, machines, and automobiles ? all unconventional ?subscribers? ? will connect to the mobile network. 0 2,000 4,000 6,000 8,000 10,000 COMPOSITE Mobile PC/Router/Tablet Smart Phones 4G Smart Phones Non 4G Feature Phones 202020192018201720162015201420132012 M Bs / D ev ic e/ M on th Figure 1. Data Demand per Device Source: Signals Research Group Page 12October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution During the same period operators will aggressively refarm spectrum from legacy technologies (2G, 3G) to LTE. LTE itself will become more capable, partly as a result of network innovations and partly as a result of device chipset improvements. Finally, operators, will acquire some addi- tional spectrum. Table 1 shows the spectrum holdings of a typical ?tier one? operator increasing from 100 MHz to 130 MHz by 2018. A tier one operator, which we define as an average of today?s two largest US operators, in the absence of any innovative small cell technologies, would need to acquire new macro cellular sites at an aggressive rate. Its site count would increase from 55,000 macro cellular sites at the end of 2013 to 102,094 macro cellular sites by the end of 2020. This site requirement reflects the commitments large operators have made to keep legacy network technologies ?alive.? It also reflects a realistic picture of the speed of device migration. Finally, it allows time for clearing incumbents out of newly acquired spectrum and readying that spectrum for service. An aggressive spectrum acquisition plan could provide additional breathing room. To achieve the required growth rate in macro cellular sites in any particular year the operator would need to begin its site acquisition process 18?24 months in advance. The required site count might ultimately be greater than the number calculated because it is likely that the new sites would not be as productive as existing sites because their locations would be less ideal than those selected decades ago. Mobile data (excluding traffic offloaded to Wi-Fi) will grow 40% per year in the United States from 2013 to 2020 Radio Layer 2013 2014 2015 2016 2017 2018 2019 2020 2G (GSM/GPRD/EDGE) 5 MHz 5 MHz 5 MHz 5 MHz 5 MHz 5 MHz 5 MHz 5 MHz 3G (WCDMA/HSPA+) 50 MHz 40 MHz 30 MHz 30 MHz 20 MHz 20 MHz 20 MHz 20 MHz 4G (LTE) 45 MHz 55 MHz 65 MHz 65 MHz 85 MHz 105 MHz 105 MHz 105 MHz TOTAL Spectrum 100 MHz 100 MHz 100 MHz 100 MHz 110 MHz 130 MHz 130 MHz 130 MHz Table 1. Spectrum Allocation of a Typical Tier 1 Operator Source: Signals Research Group Page 13October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Figure 2 shows the number of required macro cellular sites in each year. This is a calculated result based on a number of inputs: expected voice and data demand, expected improvements in LTE technology, expected increases in available spectrum, and expected refarming of spectrum by the savvy operator. The curve is lumpy because the many activities described above are happening concurrently. The growth from 55,000 sites to 102,094 sites represents a compound annual growth rate of 9.24% per year. A tier one operator?s macro site count will increase from 55,000 in 2013 to 102,094 in 2020, absent any innovative new architectures 40,000 60,000 50,000 70,000 80,000 90,000 100,000 120,000 110,000 55,000 102,094 77,736 20202019201820172016201520142013 Dense Urban In-Building Small Cell Deployments 600 MHz and 1700 MHz Spectrum Won in Auctions then Deployed, Temporarily Relieving Capacity Crunch Spectrum Refarmed Each Year from Legacy Technologies (2G/3G) to LTE Intitial Rollout of Street Light Small Cells Complete Rollout of Street Light Small Cells (DU, U, and Dense S) N um be r of S it es R eq ui re d fo r a M aj or O pe ra to r N at io na l N et w or k Street Light Small Cells Absorb Increasing Amounts of Traffic As Usage Grows Macro Cellular Sites Required for Macro + In-Building + Street Light Network Macro Cellular Sites Required for Macro Only (WCDMA/HSPA+ LTE) Network Figure 2. Macro Sites Required with and without Street Light Small Cells Source: Signals Research Group Page 14October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution This growth rate, while large, is not as large as it might be. Data demand (demand per device times the number of devices) will grow at a compound annual rate 40% per year from 2013 to 2020. Other factors (new spectrum, spectrum refarming, technology improvements, and the pres- ence of voice, which is not growing significantly) prevent the macro cellular site requirements from growing at a commensurate rate. 3.2 Logistical Feasibility Will operators actually construct this number of sites? The simple answer is no. First, it would be logistically difficult to do so. Over a recent 13-month period the two largest operators in the United States each increased their site count, on average, by 2,965 sites per year2. If our hypothetical tier one operator had 55,000 sites at the end of 2013 then this would represent an increase of 5.39% per year. The growth rate of 9.24% shown in Figure 2 would exceed the rate of construction any large operator has achieved in recent years. What then will real operators do? They will almost certainly seek to leverage small cell technology. Operators are already investing in in-building solutions (small cells and distributed antenna systems). These are very effective solutions in dense urban environments, where skyscrapers dominate the landscape. In-building systems are also present in urban areas, where large indi- vidual buildings exist. In smaller venues (your local dry cleaner or caf?) small cells could work if someone wanted to pay for them to be deployed. Will mobile operators rush in to equip your local dry cleaner with an operator-provided in-building small cell system? Probably not. In the United States 5.8% of the population lives in dense urban areas (think ?Manhattan?). Another 15.1% lives in urban areas (think ?tightly packed three to five story buildings?). The rest live in suburban, rural cluttered, or rural open areas. We glean these ratios from our geo-coded database that characterizes land mass as one of five morphologies (dense urban, urban, suburban, rural clutter, or rural open). Will extensive deployment of in-building small cells in dense urban environment solve the problem? If not, what other tools exist? Operators are experimenting with surgically deployed outdoor small cells. The problem is these small cells are few in number. While they may be indi- vidually effective, their impact on the operator?s total capacity is modest. In this paper we will explore the possibility of mass-deployed street light small cells. If an oper- ator were to deploy street light small cells in a systematic way (which we will describe shortly) in dense urban, urban, and dense suburban areas along with in-building small cells in the very specific environments in which they are effective then the required number of macro cellular sites in 2020 would be reduced from 102,094 (a macro only world) to 77,736 (a macro + in-building + street light small cell world). If a tier one operator has 55,000 sites at the end of 2013 increasing to 77,736 sites in 2020 this represents a compound annual growth of 5.07%. This is a rapid but achievable growth rate. It is in line with what operators have accomplished in recent years. The dotted line in Figure 2 reflects the impact of an aggressive deployment of small cells ? most importantly mass-deployed street light small cells. In the remainder of this whitepaper we will describe street light small cells and their economics. We will make the case that the operators and manufacturers should consider this ?radical? architecture along with other established capacity- enhancing solutions. Will operators actually construct this number of sites? The simple answer is ?no? because it will be logistically difficult to do so. Page 15October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 3.3 In-Building Solutions Operators will deploy multi-technology in-building solutions in dense urban environments where skyscrapers allow them to access large physical spaces and large numbers of people in single in-building solutions. Operators will also aggressively deploy in-building solutions in public spaces (shopping malls, airports, sports stadiums, concert venues, and train and subway stations) in urban and suburban areas. The impact of public space small cells on network capacity and network economics will be constrained by the size and number of such public spaces in each geographic area. 3.4 Surgical Deployment of Small Cells Operators will also surgically deploy outdoor small cells. These early deployments will address capacity hot spots. While labeled ?small cells? their architecture and price point in many cases resembles that of a small macro cell. Since the operator is placing them in critical locations the operator is willing to spend money to acquire sites and provide connectivity. These surgically placed small cells serve a need and enable learning. While a surgically deployed small cell may carry a significant amount of traffic in a particular venue, the impact of surgically placed small cells as a ?layer? of technology will be small because these small cells will exist in very small numbers relative to other technologies, such as macro cellular cells or in-building small cells. Operators will surgically deploy outdoor small cells to address capacity hot spots. The problem is these cells are few in number. Page 16October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 4.0 A World with Street Light Small Cells Let?s leap to the future and envision a world with widely deployed outdoor small cells. It?s not a large leap because the enabling technologies all exist today. The enablers include ruggedized outdoor small cells, wired and wireless backhaul technologies, software for self-organizing networks (SON), etc. Appendix I describes past large-scale street light small cells using other flavors of radio access. The LTE small cells with integrated backhaul needed to realize the concept have not been produced in volume. The ?gap? therefore is in cost-optimization and high volume/ low cost production ? activities that typically occur with manufacturers once a sizable market is clearly visible. By ?future? we mean 2016, 2017, and 2018 ? well within the planning horizon of any large organization. We will describe the end state then backtrack, examining the enabling technologies that make this possible and why we believe the timing could be soon. 4.1 Small, Inexpensive, Invisible An effective mass-market outdoor small cell would need to be inexpensive (a few thousand dollars) to purchase and install. A key driver of cost is the linear power amplifier (LPA) power. Most operators believe that 1?5 watts is appropriate. If the small cell has too little power then its effective coverage area will be limited. If the small cell has too much power it becomes expensive and ? arguably ? produces an unnecessary level of interference. We have also assumed that the street light small cell supports a single cellular radio access tech- nology ? LTE ? and that it operates in a single band. A street light small cell could ? in theory ? support multiple radio access technologies and multiple bands. If it did, it would begin to resemble a small macro site. Its price point would increase and its form factor would also grow. It might no longer be ?largely invisible? to the average city dweller. The street light small cell will probably include other non-cellular technologies, such as Wi-Fi. These are inexpensive, primarily because consumers expect ?opportunistic? Wi-Fi service: if they can see the signal then they will attach, but they do not expect a ubiquitous blanket of service, as they do with LTE or other cellular technologies. Our analysis suggests that a network of relatively simple LTE plus Wi-Fi sites operating in a single cellular band can have a tremendous impact. The spectacular increase in spectrum reuse associated with street light small cells means a layer of very simple inexpensive cells will deliver far more capacity than the associated macro cellular layer. There is therefore very little benefit in adding complexity to the small cell. What limits the economic impact of a street light small cell is the number of subscribers it can attract (venue-specific coverage issues), not the capacity per street light small cell. There is a dramatic increase in overall network capacity when street light small cells are deployed on a large scale. Page 17October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 4.2 Layers of the Network Table 2 shows our hypothesis of the coverage of each ?layer? of the network in each morphology. The network layers include the macro cellular network, in-building solutions, and a street light small cell network. Our desire would be for the street light small cell network and in-building solutions to absorb 100% of the traffic. We know that that is unlikely because of the low power and low elevation of a street light small cell. We have, therefore, modeled the network on a very conservative set of assumptions. It is possible, when such a network is actually deployed, that street light small cells would be far more effective at reaching obscure areas that we envisioned. That would be an excellent outcome for operators because it would mean that we underestimated the potential of street light small cells. Since we assumed that street light small cells support a single technology (LTE) all traffic from non-LTE devices will need to be carried on the macro cellular network or by in-building solu- tions. This assumption has also been part of our modeling. Since a large portion of devices are LTE-capable today and nearly all new devices will be LTE-capable, this is not a significant constraint. It does underscore the need for the other network elements. A robust solution requires a macro cellular network for coverage and, to be competitive, significant in-building deployments, primarily in dense urban areas. These technologies (macro cellular, in-building, and street light small cells) are additive rather than competitive, because they deliver coverage and capacity in different places. Operators have invested heavily in the macro cellular network. It is designed for very high availability. It is also designed to provide robust coverage over all populated areas. In-building systems provide both coverage and capacity in an in-building environment. A strong in-building signal will most likely eclipse any signal from outside. Street light small cells provide capacity. They act like sponges, pulling in traffic that would otherwise end up on the macro cellular network. Macro cellular, in-building, and street light small cells are additive rather than competitive technologies, because they deliver coverage and capacity in different places Table 2. Distribution of LTE Traffic Source: Signals Research Group Morphology Street Light Small Cells In-Building Small Cells Macro Network Dense Urban (DU) ? 15% ? Limited impact due to difficulty in reaching upper floors ? 40% ? In-building solutions are very cost effective in a high-rise ? Operators are aggressively deploying such systems ? 45% remains on macro cellular network Urban (U) ? 50% ? Effective due to low height of buildings ? 10% ? Some in-building systems, especially in public venues, but most buildings are too small to be cost effective for the operator ? 40% remains on macro cellular network Suburban (S) ? 50% ? 100% effective where small cells are deployed, due to low building heights and open spaces ? Small cells are deployed in 30% of the area to capture 50% of the traffic ? 10% ? Some in-building systems, especially in public venues, but most buildings are too small to be cost effective for the operator ? 40% remains on macro cellular network Rural (R) and Rural Open (RO) ? 0% ? Not used in any meaningful fashion, due to low population density ? 0% ? Not used in any meaningful fashion, due to low population density ? 100% remains on macro cellular network Page 18October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Street light small cells can be designed with slightly less robust specifications (number of technol- ogies, number of frequency bands, backhaul SLA, hours of battery back-up (if any, etc.) because if a single street light small cell fails traffic will revert to the macro cellular network. The advantage of street light small cells is their proximity to the subscriber, which enables complex modulation techniques and access speeds that are possible only in an excellent coverage environ- ment. There is also the possibility of using street light small cells in combination with macro cellular sites (dual connectivity), where the small cell could use its own radio downlink and a macro cellular site?s downlink in another band. We have not modeled this scenario, because it is more complex and because a simple stand-alone street light small cell has abundant capacity by itself. Once a network of street light small cells is in place such functionality is possible. The effectiveness of street light small cells, along with in-building small cells, in absorbing macro cellular data traffic can be seen in Figure 3. Street light small cells can be designed with less robust specifications than macro cells because if a single street light small cell fails traffic will revert to the macro cellular network 0 200 400 600 800 1,000 1,200 20202019201820172016201520142013 D at a Tr af fi c pe r M on th (m ill . G Bs ) Street Light Network In-Building Network Macro Network Figure 3. Street Light Small Cells Absorb Macro Cellular Traffic Source: Signals Research Group Page 19October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 4.3 Elevation: Many Implications The small cell will be deployed at a low elevation. We have assumed the average height of a street light is 28.5 feet, or 8.5 meters. This elevation is far below average building height in dense urban areas, below average building height in urban areas, and slightly below average building height in suburban areas. A macro cellular site, in contrast, is normally deployed at or above the average building height. Its radio signal skims the tops of buildings and reflections drop into the spaces in the interior of city blocks. In contrast, a street light small cell in an urban environment faces a ?wall? of adjacent buildings. It provides excellent coverage on the street and excellent coverage of the street-facing portions of buildings. Its reach into the ?interior? of a city block is much less predictable. We have assumed, therefore, that street level small cells could capture 50% of the LTE traffic in their environment. Users in the interior of the city block (rooms within buildings that are far from the street with street light small cells or outdoor spaces behind buildings) might have a stronger signal from the macro cellular network. In a dense urban area (think skyscrapers), the story is quite different. Since a street light small cell is located on the street and has a minimally directional antenna it is extremely unlikely that its signal will illuminate the interior of an office space on the 35th floor. These subscribers, if they are fortunate, will be served by an in-building system (DAS or small cell). If they are less fortunate, they will be served by a macro cellular site. Studies of where consumers use data suggest that a large percentage of data usage ? 80% to 85% ? occurs indoors. This means that the remainder is outdoors. We have assumed, conservatively, that street light small cells in a dense urban area capture only 15% of the total LTE traffic. This doesn?t sound like much, but keep reading! In a suburban area the situation is different. Buildings are lower (typically two stories, not four or five) and there is physical space between buildings, allowing radio waves to propagate into back yards and other interior spaces. We have assumed, therefore, that a dense street light network in suburbia could provide consistent area coverage everywhere. This discussion of elevation provides the detailed and largely technical rationale for the ratios shown in Table 2 and some of the deployment assumptions described later in Table 3. 4.4 Metro Area Wi-Fi Finally, most proponents of small cells like to include Wi-Fi. The incremental cost is minimal and the benefit substantial. We have assumed that a street light small cell network would include both LTE and Wi-Fi. The Wi-Fi would be offered opportunistically. We have not attempted to estimate the coverage or capacity of a wide-area Wi-Fi network. The purpose of the Wi-Fi would be to cover people on the street and people in nearby retail establishments. People in their own homes or offices would probably have their own Wi-Fi network to which their own devices would attach. We have a demand model for street-level Wi-Fi which we will discuss shortly. Figure 4 compares the economics of a street light network to the economics of the associated macro cellular network. The traffic seen by the macro cellular network is reduced to reflect the presence of in-building small cells, public venue small cells, and the street light small cell network. The street light small cell network is shown as a ?layer? with its own economics. The composite unit cost is a weighted average of the two. The more aggressively an operator can deploy street light small cells, the greater the impact of overall cost reduction. A limiting factor, of course, is that street light small cells are cost effective It is extremely unlikely that a street light small cell will illuminate the interior of an office space on the 35th floor. In the suburbs buildings are lower and there is physical space between buildings, allowing radio waves to propagate into back yards The more aggressively one deploys street light small cells, the greater the overall cost reduction Page 20October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution only in certain morphologies (e.g. dense urban, urban, and dense suburban ? not rural). Conse- quently, some operator costs (e.g. the cost of coverage in rural areas) will not decline as a result of street light small cells. Also, street light small cells attract only a percentage of the traffic in each venue, as described in Table 2. Figure 3 shows an optimized solution reflecting all these elements. Street light small cells are deployed beginning in 2016, in those areas where they are most cost effective. The objectives of the deployment are to (1) minimize macro cellular site growth and (2) reduce overall unit costs (a composite of dense urban, urban, suburban, rural cluttered and rural open morphologies and two layers of the network: macro cellular and street light small cell). 4.5 Economics Figure 4 reflects the economics of a ?top two mobile operator? in the US with the associated coverage footprint. It includes all geographic areas, since the CFO is concerned about the overall capital and operating expense budget. If a technology is impactful only in niche deployment scenarios, then it is of less interest to the CFO. Conversely, if a technology can significantly impact the bottom line, it gets attention. Mass-deployed street light small cells are interesting because they can absorb a significant amount of traffic and thereby slow the growth of the macro cellular network. In-building solutions (e.g. small cells and DAS systems in skyscrapers) and small cells in public venues are also very interesting. We reflect the presence of in-building solutions in our estimate of macro cellular growth but have not explicitly included them in the calculation of unit costs. They could be the subject of another very sophisticated whitepaper. Mass-deployed street light small cells can absorb a significant amount of traffic and thereby slow the growth of the macro cellular network. $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 COMPOSITE Street Light Network Macro Network 20202019201820172016201520142013 C os t pe r G B ($ ) o f St re et L ig ht S m al l C el l N et w or k Figure 4. Macro Cellular and Street Light Economics (LTE + Wi-Fi) Source: Signals Research Group Page 21October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution In unit cost graphs we often see a declining cost curve, reflecting increased utilization (e.g. migra- tion from a coverage-constrained to a capacity-constrained network). In this case the decline is associated with (1) a migration of spectrum from legacy radio access technologies to LTE, (2) slight increases in spectrum reflecting planned auctions, (3) modest improvements in LTE performance, and (4) significant decreases in backhaul economics as fiber-connected macro cellular sites (with spectacular capacity) become more fully utilized. The powerful finding is that a network of street light small cells, intelligently deployed, offering both LTE and Wi-Fi (more on that later) delivers capacity at a cost below that of the macro cellular network. A street light small cell network, deployed at the right time (not too soon) and in the right loca- tions (this matters) reduces macro cellular growth to a moderate level and lowers the average cost per GB for LTE traffic. The macro cellular network still grows, because there are morphologies in which street light small cells are not deployed. In our model the macro cellular network delivers 45% of the LTE capacity in dense urban areas, 40% of the LTE capacity in urban areas, 40% of the LTE capacity in suburban areas and 100% of the LTE capacity in rural cluttered and rural open areas. The macro cellular network and in-building solutions also deliver 100% of the legacy technology (2G, 3G) capacity. But recall Figure 3, which shows the Street Light Small Cells carrying 37% of all data traffic by 2020. A network of street light small cells, intelligently deployed, offering both LTE and Wi-Fi, delivers capacity at a cost below that of the macro cellular network. Page 22October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution 5.0 Deployment Decisions 5.1 Street Light Architecture We assume street light small cells are deployed systematically to ensure consistent coverage. In dense urban and urban areas we assume a ?linear? coverage of every street. Th is results in a very dense grid. It also leverages two diff erent backhaul technologies. Adjacent sites connect via a wire- less radio link. One in ?N? sites is physically connected to the wired network. We have assumed very conservatively that N=3 and that the wired connection is a DOCSIS 3 cable connection. Th e millimeter wave backhaul architecture and the bandwidth of the cable connection (selected to support a maximum 150 Mbps data rate) can actually support much higher ?N? ratios. Financially, there is a diminishing return as one increases ?N? from ?3? to a greater number. Going from N=1 to N=3 is very important. Increasing N beyond 3 is benefi cial, but the marginal benefi t dimin- ishes as N continues to increase. It is likely that street light small cells with integrated wireless backhaul (the actual hardware) will require fewer than 1 in 3 sites to be connected to a wired backhaul. Figure 5 shows the geometry of a ?linear? deployment. It is likely that street light small cells with integrated wireless backhaul will support aggregation ratios much greater than N=3 60 GHz Wireless Backhaul + Cable 60 GHz Wireless Backhaul Inter-site distance: 300 meters Figure 5. Linear Street Light Deployment (Dense Urban and Urban) Source: Signals Research Group Page 23October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution In a suburban area we can deploy using a hexagonal grid. Building heights are low and there is space between buildings. Th is geometry refl ects the view that a street light small cell can cover a geographic area. An area deployment is shown in Figure 6. 5.2 LPA Power, Device Cost We have assumed that a single ergonomic ?package? will include a 2-watt LTE radio, Wi-Fi, and a 60 GHz bidirectional (a link in each direction) backhaul. It will be designed to quickly attach to the streetlight and will be painted to match its surroundings. Finally, it will include an integrated cable modem, using the current DOCSIS 3 standard, for easy connectivity to a cable network. We could include a fi ber termination. Th e current view is that a DOCSIS 3 connection is fast enough (supporting 600 Mbps in the downstream) and that a DOCSIS 3 connection is smaller (form factor), lighter (weight), and cheaper than a fi ber connection. Finally, it is easy for a cable operator to run a dedicated strand of coax to the site, eliminating any bandwidth conten- tion, if that is a concern. Some operators like the idea of a dark fi ber connection. We discuss this possibility in Appendix II. Th e unit will cost $6,000 installed. We assume $3,500 for the LTE portion, $500 for installation, and $2,000 for the backhaul and Wi-Fi. Th e numbers could be signifi cantly less. No such product exists today, at this price point, with this degree of integration, but all the pieces do, and a real product could exist by the time specifi ed for the fi rst deployment. We assume the unit costs $6,000 installed: $3,500 for LTE, $500 for installation, and $2,000 for the backhaul and Wi-Fi. 60 GHz Wireless Backhaul + Cable 60 GHz Wireless Backhaul Figure 6. Area Street Light Deployment (Suburban) Source: Signals Research Group Page 24October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 5.3 Enabling Technologies Integrated enabling technologies include an LTE radio, an 802.11/802.11ac/802.11ad radio, an electronically steerable 60 GHz wireless mesh backhaul, self-organizing network (SON) soft- ware, and a cable modem. ?? The system concept we are describing includes a handful of important enabling technologies: ?? An integrated LTE radio ?? An integrated Wi-Fi radio (probably including 802.11ac and 802.11ad) ?? A millimeter wave mesh backhaul ?? Self-organization network (SON) software ?? A DOCSIS 3 modem ?? A weatherized enclosure that is attractive and easy to attach to existing street lights Most infrastructure manufactures today have ?outdoor small cell? in their product portfolio. Most do not have a fully integrated solution ? as described in the paper ? so it is worthwhile stepping through the various enabling technologies and discussing what this means as a package. Wi-Fi is a very inexpensive addition to any outdoor product. Including it enables the small cell to serve not only LTE subscribers but other types of users. There are a number of 802.11 stan- dards (802.11ac and 802.11ad) that operate in different bands that will become important. These are discussed in Section 7.1. They provide a significant revenue upside and position the ?box? as something beneficial to consumers and to municipalities. A wireless backhaul radio is also important. We envision a self-organizing millimeter wave mesh radio with an electronically steerable antenna. This radio enables high data rate connectivity between nearby sites. Sub-6 GHz non-line-of-site radios are also widely used, although conges- tion in lower frequency bands is a potential issue. Having wireless connectivity means that every site does not need to have a dedicated ?wired? backhaul. The combination of mesh wireless and wired (notionally DOCSIS 3) leads to very favorable deployment economics. The ?mesh? and the ?integrated? parts are also important. ?Mesh? means that a radio can be dropped into a fabric of other radios and establish connectivity. This stands in contrast to a tradi- tional macro cellular architecture where microwave technicians install and point highly direc- tional antennas and create a very deliberate backhaul network topology. With street light small cells the vision is that the installation team knows (1) how to physically attach the box to the street light and (2) how to attach power and (3) how to attach cable. Cable is needed at one out of three sites. The small cell should do most things by itself. It should figure out where it is and should then be remotely managed by a network operations center. In most cases only one or two hops is required, so the latency is negligible. At millimeter wave frequen- cies the radio link can support data rates that exceed that of the wired (DOCSIS 3) connection. The backhaul radio needs to be self-configuring. Using an electronically steerable array enables it to figure out where it is, to ?find? the next nearest site in either direction, and to establish a radio connection. It also means that the integrated unit does not look like a radio. It is simply a color coordinated bump on an existing light fixture. ?Mesh? means that a radio can be dropped into a fabric of other radios and establish connectivity. In a traditional macro cellular network microwave technicians install and point highly directional antennas Having an integrated backhaul radio means that sites can quickly be deployed with electrical power only and linked to other sites that have a hard wired cable connection. Page 25October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution The backhaul radio is beneficial, not only to reduce the cost of backhaul, but also to speed deploy- ment and to reduce the site?s connectivity requirements. Some neighborhoods (e.g. business districts) have relatively little cable infrastructure. Appendix II describes this deployment concern. Having an integrated radio means that sites can quickly be deployed with electrical power only and linked to other sites that have a hard wired cable connection. It is difficult to overstate the importance of aesthetics. If consumers look up at an otherwise attractive street light and see a collection of electronics they will be tempted to ask ?What is this?? or think ?This is ugly.? In permitting street light small cells city officials will be quick to ask ?What will residents/visitors/pedestrians think?? Cities spend vast amounts of money to bury wiring (power, telephone, and cable) underground ? largely for cosmetic purposes. Some cites have political advocacy groups whose sole purpose is to get rid of ugly wiring3. If a street light small cell contains a bunch of boxes and has objects that look like antennas (sticks, micro- wave dishes, pointed flat panels, etc.) then the company deploying such equipment will already have one strike against it. Our suggestion is that the operator includes all this functionality in a package that does not look like a radio and ? in fact ? blends into the structure to which it is attached. Finally, the integrated unit needs to include a DOCSIS 3 modem. We have selected this as the most likely choice for a wired backhaul. The exterior of the box will have a female F-type connector. In a neighborhood with aerial wiring electrical power can be dropped from above and a strand of coaxial cable (typically RG-6) can be terminated and attached. If the cable is dedi- cated (e.g. not split along the way) it can support a 600 Mbps speed. This is more than is needed for the small cell. In an urban environment, fiber extends to within a few blocks of most locations. Cable provides the last ?few hundred meters? of connectivity. Street lights, telephone poles, fiber, cable, and the associated utility economics are discussed in Appendix II. The process of a mobile operator rolling out a network of outdoor small cells is described in Appendix I. The appendices also discuss the economics of a large scale deployment. 5.4 Why 60 GHz Millimeter Wave? Microwave radio systems have been used in telecommunications networks since the 1930s. Systems operating between 2 GHz and 40 GHz are widely used for cell site backhaul. Lower frequencies are often used for rural long-haul, while higher frequencies (e.g. 40 GHz) are used in suburban/urban areas. In urban areas fiber has replaced a lot of microwave. Higher frequencies have smaller dishes and support greater bandwidths, but span shorter distances, due to various types of fading (rain, fog, etc.). A microwave engineer designs a network and files for FCC licenses. Licenses, if they are available, typically cost $1,000 or more per link. As a band becomes increasingly utilized (as they do, particularly in urban areas) it becomes diffi- cult to obtain new licenses. A 40 GHz microwave dish is relatively inconspicuous on a macro cellular site. On a street light it is much more visible. Also, it becomes very expensive to engineer a network and install lots of dishes, assuming spectrum is available. Here?s where 60 GHz (now called ?millimeter wave?) enters the scene. It is a license exempt band (ITU, FCC, and ETSI) that can support data rates over 1 Gbps on a short link. Higher data rates will become possible over time. Spectrum allocations exist in the V-Band (60 GHz) and the It is difficult to overstate the importance of aesthetics. Cities spend vast amounts of money burying wires underground. In the V-Band (60 GHz) there is no spectrum licensing cost. Page 26October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution E-Band (70/80 GHz). Regulators have adopted ?light licensing? in the E-Band where operators pay a small license fee (~$100). In the E-Band highly directional antennas are required. We have chosen the V-Band (60 GHz) as our reference because it is possible to deploy a product with an electronically steerable beam that will automatically figure out where it is in a mesh. There is no spectrum licensing cost. If we eliminate a visible antenna then the overall package becomes much more attractive. Aesthetics matter to ordinary people and to municipalities. In the solution we envision there is no ?dish? to point. Elements on the surface of the small cell form an electronically steerable array that can detect the next nearest street light small cell and focus its 60 GHz link in the appropriate direction. 5.5 Key Operating Expenses In addition to the capital investment in the radio ($6,000 per site) and the associated cost of capital the operator will incur several categories of monthly operating expenses, most notably: ?? Lease per site ?? Electrical power ?? Cable broadband connectivity (every third site in dense urban and urban regions and two out of three sites in suburban regions). The operator will also incur an allocated core network cost. This cost, however, will be similar on a per GB basis to the cost incurred by the macro cellular network. It is included in each and is therefore not influential in the decision to deploy a street light small cell layer. Metricom (described in detail in Appendix I) paid $10 per site per month for a site lease and electrical power in the year 2000. The Metricom network covered 57 million POPs, nearly 20% of the US population. In Appendix I we determined that Metricom?s expenditure represented $12.48 in site lease and $2.73 in electricity (a total of $15.21) in 2014 dollars. Google (also described in Appendix I) paid $36.00 per month in site lease in 2006 for somewhat larger sites. That lease amount would be $42.47 per site today, based on the change in consumer price index. Google purchased electricity under Pacific Gas & Electric?s A-1 schedule, which is higher than the national average. Finally, the FCC now requires telephone companies to report their telephone pole and conduit- related expenses. Based on filings early this year (see Appendix II) we calculated the average monthly cost per pole as $2.77 in depreciation plus maintenance expense. In addition, the utility must earn a rate of return on its capital. The capital cost might double this figure to $5.00 per month. We therefore have a range of numbers for monthly site lease: approximately $5.00 in utility reported costs, $12.48 from Metricom (by far the most widely deployed and the most compa- rable in form factor) and $42.47 from Google (a small number of locations and physically larger equipment). In the analysis we used $16.00 per site for urban and suburban areas and 5x$16, or $80, in dense urban areas. We didn?t want to underestimate this cost, because it is extremely important. If we eliminate a visible antenna then the overall package becomes much more attractive. Aesthetics matter to ordinary people and to municipalities. The Metricom network covered 57 million POPs, nearly 20% of the US population. Metricom spent $12.48 in site lease and $2.73 in electricity (a total of $15.21 per site per month) in 2014 dollars. The FCC requires telephone companies to report their telephone pole and conduit-related expenses Page 27October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution At the same time, we didn?t want to use the very high costs sometimes quoted for surgically deployed small cells. Costs are high with surgically deployed small cells because the operator has not made a decision to deploy in volume and therefore has not entered into serious negotiations with a municipality. Metricom (described in Appendix I) is one party that did enter into serious negotiations with municipalities in many different states and realized a favorable lease rate. We are favoring their pricing because they were engaged in a large scale deployment. We assume that our proposed street light small cell consumes 75 watts of power (significantly more than the Metricom unit) and purchases electricity at $.15 per kW-h. This results in $8.36 per site per month in electricity expense. Finally, we assume that the operator provisions each connected site with a business class cable connection with 150 Mbps in the downstream and 20 Mbps in the upstream. Comcast provided budgetary volume pricing4 for the white paper, assuming a three-year commitment. They assumed a high volume carrier-class deployment. The price includes all of the installation costs up to a limit per location. It assumes heavy utilization (consistent with a carrier-class application) at each site. The resulting cost is $200 per site per month. The trend with cable pricing is that the performance of a particular ?package? improves signifi- cantly over time while the price remains the same. We did not include any downward pricing trend in our calculations. One alternative to cable would be a city-wide rooftop 60 GHz fabric (presumably already providing retail broadband connections to homes and businesses). Such a solution would link into the 60 GHz millimeter wave radio in the small cell and would act as an alternative to the assumed cable connection. Another alternative would be a widely deployed broadband provider that could provision throughout the city using a standard physical interface (e.g. coaxial cable or CAT5e or CAT6 twisted pair). 5.6 Site Density We have considered how an operator would most likely deploy street light small cells. We ran COST 231 and Walfisch-Ikegami models over a range of frequencies, with a standard set of building penetration margins, each tailored to the morphology (extremely high margins in dense urban, with more moderate margins in successive morphologies), each assuming a 8.5 meter height and a very low gain (1 dBi) omnidirectional antenna. We have assumed a nearly isotropic antenna pattern, with the objective of keeping the small cell extremely simple and ?nearly invisible? from an aesthetic perspective. In fact, a much more direc- tional beam might be possible using patch antennas. The greater directivity would improve the radiated power. It would improve performance and would result in a greater number of devices ?choosing? the small cell. A more directional antenna would help the signal penetrate deeper into buildings in an urban area. We also considered the maximum inter-site distance associated with each wireless backhaul tech- nology. We assume the small cell uses a 60 GHz millimeter backhaul that connects to adjacent sites at inter-site distances of 300 meters. We also assume that the backhaul antenna is an elec- tronically steerable array and that the small cells are self-organizing. The operator provides line power (120 VAC) and the cells organize themselves. Every so often the street light cell needs to be physically connected to a wired broadband network. In our business model we assume that The operator provisions each connected site with a business class cable connection with 150 Mbps in the downstream and 20 Mbps in the upstream. An alternative to cable would be a city-wide rooftop network of 60/70/80 GHz radios (presumably already providing retail broadband connections to homes and businesses) The operator simply provides line power (120 VAC) Page 28October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution one in three sites in urban and dense urban morphologies has a wired connection, even though the system could support much greater aggregation ratios. 5.7 Which Frequency Band? Operators would probably choose to offer service at a relatively high frequency (e.g. 1900 MHz or 2.5 GHz or 2.6 GHz), because the distance between the street light and the subscriber will be small. Lower frequency bands will then be reserved for macro cellular sites. The street light small cells can also reuse spectrum that is used elsewhere. If the subscriber has a strong signal from the street light small cell then the small cell will provide service. Otherwise, the subscriber?s device will select another site. 5.8 Voice Traffic We have assumed ? in an abundance of caution ? that all voice is carried by the macro cellular network and/or in-building infrastructure. This is because voice is exceptionally important to the subscriber?s experience and if anything in the system architecture were to compromise voice that would be a problem. 5.9 Deployment Geometry (Linear vs. Area) In dense urban areas and in urban areas we assume that the sites will be deployed linearly along every street. This results in a very high density of sites, but also ensures excellent illumination of every street. The approach will work in very dense environments where buildings abut one another. We assume that the street will be well illuminated. In an urban environment (typically 3?5 stories) all floors will be illuminated ? at least the street facing portion. Those areas toward the interior of the block will be served with a macro cellular signal. In a dense urban area the street and nearby retail areas will be served. All other areas (upper stories and interior areas) will be serviced by in-building systems and/or macro cellular sites. In suburban areas we assume that sites will be deployed on an area basis. The separation will still be 300 meters, but we will assume that a site can serve a hexagonal area, rather than a linear road. This is possible because of the lower building height in suburbia and the separation between buildings. In suburbia we assume that half of all sites have a clear view to another site. Therefore 2 out of 3 sites have cable connections. As the network in suburbia becomes denser, nearby sites will increasingly see one another, and a lower proportion of cable connections will be needed. We have assumed ? in an abundance of caution ? that all voice is carried by the macro cellular network and/or in-building infrastructure Page 29October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 5.10 Key Assumptions Table 3 summarizes some of the key deployment assumptions. Assumption Dense Urban Urban Suburban Inter-Site Distance 300 Meters Method of Deployment Linear, Every Street Linear, Every Street Area (Hexagon) Percentage of Area within Morphology 100% 100% 30% (Densest area, with 50% of suburban population) Percentage of LTE Data Traffic 15% (Excludes in-building data usage) 50% (Assumes interior of city block is better served by a macro site) 50% (50% of population lives in densest 30% of area. Network captures 100% of that traffic) Percentage of Street Wi-Fi traffic 100% (Extrapolated from Google experience ? see Appendix I) Percentage of Residential and Business Wi-Fi Traffic 0% (This traffic is captured by home and office Wi-Fi networks) Percentage of 2G/3G Data Traffic 0% (We assume street light small cell only supports LTE) Percentage of Voice Traffic 0% (We assume all voice traffic remains on macro cellular network) Table 3. Street Light Small Cell Deployment Assumptions Source: Signals Research Group Page 30October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 5.11 Utilization and Economics Circling back to the economics, we would like to present several related views. Figure 7 shows how utilization, a driver of the economics, varies by morphology and increases over time, as a result of the increasing amount of LTE traffic. This figure includes only LTE traffic (no Wi-Fi). We expect the baseline design to include Wi-Fi but we are starting here so that we can show what happens as functionality is added. Utilization and the unit costs both improve significantly when Wi-Fi is added, as we will discover in a few pages. Street light small cells represent a spectacular increase in frequency reuse and therefore in system capacity. If they were introduced on a widespread basis too soon they would not be economic. The 2016/2017 timeframe is ?just in time.? Each year street light small cells become significantly more impactful because their utilization increases and their unit costs decline. We will discuss unit costs in detail in the coming pages. Mass deployed street light small cells represent a spectacular increase in frequency reuse and therefore in system capacity. 0% 10% 20% 30% 40% 50% COMPOSITE Suburban Urban Dense Urban 20202019201820172016 U ti liz at io n of S tr ee t Li gh t Sm al l C el l C ap ac it y (% ) Figure 7. Utilization by Morphology, LTE Only Source: Signals Research Group Page 31October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Figure 8 shows the economics of an LTE street light small cell network overlaid on top of a macro cellular network. The distribution of traffic is described in Table 2. The composite economics represents a network-wide average that includes both the macro cellular layer and the street light small cells. We assume that a lot of data traffic is not carried by street light small cells because specific clusters of subscribers are more effectively served by other network layers. In dense urban areas a lot of traffic is carried by in-building small cells and DAS systems. In urban areas traffic that originates in the interior of a city block is carried by the macro cellular network since radio signals from street light small cells might not reach those locations. We have worked hard to portray a very realistic picture of the role of street light small cells ? not to overstate their potential benefits. In this first, largely conceptual scenario, we have included LTE, but not Wi-Fi. Subsequent graphs will show the great benefit on including Wi-Fi. We assume that a lot of data traffic is not carried by street light small cells because specific clusters of subscribers are more effectively served by other network layers. $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 COMPOSITE Street Light Network Macro Network 20202019201820172016201520142013 C os t pe r G B ($ ) o f St re et L ig ht S m al l C el l N et w or k Figure 8. Economics of Street Light Small Cells, LTE Only Source: Signals Research Group Page 32October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution If we add Wi-Fi to each of the street light small cells then the economics of Figure 9 emerge. We model ?street-level? Wi-Fi traffic very differently than cellular traffic. If someone is in their home or business we assume that their Wi-Fi traffic will be carried by the home or office Wi-Fi network. We further assume that someone in a caf? is likely to connect to the caf? Wi-Fi network. People using the street light Wi-Fi network are nearby (on the street or in an adjoining retail space) and generally without any other readily available Wi-Fi connection. We have modeled the actual traffic levels by looking at the Wi-Fi traffic and number of users on Google?s Moun- tain View citywide Wi-Fi network during its heyday (~2009) then extrapolating forward to today. This network is described in Appendix I. Interestingly, there are a large number of devices that exist today for which Wi-Fi is the only means of connectivity (e.g. Wi-Fi only tablets and laptops without wired connectivity). We looked at the traffic on Google?s Mountain View citywide Wi-Fi network in 2009 then extrapolating it forward to today. COMPOSITE (LTE + Wi-Fi) COMPOSITE (LTE Only) Street Light (LTE + Wi-Fi) Street Light (LTE Only) Macro Network 20202019201820172016201520142013 C os t pe r G B ($ ) o f St re et L ig ht S m al l C el l N et w or k $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 Figure 9. Economics of Street Light Small Cells, LTE + Wi-Fi Source: Signals Research Group Page 33October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Figure 10 shows the traffic by network layer, including Wi-Fi. The first two layers (macro network and in-building network) exist today and carry 100% of the traffic. When street light small cells are deployed a portion of the LTE traffic that would otherwise reside on the macro cellular network shifts to the street light small cell network. The street-level Wi-Fi traffic captured by the street light small cell layer (the top bar) is entirely new. It represents Wi-Fi only and Wi-Fi/ cellular devices that discover the new Wi-Fi signal and attach. This traffic is not currently carried by the mobile operator. Is it reasonable to attribute a ?cost? to Wi-Fi? Who will pay for the cost of Wi-Fi traffic? We believe it is reasonable. An existing broadband provider (e.g. Comcast or AT&T) might pay a wholesale rate to add street light small cells to their list of ?Wi-Fi hot spots.? Access to branded hot-spots is often provided as a perk of the more expensive home broadband plans. It is also possible for a smartphone application to redirect traffic from the cellular network to Wi-Fi. A robust Wi-Fi network can carry traffic from multiple mobile operators ? for a small fee. Alternatively, the owner of the street light small cells could offer a ?freemium? service directly to consumers. It would be free for basic usage and fee-based for more generous/sophisticated access. Finally, the owner of the street light small cells could monetize the traffic through advertising (e.g. banner ads on a landing page or other integrated location-based advertising). The cost per GB for Wi-Fi access is so small that even a tiny amount of advertising revenue would cover the network cost. The bottom line is that there are a number of ways to monetize Wi-Fi traffic. None of the busi- ness models requires a large leap of faith to believe it is viable. It is possible for a smartphone application to redirect traffic from the cellular network to Wi-Fi. Is it reasonable to attribute a ?cost? to Wi-Fi? Who will pay for the cost of Wi-Fi traffic? The owner of the street light small cells could offer a ?freemium? service directly to consumers. The cost per GB for Wi-Fi access is so small that even a tiny amount of advertising revenue would cover the network cost. D at a Tr af fi c pe r M on th (m ill . G Bs ) 0 300 600 900 1,200 1,500 20202019201820172016201520142013 Street Light - Wi-Fi Street Light - LTE In-Building Network Macro Network Figure 10. Traffic by Network Layer, Including Wi-Fi Source: Signals Research Group Page 34October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 6.0 Neutral Host 6.1 Business Model In each of the scenarios above we have assumed a single large mobile operator, with 33.9 % market share of 2G/3G/4G voice and data. We further assumed that the system would carry 35% of the Wi-Fi traffic per person of a ?free? metro Ethernet service, similar to what Google deployed in Mountain View, California in 2006, but scaled up to carry today?s levels of traffic. We assumed only 35% because the traffic might come from a wholesale relationship where the wholesale partner provides access to a certain class of its customers, but does not provide access to the general public. The small cell operator could have multiple wholesale Wi-Fi partners, each radiating its own SSID and authenticating users via its own proprietary back-end platform. 35% is therefore very conservative. The larger the number the more it benefits the economics of the street light small cell. Free public Wi-Fi would be represented by ?100%.? The owner of the street light small cell could, of course, sell Wi-Fi access at a profit. We are assuming here that they simply recover their cost. Imagine that instead of serving a single operator the street light small cell served two or more operators. This is a ?neutral host? scenario. One party deploys the infrastructure (incurring the capital investment and the monthly operating expenses) then allows multiple operators to use the infrastructure. Some network elements, such as in-building DAS systems, are designed this way today. Mobile operators historically have had a cool or lukewarm reaction to the neutral host concept. They don?t like giving up control to a third party. We present it here because the numbers are compelling. Will mobile operators embrace it? We?ll see ? maybe, maybe not. The greatest ?headwind? to street light small cells is the lack of traffic. Each covers such a small area that despite the ?explosion in wireless data usage? street light small cells will be lightly utilized for some time to come. A neutral host scenario is financially attractive because it boosts utilization. In a neutral host scenario we assume that the owner has wholesale relationships that give it a 60% market share. If the owner had one top-tier operator plus one other modestly sized operator, he would achieve his goal. Imagine that instead of serving a single operator the street light small cell served two or more operators. This is a ?neutral host? scenario. One party deploys the infrastructure then allows multiple operators to use it. The greatest ?headwind? to street light small cells is the lack of traffic. Page 35October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution We similarly assume that the neutral host carries 70% of the traffic of a free metro Wi-Fi network. Google?s Mountain View Wi-Fi network (Appendix I), which was heavily advertised and free to the public, if it were still in operation today, carrying traffic that increased proportionately over time, would represent ?100%? of the street-level Wi-Fi. Scenario Major Operator, LTE Only Major Operator, LTE + Wi-Fi Neutral Host, LTE + Wi-Fi LTE Traffic ? We assume the operator has a 33.9% share of the overall mobile market ? The street light small cell carries 15% (dense urban) to 50% (urban, suburban) of the local LTE data traffic, as described in Table 2. ? Same as the LTE Only scenario ? We assume the street light small cell represents operators that collectively have a 60% share of the overall mobile market ? The street light small cell carries 15% (dense urban) to 50% (urban, suburban) of the local LTE data traffic, as described in Table 2. Wi-Fi Traffic ? None ? We assume that the small captures 35% of street level Wi-Fi traffic. ? This scenario is more restric- tive than the Google muni Wi-Fi network in Mountain View ? We assume that the small cell captures 70% of the street level Wi-Fi traffic 2G/3G Traffic ? None ? None ? None All Voice Traffic ? None ? None ? None Table 4. Share of Traffic for Major Operator and Neutral Host Scenarios Source: Signals Research Group Page 36October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 6.2 Economics Figure 11 shows the impact of these assumptions. Here we see street light small cell costs that are less than half the cost of the macro cellular network. Operators will have to consider if they willing to give up some control to enjoy spec- tacular economics. COMPOSITE (LTE + Wi-Fi, Neutral Host) COMPOSITE (LTE + Wi-Fi) COMPOSITE (LTE Only) Street Light (LTE + Wi-Fi, Neutral Host) Street Light (LTE + Wi-Fi) Street Light (LTE Only) Macro Network 20202019201820172016201520142013 $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 C os t pe r G B ($ ) o f St re et L ig ht S m al l C el l N et w or k Figure 11. Economics of Street Light Small Cells, Neutral Host with LTE + Wi-Fi Source: Signals Research Group Page 37October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 7.0 Technology and Incremental Revenue A street light small cell could easily support a number of radio access technologies, in addition to LTE and Wi-Fi at 2.4 GHz. Also, a street light small cell could be connected to the large network in a number of different ways. The following sections explore these possibilities. 7.1 Next Generation Wi-Fi (802.11ac) and WiGig (802.11ad) It is important to mention emerging access technologies that a street light small cell could support. 802.11ac is a new version of the 802.11 Wi-Fi standard. It was approved in January of 2014 and will be widely available in 2015. Many 802.11ac products are already available. The standard enables up to eight MIMO streams and uses complex (256-QAM) modulation. It oper- ates in the 5GHz band, providing 500 Mbps in a single link or 1 Gbps of speed in a multi-station WAN environment. A street light small cell deployed in 2016 could support 802.11ac, along with the 2.4 GHz Wi-Fi standards widely used today. Another technology, WiGig (802.11ad), is about to come to market. WiGig operates at 60 GHz. It supports data rates up to 7 Gbps. While the use cases for these Wi-Fi technologies were targeted to indoor scenarios, mobile users are growing to expect a quality of service experience in outdoor public settings similar to their indoor experiences. The radio layer of the street light small cell that provides 60 GHz backhaul could also provide end-user access via WiGig, with appropriately equipped laptops. Owners of laptops and mobile devices with these next generation wireless technologies will be gratified to use them in a wide area street light small cell network (even if the backhaul might not support the full capabilities of the access link) City officials will need to consider whether a conventional 2.4 GHz public Wi-Fi network is good enough, or whether they should impress voters by launching a public WiGig network. 7.2 Municipal Services A network of street light small cells can be used to provide a broad range of other services that are valuable to municipalities. Offering valuable synergistic services to the city free of charge is a great way to create municipal buy-in and help manage site-related costs. Potential services include broadband Wi-Fi access for city employees (police, fire, and meter maids), general purpose cameras (for still or occasional video), and traffic-related cameras (red light violations, license plate reading, etc.). The caution with cameras is that unless they are used sparingly they will consume considerable data bandwidth. Providing the city free access up to a certain data volume then charging per GB is one potential business model. An example of a seemingly off-the-wall service is gunshot detection. When a firearm is discharged in an urban area something bad has most likely happened. Police departments want to know about it as soon as possible. Unfortunately, many gunshots are not reported and others are reported but only after a few minutes. Many cities and police departments are eager to deploy the technology. It?s not cheap, in part because it requires a distributed network of broadband-connected sensors. If the sensors were bundled into street light small cells the system might become a lot cheaper. A street light small cell might support 802.11ac at 5 GHz and 902.11ad (WiGig) at 60 GHz, along with Wi-Fi at 2.4 GHz. Offering valuable synergistic services to the city free of charge is a great way to create municipal buy-in and help manage site-related costs. Page 38October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution The bottom line is that there is a broad range of functionality that could be provided by a street light small cell initiative that would greatly benefit a municipality and would make the operator extremely welcome. 7.3 Alternative Architectures Are there other ways to deploy street light small cells? The answer is ?yes.? One very different approach utilizes C-RAN/fronthaul, instead of an IP connection. The small cell is connected to a piece of dark fiber. Instead of demodulating the signal and sending packets of end-user data over the fiber the unit transmits an optical analog signal that represents the radio signal over the fiber. In some cases (e.g. CPRI) the signal is a digital sample (e.g. I/Q). In either case the backhaul connection is used very ?inefficiently? to enable the operator to have a lot of flexibility. The unit itself has fiber electronics and a set of power amplifiers. The radio waves are demodu- lated at a centralized facility. Similarly, each of the transmitted signals is generated at the ?hotel? location then sent over fiber to the site where it is converted to normal RF and fed into the power amplifier. This concept is powerful and appealing to mobile operators because it gives them complete control. They can replace the essential hardware at any time. The challenge is that you need to have a dedicated strand of dark fiber. The price and availability of dark fiber varies by geography. If an operator is deploying street light small cells surgically, this might make sense. They would utilize only a small number of strands of fiber. What if they were to deploy thousands of street light small cells in a city (as we envision)? They could no longer assume ?There?s a glut of dark fiber. It will be cheap.? Also, with such an architecture, it is difficult to share a single backhaul connection between multiple sites. In most implementations each site has its own fiber connection. While some operators are enamored with such an approach it carries certain risks, costs, and complexities. One needs to ask ?How much benefit do I get from the flexibility of C-RAN versus how much cost, complexity, and uncertainty am I willing to endure to deploy it?? Some operators will take this approach. We have assumed a radio connection between street light small cells to save money on backhaul and have assumed a cable connection because this technology is ubiquitous and available today at virtually any urban or suburban address. Another possibility is to have a backhaul connection ?from the air? via a building-top 60 GHz transceiver. In a city with such infrastructure this is a viable option, as an alternative to a cable connection. Another possibility is to use non-line-of-sight (NLOS) systems operating at sub-6 GHz for the site-to-site connections, instead of 60 GHz radios. The advantage of a lower frequency link is its ability to operate in a NLOS environment. The disadvantage of these bands is that they are congested today and bound to become more congested in the near future as the band is increas- ingly used as an access link. An operator could choose to deploy a mixture of sub 6 GHz and 60 GHz radios, depending upon the environment. An alternative approach utilizes C-RAN/fronthaul (in some cases CPRI) instead of a traditional IP connection. Another possibility is to use sub - 6 GHz radios for the site-to-site connections, instead of 60 GHz radios, or to use both bands. Page 39October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 8.0 Conclusions Operators need to actively consider how to increase capacity. The required increase is in the range of ?10 times? over the next 7 years. The excess capacity or potential capacity in a mobile opera- tor?s network will begin to disappear in two to three years. At that point in time the operator will need a concrete plan, a set of technologies that will scale and will absorb the additional demand in each morphology. The technologies that are able to deliver capacity differ greatly by venue. In-building technologies are great, if you have a large building. They are very effective in dense urban environments, but less so in suburban environments, because of the lack of large buildings. Street light small cells, if deployed at the right time, in the right venues, and in sufficient volume, appear to offer a technology solution that will substantially slow macro cellular growth. Street light small cells are especially impactful if they can also provide other forms of end-user access (e.g. 2.4 GHz Wi-Fi, 5 GHz 802.11ac, and 60 GHz WiGig) and if a single infrastructure can support more than one mobile operator. Street light small cells, if deployed today in volume, would not be economic because there is simply not enough data demand to make them economic on a broad scale, relative to a macro cellular network. That situation is changing. In 2016, 2017, and 2018 ? the target window for deployment ? mobile data will have grown enough to make this innovative infrastructure cost effective on a broad scale. In a neutral host scenario (carrying the traffic of more than one operator) this economic cross- over point occurs sooner and unit costs decline at an even more aggressive rate. Street light small cells are complementary to other types of small cells. Each provides much needed coverage of a specific venue. Street light small cells provide excellent coverage of roads and road-facing portions of a building. It is less certain that they will cover the interior of city blocks (or specifically, that devices in those locations will favor the street light small cells over macro cellular signals). Like in-building solutions, street light small cells provide a powerful tool that will add capacity in certain venues. Macro cellular networks will continue to grow, with or without street light small cells. With street light small cells that growth will be manageable. Finally, street light small cells could be viewed as an intermediate technology, bridging the gap between today?s data demand and a futuristic ?1000 times.? When will a leap of several orders of magnitude be needed? Not for a while. In the meantime, ubiquitous street light small cells could provide the needed capacity. Street light small cells will be valuable as long as people walk, drive, or mingle along streets. Street light small cells, if deployed at the right time, in the right venues, and in sufficient volume, appear to offer a technology solution that will substantially slow macro cellular growth. Equally important, they will allow the operator to enjoy a declining network cost per GB. In a neutral host scenario (carrying the traffic of more than one operator) this economic cross-over point occurs sooner and unit costs decline at more aggressive rate than in a single operator scenario Finally, street light small cells could be viewed as an intermediate technology, bridging the gap between today?s data demand and a futuristic ?1000 times.? Page 40October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution APPENDIX I: Historical Networks Page 41October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Appendix I: Historical Networks Back to the Future: Metricom Ricochet In the year 2000, five years after the Netscape browser, a small company in Cupertino, California called Metricom5 took the wireless industry by storm with a mobile data service called ?Ricochet.? The Internet was still extremely young. Many tech-savvy intelligentsia had a newly created service called ?ASDL? that connected them to the Internet (a commercial version of ARPANet) at the lightning-fast speed of 1.5 Mbps. Others, if they bothered with the Internet at all, did so over a dial-up connection. Mobile operators understood that one day ?mobility? and ?data? would be used in the same sentence and might even become important. Pioneers were working on a variety of technologies. Some (e.g. CDPD) were commercial - meaning that highly motivated early adopters could buy them. The average consumer wanted nothing to do with this stuff. Not only did smartphones not exist, but laptops were few and far between. They were expensive. They resided in the briefcases of executives, not in the backpacks of teenagers. It was with great astonishment, therefore, that mobile operators watched a Silicon Valley startup, Metricom, launch a mobile data service then roll it out over a significant portion of the United States. Metricom was not an established infrastructure manufacturer. Metricom was not a Baby Bell. They did not have billions of dollars of cash in their back pocket. They were audacious enough to use unlicensed frequencies (900 MHz ISM) to provide internet access to the masses. They offered consumers a 28kbps service for $29.95 and a 128 kbps service (faster than dial-up, slower than DSL) for as little as $44.95. Their ?cell site? did not have a T-1 connection. It did not have a six-foot diameter long-haul microwave dish. It did not reside in a prefabricated building at the foot of a 150-foot tower. It did not have air conditioning. It did not have a diesel generator or four hours of battery backup. It cost less than a million dollars. Metricom was audacious enough to use unlicensed frequencies (900 MHz ISM) to provide internet access to the masses. Page 42October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution It was a few inches in size. It cost $2,000, fully installed (a process that took 15 minutes). Figure 12 shows a picture of that cell site. A bucket truck arrives, bolts this tiny box to an existing street light, removes the light sensor, plugs in the ?cell site? then reattaches the light sensor. The cell site draws power from the street light. The cell site backhauls its traffic through a second wireless connection at 2.4 GHz. Everything about the concept was unorthodox ? yet it worked! The author of this whitepaper was a subscriber. Metricom?s cell site cost $2,000, fully installed. A bucket truck arrives, bolts this tiny box to an existing street light, removes the light sensor, plugs in the ?cell site? then reattaches the light sensor. Figure 12. Metricom Street Light Small Cell (Circa 2001) Source: http://en.wikipedia.org/wiki/Ricochet_%28Internet_service%29 Page 43October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution In a short few years Metricom rolled out its service to 17 major metropolitan markets. They covered 57 million POPs ? approximately one in six Americans, or a population the size of the England, Scotland, and Wales combined. Table 5 describes their rollout. They eventually went bankrupt. Why? Probably because they were ten years ahead of their time. Prospective mobile data users were few and far between. Metricom was also raising capital right after the great stock market collapse of 2000, when everyone who had made their fortune in technology stocks saw that fortune disappear. Metricom demonstrated that a mass produced street light small cell can be deployed quickly and inexpensively. Several KPIs emerged: ?? Coverage: with 7 sites per square mile (a system average) they provided good coverage at 900 MHz. ?? Cost per Site: they paid $10 per site per month. This is an average across 64,000 sites. They negotiated this access city by city, state by state. It included the site lease and electrical power. ?? Usage: subscribers consumed 500 MBs of data per month. This was a stunning level of usage at the time. Global Internet traffic has grown at 60% per year since 2001. If we assume a more modest 35% per year growth in usage per subscriber (since much of the growth of the Internet was increasing adoption) this would equate to 24 GBs per month today from a single mobile subscriber. Metricom provided service in 17 major metropolitan markets, covering 57 million POPs ? more than one in six Americans. The network required 7 sites per square mile, at $2,000 per site (capital) and $10 per site per month (operating expense - site lease and power). City Population Sites Coverage (Miles2) Site Investment Atlanta 1,200,000 2,812 402 $5,624,000 Baltimore 1,800,000 1,699 243 $3,398,000 Chicago 2,700,000 2,462 352 $4,924,000 Dallas/Fort Worth 3,800,000 1,371 196 $2,742,000 Denver 1,300,000 3,547 507 $7,094,000 Detroit 3,300,000 5,913 845 $11,826,000 Houston 2,600,000 3,458 494 $6,916,000 Los Angeles 11,600,000 10,285 1,469 $20,570,000 Minneapolis 1,800,000 2,743 392 $5,486,000 New York City 9,300,000 5,225 746 $10,450,000 Philadelphia 4,000,000 4,468 638 $8,936,000 Pheonix 2,100,000 3,405 486 $6,810,000 Salt Lake City 200,000 2,741 392 $5,482,000 San Diego 2,100,000 3,187 455 $6,374,000 San Francisco 5,200,000 7,166 1,024 $14,332,000 Seattle 2,400,000 1,162 166 $2,324,000 Washington DC 1,500,000 2,910 416 $5,820,000 TOTAL 56,900,000 64,554 9,222 $129,108,000 Table 5. Metricom Street Light Small Cells by City Source: SRG Analysis, http://ricochet.wikispaces.com Page 44October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Today almost every man, woman, and child (or at least teenager) has a data-enabled mobile computing device in his or her pocket. Data rates are higher, by 1?2 orders of magnitude, demanding a more robust link budget and a very different access technology. A lot has changed but a lot has not changed. It is worthwhile to spend a couple minutes studying Metricom?s monthly operating expenses. Metricom paid $10 per site per month for the site lease and electrical power in the year 2000. Adjusting for inflation that $10 fee in 2014 would be $13.86 (using a CPI for all urban consumers), $15.30 (using the S&P Case-Schiller Real Estate Index), or $15.21 (using the U.S. Energy Infor- mation Administration retail price index for electricity), depending upon the index selected. Let?s take the middle result, $15.21. The Metricom street light small cell consumed 0.22 amps at 120VAC when idle and 0.35 amps at 120 VAC at peak when active. If we assume a 20% activity factor (which is probably high), then its weighted average power consumption is .246 amps, or 29.5 watts. It therefore consumed 21.25 kW hours in a 30 day month. At the average retail price of electricity in 2014 of $.1284 per kW-h this represents an electricity expense of $2.73 per site per month. The implied lease cost in 2014 dollars is therefore $15.21 minus $2.73, or $12.48 per month. These are important reference points for our LTE street light small cell business case in section 5.5. How does our proposed ?street light small cell? deployment compare to Metricom?s on other metrics? Figure 13 shows the number of sites per square mile: 0 30 60 90 120 150 Street Light - DENSE URBAN (Linear) Street Light - URBAN (Linear) Street Light - SUBURBAN (Area) Metricom (All Morphologies) in 2000 Si te s pe r Sq ua re M ile Figure 13. Street Light Small Cell Density Source: Signals Research Group Page 45October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Metricom provided good coverage with a relatively light density of sites. Their footprint included dense urban, urban, and suburban morphologies. Google?s Wi-Fi-only coverage of Mountain View, California ? which we will discuss momentarily ? weighed in at 26 (309 sites /12 sq. miles) sites per square mile, a lot denser than Metricom, but less dense than our proposed suburban LTE density. Figure 14 shows the capital investment per square mile: Metricom is by far the least expensive, at $2,000 per site and 7 sites per square mile, or $14,000 per square mile of street light radio investment. What we are envisioning for LTE is more expen- sive ($6,000 per installed integrated radio) and is being deployed with much greater density (44 to 129 sites per square mile). It also has much more transmit power per unit (2 watts), which will give it greater coverage. Finally, the radio interface (LTE) and backhaul (60 GHz millimeter wave) are both much more capable (supporting theoretically 150 Mbps to an end user and up to 1 Gbps between sites, under ideal conditions, if permitted by the operator). The LTE access radio link could operate in a similar frequency band (700 MHz for LTE vs. 900 MHz for Metricom) or in a higher frequency band (1700 MHz, 1900 MHz, 2500MHz, or 2600 MHz, as examples) based on the preferences of the operator. Our proposed ?street light small cell? is cost effective not because the hardware is cheap (it isn?t, although it could be cheaper than we assumed) or because the sites are few (they?re not), but because data demand has soared since 2001, enabling a modern operator to monetize such an investment. Could an operator deploy a less expensive system? The answer is quite possibly, ?yes.? We have made very conservative assumptions because until a full scale system is deployed some uncer- tainty remains about the ?reach? of each street light small cell. It is possible that coverage will be Our proposed mass-deployed LTE street light small cell is cost effective not because the hardware is cheap (it isn?t) or because the sites are few (they?re not), but because data demand has soared since 2001, enabling a modern operator to monetize such an investment. 0 $100,000 $200,000 $300,000 $400,000 $500,000 $600,000 $700,000 $800,000 Street Light - DENSE URBAN (Linear) Street Light - URBAN (Linear) Street Light - SUBURBAN (Area) Metricom (All Morphologies) in 2000 In ve st m en t pe r Sq ua re M ile ($ ) Figure 14. Capital Investment per Square Mile Source: Signals Research Group Page 46October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution much better than expected and the street light small cells will capture more traffi c than expected. Th is would result in a more favorable business case than we are articulating here. Google: Free Metro Wi-Fi, an Experiment In 2006 Google made the audacious decision to deploy a free Wi-Fi network6 covering the city of Mountain View. It attracted a lot of media attention. It was also a very interesting experiment to see how people would use such a network. Figure 15. Coverage of Mountain View Wi-Fi Network Source: http://wi? .google.com/city/mv/apmap.html Page 47October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Google spent $1,000,000 in 2006 to deploy 309 sites ($3,236 per site, including a 5 GHz back- haul link). It covered 12 square miles or 31 square kilometers (95% of Mountain View). They paid the city $36 per month per site in leases. This is significantly more than Metricom paid per site. Google?s Wi-Fi network included a separate point-to-point wireless radio for backhaul. The hardware was larger and more visible than that deployed by Metricom. Google purchased bulk electrical power from Pacific Gas & Electric under their Schedule A-1 tariff.7 They took the specifications for the equipment and did a back of the envelope calculation to estimate the usage. In a single month in 2009 (during the heyday of the experiment) Google measured 16,000 unique users (22% of the population) and 15,000 GBs of usage. This translates into .203 GBs per POP in 2009 or ? if extrapolated to 2014 ? .762 GBs per POP today. This usage data is helpful because it measures the precise service a Wi-Fi enabled street light small cell will provide. The street light small cell will provide Wi-Fi not to homes or businesses but to the ?in-between? places ? people on the street or in public areas with Wi-Fi enabled devices. Since the Google service was free and heavily advertised it represented the high end of the potential demand that a street light small cell operator might experience. If we scale ?street level Wi-Fi usage? to current levels and compare it to average residential broad- band usage we arrive at the numbers if Figure 16. The business case described in section 5.11 assumes that the owner of the street light small cell network captures only 35% of the Wi-Fi traffic Google would have captured because the owner of the street light small cell network is seeking to monetize its investment by selling access to a large bidder. In section 6.0, a neutral host, we increase this share to 70%, on the premise that services would be widely available. Either way, Google spent $1,000,000 in 2006 to deploy 309 sites ($3,236 per site, including a 5 GHz backhaul link). It covered 12 square miles or 31 square kilometers (95% of Mountain View). Figure 16. Outdoor Usage vs. Residential Usage in 2014 Source: Signals Research Group 0 10,000 20,000 30,000 40,000 50,000 60,000 Fixed Residential Usage/ Household Fixed Residential Usage/ POP Free Metro Wi-Fi Usage/UserFree Metro Wi-Fi Usage/POPExpected Usage/POP We assume 35% of the ?free? usage for a widely adopted wholesale service 267 762 3,528 20,166 52,634 While people love the idea of free public Wi-Fi, usage is light relative to usage elsewhere Br oa db an d U sa ge (M Bs p er M on th ) Page 48October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Wi-Fi usage that lands ?on the street? is a small fraction of the volume of data traffic consumed in the home. Google?s Mountain View network was not enhanced over time. It eventually slowed as usage increased beyond the capacity of the original design. The system was recently decommissioned and replaced with a new Wi-Fi network with more limited coverage. Page 49October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution APPENDIX II: Enabling Technologies Page 50October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution Appendix II: Enabling Technologies Fiber and Coaxial Cables Historically a cable operator had a few fi ber connections and long stretches of coaxial cable. Along the way signals were repeated to overcome the natural attenuation of coax. When cable systems began to support data the one-way repeaters were replaced with bi-directional repeaters. Th ey amplifi ed signals in both the downstream and the upstream. Near the customer premise the cable signal was split, to accommodate multiple households. All of these elements are still part of a modern system. Th e diff erence is that now fi ber reaches deep within most neighborhoods, often within a couple blocks of any urban location. Th e close proximity of fi ber means that fewer repeaters are needed and that fewer splitters are needed. Coaxial cable, in theory, with the latest hardware, can support a downstream data rate of 1Gbps (24 channels) and an upstream data rate of 245 Mbps (8 channels). Current deployed systems with 16 downstream channels can support 600 Mbps in the downstream. We have assumed 150 Mbps in the downstream for a small cell. Clearly, there is a lot of room for growth. We have assumed that the street light small cell would include a DOCSIS 3 modem, rather than a fi ber connection, because DOCSIS 3 provides an abundance of bandwidth, is readily available in urban or suburban locations. Cable technicians like the simplicity of a coaxial connection. Our objective is to make the street light small cell physically small, expensive, and power effi cient. Th e choice of a physical connection (DOCSIS 3 over coaxial cable vs. Ethernet over twisted pair vs. fi ber) is driven by cost and form factor more than anything else. Figure 17 shows the juxtaposition of fi ber, aerial wiring, and underground wiring in a typical neighborhood in San Francisco. Fiber is shown in aqua, aerial wiring (power, phone, and cable) is shown in red, and underground conduit and vaults are shown in green. Fiber currently reaches deep within most neighborhoods, often within a couple blocks of any urban location. The close proximity of fi ber means that fewer repeaters are needed and that fewer splitters are needed. Fiber (aqua) runs every few blocks Underground conduit and vaults (green) run throughout the city Arial (bright red) is in many neighborhoods Figure 17. Comcast Hybrid Fiber Coax Network, San Francisco Source: Comcast Page 51October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Finally, Figure 18 shows the components of a typical aerial cable network. Bundles of coaxial cables are strung from pole to pole. Depending upon the geographic location a bundle of fiber might be strung from pole to pole as well. If fiber needs to be extended to a point closer to the customer it is easy to do. No trenching is required. Often the coaxial cable is split before reaching the customer premise. If a high quality connection is needed (as in a business customer) then the coaxial cable could extend from the customer premise to the nearest fiber connection without any splits. Sometimes infrastructure is placed underground. Undergrounding is common in dense urban areas. It is also common in neighborhoods where people are willing to spend money to eliminate the clutter of aerial wiring. Underground wiring costs a lot more than aerial wiring ? often ten times as much. It is safer (no risk of electrocution from fallen lines) and less prone to outages (mean time between failures). However, in the case of an outage it is more expensive and more time consuming (mean time to repair) to repair, because a huge excavation effort is needed. Many cities have a mixture of aerial and underground wiring. Installing a street light small cell in a neighborhood with aerial wiring where the street light is attached to a power/phone/cable pole is very easy. Power can be dropped from the power lines above and cable can be connected from the bundle of cable wiring. If everything is underground it is a bit more complex. There are typically access holes in the sidewalk for each utility. In a ?favorable? scenario a provider would simply snake a wire through existing conduit to the light pole, after dropping wiring down the light pole from the small cell. In a less favorable scenario it might be necessary to break one or two sidewalk tiles, add new conduit, and then replace the cement. Undergrounding is also common in neighborhoods where people are willing to spend money to eliminate the clutter of aerial wiring. Trunk Cable Guy Cable Fiser Riser Amplifier Tap Disconnected Drop Cable Trunk CableFeeder Cable Fiber Cable Feeder CableFeeder ID Tag Figure 18. Typical Cable Wiring Source: http://www.annsgarden.com/poles/poles.htm Page 52October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution If the street light has a light sensor then it already has electrical power coming up the pole. In some neighborhoods lights are ?block switched? ? turned on and off at the same time for an entire block from a single light sensor or other switch. In such a scenario an alternative electrical wire might be needed or the block might be left switched on (reverting to individual sensors) or some other arrangement established. Wireless connectivity provides a number of distinct advantages. It can be deployed very quickly. There is no monthly recurring cost. It can be provisioned in locations that only have electrical power (no cable connection). Business districts are one area where cable operators have histori- cally not deployed infrastructure. A wireless link is ideal in such an environment. Even in areas where cable plant is abundant a mixture of wireless and wired connectivity is preferable. The composite solution (3 or more sites, wirelessly connected, for every DOCSIS 3 connection) provides excellent economics and speeds deployment. It also provides some degree of backhaul redundancy. If a particular connection fails due to hardware or a localized power outage (e.g. a transformer failure at the point of fiber connection) then the traffic can be rerouted wire- lessly to the next point of interconnection. Similarly in moments of exceptionally high demand traffic can be ?load balanced? between two DOCSIS 3 connections. Also, as competitive broadband providers increasingly use rooftop wireless to reach businesses there is the possibility that street light small cells configured with millimeter backhaul could be connected directly into such a fabric. A utility typically has a construction budget associated with each new installation. Up to a certain limit this non-recurring cost (which is incurred for all customers) is included in the monthly rate the utility charges its customers. Above the limit (unusually complicated situations) the customer has to pay. In some neighborhoods lights are ?block switched? ? turned on and off at the same time for an entire block from a single light sensor or other switch. Wireless connectivity provides a number of distinct advantages. It can be deployed very quickly. There is no monthly recurring cost. It can be provisioned in locations that only have electrical power (no cable connection). Page 53October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution Poles and Lights Th e good news is that even at an inter-site spacing of 300 meters physical street lights outnumber street light small cells by a large margin. Th e city of San Francisco has 44,300 streets lights. Th e required number of street light small cells ? if everything were dense urban or urban ? would be 5,792. Th is represents a ratio of 7.65 street lights for every street light small cell, assuming 100% coverage. In fact, San Francisco, like most cities, includes parks, lakes, and a few other environ- ments that are not ?urban? morphology. Th ese areas could support a less dense street light small cell deployment to provide the same quality of coverage. Figure 19 shows the consistency of street lights in the city of San Francisco. Since some of the city?s records are not electronic this image shows only those lights that have had a problem (e.g. have burned out or fl ickered and been reported)8. An exhaustive set of all street light locations would be 3.1 times as dense. Street lights vastly outnumber the required number of street light small cells by a ratio of 7.6 times in San Francisco Figure 19. Street Lights in San Francisco Source: SRG Analysis, https://data.sfgov.org/City-Management-and-Ethics/SF_StreetLights/jvuy-quuq? Page 54October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution Figure 20 shows the high density of street lights in mid-town Manhattan near Grand Central Station. In this case we see decorative street lights, as well as other street furniture. Many decorative streetlights Newspaper & Parking Figure 20. Mid-Town Manhattan Source: Signals Research Group Page 55October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffi c could lead to an unexpected architectural solution Figure 21 shows aerial wiring in a typical neighborhood in the Bronx in New York City. Here the street light is attached to a wooden pole. Th at same pole carries electricity, telephone, and cable wiring. Adding a street light small cell would be eff ortless in this environment since each utility is a few feet away. Attach here Line Power Cable, TV, Phone 2 ? 3 story structures with open space increases the likelihood of ?over the roof? propagation Figure 21. Neighborhood in the Bronx Source: Signals Research Group Page 56October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution The FCC has taken a recent interest in telephone poles and conduits. Beginning in April of 2014 the Commission started requiring telephone companies to report their telephone pole related gross capital investment and operating expenses. We analyzed the first wave of submissions9. Out of 21 reporting entities 19 seemed to understand the questions (two provided largely mean- ingless data). Table 6 summarizes the data provided to the FCC by the 19 respondents: Depreciation ($28.81) + maintenance expense ($4.39) equals $33.20 per pole per year, or $2.77 per month. A utility also needs to earn a return on capital. Adding a capital charge will increase the effective cost. Even so, these utilities are reporting extremely small numbers. If a telecom provider is treated as another utility, sharing infrastructure at cost, or slightly above cost, this would justify low monthly fees ? which is exactly what we saw with Metricom. Metricom paid $10 per month in the year 2000. This amount was an average across a large number of cities and states, a network covering 56.9 million people. The $10 figure included not only the site lease, but also electrical power. Financial KPI Average Gross Capital Investment per Pole ($) $594.61 Depreciation Rate of Poles (%, 1/years) 4.8% Maintenance Expense/Pole/Year $4.39 Table 6. Pole Costs Reported to FCC Source: Signals Research Group Page 57October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Financial Assumptions In comparing dissimilar infrastructures it is extremely important to use a single financial metric. We typically reduce everything to a unit cost (e.g. $/GB for data or $/minute for voice). This metric includes depreciation, the cost of capital, and operating expenses. If a physical element (a cell site, a backhaul connection, etc.) is used for multiple radio tech- nologies or is used to carry voice and data then the cost is disaggregated by technology (e.g. 2G/3G/4G) and by service (e.g. voice vs. data). We ultimately arrive at a cost figure that is directly comparable to revenue. An operator may price LTE data at $10 per GB (bundle size divided by price, disregarding bundle utilization) and may deliver data for $2 per GB, resulting in healthy gross margins. We often think of Wi-Fi as free, but even ?free? Wi-Fi has a revenue component. A public Wi-Fi network might be supported by advertising. Alternatively, it could be a ?freemium? service (free for basic access, but paid for advanced services). Finally, it could be provided as part of an affili- ated offering. Comcast offers Wi-Fi hot spot access to its ?Performance and above? residential users and its ?Starter Business and above? business users. AT&T has a similar hotspot offering for broadband customers. There is a cost to providing such service that is almost certainly offset by the incremental revenue associated with the upward plan migrations of interested subscribers. While the revenue per GB for Wi-Fi is small, the cost per GB for Wi-Fi is also small, relative to LTE. Wi-Fi hardware is much less expensive and consumers are generally happy to receive service on an opportunistic basis (e.g. whenever they can see the signal), which is a much less stringent network standard than operators have for LTE, which consumers expect to be ubiquitous. We normally separate network costs from the other costs of the business (acquisition, retention, customer service, corporate overhead, etc.). Gross margins need to contribute to all these ?other? business costs. Network costs are inclusive, with a couple of exceptions. We do not include the capital invest- ment in spectrum ? even though it is large ? because it is often separated in time from the actual deployment of the network. An operator will bid on spectrum when auctions occur but may sit on that asset for years before deploying it. Spectrum can also be bought and sold, while the physical network remains largely unchanged. Also, we do not include reciprocal voice termination costs, since these are determined by regulators, and most relevant in a study of voice. When we study costs we typically see a unit cost curve that declines with increased levels of demand, as fixed costs are spread over a greater volume of traffic, and over time, as technology- related costs decline. In the current environment macro cellular network costs are declining for a number of reasons, principally because of a migration to LTE and because of increased utilization of the macro cellular fiber connections. We typically reduce everything to a unit cost (e.g. $/GB for data or $/minute for voice). In the current environment macro cellular network costs are declining for a number of reasons, principally because of a migration to LTE and because of increased utilization of the macro cellular fiber connections. Page 58October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution APPENDIX III: Revolutionary Visions Page 59October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Appendix III: Revolutionary Visions How would mobile operators evolve their network to address a level of demand that is several orders of magnitude greater than today?s? A number of visions exist in the industry. We will briefly discuss each. The Concept of ?1000x? Several companies in the industry have articulated the concept of ?1000x.? They pose the ques- tion, ?What could operators do to increase capacity per unit area by 1000-fold?? An ?Inside Out? strategy, which we will discuss momentarily, is one approach. A key question we need to ask is ?How long will it take for mobile broadband traffic to grow to 1000 times its current level?? If the answer is ?a while? then one might look for ?10x? or ?100x? solutions. Outdoor small cells are very effective in delivering ten times. It is difficult to deliver a thousand times today?s capacity from the street. At 1000x today?s demand the inter-site distance of an outdoor small cell network would be reduced to several tens of feet. At that density adja- cent cell interference would overwhelm the system, eroding potential capacity gains. However, is 1000x the most appropriate objective? There is a risk in delivering any solution to market too far ahead of demand. In between the horse-drawn carriage and the space shuttle are a number of intermediate tech- nologies, such as the automobile and the airplane. Each represents an important product space. Is it possible that the street light small cell could fill this intermediate role (greater than 1, but less than 1000)? Inside Out Qualcomm?s ?Inside Out Strategy10? imagines that small cells are located in every home and business. These small cells radiate a signal that provides capacity and coverage for mobile users at large. The signals originating inside leak through the exterior walls of the home or business and provide coverage in nearby streets. Conceptually such an architecture would produce a spectacular increase in capacity. Several questions remain: (1) when would such capacity be needed? (2) how would the homeowner be incentivized to operate a small cell in his home for others to use, and (3) who would pay for this explosion in small cells? The Inside Out discussion is primarily a technological vision of the future. It is not a forecast or a business plan. Some business model questions therefore remain unanswered. Invasion of the DVRs Comcast has announced a plan to increase its number of public hotspots in the United States from 2 million to 8 million in 201411. It will accomplish this by radiating a public Wi-Fi SSID from new Comcast DVRs. This functionality will be enabled by default, but customers can disable it. This is essentially an ?inside out? strategy using Wi-Fi instead of a licensed cellular band. Comcast can add such capability at a negligible incremental cost to new DVRs that include Wi-Fi. Since cable is a shared medium, a cable operator can provision bandwidth over an existing cable connection, without impacting its bandwidth commitment to the subscriber in whose home the DVR resides. An ?Inside Out Strategy? imagines that small cells are located in every home and business. Comcast plans to increase its number of public hotspots in the United States from 2 million to 8 million in 2014 by enabling Wi-Fi in DVRs Page 60October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Could a mobile operator or an LTE chip manufacturer do a deal with a DVR manufacturer and a cable operator to provide similar functionality in licensed cellular bands? Perhaps. Page 61October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution APPENDIX IV: Sensitivity Analysis Page 62October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Appendix IV: Sensitivity Analysis A key aspect of any complex analysis is the sensitivity analysis. It tells us what parameters would greatly improve the business case and what might cause it to fall apart. We will show and discuss key sensitivities for three scenarios: (1) Major Operator LTE Only, (2) Major Operator LTE + Wi-Fi, and (3) Neutral Host LTE + Wi-Fi. LTE Only, Major Operator This first scenario, with LTE only, is an excellent place to start to understand the economics of street light small cells. It is the most bearish scenario, but also illustrates the most important ?moving parts.? Each of the sensitivity graphs shows the economics in the final year, 2020. Sensitivities include sites per wired connection (N), morphologies covered, capital per site, oper- ating expenses per site, and inter-site distance. Figure 22 highlights the following inputs: ?? Sites per Wired Connection (N). If every site needed a hardwired backhaul connection the backhaul cost would be high. The advantage of a meshed wireless backhaul is that only one in ?N? sites needs to be connected. The first collection of bars shows the impact of increasing N from 1 to 10. ?? Morphologies. The economics of street light small cell differs by morphology, reflecting the amount of traffic each street light small cell is likely to capture. Intuitively, one would expect dense urban to be the best, then urban, then suburban. In fact, the opposite is true. In a dense urban environment, even though there are lots of people, most live and/or work in the upper floors of a high rise. Their mobile communications needs are generally served by in-building small cells or macro cells (for those without in-building solutions). Only a small percentage of the traffic is best served by street light small cells. Urban environments, with 3-5 story buildings, are friendlier to street light small cells and less likely to have dedicated in-building solutions. Dense suburban environments, ironically, are the friendliest. Their low elevation and the separation between buildings make excellent coverage possible. The economics of the solution in each area is captured in this section. ?? Hardware Costs. The cluster of bars shows what happens if any of the key cost inputs changes. The installed hardware cost is $6,000 per site ($3,500 for the LTE radio + $2,000 for the 60 GHz backhaul and enclosure + $500 for installation). We will see what happens if those costs increase. ?? Lease Cost. The lease cost is $16 per site per month in suburban and urban morphologies. In dense urban regions we assume it is five times as high, $80 per month, since real estate costs often increase greatly in the dense urban core, especially at the street level. Fortunately, dense urban areas represent a small percentage of total area of all street light small cells. In this sensitivity we see what happens if the lease cost doubles to $160, $32, and $32 for dense urban, urban, and suburban morphologies, respectively. ?? Cable Cost. We assume the monthly cost for cable broadband connectivity is $200. This purchases a 150Mbps/20 Mbps connection. While it is not an SLA connection it is business class, which means that only in the rarest of conditions would the small cell not receive that level of service at busy hour. Physically, the small cell can be connected with either a shared or dedicated strand of coaxial cable. The DOCSIS 3 connection can currently support 600 Mbps data rates. The maximum supportable speed will increase as additional features in Page 63October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution the standard are implemented. In this sensitivity we see what will happen if the cost of the connection doubles to $400 per month. ?? Inter-Site Distance. A key driver is the inter-site distance. Distances less than 300 meters cause costs to skyrocket. Distances greater than 300 meters (up to 600 meters) cause the costs to fall. 300 meters is the practical limit of millimeter backhaul. Even though 60 MHz links can travel much further we assume that this particular link is designed to optimize cost and aesthetics, as opposed to distance and data rate. If the distance is too great then our assumption about LTE LPA power should also be reconsidered. 300 meters is a conservative assumption that is technically feasible for all the key components yet enables unit costs to quickly fall below those of the macro cellular layer. $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 150m 200m Base 400m 600m200% of Cable 200% of Lease 200% of Hardware BaseALLN=7 N=10N=3N=1 $1.23 $1.02$1.02 $0.97 $0.95 $1.41 $1.36 $0.89 $1.02 $1.40 $1.07 $1.36 $2.81 $1.79 $1.02 $0.72 $0.49 St re et lig ht C os t p er G B ($ , Y ea r 2 02 0) Sites per Wired Connection (N) Intercell SpacingCost Assumptions Dense Urban Suburban Urban Morphologies Figure 22. Major Operator, LTE Only, Sensitivities Source: Signals Research Group Page 64October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution LTE + Wi-Fi, Major Operator Figure 23 shows the same ?major operator.? The difference is that the street light small cells are now enabled for Wi-Fi. We assume that the system serves only users on the street or in street- facing venues (e.g. caf?s) without competing Wi-Fi systems. Even though this is a very simple and modest change in assumptions the cost per GB drops from $1.02 in the base case to $0.74. We also see that the relative economics of different morphologies changes. Dense urban suddenly becomes extremely attractive, because there are a lot of people on the street in a dense urban environment who will take advantage of a Wi-Fi connection if it is available. $0.00 $0.50 $1.00 $1.50 $2.00 600m400mBase200m150m200% of Cable 200% of Lease 200% of Hardware BaseSuburbanUrbanDense Urban ALLN=10N=7N=3N=1 $0.88 $0.74 $0.71 $0.70 $0.74 $0.56 $0.88 $0.71 $0.74 $0.99 $0.77 $0.97 $1.92 $1.25 $0.74 $0.55 $0.40 St re et lig ht C os t p er G B ($ , Y ea r 2 02 0) Sites per Wired Connection (N) Morphologies Intercell SpacingCost Assumptions Figure 23. Major Operator, LTE+ Wi-Fi, Sensitivities Source: Signals Research Group Page 65October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution LTE + Wi-Fi, Neutral Host Figure 24 shows an LTE + Wi-Fi enabled ?neutral host.? This is by far the most favorable scenario. The economics of street light small cells are fundamentally limited by the amount of traffic they can attract. By allowing a single street light network layer to support more than one operator the economics improves significantly. The composite cost per GB has declined from $1.02 (major operator, LTE only) to $0.74 (major operator, LTE + Wi-Fi), to $0.50 (neutral host, LTE + Wi-Fi). A neutral host architecture, if acceptable to the mobile operator, is by far the most cost-effective. As we look at the second and third set of sensitivities we reach a stunning conclusion: there is almost nothing you can do to cause the business case to collapse. Hardware costs can greatly exceed their target. Lease costs can increase greatly. Cable connection costs can go wild. The conclusions cut across morphologies. If sites are deployed too densely (e.g. at 150 meter or 200 meter spacing) then costs do increase, primarily because the traffic per site is greatly reduced. The one requirement (reflected in Figure 23 and Figure 24) is that the street light small cell must capture a minimum amount of traffic. With traffic levels increasing 40% per year this problem is ultimately solved with the passage of time. Hopefully this sensitivity analysis provides a helpful perspective on how key inputs influence the business case. St re et lig ht C os t p er G B ($ , Y ea r 2 02 0) 150m 200m Base 400m 600m200% of Cable 200% of Lease 200% of Hardware BaseALLN=7 N=10N=3N=1 Sites per Wired Connection (N) Intercell SpacingCost Assumptions Dense Urban Suburban Urban Morphologies $0.00 $0.20 $0.40 $0.60 $0.80 $1.00 $1.20 $0.57 $0.50 $0.48 $0.47 $0.50 $0.39$ 0.57 $0.48 $0.50 $0.63 $0.52 $0.62 $1.14 $0.77$ 0.50 $0.39 $0.31 Figure 24. Neutral Host, LTE + Wi-Fi, Sensitivities Source: Signals Research Group Page 66October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution Notes 1. Key sources include: Cisco VNI Forecast and Methodology, 2013-2018 http://www.cisco.com/c/en/us/solutions/collateral/service-provider/ip-ngn-ip-next-generation-network/white_paper_ c11-481360.html Cisco VNI Global Mobile Data Traffic Update, 2012-2018 http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white_paper_ c11-520862.html Sandvine Global Internet Phenomena Report https://www.sandvine.com/downloads/general/global-internet-phenomena/2014/1h-2014-global-internet-phenomena- report.pdf Ericsson Mobility Report (including custom data sets) http://www.ericsson.com/ericsson-mobility-report UBS Investment Research: US Wireless 411 http://www.ubs.com/global/en/investment-bank/institutions/securities-research.html 2. UBS Investment Research: US Wireless 411 http://www.ubs.com/global/en/investment-bank/institutions/securities-research.html 3. For example: http://www.sfundergrounding.org/ 4. Courtesy of Derek Griffin, Advanced Product Specialist, Comcast ( , 925-667-7983) 5. Wikipedia: Ricochet Internet Service http://en.wikipedia.org/wiki/Ricochet_%28Internet_service%29 Metricom Asset List http://ricochet.wikispaces.com/file/view/assetlist.pdf/30141986/assetlist.pdf Metricom Hardware http://ricochet.wikispaces.com/Hardware Other Information http://ricochet.wikispaces.com/ 6. Wikipedia: Google Wi-Fi http://en.wikipedia.org/wiki/Google_WiFi Contract with the city of Mountain View https://www.documentcloud.org/documents/749673-google-wifi-agreement-executed.html Coverage Map http://wifi.google.com/city/mv/apmap.html 7. PG&E Schedule A-1, Current and Historical http://www.pge.com/tariffs/electric.shtml#COMMERCIAL 8. https://data.sfgov.org/City-Management-and-Ethics/SF_StreetLights/jvuy-quuq? 9. Article on New Requirements http://www.broadbandlawadvisor.com/2014/04/articles/pole-attachments-2/ telephone-pole-owners-file-cost-data-with-fcc/ FCC Database http://fjallfoss.fcc.gov/eafs7/adhoc/quickpull/QuickPull.cfm Page 67October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution 10. Qualcomm Inside-Out Concept https://www.qualcomm.com/1000x/small-cells Qualcomm 1000x Challenge https://www.qualcomm.com/1000x 11. Comcast - 8 Million Hotspots in 2014 http://corporate.comcast.com/news-information/news-feed/comcast-to-reach-8-million-xfinity-wifi-hotspots-in-2014 Page 68October 2014 www.signalsresearch.com Street Light Small Cells ? a revolution in mobile operator network economics How exponentially growing smartphone data traffic could lead to an unexpected architectural solution www.signalsresearch.com