West Virginia Broadband Connectivity Final Report Spring 2017 George Mason University Department of Systems Engineering and Operations Research SYST 699 Ahmed Saeed Bao Nguyen Peter Gardner
Summary Broadband internet is essential for providing online resources to those who need it. The availability and/or use of broadband is sparse in rural regions of West Virginia. The digital divide will continue as these rural areas lack the ability to access the internet, and the gap in economic conditions and education will continue to increase as well. Increased broadband adoption rates are correlated to economic benefits which will benefit both the users and service providers. The primary goal of this project is to understand the options and issues of broadband availability in the targeted southern West Virginia counties. As the primary stakeholders of this project are the citizens of the selected counties, a comprehensive stakeholder analysis was conducted. The stakeholder analysis looked at key demographic data to identify the current economic conditions of the region. To grasp the current broadband coverage conditions of the region, a coverage analysis was conducted in order see the actual coverage. Analysis of the actual coverage, topography, and distribution of service to addresses was also performed. Along with the actual coverage, this project aimed to identify the main commercially available broadband technologies that can serve the rural residents. An analysis of wired and wireless expansions was performed. While identifying potential solutions and finding coverage gaps, we discovered that availability of broadband options was less of an issue than originally thought. The discovery led to a shift in focus to look at the actual broadband adoption rate. With the expanded scope, the problem then looked to be more of a socio-economics issue rather than purely technological availability. We conducted several case studies and a sensitivity analysis to observe the effects of broadband adoption on economic factors which includes population, income, overall employment, and provider costs. 1
Table of Contents Summary 1 Table of Contents 2 I. Introduction 5 1.1. Background 5 1.2. Scope 6 Technological Scope 6 Geographic Scope 6 1.3. Stakeholders 8 Stakeholder Demographics 9 Stakeholder Tensions 11 Demographics 11 II. Assessment 12 2.1. Available and Future Technologies 12 DSL 12 Cable 12 Fiber 13 Cellular 13 Fixed Wireless 13 Satellite 14 Future Technology 14 Low Orbit Subsidized Satellite 14 Balloon 4G LTE (Project Loon) 15 2.2. Current Status 16 GIS Introduction 16 Wired Connectivity 16 Methodology and Data 16 Results 17 Wireless Connectivity 18 Methodology and Data 18 Results 19 Overall Connectivity 20 Adoption Rate 21 2.3. Economic Model 23 2
Model Development 23 Model Behavior and Design 23 Demographic Data Integration 24 Sensitivity Analysis 26 Wireless Model 26 Wired Model 27 2.4. Case Studies 29 Huff Creek Valley - Wireless Expansion 29 Geography and Demographics 29 Wireless Expansion 30 Economics 32 Northwest Logan County - Wired Expansion 33 Geography and Demographics 33 Wired Expansion 33 Economics 34 Case Study Summary 35 2.5. Alternatives Analysis 36 Comparison of Wired vs Wireless Expansion Simulations 36 III. Conclusion 38 Future Work 39 Acknowledgements 39 Appendices 40 A. Initial Requirements Elicitation 40 B. GeoTel Data Validation 41 C. Demographics of Counties 43 D. Detailed Data for Alternative Solutions 46 D.1 - Current Technologies 46 DSL 46 Cable 46 Fiber 46 Cellular 46 Cellular Repeaters 47 Fixed Wireless 48 Signal Booster Implementation - Fixed Wireless 48 Approximate Cost Breakdown of a signal booster installation* 48 Satellite 49 D.2 - Future Technologies 49 Internet.Org / Facebook Free Basics 49 3
Project Loon 50 SpaceX Satellite Constellation 50 OneWeb satellite constellation 51 AirGig by AT&T 51 Dynamic Spectrum and White Spaces 52 D.3. Installation Costs 53 E. Modeling Data 55 E.1. Model Variables 55 E.2. Sensitivity Analysis Results 55 Northwest Logan (Wired) 55 Huff Creek Valley (Wireless) 56 E.3. Alternatives Analysis Wired vs Wireless Model Outputs 57 Wired Model Outputs 57 Wireless Model Outputs 58 E.4. Case Study Outputs 59 Huff Creek Valley (Wireless) Model Outputs: 59 NW Logan County (Wired) Model Outputs 59 E.5. Model Variables Definitions - Vensim Exports 60 Wired Model 60 Wireless Model 63 E.6. Model Variables Verification Income and Jobs Regression 65 E.7. Model Verification and Validation 65 E.8. Wired Model Assumptions 66 E.9 Wireless Model Assumptions 66 E.10. Wired Model Behavior and Design 67 E.11. Wireless Model Behavior and Design 67 F. Project Plan 69 G. Risk 70 H. Earned Value Management 72 Sources 75 4
I. Introduction 1.1. Background In this digital age, the internet and all of its benefits can easily be taken for granted. The many benefits of broadband internet are often unnoticed as they are seamlessly intertwined with social and economic progress. While urban areas are often the first regions to utilize the fastest broadband options, rural areas struggle to provide even the bare minimum of broadband internet services. This project aims to address the issue of broadband availability in rural areas of Logan and Wyoming counties of West Virginia. Broadband internet allows users to utilize online services that can potentially improve their quality of life, save time, and improve their education. Without it, many economic opportunities in today s world are not present who can operate an online business without the internet? And without sufficiently fast internet access, it is difficult to fully participate in the cultural life of America. These factors tend to encourage people to move away from underserved areas to places like cities, where broadband internet is readily available, and the economy is more vibrant. Additionally, from an educational perspective, an increase in the availability of broadband internet for homework and research can significantly improve the overall aptitude of students who previously did not have access to broadband internet. It also allows students and citizens to access a of online resources that opens up a world of education and economical opportunities. Without internet to assist with school work, students of rural regions will fall further behind students who do have more access to information and resources. This perpetuates lower education levels, and thus economic stagnation. 5
1.2. Scope Technological Scope As of 2015, the FCC defined broadband internet as data transmission with a minimum of 25 megabits per second (Mbps) download, and 3 Mbps upload. Prior to the change, the requirements for broadband were 4 Mbps download and 1 Mbps upload. To maximize the amount of options available, we will use the former definition of broadband. Where each broadband service is actually offered depends heavily on the location of the customer. There are a number of different mediums for broadband access, however the geography of the region alters the feasibility of some services. As part of this paper, we will evaluate the multiple commercial alternatives available for consumers, evaluate the key issues they face in the rural environment, and provide an alternatives analysis of expansion of wired and wireless infrastructure. Geographic Scope Figure 1.2.1 6
Originally, this project was to focus on McDowell, Mercer, and Wyoming Counties, at the southern end of West Virginia (fig. 1.2.1). However, due to difficulty in obtaining specific population data about McDowell and Mercer Counties, they were removed from the scope, and Logan county was added. These counties, along with the rest of West Virginia, are mountainous. The southeastern edge of Wyoming County is the westernmost ridge of the ridge and valley portion of the Appalachians; from there northwest all the way to Ohio (including the rest of the two counties) is dissected plateau rugged terrain with steep valleys (fig. 1.2.2). Significantly, the roads, along with almost all human settlement, tend to run along the bottoms of the valleys, leaving the ridges and mountain tops themselves mostly empty of human structures and habitation. Figure 1.2.2 Topography of Logan and Wyoming Counties 7
1.3. Stakeholders Citizens of Logan, Wyoming, [McDowell, and Mercer] Counties Local, State, and Federal Government Local Internet Service Providers Advancing Appalachia Initiative Shepherd University Dr. Michael Hieb The primary stakeholders of this project are the citizens of the aforementioned West Virginia counties. As they are the customers that can utilize broadband internet services, their potential benefits are the primary focus for this endeavor. The homework gap is a nationwide issue which is prevalent in areas that lack widespread adoption of broadband [2]. Schools and students will benefit significantly from the availability and adoption of broadband as it will improve the access to resources. Internet service providers in the area are also key stakeholders that may benefit from the research and conclusions found in this project. With the data provided in this document, internet service providers may be able to reassess expansion plans or marketing strategies. The service providers have the capability to make a difference by expanding wired and wireless services to those who don t have access. Government at the local, state, and federal levels are all included as stakeholders that can benefit from the output of this project. With the expansion of wired infrastructure, local government facilities may benefit from the increased broadband availability, bandwidth capacities, and overall throughput. At the state level, this could result in an overall increase in revenue or economic gains from increased wages and taxes from the result of increased broadband adoption. Subsidies for broadband expansion at the state and federal level will also cause their involvement. The FCC is specifically a key stakeholder for the federal level because of the research their programs that are aimed at improving broadband nationwide, as well as their Connect America Fund which provides qualifying areas and ISPs with subsidies. [20] Reconnecting McDowell is a collaborative effort that aims to improve the educational status in McDowell county. An effort in March of 2014 involved a heavy investment in providing young students with laptops as an effort to provide educational and technological support, and to help foster long term broadband usage. [44] The Advancing Appalachia Initiative and Shepherd University have been involved in research and progress of efforts to improve the broadband conditions in rural West Virginia. From George Mason University, Dr. Michael Hieb has served as the project sponsor. The results of this project looks to provide these parties with valuable data regarding the current status of the wired and wireless infrastructure coverage areas, along with potential coverage gaps or issues. 8
Stakeholder Demographics Wyoming and Logan counties are both rather sparsely populated, and rather poor. There are no major highways running through them, and only a few small towns (fig. 1.3.1). Figure 1.3.1 Logan and Wyoming Counties From the US census and Bureau of Labor statistics, we were able to obtain key demographic data points for each county (details of which can be found in Appendix C). Since the one of the goals of the project is to observe the potential providing more broadband access to consumers, key attributes to be closely monitored include total population, the population change rate, total jobs, unemployment rates, and poverty levels. Population decline in the southern counties of West Virginia is undeniably evident from the census data (fig. 1.3.2). One main reason for this is the decrease of coal usage in the recent decades, which caused a significant population and economic decline. Due largely to the departure of coal jobs, unemployment has risen and average incomes have stagnated. Both counties exhibit the same general demographic attributes of low incomes, high unemployment, and large percentage in poverty. 9
Figure 1.3.2 - Population Trend of Target Counties 1970-2010 A noteworthy demographic relevant to the adoption of broadband is the age group. The younger age groups are correlated to higher broadband usage and adoption [42]. A large proportion of the 18-24 age groups are not residing in these counties (see Appendix C.5). The age distribution shows large youth populations as well as large older adults that end up residing in the region. This infers that the majority of young adults are migrating to other regions to find work, which hints at the poor economic conditions. For the purpose of this project, the effects of different age groups and broadband adoption will be voided but can be expanded upon in further work. Incomes in this region are relatively low when compared to the rest of the state of West Virginia. Broadband adoption has been positively correlated with higher incomes [1], which may account for a good proportion of the lack of adoption from the citizens in the area. One of the major benefits of higher broadband is an increase in the per capita income, as well as an increase in overall employment for the affected regions. With an increase in disposable income, these citizens can invest in broadband services which theoretically will improve their incomes as the relationship between income and broadband adoption are positively correlated. The internet service providers (ISPs) for the region have an uphill battle to provide broadband services for this rural region. In general, rural regions provide an economically poor return on investment due to customers being geographically further apart and requires ISPs to cover larger areas. On top of population density problem, the ISPs in this region face the mountainous terrain of the Appalachian mountains and thick vegetation of West Virginia. The physical geography of this region pose possible issues for the construction and expansion of wiring, and undoubtedly causes issues for the coverage from cellular towers. Despite the poor historical adoption rate, several ISPs are continuing to build out cable, fiber, and wireless networks in these rural areas. Shentel is one of the leading providers in the area, and in 2016 alone they spent $200M in network upgrades and expansion. 10
In general, the ISPs are aptly aware of the digital divide in rural areas and are attempting to make financially viable efforts to provide services. While ISPs often have to shoulder the burden of service expansions, the Federal Communication Commission (FCC) does provide subsidies that can be applied for via the FCC Connect America Fund [19]. Those subsidies are essential in providing financial support to local ISPs in their efforts to expand wired and wireless services to rural areas in need. Stakeholder Tensions In some cases, the stakeholders may actually oppose each other in ways that can hinder the growth and progress of broadband expansion. One prime example is the set up of a cellular grade tower within a residential neighborhood. The service provided could undeniably benefit both the ISPs with revenue and consumers with broadband, however zoning laws or push back from local residents would ultimately hinder the implementation. Fixed wireless providers often use unlicensed frequency bands, which are relatively unaffordable for smaller providers such as Gigabeam Networks [39]. The larger providers such as AT&T and Verizon have purchased several frequency bands that fixed wireless providers use, and ultimately interlock both parties in short term progress hindering competition. Demographics The average incomes in the region are significantly lower than that of more urbanized areas even within the state of West Virginia. The average household income for Logan county, for example, is $35,465. This poses a large economic problem for the ISPs as broadband service is often viewed as an unnecessary expense for many citizens. Many citizens in the region reported that the cost of broadband is the primary factor in their reasons for not adopting, which is consistent with the results of a national study conducted by the Pew Research Institute [2]. As cost is a factor, so is the need for broadband services. Many users report that they simply don t need the actual service as they don t rely on internet services aside from what can be provided from a smartphone. 11
II. Assessment 2.1. Available and Future Technologies The commercial broadband service alternatives for this region are typically those that you would see in urban areas. However, due to the mountainous terrain and thick vegetation, the usage and expansion of such commercial broadband services incur some type of impediments. While there are a number of different technologies currently available, this project will also acknowledge future technologies that may be able to provide access for these rural communities. Detailed analysis will only be conducted for commercially available wired (cable/fiber) and wireless (cellular) commercial alternatives due to the heavy focus from ISPs and available data for those specific options. This immediate section will preface the major issues associated with each of the major available commercial alternatives. A more detailed analysis of each type of technology used can be found in Appendix D. DSL As indicated by the urban movement to abandon DSL, the DSL technology is considered a poor long term choice as the copper infrastructure becomes unsupported in the future. For now, ISPs are continuing to maintain but not expand DSL technologies due to a high subscriber rate in these rural areas. For DSL to function quickly and reliably, the end user s loop lengths (distance from user to Central Office (CO) hub) must be within the maximum distance supported by the service provider. For reference, the maximum distance Verizon supported is 18,000 ft, or 3.4 miles. The maximum speed of DSL is inversely proportional to the loop length, so as the length increases the transmission speeds decrease and reliability drops. In regards to its availability in rural areas, DSL is technically available to most homes with a phone line. Performance is the key issue with DSL as many subscribers report that their actual service is much slower than the advertised speeds. Cable Cable internet requires a physical coaxial cable to be connected from ISP s remote terminal to the end user s residential termination point. It is common in many home and makes it a very good last mile medium Terrain in rural areas can make it difficult to expand the wired cable infrastructure especially if the ISP is aiming to use direct burial cable. If available, cable provides good speeds for a moderate cost. A majority of the service providers are continuing to expand cable service to homes. With the advent of the new DOCSIS 3.1 standard, the cable medium should be able to reach download speeds of over 150 Mbps. Cable broadband is being expanded along with fiber, and the longevity of this technology is not an issue. 12
Fiber Like cable, fiber requires a physical fiber cable to be connected from the ISP s distribution fiber hub to the end user s residential termination point. Fiber is has actually become less costly to run than cable and shares the same issues for expansion. Its service is more expensive than other wired alternatives. As the cost of fiber is continuing to drop, service providers are actually installing fiber directly to the homes as they continue to expand their infrastructure. The theoretical upper throughput limit of the technology is currently around 10 Gbps at 550m, and is the ideal wired medium for the future. Cellular Due to the wide theoretical coverage areas and high data transmission speeds, cellular technology is an overall great solution for rural customers assuming they have decent reception to their residence. However, the reality is that the mountainous terrain of West Virginia significantly reduces the practical operating range of each cell tower and thus service dead zones are common. Because of the terrain, the tower setup can be more difficult than most typical installations and is likely to cost more. For smaller installations, a small cell type of setup can be implemented. This still ideally requires a fiber or cable backhaul, but can provide cellular service to a small community depending on the topography and wiring availability. Ideally, 4G LTE and the advent of 5G can provide users with wireless broadband service. However, even with unlimited plans the speed throttling enforced by the ISPs will deliver service at an unacceptably slow rate (3G speeds). Standard data plan capacities also cause the service to become significantly more expensive than wired broadband for any comparable service. Improvements in the cellular services by utilizing distributed antenna systems or other means, along with the availability of more affordable cellular plans that can provide the same level of service of wired broadband could potentially change the long term infrastructure landscape. Some regions in the US have used micro cell sites, which provide improved signal strength to a small immediate area. Micro cells are popular in many urban locations, however can be applicable to rural areas if setup properly. These micro cells fall under a particular topology called Distributed Antenna System (DAS), which are implemented to provide coverage to buildings or small geographic areas. Fixed Wireless Fixed wireless is wireless broadband using a stationary directional antenna on the customer side, pointed directly at the nearest available service tower. Line of sight becomes the primary factor in the success or failure of a fixed wireless installation. The terrain in West Virginia is mountainous and full have thick vegetation, which hinders the ability for many rural users to utilize this technology. In flat regions with little to no physical obstacles, fixed wireless can reach 13
customers up to 35 miles away. Fixed wireless providers such as Gigabeam Networks are relatively small and rely on utilizing unlicensed frequencies to reduce their operating costs. Larger companies such as AT&T have purchased frequency bands which forces direct competition. Expansion of fixed wireless towers can be hindered by that competition and the lack of customers within operating range. At the time of this writing, AT&T is currently preparing for a rollout of fixed wireless to 17 states in the US, of which West Virginia is not included. Satellite Geosynchronous satellite internet provides a feasible solution for rural users to get access to the internet and is the most available alternative for many users. The design of these geosynchronous satellite internet systems is inherently flawed in regards to its latency, which will cause packet loss and a poor internet surfing experience. Along with those issues, because of its operating range in the higher frequency Ku/Ka band, it lacks penetration power and its susceptibility to inclement weather also creates a problem for this service. Lastly, the cost of satellite internet is high and subsequently is not a viable option for many low income rural customers. Future Technology There are a number of future technologies that are aimed at bridging the digital divide worldwide. Most of the the endeavors are focused on providing internet to the world s most desolate regions that lack any broadband options. While rural West Virginia does have availability to several broadband options, some of these future technologies help bridge the gap. Details and specifications of each future technology identified in this study can be found in Appendix D. Because of the longevity, maturity, and available specifications of the low orbit subsidized satellites and 4G balloon projects, those two will be the only future options that are considered in the comparisons with commercial alternatives. Low Orbit Subsidized Satellite Several large companies intend on providing subsidized broadband access via low orbit satellites to remote areas all over the world. The low orbit constellations will provide much lower latency than conventional geosynchronous satellite systems and higher throughput. The ground unit will be a user terminal that is purchased by the user or small community. The major leader in this effort with clear results and estimates is currently OneWeb. OneWeb is a collaborative effort that is sponsored by large companies like Virgin Group,Airbus, Softbank, and many others. SpaceX has also recently entered the space with their own plans for a satellite constellation. Early estimates predict that the OneWeb service will be available for usage by 2019 for select educational markets, and full commercial availability by 2027. 14
Balloon 4G LTE (Project Loon) By mounting solar powered radios onto balloons, project Loon aims to cover targeted regions with 4G coverage. Balloons are launched at strategic locations that utilize known air current patterns to establish estimated coverage paths. The project has been in the testing phase since 2013, with published estimate of commercial deployment. Generally, it has been met with positive reception in its testing regions. Table 2.1.1 Technology Summary Table 2.1.1 compares the commercially available alternatives, along with key future technologies to display key attributes. Cable and fiber provide the best speed/cost ratios, although their costs are higher than some of the other options. Satellite provides the poorest performance in all categories. Balloon 4G LTE and subsidized low orbit satellite offers promising speeds and a great speed/cost factor, however are still not commercially available. Subsidized satellite service offered by OneWeb has an early estimate of coverage, yet consideration of this option as a viable long term solution should be taken with caution. Latency of the alternatives also are a key attribute to be considered. Typically, cable and fiber provide the best latency while satellite provides the worst. Poor latency results in packet drops and overall unreliability, as well as an overall unpleasant internet usage experience. Generally, online applications will experience moderate issues when latency speeds exceed 150 ms, and significant failures occur after 500 ms. Robust and consistent latency speeds are also critical to ensuring reliable service as well, and can influence service retention. Appendix figure D.4 provides further comparisons between the various service types and their associated latencies. 15
2.2. Current Status GIS Introduction In order to discover the various spatial relationships of addresses and connected areas and such, we used Geographic Information Systems (GIS) software. In particular, three functions were important for the analysis: buffer, viewshed, and count points within polygon, in addition to basic functions such as finding unions and intersections. With buffering, all the area within a given distance of a line, point, or polygon is put into a new buffer polygon, which can then be dealt with and manipulated as any other shape. In a viewshed analysis, for a given point, using elevation data, all the areas which are visible from that point are calculated. This can be done for one point, or in aggregate, with the number of points that can see each area as a variable output. Shentel provided us with address data for Wyoming and Logan counties, as a set of point coordinates. The number of these points in each polygon could then be counted. Wired Connectivity Methodology and Data The primary data source used for the wired connectivity analysis was GeoTel, which provided polygons for the the two counties in our scope, as well as the two de-scoped counties, categorized by connection type and ISP. Shentel also provided us with the locations of their fiber and cable lines in Wyoming and McDowell counties. In order to verify the GeoTel data, we approximated the coverage polygons by drawing a 500-meter buffer around the Shentel wires. While 500 meters is, in fact, a fairly long distance to connect a house, there is uncertainty in exactly where along the roads the wires are, and exactly where, in relation to the road, the buildings are at each address, that a larger distance than strictly. The wide buffer allows for this. Additionally, the population is so clustered around the roads that the width of the buffer had very little effect on how many addresses were included in it. When the Shentel-derived polygons were compared with the GeoTel polygons for these two counties (see Appendix B), the results were, in fact, quite close; the only major difference was an area in northwest Wyoming County, which GeoTel showed as empty. However, on inquiry, Shentel reported that that area is in the process of being connected, and coverage has not yet 16
begun. This area was then manually added to the GeoTel map, and can be seen in fig. 2.2.1 as Fiber coverage imminent. Because many of the GeoTel polygons are non-contiguous, and in many cases, even addresses across the street from a GeoTel polygon are excluded, we also made a 500-meter plausible expansion buffer around each polygon, as an estimate for not only areas that might be covered in principle though not yet hooked up, but also the easiest areas for wired connectivity to expand into in the immediate future. Results Figure 2.2.1 Wyoming County is largely covered by cable. Logan County is somewhat more sparsely covered, again mostly by cable, though Asymmetric DSL (ADSL) is available in most areas (fig. 2.2.2). Fiber is not currently available in most of the area. 17
Current Wired Reasonable Expansion Wyoming Logan Total Wyoming Logan Total Addresses 13004 17680 30684 13004 17680 30684 Cable 7725 4827 12552 9917 9611 19528 Good Connection Fiber 154 25 179 288 72 360 SW Wyo. Expansion 0 0 0 1269 0 1269 Total Cable/Fiber 7879 4852 12731 11474 9683 21157 No Cable/Fiber 5125 12828 17953 1530 7997 9527 DSL Total ADSL-total 5863 12695 18558 5863 12695 18558 ADSL-only 847 8047 8894 162 4231 4393 Connected 8726 12899 21625 11636 13914 25550 Not Connected 4278 4781 9059 1368 3766 5134 Cable/Fiber 61% 27% 41% 88% 55% 69% Percent ADSL-only 6.5% 46% 29% 1.2% 24% 14% Connected 67% 73% 70% 89% 79% 83% Not Connected 33% 27% 30% 11% 21% 17% Table 2.2.1 When cable, fiber, and DSL are all taken into account, a pretty consistent 70% of the addresses in these two counties are currently in a covered area, which can be expanded to 80 90%. Wireless Connectivity Methodology and Data Like with wired connectivity, the major data source for wireless was GeoTel. In this case, the polygons were given by ISP and part of the spectrum used. In order to verify this data, we created viewsheds using FCC data for the location of each cell tower in the area, and USGS elevation data. Unfortunately, the results were not helpful; while there was a general correlation between the number of cell towers in view and the density of GeoTel wireless polygons (see Appendix B), the correlation was weak enough so as not to be very useful in any but the most general sense. On one hand, there are reductions in signal strength due to vegetation and buildings, and on the other, there are unexpected increases due to multipath signals, where the signal bounces off of objects to go places outside of the usual line of sight. When these two opposing factors are combined, the resulting coverage area calculations are far too complicated, and rely on too much unavailable data, for our research. 18
Results Figure 2.2.2 As seen in figure and table 2.2.2, AT&T is the dominant provider here, covering over 90% of the area. Verizon has a significant presence in central Logan County as well; Wyoming County is less thoroughly covered, and gets a little spillover from nearby areas for Verizon and US Cellular coverage, as well as WVVA.NET fixed wireless. Wyoming Logan Total Wyoming Logan Total AT&T 10801 17320 28121 83% 98% 92% Verizon 130 6566 6696 1.0% 37% 22% US Cellular 27 0 27 0.2% 0% 0.1% WVVA.NET 32 0 32 0.2% 0% 0.1% No Coverage 2087 360 2447 16% 2.0% 8.0% Addresses 13004 17680 30684 Table 2.2.2 19
Significantly, the GeoTel data is on a scale of tens of meters; even in comparatively flat areas, there are cellular dead zones much smaller than this; there is no available data for us to analyze that level of detail. Between the terrain and the dense forests, there are certain to be many dead spots with no coverage in the counties, so the wireless coverage is a best-case scenario. While this is a significant problem for cell phones (and a problem which is beyond the scope of this report), it is probably a somewhat smaller issue for domestic wireless broadband; a stationary wireless receiver can be positioned, in principle, anywhere on a given property, and connected via Ethernet cable or WiFi to the rest of the house, in the same manner as terrestrial wireless. Overall Connectivity Combining wired and wireless coverage onto one map, we find that by far, the majority of the area is at least nominally covered (fig. 2.2.3). The main gaps in coverage are along the northeastern end of the Logan-Wyoming border, which is sparsely populated. Figure 2.2.3 20
Cable/Fiber Wireless Wyoming Logan Total Wyoming Logan Total Now Available 6908 4852 11760 53% 27% 38% Soon Available 1754 4802 6556 13% 27% 21% None Available 1280 7673 8953 10% 43% 29% Now None 971 0 971 7.5% 0% 3.2% Soon None 1841 29 1870 14% 0.16% 6.1% None None 250 324 574 1.9% 1.8% 1.9% Addresses 13004 17680 30684 Table 2.2.3 As can be seen in table 2.2.3, only eight percent of the addresses in these two counties are currently without any form of broadband internet access, at least nominally, and this should be able to be shrunk to less than two percent without extraordinary difficulty. This is a rather remarkable result, which is not at all consistent with the stereotype and perception of rural West Virginia as a place without broadband internet access. Adoption Rate The disconnect in the adoption rate of broadband was the critical discovery of this project. Adoption rate, as defined by the Organization for Economic Cooperation and Development (OECD), is the amount of population that adopts broadband divided by the population with access. While the coverage summary in table 2.2.3 indicates nearly 98% of total wired and wireless coverage for Wyoming and Logan counties, the actual adoption rates appear to have no correlation. From Shentel, the largest provider in the area, we were able to ascertain that the region had approximately 33%, while Logan had a total of 14% [38]. These values are from Shentel only, and actual values may vary slightly. By comparison, Washington D.C. touts an adoption rate of 76% [40]. Discussed later in this paper will be the links between broadband adoption and economic growth. Due to the late discovery of adoption rate as a key factor, the scope of the project was slightly altered. In section 2.3, we discuss the economic model effort that aimed to address the effects of broadband adoption on a local economy and demographics. The key objectives of the economic models were to provide a visualization of relationships, and to provide a relative estimate of affected demographics and cost values. From a nationwide survey conducted by the Pew Research Institute [43], the leading reasons for not adopting home broadband services include monthly cost, device cost, and smartphone being sufficient for their needs. A major note in that study is that service unavailability only accounted 21
for 5% of the responses. More specifically to the region, a similar study conducted by the L.R. Kimball for the state of West Virginia [13], the main reasons for not having service included cost, poor connection quality, and also service unavailability. 22
2.3. Economic Model Model Development As an original goal of the project, the relationship between the expansion of the service and costs were to be displayed. The economic analysis component was originally aimed to provide the financial and economic issues related to the availability and expansion of broadband. As we discovered that adoption rate had become a critical factor in the overall project, the model was then aimed to incorporate the adoption rate along with the downstream effects on population, employment, and per capita income. Not only does it provide a relative estimate for the broadband adoption effects, it also serves as a valuable tool in visualizing relationships between the tested variables. This project used a dynamic systems model to accomplish the goals previously mentioned. Many studies indicate that there is a positive correlation between the adoption of broadband and economic improvement. This project uses the aforementioned assertion of broadband adoption and increased economic indicators as a basis for the development of the economic model. The key indicators that this model aimed to predict are the total employment, population, and per capita income. The development of the model component was dependent on the validation of the links between the variables, as well as the general behavior of key components within the model. Due to the extreme complexity of the issue, the model was scaled down to incorporate only a few key variables to be used. Another precaution that was taken for clarification is the separation of wired and wireless expansions, which was addressed by creating separate models for each type. With data provided from our interview with Shentel personnel [37], we were able to ascertain the general adoption rate behavior to be modeled for the wired expansion, as well as wireless expansion patterns. Model Behavior and Design Each of the wired and wireless models are designed to evaluate the effect of broadband expansion into an area which has inadequate or unavailable broadband service. To localize and bound the model space, we decided on a design that effectively only models the immediately affected population. Using the wireless model as an example, the total population of the model is simply the amount of users that would potentially be able to acquire broadband service with the introduction of a new cell tower. Likewise, the overall effects of the increase in income and jobs available are representative of the overall adoption benefits for the affected population in the immediate area. To simulate the variance in customer retention, a gaussian distribution was utilized to produce a pseudo random number generation that partially drives the customer increase variable. The initial causal loop diagram provided a basis for the general structure of the wired and wireless models, but further investigation and behavioral differences caused the actual structures to vary once developed. An external broadband adoption variable was set in 23
place to acknowledge the growth of adoption from other sources. For simplicity, this is set to 0 for this project, but can be further improved upon in later work. Details for each of the individual model behaviors can be found in appendix sections E.10 and E.11. Verification of the model behavior and outputs can be found in appendix section E.7. Figure 2.3.1 - Initial Causal Loop Diagram As indicated in the Brookings study [4], for every 1% of broadband penetration there is an increase of 0.2-0.3% of employment. From the a collaborative study conducted by Oklahoma State, Mississippi State, and University of Texas, for every 10% of regional broadband adoption rate the per capita income increases by approximately 1.2%. The availability of training and education is linked to increasing adoption rate by approximately 9% over the course of 5 years, which was extrapolated to 1.8% annually for the purpose of this model [5]. Subsidy value scales positively with the likelihood of broadband adoption, where for every $10 of subsidy available the adoption rate likelihood increases by 3% [6]. After the increase in broadband adoption, there is an approximate lag period of 24 months before the economic effects are evident in income gains [7]. Costs associated with expansion were provided by Shentel [38] and tower construction documentation [9]. The calculations for expansion costs used in the model can be found in Appendix D.3. Demographic Data Integration Each county has different demographics, and those differences are incorporated in the model by altering constant placeholders within the calculation equations. Specifically, the variables that are affected are the initial values of the total population, population change rate, total customers, jobs, and average income. For the wired model, the total eligible customers varies due to variance in the household size in each county. The wireless model is isolated from that variance as it looks at the entire population rather than households. Population density also affected the wire model, as it changes the amount of customers available (total population) based on each linear mile of wire that is expanded. 24
Figure 2.3.2 - Wired Economic Model Figure 2.3.3 - Wireless Economic Model 25
Figure 2.3.4 - Broadband Adoption from Wireless (Left) and Wired (Right) Models In figure 2.3.4, the broadband adoption for each of the models are displayed from simulations with identical initial conditions. The major result to be acknowledged is the long term difference between the overall adoption rates. This is due to the nature of the adoption rate calculation, which is simply the amount of population with service divided by the total population. A wired broadband plan provides access for a whole home, whereas typically a wireless plan usually only provides broadband access for a single user. In the wired model, we can see that there is an initial drop in the adoption rate. This is an artificial representation caused by the calculation of the adoption rate which takes into account the amount of existing customers divided by the amount of customers who now have access. Once the wiring is completed, there is a lag time of one calculation step where the population available is now significantly higher due to the expansion and the adoption spike hasn t taken effect. Sensitivity Analysis Using the wired and wireless economic models, we performed a sensitivity analysis to investigate the impact of changes in the model variables on a base-case. In this case, we want to see the impact that the subsidized cost and the training available will have on the model outputs. We determined the base case values, where the subsidized cost and training are set to be zero. The value of base-case are summarized in table 2.3.1. Tables 2.3.1 Wireless (Left), Wired (Right) For the sensitivity analysis, we used the values from the case studies, which will be discussed below in section 2.4. Wireless Model First, to examine the sensitivity of the wireless model, we used the geographic and demographic conditions in Huff Creek Valley, in Wyoming County. We set the training available to be zero and increased the subsidized cost from 0% to 3%, 6%, 9% and 12%. The effects of the above changes are summarized in tables in Appendix E.2, and below in fig. 2.3.5. Without available training, for every 1% increase in subsidized cost, the total jobs increases 1.9%, the average income increases 0.54%, the adoption rate increases 1% and finally the cell revenue increases 4%. When training is available, however, for every 1% increase in subsidized cost, the 26
total jobs (employment) increases 2.8%, the average income increases 0.77%, the adoption rate increases 2% and finally the cell revenue increases 6%. Figure 2.3.5 Comparing these, we notice when training is available, every 1% increase in the subsidized cost, the total jobs increases 0.84% more than when training is not available. Similarly, with training available the average income increases 0.25% more than when training is not available. Also with training the adoption rate increases 1% more than when training is not available. Lastly, with training available the cell revenue increases 1.8% more than when training is not available. Wired Model For the wired model, we used the values for Northwest Logan County. We have performed similar runs where we set the training available to be zero and increased the subsidized cost from 0% to 3%, 6%, 9% and 12%. The effects of the above changes are summarized in Appendix E.2, and fig. 2.3.6. 27
Without training, for every 1% increase in subsidized cost, the total jobs increases 6.6%, the average income increases %1.67, the adoption rate increases 5% and finally the cell revenue increases 15%. But when training is available, for every 1% increase in subsidized cost, the total jobs increases 8.5%, the average income increases 2.15, the adoption rate increases 6% and finally the cell revenue increases 19%. Figure 2.3.6 Comparing these two again, we notice when training is available, every 1% increase in the subsidized cost, the total jobs increases 2% more than when training is not available. Similarly, with training available the average income increases 0.48% more than when training is not available. Also a similar pattern can be seen in the adoption rate, when training is available the adoption rate increases 1% more than when training is not available. Lastly, with training available the cell revenue increases 4% more than when training is not available. 28
2.4. Case Studies To assess the financial and economic impact of broadband expansion in the region, we selected two locations which could be evaluated using the wireless and wired models, respectively. Of these two case study areas, shown in figure 2.4.1, Huff Creek Valley, in Wyoming County, lacks wireless coverage, and Northwest Logan County lacks most wired coverage. We were unable to identify any significantly large area that lacked both wireless and wired coverage. Figure 2.4.1 Huff Creek Valley - Wireless Expansion Geography and Demographics Huff Creek Valley is located in the northwest of Wyoming County. Because of its location and the steepness of the surrounding terrain, the valley is almost entirely shadowed from the cell towers surrounding it (fig. 2.4.2). Because of this, it lacks all wireless coverage (though cable broadband is available in most of the valley). 29
There are about four hundred addresses in Huff Creek Valley, mainly in the west and center of the valley, all of which lack any wireless connectivity, according to the GeoTel data. The area includes parts of the small unincorporated communities of Lacoma and Cyclone. Figure 2.4.2 Wireless Expansion In order to fill the void of wireless coverage in Huff Creek Valley, three potential tower sites were chosen. For each of these sites, a viewshed was run, assuming a thirty-meter tower (fig. 2.4.3). The third site, shown in blue, was chosen as the optimal site, as its viewshed covered the largest portion of Huff Creek Valley. 30
Figure 2.4.3 Because viewshed is a mediocre approximation of wireless coverage, a 500-meter buffer was then created around this viewshed to account for multipath signals. This buffer is the estimated coverage area (fig. 2.4.4). 31
Figure 2.4.4 Of the 404 addresses in Huff Creek Valley that are not currently within the wireless coverage area, 314 78% are within the estimated coverage area (table 2.4.1). Additionally, of those who are not covered by the new tower, only five addresses lack cable access. Covered by new tower 314 78% Not covered by new tower 90 22% Total 404 Table 2.4.1 Economics New tower cost $225k Predicted 10-year revenue $1.23M Predicted 10-year adoption rate 29% Table 2.4.2 32
With the installation of the new cellular tower, a total of 314 addresses were introduced as the population of viable customers. This resulted in an overall adoption rate of 29% over 10 years, and a total revenue of $1.23M for the ISPs from this market alone (table 2.4.2). The data shows a lower yield of broadband adoption compared to wired expansion, yet a higher yield in overall revenue because of the low population coverage a single broadband device provides and the higher individual cost. Northwest Logan County - Wired Expansion Geography and Demographics Northwest Logan County has a small population, spread over many valleys exactly the situation least conducive to wired broadband connection. And indeed, there is almost none; only DSL connection is available, and that only in parts of the area. There are approximately 1500 addresses in Northwest Logan County. The unincorporated communities of Shively, Halcyon, and Whirlwind are in this area, though they do not correspond to any significant concentrations of population. Wired Expansion Aside from the DSL coverage, Northwest Logan County does contain one bit of wired infrastructure: a fiber line, on the western edge of the area (owned by Lumos), which is currently being used only as a pass-through. Because the line is already there, connecting it to the local community is possible, in principle. To supplement the fiber line, we drew two additional lines on the map, as cable expansion. Line 1 runs from the northern end of the fiber pass-through to the southeast, and line 2 from the south end to the northeast. 500-meter buffers were then drawn around the fiber line and the two new lines, to identify nearby addresses. There is some overlap between the two wires. This is shown in fig. 2.4.5. 33
Figure 2.4.5 While currently, no one in the area has any cable or fiber connection, with all three wires connected, nearly half would (table 2.4.2). As is; no fiber connection With existing fiber line connected No new wire New wire 1 New wire 2 Both new wires Cable/Fiber 0 0% 176 12% 420 28% 548 36% 703 46% ASDL only 426 28% 381 25% 307 20% 274 18% 209 14% No connection 1088 72% 957 63% 787 52% 692 46% 602 40% Table 2.4.2 Economics Naturally, additional wire routes could no doubt be productive as well, though with diminishing returns; the new wires do carry a significant cost (table 2.4.3). However, our model predicts that the revenue from the newly-connected areas should be sufficient to cover the costs of expansion within a few years. 34
New wire 1 New wire 2 Both new wires New wire length (km) 17.25 12.87 28.22 New wire cost $323k $241k $529k Wire cost/km $19k Predicted 10-year revenue $1.6M Predicted 10-year adoption rate Figure 2.4.3 - Summary of Economic Model Outputs 44% Case Study Summary The models were conducted using an optimistic case of subsidy inclusion (3%) and training availability, which may cause a discrepancy from expected adoption rates and revenue estimates from proprietary studies conducted by the ISPs. Using the 10 year summaries of the wired and wireless models, it can be concluded that the investment of these respective infrastructures will yield a positive return. Again, based upon adoption rates alone it appears that the wired infrastructure will yield a higher outcome adoption rate, with a lower margin of return. The larger gains in the average income and employment of the wired implementation can be seen by comparing the model outputs in appendix figures E.5 (Wired) and E.4 (Wireless). This is relatively consistent with the expected behavior as mentioned in section 2.3 when comparing the economic two models. 35
2.5. Alternatives Analysis The two main commercial alternatives that we are analyzing are the cable/fiber (wired) and mobile cellular (wireless). Cable/fiber and cellular have the most potential to improve the adoption rate and landscape because the local service providers have the existing infrastructure and financial prospect to support such expansions. While other alternatives and future technologies could be considered, the behavioral and fiscal data available for only these two alternatives is sufficient for use with the economic models developed. In reality, there are no areas of significant population in this region that lack access to both wired and wireless broadband connectivity. For this analysis, we are putting the initial values from the wireless analysis of Huff Creek Valley into the wired model for comparison. This is simulating a region with Huff Creek Valley s population, with a cable expansion of 12 km to provide wired service. To compare the available alternatives, we utilized each of the economic models with the same initial values to simulate an identical start point. For the wired expansion, we are using cable ($75k/Mile) rather than fiber ($30k/Mile) to simulate the more costly scenario. It is noted that these expansion values do not reflect prior costs to lay backhaul wired infrastructure to the initial site of installation. Comparison of Wired vs Wireless Expansion Simulations Initial Value Wireless Wired BB Adoption Rate 0.1 0.29 0.39 Total Jobs 201 348 342 Average Income $19,009 $23,169 $23,773 Income Gain 0% 21.80% 25.10% Total population 745 825 842 Expansion Costs 0 $225k $625k Revenue 0 $1.23M $715k Table 2.5.1 Summary of 10 Year Simulation Results The results of the 10 year simulation show a noticeably higher broadband adoption rate for the wired model. This is also evident in the downstream effects of the broadband adoption on the income, jobs, and population. As mentioned earlier, the adoption rate for wired broadband is generally higher due to a single service connection providing for a complete household, while a single mobile broadband only covers a single user. However, from a financial perspective the 36
cellular expansion would yield a higher long term revenue with a significantly lower cost of expansion implementation. The wired revenue accounts only for broadband service, which reflects the economic opportunity for the ISPs to provide cable television. 37
III. Conclusion Broadband internet has been proven to improve economic conditions in other rural regions of the United States, and can undoubtedly provide similar benefits in rural West Virginia. The issue of the broadband adoption is a prevalent concern that many parties have tried to solve. While the results of this project has proven that the actual broadband availability in the region is relatively adequate based on current coverage estimates, the broadband adoption rate is the critical factor to be focused upon. To address the adoption rate, there should be a focus on the training and technical education of the rural customers. The Reconnecting McDowell initiative is an example of an effort that aims to improve the economic conditions by providing technology and information to those who can benefit. The availability of subsidies for broadband from the FCC has existed for decades. Providing the eligible residents of this region with information on how to take advantage of subsidies, along with technical training and education should improve the overall adoption rate as indicated in the sensitivity analysis. The increase of subsidy availability should be lobbied along with infrastructure aiding legislation. Compared to urban areas such as Fairfax County, wired broadband availability in rural West Virginia is inadequate and requires more attention. Based on the results of this project s models, the wired buildout should generate a higher broadband adoption rate when compared to new cellular installations. Cable and fiber infrastructure should be able to provide a future proof delivery medium for ISPs, providing a better return on investment and incentive to continue to expand. Cellular broadband has the potential to provide wireless broadband services to many customers with a larger margin of profit. Utilizing microcell structures or distributed antenna systems for small communities could provide improved service and higher adoption rates as well. Due to current mobile plan data throttling and monthly capacities, mobile broadband services become too expensive and slow to be a true replacement for wired broadband services. By improving the service availability and service plan structure could result in mobile broadband being used as a viable home broadband alternative. Consumers should find the best broadband option available based upon reliability, speed, and cost in order to utilize internet services and benefit from access. Our analysis from shows that the best wired solution in the near term is cable or fiber if available as they continue to build out infrastructure. We have also analyzed future options and find that subsidized low orbit satellite and balloon wireless could provide valuable broadband service as it becomes commercially available. 38
Future Work Results of this project indicated that there should be more of an effort towards the adoption rate issue. To improve upon the adoption rate, there should be further investigation into government (local, state, and federal) subsidies or programs that either consumers and/or ISPs can be eligible for receiving. The issue of low rural broadband usage is not isolated to West Virginia. There have been other successful initiatives conducted around the world, and their results could provide valuable insight on how to better improve the region s broadband status. Service providers around the world continue to produce possible solutions to help bridge the digital divide. Closely following these solutions and providing support when possible may help bring a viable solution to commercial availability for the consumers of rural West Virginia. Acknowledgements Most geographic data was acquired from GeoTel, Shenandoah Telecommunications (Shentel), the US Geological Survey (USGS) and Federal Communications Commission (FCC). Provider costs, adoption rates, and general information were provided by Shentel and GigaBeam Networks. Modeling support was provided by Dr. Syed Abbas Zaidi of George Mason University. 39
Appendices A. Initial Requirements Elicitation UID Functional Requirement Description Implemented In 1 General System Requirements 1.1 System shall address the internet conditions for Mercer, Wyoming, and McDowell Counties of West Virginia. The named counties will be referred to as the "target regions" for this project. 1.2 System shall list the existing Internet Service Providers in the target regions. 1.3 System shall provide the limitations of bandwidth, upload speed, and download speed of each Internet Service Provider available. 1.4 System shall provide at least one internet service solution for all households in the target region. GIS Model Alternatives Analysis Alternatives Analysis Alternatives Analysis 2 Analytics Requirements 2.1 System shall create a matrix correlating the available services against monthly cost and download speeds. 2.2 System shall use a quantitative metric to compare different internet service alternatives. Economics Analysis Economics Analysis 3 Technical Requirements 3.1 System shall provide a Geographic Information System(GIS) model displaying the target regions with current existing wired and wireless infrastructure. 3.2 System shall provide a Geographic Information System(GIS) model displaying the target regions with a proposed expansion. GIS Model GIS Model 4 System Security Requirements 4.1 System shall provide internet service options that provides a direct connection to the service provider. Alternatives Solutions Analysis 40
B. GeoTel Data Validation Figure B.1 The GeoTel data and Shentel-derived buffers largely line up. The two main areas where they don t are in Huff Creek valley, where there are actually some very small GeoTel polygons, and in southwest Wyoming County, where Shentel service is due to begin very soon. 41
Figure B.2 Here, the GeoTel purple coverage polygons are barely correlated with the green FCC/USGS/viewshed polygons. Not only does each tower have some unidentified number of antennas on it, and thus can be participating in multiple spectrum/isp combinations, multipath signals and vegetation shadowing complicate everything beyond our ability to simulate. 42
C. Demographics of Counties Figure C.1 McDowell County Demographics Figure C.2 Mercer County Demographics 43
Figure C.3 Wyoming County Demographics Figure C.4 Logan County Demographics 44
Figure C.5 and C.6 Wyoming and Logan Age Histograms Figure C.7 and C.8 McDowell and Mercer Age Histograms Figure C.9 Region 1(WVA) Broadband Survey Results 45
D. Detailed Data for Alternative Solutions D.1 - Current Technologies DSL (A)DSL stands for (Asymmetric) Digital Subscriber Line. Like dial-up, DSL utilizes the phone line to communicate from the end user and the central office (CO), also referred to as hubs or point of presence. While it uses the antiquated phone line infrastructure, using newer frequency modulation techniques DSL can provide download speeds up to 25 Mbps. The primary driver of high speeds and reliability for DSL is the loop length. As the loop length increases, the speeds decrease. As service providers move away from hardwired copper lines, central office hubs are becoming more obsolete and unsupported in urban areas. In the rural areas, DSL usage is relatively high and many providers continue to support this technology. Cable For cable broadband, the end user utilizes an external modem which communicates with the service provider s hub in a multi-drop configuration with other end users. The service provider s hub is usually connected via fiber optics to provide larger pipe, while the end user is connected to the hub via a coaxial cable. Currently, cable is one of the most reliable technologies for wired connections and it provides speeds up to 100 Mbps. With the advent of the new DOCSIS 3.1 standard, the cable medium should be able to reach download speeds of over 150 Mbps. Cable broadband is being expanded along with fiber, and the longevity of this technology is not an issue. Fiber Fiber optics is currently the fastest commercially available technology for consumers. The basis of the fiber optic technology is the transmission of light pulses via fiber optic cable between fiber transceivers. In commercial deployments, users have reported speeds of up to 500 Mbps download using multimode fiber. Theoretically, current multimode fiber can transmit up to 10 Gbps with a loop length of 550 m, and 100 Mbps at 2000 m. Most modern systems currently utilize fiber for their backend, but for the last mile to the consumer other types of technology are used. Fiber technology offers tremendous speeds for consumers if the service available. Not only is the speed impressive, but due to the nature of the technology its environmental resilience and reliability is also noteworthy. By utilizing other communication mediums such as cable or DSL for the last mile to the consumer, fiber offers the highest performance for those who have access. Cellular Cellular technology is relatively commonplace knowledge for most end users. Cell towers are outfitted with radio antennas connected to the service providers (via wired backhaul) that connect the remote cell phones to the service providers. The cell towers conventionally serve in a 46
hub and spoke configuration with 360 degrees of coverage by utilizing several sector antennas. The coverage tends to vary based upon altitude and physical objects between the tower and end user, as well as the actual radio frequency used and its ability to penetrate objects and vegetation. Because this technology is based on RF, the line of sight between the two devices is fairly critical for the connection quality. The signal strength and integrity will degrade when the radio waves can t propagate clearly between the sender and receiver. Cellular technology has evolved through several iterations within the last few decades, which includes HSPA(+), 2G/3G/4G/5G LTE increasing in speed and capacity in each new standard. Cellular towers are positioned in strategic positions in order to maximize the service provider(s) coverage areas. Although towers realistically serve the surrounding areas in a circular pattern originating at the towers, cellular mapping networks are actually represented with hexagonal coverage areas in a honeycomb-like pattern. On average, 4G LTE mobile users can see 10 Mbps download speeds and peaks of 25 Mbps, and bandwidth capacities depend on the service provider and monthly plan terms. Unlimited plans tend to include a type of throttling after reaching a certain bandwidth, and are usually costly. The frequencies that 4G cellular radios operate vary between 700 MHz up to 2.1 GHz depending on the provider. That operating range allows for relatively decent penetration through vegetation and thin structural walls. However, the higher the frequency the less penetration power and coverage distance. Customers can try a cellular signal boosting device. This would require a cell phone plan OR a 4G Broadband router for their devices at home. Theoretically, this should be able to help by using a directional antenna aimed at the nearest cellular tower. In doing so, the signal strength is focused and should improve the signal. The antenna would be mounted above the home or at some high point on the potential customer s property. Installation for the device will depend on the service provider or on the customer themselves. Cellular Repeaters To improve coverage area and signal strength for users, a cellular repeater can be used. On the consumer side, there are products that customers can install at their home using an external antenna to reach the nearest cell tower and improve their connection quality. From the ISP side, repeaters can be installed at strategic locations to mitigate coverage gaps. A common installation technique is to use existing utility poles. The remote unit acts as an extension of the original network by using focused signals via an external (often directional) antenna towards the desired connection point. One of the strengths of this solution is that they can use the existing infrastructure of the power and utility poles. The vegetation and elevation changes causes issues for the line of sight between the remote unit and the cell tower. To combat the line of sight issue, repeaters can be mounted in regular intervals on the utility poles that create a daisy chained network. 47
Fixed Wireless Fixed wireless is essentially wireless broadband using a stationary directional antenna on the customer side, pointed directly at the nearest available service tower. This technology has the same strengths and weaknesses of mobile broadband, however by utilizing a fixed directional antenna the signal strength and reliability increases along with speed. Gigabeam in WVA provides 10 Mbps for $39, up to 50 Mbps for $79 with an unlimited data cap. A 2015 study showed Verizon provided a similar service with a 30 GB data cap. Signal Booster Implementation - Fixed Wireless To implement a 4G Cellular signal booster, several components are required. First, a high gain cellular antenna must be mounted on the residence pointed at the nearest cell tower that provides the best signal strength. Typically, the antennas used are high gain yagi or panel type antennas. In this example, we are using AT&T as the service provider as the maps indicate they have theoretical coverage of most of these areas. AT&T uses a dual band technology to provide LTE service by utilizing frequencies in the 700 MHz range and 1.7 GHz Advanced Wireless Spectrum (AWS) range. For best results, the antenna to be utilized should be compatible with both bands. When mounting the antenna, the antenna should be placed as high as possible and clear from any obstructions that can cause internal reflection or infraction into the fresnel zone. The antenna is then connected to the cellular router ideally using low loss shielded coaxial cable. The length of cable is proportional to the signal loss from the transmitter, so the cellular modem or repeater should be kept relatively close to the antenna. For reference, at 1700 MHz the cable induces approximately a 5.3 db/ft loss. Approximate Cost Breakdown of a signal booster installation* Cellular Antenna (700MHz, 1700MHz)- $100 Shielded Coax Cabling (LMR400-20 Ft): $20 Signal Booster (4G LTE, 1-2 Devices): $400 (Optional) Cellular Router (4G LTE): $120 Labor (4 Hours @ $50/Hour) - $200 Total Cost : $840 / Installation *Prices obtained from commonly used products for residential grade usage of May 2017 Key Assumptions: Assuming a minimum of -100dB for usable signal exists at the site of installation. There exists a location on the dwelling or property that is appropriate for an external antenna to be mounted. This type of implementation should be further detailed with a proper link budget analysis to account for the specific antenna/device signal gains and path losses. 48
To protect the consumer from a poor investment in an unreliable service, a feasibility or speed test should be implemented at the site of installation. For the ISP s financial protection, an annual contract may be enforced in order to receive such service as often seen with cellular or broadband plans. Satellite Satellite technology uses RF in order to communicate from the sender to receiver. The end user will have a satellite dish installed at the desired location, and will be pointed in a fixed direction towards a satellite constellation. The satellite constellation are generally composed of geosynchronous orbiting satellites. Because it operates in a much higher frequency band, the signal is significantly more susceptible to interference by clouds and precipitation. The path required for a round trip transmission is relatively long, which causes latency and reduces the overall transmission speeds. Latency issues will cause problems for many applications, packet loss, and an overall compromised broadband experience. Most satellite internet providers advertise speeds of up to 10 Mbps, however many users report roughly 6 Mbps or less. Improvements to satellite technology are slower to materialize due to the unavailability to physically upgrade satellite hardware and the cost of deploying new satellites. D.2 - Future Technologies Internet.Org / Facebook Free Basics Project Sponsor: Facebook, Samsung, Ericsson, MediaTek, Nokia, Qualcomm, Opera Software Project Status: Ongoing Technology Used: Amos-6 Satellite(Ka Band), Aquila UAV Drone(Laser) Start Date: August 2013 Current Coverage Areas: Countries in Africa, Asia, Latin America Future Coverage Areas: TBD Future Coverage Date: TBD Download/Upload Speeds: Bandwidth Capacity:? Issues with Deployment: Amos-6 Satellite destroyed on launch. Net Neutrality Issues in India Links: https://info.internet.org/en/story/connectivity-lab/ Project Loon Project Sponsor: Google, X Project Status: Ongoing Technology Used: High Altitude Balloon (4G LTE) - Launched using predictive wind current modeling to spread coverage. Each Balloon covers ~ 40km radius 49
Start Date: June 2011 Current Coverage Areas: New Zealand, Chile, Argentina, Australia, Sri Lanka, Vatican City?, India, Indonesia? Future Coverage Areas: WorldWide Future Coverage Date: TBD Download/Upload Speeds: 10 Mbps Down, Unknown Up - 4G Speeds Latency: 150 ms Bandwidth Capacity:? Issues with Deployment: Uncontrolled landing locations Links: http://cleanleap.com/update-googles-project-loon https://x.company/loon/ https://www.technologyreview.com/s/534986/project-loon/ https://arstechnica.com/information-technology/2014/06/google-to-deploy-180-low-orbit-sate llites-that-provide-internet-access/ SpaceX Satellite Constellation Project Sponsor: SpaceX, Google, Fidelity Investments Project Status: Starting - No deployment yet Technology Used: Satellite - Non-Geostationary orbit (Ku and Ka Bands), Low orbit Non-geosynchronous V-Band Satellites, User Equipment (Terminal Transceiver) to cost ~ $200 Start Date: 2015 End Date: NA Current Coverage Areas: NA, Testing to start in 2017 - Initial ground stations will be in Washington and California. Future Coverage Areas: WorldWide. 2 Sets of Satellites - V-band low-earth orbit(vleo) with 7518 satellites, Non-geostationary orbit satellites using the Fixed-Satellite Service in Ku and Ka bands with 4225 satellites. Future Coverage Date: TBD, Aiming for 2020 (Early Estimate) Download/Upload Speeds: Up to 1 Gbps, Unknown Up - 4G Speeds Latency: 25-35 ms Bandwidth Capacity: 1 Gbps/User, each Satellite can provide downlink capacity of 17-23 Gbps. First 800 satellites to provide complete US connectivity. Issues with Deployment: SpaceX acknowledges this is not a profitable business model. Links: https://arstechnica.com/information-technology/2016/11/spacex-plans-worldwide-satellite-int ernet-with-low-latency-gigabit-speed/ OneWeb satellite constellation Project Sponsor: Virgin Mobile, Airbus, Softbank, Coca Cola, Intelsat Hughes..many others Project Status: Ongoing 50
Technology Used: Circular Low-Earth Orbit (~750-mile Altitude) in Ku band. Initial 648 satellite capacity has been sold, considering adding 1972 satellites. Satellites will operate in 18 polar orbits. Ground Unit (~$250) to transmit to users. Start Date: 2014 Current Coverage Areas: N/A Future Coverage Areas: Worldwide Future Coverage Date: 2019-2020 (Aggressive Estimate). Aimed to cover disconnected schools by 2022, all users by 2027. Download/Upload Speeds: 50 Mbps Latency: 30 ms Bandwidth Capacity: Each satellite can generate 6 Gbps of throughput. Ground unit can provide internet at 50 Mbps. Issues with Deployment: Requires $3 Billion in capital before full deployment in 2019-2020. End of life concerns as space debris. Links: http://oneweb.world/#home https://qz.com/772972/the-speed-coverage-cost-and-future-of-satellite-internet/ AirGig by AT&T Project Sponsor: AT&T Project Status: Trials starting Fall 2017 Technology Used: RF Repeater on utility poles Start Date: Announced October 2016 End Date: NA Current Coverage Areas: NA, Testing to start in 2017 Future Coverage Areas: Rural areas, worldwide anywhere with utility poles Future Coverage Date: Trials in Fall 2017 Download/Upload Speeds: Up to 1 Gbps, Unknown Up - 4G Speeds Latency: Fiber Speeds, < 25 ms Bandwidth Capacity: Network capable of > 1 Gbps, users at 4/5G Speeds Issues with Deployment: Requires permits and rights to utility poles in line. Links: http://www.telecompetitor.com/att-airgig-trials-with-power-companies-expected-in-2017/ http://about.att.com/newsroom/att_to_test_delivering_multi_gigabit_wireless_internet_spee ds_using_power_lines.html Dynamic Spectrum and White Spaces Project Sponsor: Microsoft Project Status: Ongoing Technology Used: Directional Antennas using UHF Frequencies. Utilizes White Space frequencies not used by OTA television channels Start Date: 2009 51
End Date: NA Current Coverage Areas: Tested in various countries in Africa, Asia, UK, and North America. Commercially available in Ghana Future Coverage Areas: Rural areas worldwide Future Coverage Date: Download/Upload Speeds: Up to 30 Mbps,12-16 with non-line of sight (nlos) Latency: NA Bandwidth Capacity: NA Issues with Deployment: Areas with difficult terrain or thick vegetation still poses financial issues for backhaul expansions. Figure D.3 Latency vs Technology Summary Figure D.4 Download Speed vs. Cost Summary 52
Figure D.5 Consumer Recommendations by Time Frame D.3. Installation Costs Wired Cable: Installation Cost Per Mile = CPM c (wire per mile), Total Miles = TM c Cost of hookup(termination), limited to 300 = CH Number of Cable Hookups = NC Cost of Cable Equipment = CE c Wired Fiber: Installation Cost Per Mile = CPM f (wire per mile), Total Miles = TM f Cost of hookup(termination), limited to 300 = FH Number of Fiber Hookups = NF Cost of Fiber Equipment = CE f Wired Expansion Cost = Σ c,f ( TM c * CPM c + NC* CE c )+ ( TM f * CPM f + NF* CE f ) *Assuming CPM c and CPM f also includes cost of junction boxes and equipment Wireless Cell Tower: Number of Towers = NC Average ~ $30,000 Per Mile = CPM (cost per mile), Total Miles = TM Average Cost of tower construction ~$200K = TC Average cost of Cell equipment = $15K = CE i = Each unique cell tower instance Wireless Expansion Cost = Σ ( TM i * WPM i + TC i + CE i ) 53
E. Modeling Data E.1. Model Variables Table E.1 E.2. Sensitivity Analysis Results Northwest Logan (Wired) Table E.2 54
Huff Creek Valley (Wireless) Table E.3 Table E.4 55
E.3. Alternatives Analysis Wired vs Wireless Model Outputs Wired Model Outputs Figures E.1 Figures E.2 56
Wireless Model Outputs Figures E.3 57
E.4. Case Study Outputs Huff Creek Valley (Wireless) Model Outputs: Figures E.4 NW Logan County (Wired) Model Outputs 58
Figures E.5 E.5. Model Variables Definitions - Vensim Exports Wired Model (01) Average Income= INTEG ( Average Income * STEP(1, 24) * (Income Gain Pct/120) * (DELAY1(BB Adoption Rate,24 )), 18614) Units: **undefined** Average income increases proportionally to the income percentage and broadband adoption rate for the local area. (02) BB Adoption Rate= (DELAY1(Total Wired Customers,1))*2.19/DELAY1(Total Population,1) + 0*External BB Adoption Units: **undefined** Does not take into account existing broadband customers of other types. Using Average Household size(mercer=2.06,mcdowell=1.95,logan =2.19,Wyoming=2.45) (03) Customer Satisfaction= RANDOM NORMAL(0, 100, 50, 25, 90 ) Units: **undefined** (04) Eligible Customers= IF THEN ELSE((Total Population * 0.45) - Total Wired Customers > 0, (Total Population * 0.45) - Total Wired Customers, 0 ) Units: **undefined** Based on households, Wyoming ratio of households to population =.40 (05) Expansion costs= INTEG ( 300*(Total Wired Customers-DELAY1(Total Wired Customers,1)) + 30000*(Miles Expanded -DELAY1(Miles Expanded, 1 )) -0*ISP Expansion Costs, 0) Units: **undefined** 75000*Miles Expanded + Total Wired Customers*300 - ISP Expansion Costs*0 (06) External BB Adoption= 0 59
Units: **undefined** (07) FINAL TIME = 120 Units: Month The final time for the simulation. (08) Income Gain Pct= 1.2 * STEP( 1, 6 ) Units: **undefined** Income Gain per Mbps, in percentage Based on broadband penetration, see OEDC study. http://www.nardep.info/uploads/broadbandwhitepaper.pdf every 10% penetration = 0.9-1.5% gain per capita - Use 1.2 right in the middle. (09) INITIAL TIME = 0 Units: Month The initial time for the simulation. (10) ISP Expansion Costs= Expansion costs Units: **undefined** (11) ISP Revenue= INTEG ( Wired Revenue+ISP Revenue*0, 0) Units: **undefined** (12) Miles Expanded= INTEG ( IF THEN ELSE( Miles Expanded < 25, +12.5, 0 ), 0) Units: **undefined** 1 year to do 300 miles of hard wire = 25 Miles Per Month at max capacity. Use 12.5 as estimate. (13) Monthly Cost= 65 Units: **undefined** $65/ Month is the average cost of a basic cable broadband plan (14) Monthly Wired Revenue1= Monthly Cost * Total Wired Customers Units: **undefined** (15) Population Change Rate= -1.5 + ((Total Jobs - DELAY1(Total Jobs, 12)) / Total Jobs)*100*0.27 Units: **undefined** Calculates the change rate of population. Hard coded Values for percentage change,and then added with the job difference percentage that would offset some of the persistent historical population drop. Value of (0.27) would be the ratio of persons affected per population ( approximate current employment value). (16) SAVEPER = TIME STEP Units: Month [0,?] The frequency with which output is stored. 60
(17) Subsidized Cost= 0.03 Units: **undefined** Increasing factor of customer growth to account for subsidized cost Every $10 in subsidy 3% annual increase in adoption For this value, put in percentage, 3% = 0.03 (18) TIME STEP = 1 Units: Month [0,?] The time step for the simulation. (19) Total Jobs= INTEG ( Total Jobs*((0.0025/12)*BB Adoption Rate*100) + Average Income * 0 - (0.055 /12)*Total Jobs*0, 414) Units: **undefined** For every 1% of penetration, employment increases 0.2-0.3% per year. In the Brooking institute study, penetration was used as a proxy for adoption. https://www.brookings.edu/wp-content/uploads/2016/06/06labor_cran dall.pdf (20) Total Population= INTEG ( Total Population*(Population Change Rate/1200) + (Miles Expanded-DELAY1(Miles Expanded, 1 ))* 27 * 2.19, 20) Units: **undefined** 30 houses per mile passthrough in rural areas, communities every 30 miles Total Population*(Population Change Rate/1200) + IF THEN ELSE( MODULO(Miles Expanded,25 ) = 0, 1, 0 )*25*0 (21) Total Wired Customers= INTEG ( IF THEN ELSE(Customer Satisfaction < 5, -Eligible Customers*0.01/12,Eligible Customers *(0.01+Subsidized Cost+Training Available )/12 ) + (IF THEN ELSE( (Miles Expanded-DELAY1(Miles Expanded, 1 ))>0,1, 0)*(Eligible Customers/10)), 1) Units: **undefined** Customer Satisfaction will drive the rate at which users pick up the service or not - serves as a pseudo random variable for the system to help simulate real world variance. Broadband adoption for wired customers spikes for a small amount when service becomes available, then tapers off into slow growth. IF THEN ELSE(Customer Satisfaction < 45, -Eligible Customers*0.01/12,Eligible Customers*(0.01+Subsidized Cost+Training Available )/12 ) + (IF THEN ELSE( MODULO(Miles Expanded,25 )=0,1,0)*2.5)*0 (22) Training Available= 0.018 Units: **undefined** With training, 1.8%/year, or 9%/5 years (23) Wired Revenue= INTEG ( Monthly Wired Revenue1 + ISP Revenue*0, 0) 61
Units: **undefined** NOTE that this wired revenue is only for broadband. For ISPs, they may be able to incur greater profits by providing cable television on top of the wired infrastructure. Wireless Model (01) Average Income= INTEG ( Average Income * STEP(1, 24) * (Income Gain Pct/120) * (DELAY1(BB Adoption Rate,24 )), 19009) Units: **undefined** Per Capita (02) BB Adoption Rate= Total Cell Customers/DELAY1(Total Population,1) + External BB Adoption*0 Units: **undefined** Defined as total cell customers /total population. Does not take into account existing broadband customers of other types. (03) Cell Revenue= INTEG ( Monthly Cell Revenue1+ ISP Revenue*0, 0) Units: **undefined** (04) Cell Tower Lease Cost= 1800 Units: **undefined** (05) Customer Satisfaction= RANDOM NORMAL(0, 100, 50, 25, 50 ) Units: **undefined** (06) Eligible Customers = Total Population * 0.75 - Total Cell Customers Units: **undefined** (07) External BB Adoption= 0 Units: **undefined** (08) FINAL TIME = 120 Units: Month The final time for the simulation. (09) Income Gain Pct= 1.2 * STEP( 1, 6 ) Units: **undefined** Income Gain per Mbps, in percentage Based on broadband penetration, see OEDC study. every 10% penetration = 0.9-1.5% gain per capita - Use 1.2 right in the middle. (10) INITIAL TIME = 0 Units: Month The initial time for the simulation. (11) ISP Revenue= INTEG ( 62
Cell Revenue+ISP Revenue*0, 0) Units: **undefined** (12) Monthly Cell Revenue1= Monthly Cost * Total Cell Customers - Cell Tower Lease Cost Units: **undefined** (13) Monthly Cost= 71 Units: **undefined** $71/ Month is the average cost of a cell phone bill in 2016 (14) Population Change Rate= -0.8 + ((Total Jobs - DELAY1(Total Jobs, 12)) / Total Jobs )*100*0.4 Units: **undefined** (15) SAVEPER = TIME STEP Units: Month [0,?] The frequency with which output is stored. (16) Subsidized Cost= 0.03 Units: **undefined** Increasing factor of customer growth to account for subsidized cost Every $10 in subsidy 3% annual increase in adoption For this value, put in percentage, 3% = 0.03 (17) TIME STEP = 1 Units: Month [0,?] The time step for the simulation. (18) Total Cell Customers= INTEG ( IF THEN ELSE(Customer Satisfaction < 40, -Eligible Customers*0.01/12, Eligible Customers *(0.01+Subsidized Cost+Training Available)/12 ), 93) Units: **undefined** Customer Satisfaction will drive the rate at which users pick up the service or not. Setting the bar at 50% is relatively low in most areas however here it may be reasonable considering the area's history of very poor service. (19) Total Jobs= INTEG ( Total Jobs*((0.0025/12)*BB Adoption Rate*100) + Average Income * 0-0* Total Jobs*(0.055/12), 201) Units: **undefined** For every 1% of penetration, employment increases 0.2-0.3% per year. In the Brooking institute study, penetration was used as a proxy for adoption. https://www.brookings.edu/wp-content/uploads/2016/06/06labor_cran dall.pdf (20) Total Population= INTEG ( Total Population*(Population Change Rate/1200), 745) 63
Units: **undefined** (21) Training Available= 0.018 Units: **undefined** With training, 1.8%/year, or 9%/5 years E.6. Model Variables Verification Income and Jobs Regression Figures E.6 Table E.5 Source for fig. E.6 E.7. Model Verification and Validation The economic models were developed using known relationships between key variables. There are also several links that were derived from logical connections and simple calculations. Justification for quantitative variables and proportions are verified by using the known relationships as stated in the associated documents and studies mentioned in the model design. Please see Appendices E1 and E.5 for complete variable details. From a cursory standpoint, the models outputs were within reason and behaved as expected. 64
For a semantic and logistical check, the models were verified by George Mason University s Dr. Syed Abbas Zaidi. The general structure of the model was determined to make logistical sense and did not violate any structural rules. Mathematically, the models were compared against known linear relationships between broadband adoption and the associated variables. For example, we obtained an average of the adoption rate in a simulated 10 year run, and we plotted the estimated values along with the model outputs. As seen in figure E.6, for income, the estimates stayed within 5% of difference, while employment ranged between 1 and 13% towards the end of the simulation. The models were sent out to contacts at Shentel and AT&T who attempted to provide a contact that could validate and deliver feedback. Currently, the model has not been validated by these parties and are still awaiting a response. The parties of Shentel and AT&T may not be able to provide a validation due to proprietary data and legal issues. E.8. Wired Model Assumptions Extrapolated Population Change Rate based on historical values of targeted region s actual population change rate. a. Population change rate includes the overall birth, death, and migration rates. This model is specifically aimed at simulating the effect of expanding wired infrastructure. b. The Total Population is the population that is affected by the new wiring in each step size. Total Population drives the eligible customers which is the pool of eligible new customers for the ISP. c. The Wired Revenue is only calculating new revenue gained from the addition of the expanded cable/fiber and new customers. i. The wired revenue only accounts for wired broadband plans, and no cable television bundles. Using fiber expansion cost of $30K /Mile. Total jobs differential(annual change) is used to offset the population change rate. d. The differential percentage only affects the percentage of employable population. e. Using the existing employment rate of the target county to determine employable population. f. One job added equates to one person staying in the area. Average Income has no impact on total jobs - Logically there may be a connection, however there has been no direct research that quantifies any type of relationship pertinent for this model. The modeled region has low existing broadband adoption. Using 10% as an initial level to simulate a conservative start point. The approximate growth rate with no external factors such as training and subsidies is 1% annually. The household has at least one wired/wireless broadband device ( Computer or wireless device) E.9 Wireless Model Assumptions Extrapolated Population Change Rate based on historical values of targeted region s actual population change rate. a. Population change rate includes the overall birth, death, and migration rates. i. This model is specifically aimed at simulating the effect of adding a single cell tower. 65
b. The Total Population is the population that is affected by the addition of the tower. Total Population drives Eligible Customers, which is the pool of eligible new customers for the ISP. c. The Cell Revenue is only calculating new revenue gained from the addition of the tower and new customers. Cell tower lease cost is implemented to display the worst case if the cell tower is owned by an external party. Total jobs differential(annual change) is used to offset the population change rate. The differential percentage only affects the percentage of employable population. d. Using the existing employment rate of the target county to determine employable population. e. One job added equates to one person staying in the area. Average Income has no impact on total jobs - Logically there may be a connection, however there has been no direct research that quantifies any type of relationship pertinent for this model. The modeled region has low existing broadband adoption. Using 10% as an initial level to simulate a conservative start point. The gaussian random number generator simulates customer satisfaction. Using a low threshold for customer satisfaction to induce a drop in customer take-up due to poor service as a relative norm in the region. The approximate growth rate with no external factors such as training and subsidies is 1% annually. The customer uses a wireless broadband device (Router or cellular phone/tablet) E.10. Wired Model Behavior and Design The key difference between the wired and wireless model is the rate of expansion and adoption rate increase behavior. The rate of expansion is derived from a estimate of the fiber/cable expansion distance covered within a year. The modeled rate of expansion lowered by approximately 50% to accommodate for unforeseen delays in construction and to provide a conservative estimate. The estimated amount of wiring that was estimated for one expansion team in Shentel is approximately 300 miles in one year. When the service provider performs estimates for expansion, they consider the total number of house passings to evaluate the potential adoption rates. This evaluation methodology is reflected by increasing the total population (total people who can now adopt broadband) by the approximate population density for the area per linear mile. The known behavior of broadband adoption is known to spike initially when the wired broadband is available to the affected population. After the initial spike, the adoption rate tends to level out and slowly grow. [3] To mimic this behavior of adoption rate, the total wired customers is increased with a pulse function multiplied by a fraction of the total increased population to match the approximate initial adoption rate. The calculation performed for expansion is proportional to the number of total miles expanded, and the calculation for ISP revenue is simply the product of the total customers multiplied by the broadband plan cost. The calculation method for the effects of adoption rate on income, population, and job growth are exactly the same as the wired model. E.11. Wireless Model Behavior and Design R elative to the wired model, the wireless model is relatively simpler in its functional concept. What the model is simulating is the implementation of a cell tower that will provide adequate mobile broadband service to a local area. The initial total population is the amount of customers 66
that will have enough signal strength to warrant purchasing cellular service. The cost of expansion is simply set to the initial cost of installing a cell tower. The adoption rate pattern for a wired tower is reported to be slow and steady, unlike the initial steplike function of the wired expansion model. The number of eligible customers is approximately 75% of the total initial population, which attempts to account for small children and other factors that prohibit a member of the population from being an eligible customer. The revenue for the ISP is simply the total sum of the cell tower costs and the revenue from cell subscriptions. Based on mathematics, the wireless model should yield a higher revenue than the wired model. 67
F. Project Plan Figure F.1 - Project Schedule Figure F.2 - Work Breakdown Structure 68