Rural Broadband Trials Laikipia County Kenya

For the Communications Authority of Kenya

Summary findings of 12 month trial of television white spaces technologies

August 2014

2

3

EXECUTIVE SUMMARY

On 19 August 2013, the Communications Authority of Kenya (‘CAK’ or the ‘Authority’) issued Microsoft East Africa Limited (‘Microsoft’ or ‘we’) with a trial authorization to conduct a trial utilizing television white spaces (‘TV white spaces’) technologies in the vicinity of Nanyuki and Kalema (the ‘2013 Trial Authorization’). Consequently, a series of pilots and projects were undertaken in defined rural parts of Kenya. These projects relied and continue to rely on a combination of wireless technologies designed to operate on a license-exempt or unlicensed basis, including Wi-Fi and TV White Space base stations and end user devices. To maximize coverage and bandwidth, while keeping costs to a minimum, these radios use several complementary spectrum bands available to license- exempt devices, including 13 GHz, 5 GHz, 2.4 GHz, and unused UHF TV band spectrum (also referred to generally and herein as ‘TV white space’).

The TV white space technology is being used for point-to-multipoint “last mile” broadband connectivity. These networks typify the transition from homogenous to heterogeneous networks. The diagram below is the network deployed in the vicinity of Nanyuki in rural Laikipia and Nyeri Counties, Kenya (and discussed below), leveraging different low-cost wireless technologies and connected to a long-haul fibre network.

4

To reduce operating expenditures and to address the huge challenge of delivering broadband access to citizens who lack access to electricity where they live, this project leverages solar energy, which powers base stations and also enables consumers to charge devices.

The Need

The Mawingu project seeks to connect unserved and underserved communities in rural Kenya with affordable high-speed solar-powered broadband. This model could prove a template for meaningfully connecting the world’s unconnected billions of people.

While Kenya has one of the highest rates of Internet usage in Sub-Saharan Africa, broadband Internet access continues to lag behind other nations. This contrast underlines the dramatic underutilization of the broadband Internet access across Sub- Saharan Africa.

The problems of broadband, or even just basic wireless connectivity in rural areas of Kenya are even more acute. Mobile broadband is not universally available and even those areas that can receive a signal are likely to be hindered by a lack of capacity in the networks and low disposable incomes making access prohibitive for families and communities.

The Project

Upon issuance of the Trial Authorization, Microsoft, together with our partners, embarked on the Mawingu Project. The Mawingu Project is a pilot project that aims to provide affordable access to fast broadband connectivity in semi-urban and rural areas in and near Nanyuki, Kenya. The project utilizes TV white space spectrum to deliver connectivity to solar powered Internet kiosks, or ‘solar cybers’, in rural communities.

To date the Mawingu Project provides at absolutely no charge connectivity to the following locations in the Nanyuki area:

1. Kenya National Library Services Nanyuki Branch – 10 computers with dozens of users daily; 2. Male Primary School – 315 students; 3. Nanyuki Red Cross Office – 7 staff members; 4. Tambuzi Farm – 35 employees; 5. Tithigi Boys Secondary School – 280 students; 6. Thome Boys Secondary School – 260 students; 7. Male Secondary School – 220 students; 8. Gakawa Secondary School – 310 students;

5

9. Laikipia County Government Office – the Executive, Assembly and all staff members as well as visitors on the premises; 10. Solar Cyber run by Mawingu Networks Limited providing electrical charging services for phones and torches and high speed Internet access to the community for free – 20 to 25 persons per day; and 11. Cape Chestnut, a restaurant and business centre – all guests and customers within the premises.

Positive Impact

In our view, the Mawingu Project has been extremely successful and we would highlights and summarize the following achievements:

 Technical Results: o The Mawingu project has successfully demonstrated the technical viability of this model of delivery, with interference free point to multi-point coverage of up to 14 kilometres from TVWS base stations operating at only 2.5 Watts power (EIRP measurement); o In total this provides approximately 235 km2 of TVWS coverage using multiple 90 degree base station sector antennas; o To date, we have achieved speeds of up to 16 Mbps on a single 8 MHz TV channel at distances of up to 14 kilometres; o There have been no reports of interference from any concerned parties; o There has been no interference with the equipment or network during the test period which helps the project achieve a 99.8% availability; o At Gakawa Secondary School, they are averaging between 7 and 10 GB of data consumption per day; o We have demonstrated that the technology can support various media protocols such as streaming videos, emails, FTP, Skype voice and video conferencing, and high speed VPN services; o We have delivered this technology in areas with no source of electricity. In fact, the majority of the endpoints are running on standalone solar power systems – a clean and renewable source of energy.

 Community Benefits: o For all of the above establishments we provide all the infrastructure including connectivity infrastructure; o Microsoft has furnished ICT labs at Gakawa Secondary School, Male Primary School, and Male Secondary school using Window Multipoint Servers and teacher Window 8 tablets, along with relevant education software, applications, and content (including Kenya national examination preparation); o We have also provided training to all the teachers and administrators with regard to their computer technology know how for free;

6

o Early anecdotal evidence shows that Internet access is enabling community members, including farmers and other small businesses, to be more productive.

 Local Support: o We have tremendous support from the community and leadership in the project area, including leadership in education, health, and county government; o The whole network has been installed and is maintained by a Kenyan systems integrator based in Nanyuki; o Solar cyber bandwidth and charging stations will create employment, income and business opportunities for local entrepreneurs.

 Regional and Global Impact: o There has been immense positive attention for CAK and the Ministry of ICT and press coverage of the project not only locally but also internationally. o This project has served as a basis for other regulators in Africa and around the world to permit access to TV white spaces, first on a trial basis.

With all of these accomplishments, the project partners believe that tremendous benefit could be derived from continuation of this project. Microsoft, on behalf of the project partners, has asked CAK to extend the Trial Authorization for another 24 months. Extension of the 2013 Trial Authorization will enable the project partners to expand the trial network to cover more end points, which will give the partners, along with CAK, a better technical understanding of interference characteristics. Additional time, likewise, will enable to project partners to conduct further radio and network stress and load testing, which will help us understand the network’s overall capacity. This additional time also will enable the project partners to conduct a comprehensive Monitoring and Evaluation program assessing the socio-economic impact of the project, which is just now commencing with baseline surveys and measurements and will be conducted over the next 18 months. The result of the Monitoring and Evaluation (M&E) program will be shared with the Authority in due course. Extension of Microsoft’s 2013 Trial Authorization is particularly appropriate in light of the Government of Kenya’s goals for universal and affordable broadband access, small business empowerment and employability, distance learning, and delivery of government services. Should CAK adopt and implement regulations allowing devices to transmit on TV white space frequencies during the term of the request extension, those regulations would supersede.

7

In summary, the TVWS trial in Laikipia County, Kenya, has been extremely useful in gaining insight into the viability of using white space spectrum for rural broadband provision, and has successfully demonstrated the potential of white space technology through the use of real-world applications such as video streaming and video conferencing. The provision of adequate broadband connectivity in rural areas is clearly very important in today’s society, and the benefits of using white space spectrum are applicable not only to remote parts of Kenya but also to remote areas in many other parts of the world.

8

Contents EXECUTIVE SUMMARY ...... 3 TABLE OF FIGURES ...... 10 Acknowledgements...... 12 GLOSSARY OF TERMS ...... 13 1. PROJECT BACKGROUND ...... 16 1.1. AN OVERVIEW OF ‘TV WHITE SPACES’ ...... 18 1.2. PROJECT AIMS ...... 19 1.3. IMPLEMENTING THE PLAN ...... 20 1.3.1. Conceptualization and Design ...... 20 1.3.2. Procurement and Installation ...... 22 1.3.3. Monitoring and Evaluation ...... 22 1.4. BENEFITS OF LICENSE-EXEMPT MANAGED ACCESS TVWS ...... 23 1.5. CREATING THE REGULATORY FRAMEWORK ...... 26 1.6. TECHNICAL CONSIDERATIONS...... 26 1.6.1. Coverage and Data Throughput Potential ...... 26 1.6.2. Databases ...... 27 1.6.3. Relationship between TVWS and the Digital Migration Process ...... 27 2. INTRODUCTION TO THE KENYAN TVWS TRIAL ...... 29 2.1. THE TRIAL PARTNERS ...... 29 2.2. THE TRIAL NETWORK...... 31 3. NETWORK INSTALLATION AND CONFIGURATION ...... 36 3.1. BACKHAUL AND DISTRIBUTION...... 36 3.2. WHITE SPACE RADIO EQUIPMENT ...... 37 3.2.1. Cape Chestnut base station ...... 43 3.2.2. Tambuzi base station ...... 47 4. DATA THROUGHPUT MEASUREMENTS ...... 51 4.1. DATA THROUGHPUT ...... 53 4.2. USAGE STATISTICS ...... 54 4.2.1. YouTube HD Streaming Tests...... 55

9

4.2.2. ‘Virtual Village’ via Skype ...... 56 5. SURVEY OF TRIALISTS ...... 57 5.1. BEN’S AGENT OUTLET ...... 57 5.1.1. Agent activity report from September 2013 ...... 59 5.1.2. Activity report for December 2013 holiday period...... 63 5.1.3. Activity report from 20th May 2014 ...... 64 5.2. LAIKIPIA COUNTY RED CROSS (KENYA) OFFICES ...... 67 5.3. LAIKIPIA COUNTY GOVERNMENT...... 70 5.4. LOCAL SCHOOL ICT LABS ...... 71 6. NEXT STEPS: EXTENDING THE WHITE SPACE NETWORK ...... 75 7. CONCLUSIONS ...... 78 8. FURTHER REFERENCES AND READING ...... 81

10

TABLE OF FIGURES Figure 1 - Graphic illustration of licensed transmissions at certain frequencies, with ‘white spaces’ between them...... 18 Figure 2 - Some of the ‘white space’ spectrum may be utilized by license-exempt devices interleaving their transmissions with those of licensed users...... 19 Figure 3 - Economic value generated by Wi-Fi through fixed broadband value enhancement...... 24 Figure 4 - Value of Wi-Fi to mobile network operators in terms of additional base stations which would have to be built in the absence of the operator off-loading traffic through Wi-Fi...... 25

Figure 5 - Table of trial network nodes...... 33 Figure 6 - Map showing the physical locations of the 3 Base Stations and their associated CPE...... 34 Figure 7 - Approximate coverage area of all 3 Base Stations located at Cape Chestnut, Tambuzi, and Gakawa. The total area covered with a RX level of -80 dBm or greater is irradiated from 3 x 90 degree sectors is approximately 235 Km2...... 34 Figure 8 - Radio Mobile coverage predictions for all 3 base station sites: Cape Chestnut, Tambuzi Flower Farm and Gakawa Secondary School...... 35

Figure 9 - Picture and set up of Kenya Power Mast...... 37 Figure 10 - Adaptrum ACRS2...... 37

Figure 11 - GW300 EAR/CAR ...... 38 Figure 12 - A typical CPE installation at trialists' premises ...... 39 Figure 13- Gakawa Secondary School base station ...... 40 Figure 14- Radio Mobile coverage pattern for Gakawa Secondary School approximately 85 KM2 at RX -80dBm, 8 metre mast with 10 dBi gain log periodic 90 degree sector. ... 41 Figure 15- Radio Mobile path profile and path loss: Gakawa Secondary School to Male Primary School...... 42 Figure 16- Screen shot of 6 Harmonics Thome school’s EAR -75 dBm rx ch 35 (585 MHz)...... 42 Figure 17 - 6 Harmonics CAR 30 white-space base station on an 18 m mast at Cape Chestnut Nanyuki. Ch 49 (698 MHz) the 5.4 GHz 25 dBi backhaul antenna to Kenya Power 133 kv switching station...... 43 Figure 18 - Radio Mobile coverage pattern: approximately 104 km2 at rx -80 dbm...... 44

11

Figure 19 - Cape Chestnut to Red Cross Nanyuki HQ...... 45 Figure 20 - Screen shot of 6 Harmonics CPE/EAR at the Red Cross Nanyuki rx -75 dBm ch 35 (585 MHz)...... 45 Figure 21 - Spectrum plot, Cape Chestnut to Red Cross Kenya (20 MHz span)...... 46 Figure 22- 18 m lattice mast: 1 x 5.8 GHz link to Kenya Power 1 x 5.4 GHz link to Gakawa. 1 x 6Harmonics base station with 10 dBi 90 degree v pol antenna tx power 20 dBm. .... 47 Figure 23- Radio Mobile coverage pattern: approximately 47 km2 at -80 dBm rx...... 48

Figure 24- Radio Mobile path profile and path loss: Tambuzi to Thome school...... 48 Figure 25 - Screen shot of 6 Harmonics Thome school’s EAR -75 dBm rx ch 35 (585 MHz)...... 49 Figure 26- Spectrum plot, Tambuzi to Thome school, (20 MHz span, Preamp ON)...... 50

Figure 27- Measured data throughput at trialists' premises...... 54 Figure 28- Screen shot of Meraki Cloud controlled MR-12 Traffic analytics ...... 55 Figure 29 - Benson Maina holds a TV white spaces antenna in front of the Mawingu white spaces broadband pilot container near Nanyuki, Kenya...... 57 Figure 30- Students conduct research at the Mawingu white spaces broadband pilot container...... 58 Figure 31- Laikipia county Red Cross offices, Nanyuki...... 67

Figure 32 - Home fire disaster, Nanyuki (1 of 2)...... 68 Figure 33 - Home fire disaster, Nanyuki (2 of 2)...... 68

Figure 34 - Victims receive donations via Laikipia county Red Cross...... 69 Figure 35 - Danilo Legei, Director of ICT & Research Services of the County of Assembly of Laikipia...... 70 Figure 36- ICT Lab launch event at local schools...... 73 Figure 37 - - RIchard Thanki instructing Charles - school care-taker at Gakawa School - how to use the Internet...... 74

Figure 38 - Proposed radio mast locations for Ol Pejeta conservancy project...... 77 Figure 39 - NASA's 'world light pollution map'...... 79 Figure 40 - 'Heat map' of world's Internet usage...... 80

12

Acknowledgements

The project partners wish to express their thanks and appreciation to the Government of Kenya, with special thanks going to the County Government of Laikipia for its ongoing support and commitment in helping make the project an overall success and to the people of Laikipia for always extending a warm welcome to numerous delegations of visitors ranging from politicians, journalists, engineers, donors, and other well-wishers.

Asante Asana!

13

GLOSSARY OF TERMS

ACS Adjacent Channel Selectivity. AGC Automatic Gain Control. BS Base Station. CAK Communications Authority of Kenya. C/I Ratio of Carrier power to Interference power. COM Commercial multiplex. Coupling Gain/Loss A measure of the extent to which electromagnetic energy radiated by a WSD antenna gets picked up by a DTT antenna. Coupling gain/loss is typically measured in decibels (dB). CAR Core Adaptive Radio CPE Customer Premises Equipment. dBm Unit of power, measured in decibels relative to 1 milliwatt (mW). dBµV/m Unit of electric field strength, measured in decibels relative to 1 microvolt per metre (µV/m). Downlink Radio link from BS to CPE. DTT Digital Terrestrial Television. DVB-T Digital Video Broadcasting - Terrestrial. (European DTT standard, adopted by numerous countries worldwide.) EAR Edge Adaptive Radio EIRP Effective/Equivalent Isotropic Radiated Power – The power that an isotropic antenna would need to radiate in order to produce the same power as an actual antenna in its direction of maximum power transmission. FCC Federal Communications Commission (US regulator). FPGA Field-Programmable Gate Array GPON Gigabit Passive Optical Network HP Horizontal Polarization. IDA Infocomm and Development Authority of Singapore IP Internet Protocol.

14

ITU International Telecommunication Union. IXP Internet exchange point MAC Medium Access Control – Mechanism by which a device’s access to the radio link is controlled. MIMO Multiple Input, Multiple Output – The use of multiple antennas to improve communications performance. Ofcom Office of Communication (UK regulator). PHY Physical layer – Mechanism by which signals are modulated and transmitted on a radio link. PMSE Programme Making & Special Events – e.g. wireless microphones and in-ear monitors used in theatres, sports venues, music festivals, etc. POE Power Over Ethernet. Protection Ratio The minimum permitted ratio between a wanted DTT signal and an interfering signal for stable TV reception. PSB Public Service Broadcast. RAN Radio Access Network RTT Round-Trip Time – the time interval between a data packet being transmitted and a corresponding acknowledgement being received. SFN Single Frequency Network TCP Transmission Control Protocol. TDD Time Division Duplex – Transmit and Receive signals are time division multiplexed in order to emulate full- duplex communication on a half-duplex physical channel. TVWS TV White Space – White spaces in the TV band (470- 790 MHz for ITU Region 1). UDP User Datagram Protocol. UHF Ultra-High Frequency (300 MHz – 3 GHz). UKPM UK Planning Model. Uplink Radio link from CPE to BS. UNESCO United Nations Educational, Scientific, and Cultural Organization. VoIP Voice over IP – the transmission of speech over IP networks.

15

VP Vertical Polarization. White Spaces Portions of radio spectrum which are not used by existing licensees at all times or in all locations. WiMAX Worldwide Interoperability for Microwave Access WSD White Space Device.

16

1. PROJECT BACKGROUND Internet usage has been rising exponentially in recent years, as developments in technology have increased data rates and expanded connectivity in many parts of the world. This has spawned new applications, based around Internet connectivity, which large numbers of people have been embracing in their personal and professional lives. Many people now rely on social networking and Internet-based video-conferencing applications to keep in touch with friends and family, or to search for all manner of information that would previously have been unavailable or difficult to access for the majority of the population. Many businesses have transformed the way in which they operate, through innovative and effective use of the Internet as a powerful tool for interacting with customers and suppliers. Governments and not-for-profit organizations are using the Internet to reach people and interact with them in ways which previously would not have been possible.

Despite all of this, there are still large numbers of people around the world (in fact a significant proportion of the world’s population) for whom Internet access is slow and cumbersome, unaffordable, or even non-existent, and for whom many of the above- mentioned applications are therefore unavailable. This is particularly true in semi-urban, suburban and rural areas, where the average cost of deploying and maintaining broadband infrastructure and available incomes can make deployment of broadband Internet access commercially unprofitable. Although, according to ITU statistics, close to 3 billion of the world’s population is now using the Internet, over 4 billion of the world’s population is not yet online.

Even with availability of technology, broadband access often remains unaffordable for at least half the world’s population. In many emerging markets the cost of broadband access often exceeds consumers’ ability to pay for it. The problem is particularly acute in growth markets, such as in Latin America, Africa, and South and Southeast Asia, and especially in areas which lack reliable electricity or are altogether off-grid. For example, more than 80% of Africans would need to spend more than 5% of their annual income to purchase wireless broadband access – making broadband unaffordable to most Africans. In addition to creating a barrier to consumer markets, these connectivity and cost challenges also undermine governments’ goals around getting more public sector institutions and small and medium sized enterprises (SMEs) online.

This translates to a low rate of broadband utilization even where it is available. For example, according to Communications Authority of Kenya statistics, as of December 2013, the number of broadband subscriptions in Kenya stands at only 1,435,267 or just 3.2% of the population. Many governments and other organizations such as the ITU and UNESCO have recognized and acknowledged that Internet connectivity is essential for

17

the prosperity and survival of such communities, and have committed to various targets aimed at improving coverage and data rates.

In many parts of the world, including in Kenya, telecommunications network operators are deploying fibre-based high-speed broadband infrastructure. The cost of deploying fibre-based last mile infrastructure can be several thousand US dollars per home. Moreover, many homes, particularly those in rural areas, are situated many miles from any type of telecommunications infrastructure or utility power. As a result, extending the fibre-based last-mile infrastructure can only be part of the solution.

One approach to solving this problem is to use wireless radio links instead. A variety of wireless technologies can be utilized for last mile access, or even long distance connections. Each of the technologies comes with its own strengths and weaknesses, and if provided by a commercial or even community network operator they will consider a variety of factors – population density, geography, spectrum availability, desired throughput, assumptions about data consumption, etc. – when determining which mix of technologies and techniques are needed to serve a particular population.

One of the key opportunities to emerge in the last few years is the availability of new wireless devices and networks leveraging and using the so-called TV white spaces spectrum (operating on UHF frequencies). The combination of recent developments in spectrum management policy, enabled by ever smarter radio systems, and the potential spectrum availability emerging from the switch-over of TV broadcast transmissions from analogue to digital in Europe and the USA, has meant that interest in so-called ‘white space’ spectrum is gathering pace around the world and technology solutions and networks are being developed to allow access to spectrum that is more open and available for communities to share and use than ever before. Delivering broadband to unserved and underserved communities is one of the key applications that could potentially benefit from making use of white space spectrum due to its range (UHF frequencies can travel long distances) and reach (UHF has excellent penetration into buildings and diffracting around hills and objects). A not-insignificant bonus is that the latest hardware comprising these networks is designed for very low power operation and therefore capable of being fully powered with photovoltaic (i.e. solar) cells. This creates the opportunity to provide ultra-cheap high capacity wireless broadband in areas that only a few years ago would not have been considered economically viable to serve: unable to support a sufficient return on investment (ROI) for any credible and viable business and technology plan.

18

1.1. AN OVERVIEW OF ‘TV WHITE SPACES’

‘TV White Spaces’ are portions of the UHF radio spectrum which are not assigned to licensed operators (often termed the primary user) or are otherwise unused at all times or in all locations. Figure 1 illustrates the concept, showing unused ‘white spaces’ between licensed transmissions. With demand for wireless connectivity increasing, the exploitation of white space is an attractive way of making more efficient use of radio spectrum simply by sharing the spectrum such that if not used in one location of a country by the primary user, then in that geography it can be redeployed and used by secondary users and will not interfere with the primary user.

Licensed transmissions ‘White Spaces’ (unused spectrum)

f

Figure 1 - Graphic illustration of licensed transmissions at certain frequencies, with ‘white spaces’ between them.

In many countries, analogue television broadcasts have been switched off and replaced by more spectrally efficient digital television transmissions. Two potential benefits of this transition are freeing up or clearing spectrum for reallocation to other licensed uses and to enable additional further opportunistic access to interleaved spectrum between licensed users, such as broadcasters. These UHF frequencies have good propagation and building penetration characteristics, and of course this is the reason they were first chosen for TV broadcasting. These characteristics make TV white space eminently suitable for use in rural broadband applications, where transmission links may be several kilometers in length and may involve challenging terrain such as hills, foliage, and water.

The US regulator (FCC) has made certain parts of the TV band available for use on a license-exempt basis. The UK regulator (Ofcom), the Singaporean regulator (IDA), and the Canadian regulator (Industry Canada) are in the final stages of rule-making. This represents a ground breaking development in the management of spectrum, as it involves unlicensed transmissions being interleaved with those of licensed users such as TV broadcasters and wireless microphone (PMSE) users. Several other regulators around

19

the world are actively considering similar approaches. Figures 1 and 2 illustrate how white space transmissions can exist alongside licensed transmissions.

Allowing license-exempt devices to interleave their transmissions with those of licensed users does, however, presents challenges in ensuring that such unlicensed transmissions will not adversely interfere with the licensed transmissions. The approaches being pursued by the FCC, Ofcom, IDA, and Industry Canada differ slightly, but all involve the use of a regulator-approved database which White Space Devices (WSDs) will need to consult before being allowed to access the spectrum.

Licensed transmissions ‘White Space’ transmissions

f

Figure 2 - Some of the ‘white space’ spectrum may be utilized by license-exempt devices interleaving their transmissions with those of licensed users.

1.2. PROJECT AIMS

With broadband connectivity clearly an important social and economic issue in Kenya, a collaborative R&D project was set up to install a trial network using ‘TVWS’ Dynamic Spectrum Access UHF radio spectrum to deliver broadband connectivity to rural communities in and near Nanyuki and Kalema. The first phase of this project has focused on a deployment in and near Nanyuki.

The project aims include:

 To assess the technical performance and viability of white space spectrum as a means of delivering broadband Internet access into a typical Kenyan rural environment. Performance was measured on a number of levels, ranging from data rates and latencies through to user experience while leveraging applications such as video-conferencing and video streaming. These technical metrics, in turn, will help the Kenyan Government in understanding the overall viability of this

20

new technological approach. A related project aim is to provide a rich set of data for researchers at organizations and companies such as: DFID, USAID, Google, Intel, Facebook, Cisco, MediaTek, Qualcomm, the Alliance for Affordable Internet Access and the World Bank, to name but a few, who are all tracking what TVWS and other wireless and wireline technologies can yield in both social benefit and economic impact.  To work with Southampton University in the UK in generating strong monitoring and evaluation (M&E) data in assessing the cost and benefits of opening up the UHF band for Dynamic Spectrum Access.  To work with the Strathclyde University, Department of Electronic and Electrical Engineering in the UK to facilitate knowledge transfer and information sharing with their white space networks on the Isle of Bute (for rural broadband to unserved communities on the island) and an Ofcom TV white space test bed project running in Glasgow, UK.  To evaluate TVWS radio performance to be sure this technology can serve large areas of relatively low population density in various terrain types with varied levels of RF clutter. Further, it is important for the TVWS base stations to be self- powered, and not to require any connection to the electrical grid.  Finally, as this was to be a real network with real people using the broadband services provided to them during the trial period, usage statistics would be collected and an end user survey carried out, in order to gain insights into how broadband connectivity was being used in impacted communities.

1.3. IMPLEMENTING THE PLAN

The project had three major stages/phases over approximately a 2 year time period.

1.3.1. Conceptualization and Design White spaces technology has been in research and development for about fifteen years, and has been extensively tested in trials and pilots conducted by Microsoft and others. Microsoft Research was involved in some of the first white spaces trials in the United States; for example, demonstrating these technologies on Microsoft’s corporate headquarters in Redmond, Washington, United States of America, providing campus- wide coverage under an experimental license from the US FCC for the last six years. Microsoft also developed the first prototype TV white space database. Microsoft participated in the US FCC’s first field trials of TVWS prototype radios which provided critical insight into development of the FCC’s initial regulations. Microsoft, likewise, participated in a major technology trial in Cambridge, United Kingdom, that involved

21

partnering with 16 companies and demonstrating the feasibility of indoor and outdoor urban and rural connectivity. Microsoft joined 15 companies in Singapore assessing the commercial feasibility of using TV white space technologies to enable various different use cases.

For all of this progress in demonstrating this technology’s feasibility and for all of its promise, as of 2011, nobody had demonstrated, let alone deployed, TV white space technology on the African continent. All of that changed when the Communications Commission of Kenya (now the CAK) provided Microsoft with authorization to conduct a TV white space demonstration at the Internet Governance Forum conference held at the UN Compound in Gigiri. At that event, Microsoft and Adaptrum, a pioneer in TV white space radio technologies, successfully established a broadband connection on unassigned UHF frequencies streaming HD video and showcasing other applications. The Gigiri demo was the first time TV white space technology was demonstrated on the African continent.

The Gigiri demonstration generated significant interest in Kenya for a trial project. Shortly after the Gigiri event, Microsoft teamed up with the University of Strathclyde on another live TV white space demo at ITU Telecom World 2011 in Geneva, Switzerland and to report on the UK Governments Technology Strategy Board (TSB) Isle of Bute TVWS project. Since 2011, the University of Strathclyde has participated in a highly successfully TV white space trial on the Isle of Bute, Scotland, UK in collaboration with British Telecom (BT) and the British Broadcasting Corporation (BBC), delivering wind- powered (i.e., using renewable energy) broadband access to rural communities. At the ITU conference, much interest was shown on Microsoft’s stand from regulators in general but the greatest interest shown was from the African delegates, with several of them asking if they could collaborate in trialing some pilot tests in their respective countries.

Soon after the ITU Telecom World demo, Microsoft began discussions with the Kenyan Ministry of Information and Communication about what TV white space technology could mean for Kenyans. In particular, the Ministry of ICT asked for a better understanding of TV white space’s potential in helping the Government rollout ‘E- Governance’ services all of its citizens, especially at the county level. The Ministry was also responsible for the ‘broadband task force’ as part of the Government’s 2030 longer term development goals and therefore was interested in the role TV white space could play in helping to increase digital inclusion, in support of education, healthcare, agriculture, and sectors critical to the Kenyan economy.

Many of the Kenyan Government’s key policy goals align well with Microsoft’s own interests. As a company founded on the notion of democratizing access to technology,

22

Microsoft has been particularly interested in promoting new technologies, such as TV white spaces and other forms of dynamic spectrum access, that can help reduce the cost of Internet access and allow billions more of the world’s inhabitants to get online. This is reflected in Microsoft’s 4Afrika Initiative which aims to help address common issues shared across the African Continent including access, employability, and competitiveness. Microsoft believes deeply in the potential of technology, such as TV white spaces, to change Africa, but equally believes deeply in the potential of Africa and Africans to change technology for the world. For that reason, all of Microsoft’s pilot projects in Africa, including the Mawingu project, involve partnerships with African entrepreneurs, governments, and local communities.

1.3.2. Procurement and Installation In 2012, Microsoft, the Kenyan Ministry of Information and Communications, including CCK and the ICT Board, along with Indigo Telecom (the predecessor to Mawingu Networks on this project), agreed to support a rural broadband pilot leveraging TV white space technology using the latest FPGA (field-programmable gate array) instantiation of FCC approved TVWS Software Defined Radios. The project moved into a phase of procurement when CCK (now CAK) issued a Letter of Authorization permitting Microsoft in partnership with an appropriate Tier 3 network operator to conduct trials in Laikipia and Kajiado counties. It took approximately 4 months for the supply and delivery of hardware and another 2 months of installation. All network installation work was carried out by a telecom integration firm based in Nanyuki and the TVWS trial network was delivered as a ‘turn-key’ project.

1.3.3. Monitoring and Evaluation A monitoring and evaluation (M&E) exercise is intended to be an important part of the process of the TVWS trial taking place in Kenya. ECS Partners, of the University of Southampton, have been retained by Microsoft to design and conduct an M&E exercise focusing on the work that Mawingu is performing in Laikipia County. The baseline household and schools survey will be taking place in September 2014, shortly before the planned rollout to an additional 20 schools connected with TVWS spectrum and the launch of Mawingu Solar Cyber agency model using other license-exempt bands. Consistent with the trial authorization, it is critical to note that no TV white space links will be used in support of commercial offerings.

The monitoring and evaluation program seeks to understand the economic and social impact of broadband connectivity services delivered over TVWS spectrum in a rural Kenyan context. Specifically, the M&E task will address this question: what impact does the introduction of affordable connectivity services have in an environment which has none or very limited options for connecting to broadband Internet and the multitude of

23

services that this entails? The major focus of M&E will be to look at the constituencies of households living near Solar Cyber kiosks and schools that will be connected to the Internet using TVWS or other license-exempt spectrum.

With households, M&E will look at social conditions and the existing use of telecommunications in the region in which the project is taking place. Of particular interest will be the uses to which such communities put technologies and services that have largely been confined to urban areas. For example, to what extent and how will Internet access be used to further economic activity, the gaining of health, agricultural and citizenship information and for the purposes of general education?

With schools, M&E will attempt to understand the existing approaches to education, the level of resources present and the effect that broadband over TVWS has on the lives and learning processes of the learners and the ability to teach, job satisfaction and career development of the teachers.

The M&E task will run for approximately 18 months tracking the progress of the project. It will begin with a baseline household survey and focus group discussions in September 2014 and the first report, describing the existing state of telecommunications usage, needs and attitudes will be available in December 2014. Depending on the progress with the network there will be a first impact report released six months later, drawing on repeat surveys and focus groups as well as the analysis of network data to provide a detailed view of clusters of usage and users.

1.4. BENEFITS OF LICENSE-EXEMPT MANAGED ACCESS TVWS

Kenya has one of two largest Internet-based economies in Africa. In 2013, around 2.9% of Kenya’s gross domestic product (GDP) was created by the Internet economy, worth Sh99.8 billion. Africa as a whole is likely to see 10% year-on-year increases in the value of this contribution. Given Kenya’s relatively high starting base this growth has the potential to see Kenya become a strong leader of the African Internet economy.

Wireless access to the Internet is, and will be, a key enabler of this growth. In particular, the use of license-exempt spectrum including Wi-Fi is an important means of accessing the Internet for consumers and businesses. One of the challenges for use of Wi-Fi (primarily a local area network technology) in Kenya is the lack of ubiquitous last mile infrastructure.

24

A number of studies have analyzed the potential value to be derived from allowing license-exempt managed access to broadcast spectrum1.

According to Richard Thanki, an economist from the University of Southampton, unlicensed or license-exempt regimes decrease the costs of delivering broadband and increase the quality of the product, but these benefits have, to date, been limited “by the lack of a harmonized globally available broadband-capable band of license-exempt spectrum in sub-1GHz spectrum.” Thanki concludes that TVWS represents an opportunity to create such a band which will assist in meeting future demand, delivering universal and affordable connectivity to people, facilitating machine-to-machine connections and developing robust and adaptable networks.

Thanki quantifies the value of Wi-Fi to fixed broadband and mobile network operators by country. The table below shows economic value generated by Wi-Fi through fixed broadband value enhancement for selected countries:

Country Population GNI per Total fixed Total Wi-Fi Evenly scaled GNI scaled (millions) capita broadband connections annual economic annual economic (USD) connections (millions) value value (millions) (USD million) (USD million) Brazil 192.4 9 390 13.85 11.77 2 613.7 517.9 China 1 347.4 4 270 126.65 107.65 23 899 2 153.4 India 1 210.2 1 330 10.89 9.26 2 055.3 57.7 Russia 143 9 900 15.73 13.80 2 968.9 620.2 Kenya 50.6 6 090 0.76 .64 143.2 18.4 United 62.3 38 370 19.56 16.63 3 691.4 2 988.8 Kingdom United States 313.2 47 390 82.28 70.02 15 545 15 545 Figure 3 - Economic value generated by Wi-Fi through fixed broadband value enhancement.

Thanki presents the value of Wi-Fi to mobile network operators in terms of additional base stations which would have to be built in the absence of the operator off-loading traffic through Wi-Fi. The figures for Kenya are set out below:

1 See e.g. “The Economic Significance of License-Exempt Spectrum to the Future of the Internet” by Richard Thanki, http://www.wirelessinnovationalliance.org/index.cfm?objectid=DC8708C0-D1D2-11E1- 96E9000C296BA163 (accessed August 2013); “Efficiency gains and consumer benefits of unlicensed access to the public airwaves: The dramatic success of combining market principles and shared access” by Mark Cooper, http://www.markcooperresearch.com/SharedSpectrumAnalysis.pdf (accessed August 2013); “The Case for Unlicensed Spectrum” by Paul Milgrom, Jonathan Levin and Assaf Eilat, http://www.stanford.edu/~jdlevin/Papers/UnlicensedSpectrum.pdf (accessed August 2013) (and sources cited therein); “The economic value of license exempt spectrum, A final report to Ofcom from Indepen. Aegis and Ovum”, December 2006, http://www.aegis-systems.co.uk/download/1818/value.pdf (accessed September 2013); “Perspectives on the value of shared spectrum access”, Final Report for the European Commission, February 2012, http://ec.europa.eu/digital-agenda/sites/digital- agenda/files/scf_study_shared_spectrum_access_20120210.pdf (accessed September 2013).

25

Country Population Total mobile Additional urban cell sites needed (millions) data using 900MHz (assuming 10 MB using 2100MHz (assuming 10 MB per connections per user per day) user per day) (millions) Brazil 192.4 20.3 9,249 28,928 China 1347.4 26.9 25,647 80,703 India 1210.2 10.8 15,016 47,333 Russia 143 24.8 12,671 39,679 Kenya 38.6 2.1 295 899 United 62.3 34.8 4,282 14,013 Kingdom United States 313.2 169.1 54,400 174,372 Figure 4 - Value of Wi-Fi to mobile network operators in terms of additional base stations which would have to be built in the absence of the operator off-loading traffic through Wi-Fi.

It can be seen that Kenya would need far fewer additional cell sites that other countries in the absence of Wi-Fi. However, this should be a source of alarm rather than relief. Kenya has far fewer fixed connections in operation than any of the other comparator countries and so a much smaller quantity of Wi-Fi. Whereas in the other countries Wi-Fi carries much of the data traffic from smartphones (more than 80% in some European nations), in Kenya almost all of the data will be transferred over the mobile networks. Therefore, for any given density of cell sites, end users in Kenya will experience a much lower quality of service than in the other countries. In addition, due to the substantially higher costs of data transport over mobile networks than highly depreciated fixed networks, the Kenyan consumer will also pay substantially higher costs for data. Due to the presence of extensive Wi-Fi, mobile operators in other countries are able to offer their customers higher quality services at lower prices than Kenya could manage. As such, it is clear that Kenya needs more low-cost last mile access solutions on which Wi-Fi can be connected.

Mark Cooper sets out empirical evidence showing that spectrum allocated for unlicensed use has been more effective in encouraging the development of innovative uses2. Supporting low-powered license-exempt uses of spectrum allows greater reuse and sharing of the spectrum and encourages third-party innovation while also promoting investment in complementary technologies.

License-exempt wireless technologies currently contribute significantly to the Kenya economy by expanding network reach and improving network management. In the

2 Cooper, Mark, SENIOR ADJUNCT FELLOW, and SILICON FLATIRONS. "Efficiency Gains and Consumer Benefits of Unlicensed Access to the Public Airwaves." Available at SSRN 2030907 (2012).

26

future, there will be even greater demand for sub-1-GHz license-exempt spectrum (i.e., in the TV white space UHF bands) to address the exponential increase in consumer demand for broadband access, support the expansion of cellular offload, and network the millions of devices that will compose the coming Internet of Things (IoT). As noted above, the TV broadcast spectrum below 1 GHz has better propagation characteristics than spectrum above 1 GHz, enabling signals to travel further and penetrate walls and irregular terrain. As a result, it is uniquely well-suited for non-line-of-sight broadband communication. Therefore, the potential new opportunities for Kenyan innovation and economic contribution derived from providing license-exempt access to the TV white spaces is immense.

1.5. CREATING THE REGULATORY FRAMEWORK With the demands on the radio spectrum becoming more intense, it is necessary to use this unique resource as efficiently as possible. One way forward is to apply innovative and flexible license authorization schemes like shared spectrum access.

In Kenya, regulators can enable the use of TVWS by designating the 470-694 MHz band as “interleaved” or shared spectrum. Prior to the completion of digital dividend efforts, vacant spectrum between 694-790 MHz should also be made available for shared use. This approach would allow devices to use available channels in this band when doing so will not cause harmful interference to licensed users.

In particular, CAK should establish the technical parameters for TVWS operation and establish procedures for TVWS devices to access available spectrum.

Most TVWS devices today utilize a white space or geolocation database to determine which channels are available for use. In general terms, TVWS devices are required to register their location with the TVWS database. The database uses this location information and the time of the request to determine and communicate to the TVWS device what channels are available for use in that location at that time. For the Mawingu trial, the TV white space devices were tuned to UHF channels known to be vacant in the Nanyuki area. This was verified through spectrum scanning. At CAK’s direction, a TVWS database was not utilized.

1.6. TECHNICAL CONSIDERATIONS

1.6.1. Coverage and Data Throughput Potential The coverage/throughput combination for a TVWS device is determined by:

• The transmission power used by a device in a given location;

27

• The height and gain of the base station and client antennae; • The specific channels of the UHF band used; • Varying levels of absorption of the signals between base station and its client or clients, which may be impacted by the presence of foliage or varied terrain; and • The level of interference and noise that is present in the adjacent protected terrestrial TV channel.

1.6.2. Databases A white spaces or geolocation database provides TVWS devices with operating parameters for any given location. These include:

• Available white space channels (i.e. frequency bands); • Transmission power limits applicable to the use of these channels, taking into account the location, height, and configuration of the device; • Duration for which the white spaces channels will be available (if necessary); and • Timeframe within which the device must re-register in order to receive updated information.

The methodology and algorithms employed by the TVWS database take into account the known transmission characteristics - power output, frequency, antenna height, type, location and orientation of existing TV transmitters3 , and can also take into account the topography of the area, in order to model the radio propagation of every channel transmitted in that area. The database is then able to determine which channels are available for use.

Overall, a TVWS database allows TVWS devices to be used on a license-exempt managed access basis while mitigating the potential for harmful interference. Allowing license- exempt managed access use of devices increases the utility of TVWS by effectively lowering the barrier to entry for competition and to the introduction of new technologies and applications.

1.6.3. Relationship between TVWS and the Digital Migration Process In Kenya, a significant portion of television broadcast spectrum is unused at any one place. This is because analogue TV transmitters require significant physical or spectral separation to avoid interfering with each other and because there simply are not that

3 Current practice is for the data on existing transmitters to be sourced from the communications regulator in each respective country. The integrity of the data and the range of parameters provided has an impact on the accuracy of the analytical results of the database.

28

many broadcast licensees in most markets (i.e., there is more unassigned broadcast spectrum than is needed to prevent broadcasters from interfering with one another).

While there will be a reduction in the available white space during and after the migration to Digital Terrestrial Television (DTT) and the subsequent digital-to-digital migration necessary to ensure optimal usage of the assigned Digital Dividends, there will still be a substantial amount of spectrum unused at any given location. Substantial spectrum will be available especially if a Single Frequency Network (SFN) is implemented.

Enabling use of TVWS does not need to be tied to the digital migration process and TVWS technology can be utilized before, during and after digital migration. Further, database technology can instantaneously direct devices to use whatever TVWS channels are available at the time; even as television broadcast stations are being relocated.

29

2. INTRODUCTION TO THE KENYAN TVWS TRIAL

2.1. THE TRIAL PARTNERS

Government of Kenya (CAK) - The Communications Authority of Kenya (CAK) controls access to all of Kenya’s licensed radio spectrum.

The CAK granted Microsoft a Trial Authorization for UHF spectrum needed to operate TVWS radios.

Ministry of ICT - The Ministry of ICT has responsibility of formulating administering, managing and developing the Information, Broadcasting and Communication policy.

Mawingu Networks, as Microsoft’s implementing partner, has applied for a Tier 3 license with CAK.

Microsoft - For over a decade, Microsoft Research and various other researchers around the world have been working to perfect dynamic spectrum management techniques. Numerous field trials have validated the feasibility of using these technologies to deliver broadband access. With regulations now in place in the United States and other regulators around the world working on policies and practices to enable dynamic spectrum access (DSA), we are now starting to see the first commercial deployment of white space networks.

As part of an effort to achieve needed regulatory change, Microsoft is supporting some of these deployments, including reference materials, global pilots, and various Microsoft Research projects related to spectrum technologies.

Microsoft has since early 2013 supported the trial initiative by procuring and providing access to key capital items for network deployment.

County Government of Laikipia - 2013 saw the devolution of political power in Kenya to the County Government level.

Microsoft and Mawingu Networks have been working closely with the Governor of Laikipia, Joshua Irungu and the Minister of ICT and Education, John Akaale, to prepare the government staff for effectively using the Internet to deliver government services to manage county residents. To this end, Microsoft has provided a limited quantity of Office 365 trial accounts for the county staff.

30

The TVWS radios used to connect the county assembly and government offices in Nanyuki were made available by Microsoft and installed by Mawingu Networks and their local network integration partner, Solaris Ltd.

Tambuzi Flowers – Tambuzi Flowers is a Kenyan export flower grower with green house operations in the Nanyuki area. Tambuzi has a strong social mission as part of its corporate mandate and have been supporting the TVWS trial by facilitating access to local government and schools.

USAID/NetHope – USAID, through its Power Africa initiative and implementing partner NetHope, has contributed photo-voltaic power systems to the project and has visited the trial site to gain first hand impressions of the conditions on the ground and to better understand potential impact on the communities served by dynamic spectrum broadband.

Adaptrum Inc. – Adaptrum is a Silicon Valley based company developing innovative wireless broadband technology using vacant VHF/UHF TV channels or TV White Spaces. Adaptrum was an active participant in the FCC TV White Space proceeding from 2008 to 2010 and made significant contributions leading to the successful conclusion of the Rulemaking.

Adaptrum’s TV White Space product was among the first TV White Space devices certified by the FCC together with leading TV White Space Database providers. Adaptrum is launching its second generation ACRS 2.0 TV White Space product with market leading performance and cost. ACRS 2.0 has entered volume production since the beginning of 2014.

6Harmonics - A Canadian company developing Adaptive Radio Systems. Its branded GWS access platform is a family of OFDM-based cognitive solutions for wide area broadband wireless networks using unlicensed spectrum or dynamically accessing other spectrum. At present time there are seven (2 base stations and 5 CPEs) 6Harmonics TVWS radios operating as part of the Nanyuki trial network.

GreenBridge Computing – GreenBridge’s mission is to “sustainably bridge the digital divide” for millions of students by connecting schools and libraries with top quality computer workstations at a fraction of the cost of conventional PCs.

As a triple-bottom-line social enterprise, GreenBridge Computing delivers value to its shareholders, the communities served, and our planet. GreenBridge does this by creating affordable technology access, reducing energy use by up to 80% – and donating the equivalent of 5% of their profits to schools in low-income communities.

31

Microsoft paid for the installation of four GreenBridge Computing-based school ICT labs as part of the Nanyuki trial.

Jamii Communications – Jamii Telecommunications Limited (“JTL”) is a Kenyan company incorporated on 20th April 2004 under Certificate of Incorporation No. C100623. JTL is Microsoft’s designated Tier 2 network operator partner for this project.

JTL is licensed by the Communications Authority of Kenya as a Telecommunications Operator under the new Unified Licensing Framework. JTL offers data communications services to telecommunications operators, corporate, Media Houses and Government in East Africa using Terrestrial fibre optic cable, Satellite and Broadband wireless networks (WiMAX).

JTL offers Carrier of Carriers Network Service, Internet Backbone Services, International Data and Voice Services, Satellite Solutions (SCPC, DVB, and VSAT), Broadcast Signal Distribution and Public Data Network services. JTL is providing backhaul to the Nanyuki trial through its optic fibre network termination in Nanyuki.

University of Strathclyde – the University of Strathclyde’s Department of Electronic and Electrical Engineering comprises over 65 members of academic staff and is one the largest in the UK. Staff from the Centre for White Space Communications provided prototype Wi-Fi-based white space radio equipment for the test bed on Bute, Scotland, and were a core partner in the practicalities of installing and configuring the radio network alongside British Telecom and BBC. Currently Strathclyde is running an Ofcom (UK communications regulator) test bed in partnership with Microsoft, 6Harmonics, Mediatek, Adaptrum and the Scottish Government on its City Centre campus in Glasgow, and ensuring knowledge exchange and sharing of best practices with the Kenya project. The University of Strathclyde’s latest project in Glasgow featured the first in the world triple mode radio prototype implementing the new 802.11af Wi-Fi standard for TV white spaces.

2.2.THE TRIAL NETWORK

The Trial’s Radio Access Network or RAN layer consisted of 3 x TVWS base stations supplied by two different hardware vendors located in and around Nanyuki as indicated in the map below in Figure 6. The locations were selected based on proximity to the base stations, local IT infrastructure along with other connectivity requirements at each location. Each of the trial sites where a TVWS CPE is installed receives a best effort Committed Information Rate (CIR) of 2.5 Mbps with most end point/users being able to

32

enjoy burst speeds of 8 to 16 Mbps and being able to download well in excess of 10 GB worth of traffic per day. When this performance is combined with the fact that each base station radio operates at below 1 Watt EIRP trial, highly efficient spectrum re-use can take place. This in turn lends itself to the use of a TVWS data base and increases the overall spectrum efficiency, yielding more ‘Mbps’ per Km2/ captor for money spent, for spectrum occupied. During the course of the trial, there have been no reported cases of interference and no visible increase to the local noise floor of the UHF band (which cannot be said of the 5.8 GHz Band B, particularly in and around Nanyuki, where the project has seen an increase in overall levels by as much as 10 dB).

The following series of diagrams depict the Mawingu network site locations (Figure 5), the transport network layout (Figure 6), the estimated coverage area of the TVWS sectors (Figure 7), and the TVWS coverage area as predicted by the open-source Radio Mobile network planning tool (Figure 8).

33

SNR Mbps

No Site Type Site Name Technology Km Noise RX/TX

Fibre POP Base 5.8GHz + MR-12 + CIR 20/50 Mbps 1 Kenya Power n/a n/a Station Fibre GPON Burst

5.8GHz + 6H (698 55 Mbps Back- 2 Mini Base Station Cape Chestnut 1.1 Km n/a MHz) + NOC haul

16/6 Mbps : 3 CPE Red Cross 6H 1.7 Km -74 dBm/-101 dBm 7/3 MCS

4 CPE Town Hall 6H 0.8 Km -48 dBm/-96 dBm 11/4 Mbps :

13/6 Mbps : 5 CPE Laikipia Library 6H 0.6 Km -46 dBm/-98 dBm 11/3 MCS

Gakawa 5.8GHz + ACRS1 (570 26 Mbps Back- 6 Mini Base Station 6.8 Km n/a Secondary MHz) haul

8/8 Mbps (64 7 CPE Dispensary Adaptrum 7.4 Km -67 dBm/-102 dBm QAM)

7.3/6 Mbps (64 8 CPE Male Secondary Adaptrum 12.8 Km -74 dBm/-103 dBm QAM)

8/6 Mbps (QAM 9 CPE Male Primary Adaptrum 12.6 Km -73 dBm/-102 dBm 64)

5.8 GHz + 6H (586 38 Mbps Back- 10 Mini Base Station Tambuzi 12.1 Km n/a MHz) haul

6/6 Mbps : 3/3 11 CPE Thome School 6H 6.0 Km -75 dBm/-92 dBm MCS

16/6 Mbps : 7/3 12 CPE Tigithi School 6H 5.6 Km -74 dBm/101 dBm MCS

13/14 Mbps : 12 CPE Mawingu Ben 6H 0.2 Km -24 dBm/101 dBm 5/6 MCS

Figure 5 - Table of trial network nodes.

34

Figure 6 - Map showing the physical locations of the 3 Base Stations and their associated CPE.

Figure 7 - Approximate coverage area of all 3 Base Stations located at Cape Chestnut, Tambuzi, and Gakawa. The total area covered with a RX level of -80 dBm or greater is irradiated from 3 x 90 degree sectors is approximately 235 Km2.

35

Figure 8 - Radio Mobile coverage predictions for all 3 base station sites: Cape Chestnut, Tambuzi Flower Farm and Gakawa Secondary School.

The Radio Mobile coverage predictions shown in Figure 8 indicates over 235 Km2 of coverage from three TVWS sectors, each transmitting 1 Watt EIRP from 90 degree 10 dBi UHF sector antennas at heights of 8 to 18m. The aggregate throughput achieved was approximately 3 x 16 Mbps.

36

3. NETWORK INSTALLATION AND CONFIGURATION .

3.1. BACKHAUL AND DISTRIBUTION

Initially, the trial’s main upstream Internet connection was established using fibre provided by Kenya Power Ltd. However, during the course of the project JTL Telcom Ltd offered better commercial terms for fibre connectivity, so Mawingu switched fibre providers. With JTL, a high speed GPON service was activated as part of their national GPON roll-out, and this service delivers a Committed Information Rate of 20 Mbps bursting to 50 Mbps. Round trip times to the Kenya IXP are approximately 4-5 ms and the service is presented on an FX 100 Mbps full duplex L3 ethernet interface. Since activating this service, there has been no fibre down-time. JTL’s service is terminated at the Kenya Power 133 KVa substation located on the outskirts of Nanyuki, shown in Figure 9. A Ubiquity point to multipoint 5.8 GHz Rocket 5M radio using a 20 dBi antenna at 35m on Kenya Power's Telecommunications mast provides connectivity to both the Tambuzi and the Cape Chestnut base stations. Each remote base station connects to this access network via 25 dBi antennas, 5.8 GHz links, running at between 30/40 Mbps. The project uses a Mikrotek gateway router which in turn is connected to a Meraki MR12 cloud controller where all of the network’s IP addressing is managed via DHCP. At Cape Chestnut, the 5.8 GHz Ubiquity link is rolled over into a 6Harmonic base station. At Tambuzi, farm a 5.8 GHz link is rolled over into a 6Harmonic base station and an additional 5.4 GHz microwave link to Gakawa School which then connects to an Adaptrum TV White Space base station.

37

Figure 9 - Picture and set up of Kenya Power Mast

3.2. WHITE SPACE RADIO EQUIPMENT

Two types of white space radio equipment are intended for use in the trial:

Prototype/Pre-Production Adaptrum ACSR1 equipment was originally tested at the start of the trial under supervision of CAK engineers. These radios were certified by the US FCC’s Office of Engineering and Technology and are now being replaced with newer Adaptrum ACRS2 production units pending the CAK type approvals process.

Figure 10 - Adaptrum ACRS2

38

The 6Harmonics GW 300 core adaptive radio (CAR, which is the basestation radios)/edge adaptive radio (EAR, which is the CPE unit) hardware based on Wi-Fi 802.11 configured to operate within a single 8 MHz-wide TV channel.

Figure 11 - GW300 EAR/CAR

CPE units comprising a white space transceiver and aerial were installed at trialists’ premises, as shown in Figure 12. The white space transceivers were mounted externally and connected to an Ethernet cable which was fed into the premises.

Inside the premises, a small Ethernet/Wi-Fi router was supplied to connect end user devices. In the case of Ben’s ‘Solar Cyber’ Agency, solar power equipment was included. This consists of 2 x 250 Watt Solar panels with a suitable MPPT DC to DC inverter, 2 x 220 Amp/hour AGM 12 volt batteries capable of generating and storing 2 to 2.5 Kwh of power a day.

The Ethernet cable was also used to deliver electrical power to the externally-mounted white space transceivers, using Power-Over-Ethernet (POE).

39

Figure 12 - A typical CPE installation at trialists' premises

Figure 12 above the red circle highlights the 10 dBi log periodic CPE antenna and one of the two 250 watt solar panels on the roof of a ‘Solar Cyber’ agency.

Three TVWS Base Station masts were supplied and installed during the trial period, including the use of Kenya Power’s 60 m SCADA and PMR mast which is located at their main 133 KVa HV switch park just south of Nanyuki. This site was also used as the location for the gateway/backbone router where the Meraki MR 12 (a cloud managed network controller appliance) accessed JTL’s fibre POP via a standard GPON service on an FX 100 Mbps full duplex L2/3 Ethernet port.

Ben’s Community ‘Solar-Cyber’ iKioski is capable of serving ‘Power and Packets’ for up to 8 customers simultaneously from 6 x recycled laptops and 2 x Samsung Windows 8 tablets, with both the TVWS back haul CPE and Wi-Fi/Hot Spot all running off a small AC/DC 600 watt inverter. With typical duty cycles of usage at the iKioski consuming less than 50% of the average 2.4 KWh of power generated per day, this leaves the balance for phone and torch charging which pays for the leasing/financing of the ‘Solar Cyber’ agency as a local small business.

The local community has been supportive of the initial deployment from day one and now rely on Mawingu connectivity as they go about their day to day lives. At the beginning of the trials the maximum number of simultaneous client devices on the network averaged 20 to 30 a day but within 12 months this number has grown to over 220 unique MAC address/devices and is growing steadily. The ‘peak traffic’ in July reached 1.2 TB and this is also increasing.

40

At Gakawa Secondary School, as shown in Figure 13, the installation consists of an Adaptrum ACRS2 White-Space base station on an 8 m mast. In the figure below the 5.4 GHz backhaul antenna connects to the Tambuzi Farm mast.

5.4 GHz Microwave Backhaul Antenna

10 dBi 90 degree UHF Sector

Antenna

Figure 13- Gakawa Secondary School base station

Figure 14 depicts the predicted RF coverage of the TVWS sector at Gakawa secondary school.

41

Figure 14- Radio Mobile coverage pattern for Gakawa Secondary School approximately 85 KM2 at RX -80dBm, 8 metre mast with 10 dBi gain log periodic 90 degree sector.

Figure 15 shows the radio line of sight analysis between Gakawa Secondary and Male Primary schools.

42

Figure 15- Radio Mobile path profile and path loss: Gakawa Secondary School to Male Primary School.

Figure 16 shows the 6Harmonics radio monitoring display capability.

Figure 16- Screen shot of 6 Harmonics Thome school’s EAR -75 dBm rx ch 35 (585 MHz).

43

3.2.1. Cape Chestnut base station

5.4 GHz Microwave Backhaul Antenna

10 dBi 90 degree UHF Sector

Antenna

Figure 17 - 6 Harmonics CAR 30 white-space base station on an 18 m mast at Cape Chestnut Nanyuki. Ch 49 (698 MHz) the 5.4 GHz 25 dBi backhaul antenna to Kenya Power 133 kv switching station.

44

Figure 18 depicts the predicted RF coverage of the TVWS sector at Cape Chestnut.

Figure 18 - Radio Mobile coverage pattern: approximately 104 km2 at rx -80 dbm.

Figure 19 shows the radio line of sight analysis between Gakawa Secondary and Male Primary schools.

45

Figure 19 - Cape Chestnut to Red Cross Nanyuki HQ.

Figure 20 shows the 6Harmonics radio monitoring display capability.

Figure 20 - Screen shot of 6 Harmonics CPE/EAR at the Red Cross Nanyuki rx -75 dBm ch 35 (585 MHz).

46

Figure 21 shows the spectrum analyzer output for the link between Cape Chestnut and the Red Cross health care site.

Figure 21 - Spectrum plot, Cape Chestnut to Red Cross Kenya (20 MHz span).

47

3.2.2. Tambuzi base station

2 x 5.4 GHz Microwave Backhaul Antennas 10 dBi 90 degree Gawaka School UHF Sector Kenya Power Antenna

Figure 22- 18 m lattice mast: 1 x 5.8 GHz link to Kenya Power 1 x 5.4 GHz link to Gakawa. 1 x 6Harmonics base station with 10 dBi 90 degree v pol antenna tx power 20 dBm.

Figure 23 depicts the predicted RF coverage of the TVWS sector between Tambuzi and Thome.

48

Figure 23- Radio Mobile coverage pattern: approximately 47 km2 at -80 dBm rx.

Figure 24 shows the radio line of sight analysis between Tambuzi and Thome.

Figure 24- Radio Mobile path profile and path loss: Tambuzi to Thome school.

49

Figure 25 shows the 6Harmonics radio monitoring display capability.

Figure 25 - Screen shot of 6 Harmonics Thome school’s EAR -75 dBm rx ch 35 (585 MHz).

50

Figure 26 shows the spectrum analyzer output for the link between Tambuzi and the Thome.

Figure 26- Spectrum plot, Tambuzi to Thome school, (20 MHz span, Preamp ON).

51

4. DATA THROUGHPUT MEASUREMENTS

Data throughput on the network was measured and analyzed in a number of ways. First, the white space equipment was tested in the laboratory with base stations and CPEs connected using cables and attenuators so that RF signal levels could be varied in a controlled manner and the results measured. Having completed this work at University of Strathclyde (UK) and down in Nanyuki, it was confirmed that all the hardware installed in the network performed as per the manufacturers’ technical performance data. This was followed by field measurements, with both the Adaptrum and 6Harmonics equipment installed in a Land Cruiser used as a mobile survey vehicle in taking measurements and throughput performance tests at various locations across the project’s base station coverage area. Such testing confirmed the accuracy in the original Radio Mobile predictions and was found to be within 2 to 3 dB of measured receive signal strength. This helps in giving the radio access networks’ RF planners a reasonably accurate tool to model coverage which is invaluable in the overall network design.

Factors Affecting End-to-End Throughput

The achievable end-to-end throughput is a function of several factors:

• Path loss and multi-path effects associated with the radio channel; • Physical Layer capabilities of the radios, e.g. modulation scheme, PA and LNA performances etc. • Efficiency of the MAC layer in its ability to share the radio channel among different users; • The Internet transport protocol in use, e.g. TCP or UDP (see notes in box below)

52

TCP vs UDP

Internet applications commonly use one of two transport protocols: Transmission Control Protocol (TCP) or User Datagram Protocol (UDP).

With TCP, each data segment is checked for errors at the receiver, and an acknowledgement is sent back to the source to confirm correct reception of the segment or a block of segments. If the source does not receive an acknowledgement within a certain time, it will assume that the data was not successfully delivered and will schedule a retransmission. TCP is typically used for non-real-time, error-sensitive applications in which the correctness of the data is important, even if this requires multiple transmissions. Examples include file transfer, web browsing, and email.

With UDP, no acknowledgements or retransmissions occur. If a segment is received with errors, the receiver handles this situation in an application- dependent way, and the source receives no information about whether the segment was delivered successfully or not. UDP is typically used for real-time applications in which the time of arrival of segments, rather than their correctness, is critical. Examples include streaming applications such as Voice over IP (VoIP) or video-conferencing, in which occasional packet loss can be tolerated.

Because UDP does not involve the overhead of acknowledgement and retransmission that exists in TCP, it can offer higher throughput than TCP, albeit at the expense of reliability.

An important factor which must be borne in mind when using TCP is that successful data transmission relies on successful acknowledgements in the reverse direction, so any asymmetry in the radio link that causes acknowledgement packets to be lost could have a negative impact on TCP performance, even in situations in which a strong link exists in the forward direction.

Another important factor is the Round-Trip Time (RTT); i.e., the time between transmission of a packet and reception of the corresponding acknowledgement. RTT has two components:

53

• Delays caused by buffering and processing within radio equipment. (The typical latency found in current radio technologies is between 5 ms and 100 ms.) • The two-way propagation delay of the radio path. (For low-power TVWS applications, where the range is generally less than 10 km, the two-way propagation delay will be less than 60 µs.)

Radio equipment latency will therefore tend to dominate the overall RTT for low-power TVWS applications.

When using TCP, a ‘window size’ is specified by the receiver and communicated to the transmitter. This essentially enforces an upper limit on the amount of data that the transmitter may send in the absence of corresponding acknowledgements being received from the receiver. Once the window has been filled, the transmitter is not permitted to send any further data until either an appropriate acknowledgement is received for data that has already been sent or an acknowledgement time-out is reached.

For a communications link with a data capacity of C bits per second and a round-trip time of D seconds, a continuously-transmitting transmitter would transmit at least C×D bits of data before receiving an acknowledgement. This implies a window size, W, of at least C×D bits. If, however, the window size is smaller than C×D bits, the transmitter will be forced to suspend transmission when the window becomes full, and will need to wait until an appropriate acknowledgement is received before resuming transmission. In this situation, the average data throughput rate is less than C and the link utilization is sub- optimal.

It is important, therefore, to ensure that the TCP window size is configured appropriately for the RTT that exists within the network if optimal data throughput is desired. For maximum throughput, the window size needs to be greater than or equal to the bandwidth-delay product. Some applications and operating systems are capable of automatically adjusting or ‘auto-tuning’ the TCP window size to take account of link characteristics, but not all are capable of doing this. Furthermore, in order to achieve high data throughput on a high-delay link, a relatively large TCP window size is needed, and this consumes more buffer memory in the sending and receiving devices.

4.1. DATA THROUGHPUT Figure 29 we show the data throughput that has been achieved at a number of sites, with different radios, distances and noise floors.

54

Mbps SNR Noise floor No Site Name Technology Km dBm dBm RX/TX

1 Red Cross TVWS 1.7 Km -74 -101 16/6 Mbps : 7/3 MCS

2 Town Hall TVWS 0.8 Km -48 -96 11/4 Mbps :

3 Laikipia Library TVWS 0.6 Km -46 -98 13/6 Mbps : 11/3 MCS

4 Dispensary TVWS 7.4 Km -71 -102 8/8 Mbps (64 QAM)

TVWS 12.8 -103 5 Male Secondary -74 7.3/6 Mbps (64 QAM) Km

TVWS 12.6 -102 6 Male Primary -73 8/6 Mbps (QAM 64) Km

7 Thome School TVWS 6.0 Km -75 -92 6/6 Mbps : 3/3 MCS

8 Tigithi School TVWS 5.6 Km -74 -101 16/6 Mbps : 7/3 MCS

9 Mawingu Ben TVWS 0.2 Km -24 -101 13/14 Mbps : 5/6 MCS

Figure 27- Measured data throughput at trialists' premises.

4.2. USAGE STATISTICS

All end points on the network were tracked and authenticated on to the network via a centrally controlled cloud based managed service via a Meraki MR-12 router located at the Kenya Power 133 KVa Sub Station where it in turn is connected to a JTL upstream feed with public IP addressing delivered via a standard GPON service at a Point Of Presence in Nanyuki. Throughout the duration of the trial, this proved to be a highly available service, with round trip time from Nanyuki to Nairobi being around 2 to 3 ms and then on up to London in around 100 to 150 ms. This locally available service in Nanyuki is comparable to some of the best Internet suppliers can offer businesses in the UK and US.

Figure 28 shows daily traffic measured over a one-month period from July 6th to 5th August 2014. 730,000 URL requests were generated over the month, and total usage for the month was 1.23 TB, over 197 ‘unique client’ logins. 33% of the gross traffic was

55

YouTube/Google video in supporting 1,713,406 traffic flows for a total elapsed time of 43 days. The network also shows it’s capable of regularly bursting up to 24 Mbps. At the time of writing this report JTL, Microsoft and Mawingu Networks are working on doubling this burst capability to 50 Mbps as it’s clear that the network as built is currently ‘back-haul constrained’ and the project partners feel that even higher network utilization statics can be gathered. The gross traffic of 1.23 TB (1TB = 1 Terabyte = 1000 Gigabytes) divided by the number of end points/CPE’s equates to an average of 120 GB per end point per month which is approximately 10x higher than the average household in the UK. At the average wholesale price of bandwidth in Kenya via a 3G network this would equate to approximately US$500 to US%600 worth of traffic/utilization.

Figure 28- Screen shot of Meraki Cloud controlled MR-12 Traffic analytics

4.2.1. YouTube HD Streaming Tests

Assessing and demonstrating the ability of TV-band white space to provide a broadband link that can support the latest applications was an aim of the trial. Streaming YouTube HD (High Definition) - 1080p video requires a very stable, low latency and low jitter connection.

56

To ensure high-quality, uninterrupted programme playback, the trialists would have to have an Internet connection that delivers at least 2 Mbit/s of sustained bandwidth for Standard Definition (SD) content, and at least 3 Mbit/s of sustained bandwidth for High Quality (HQ) or High Definition (HD) content.

4.2.2. ‘Virtual Village’ via Skype

During the extent of the trial many HD Skype two-way interactive video conferencing calls/demos were made, into offices and auditoriums both locally and abroad with great success. To have smooth and fluid HD video sessions the latency and jitter of the network must be reliable and consistent. For comparison purposes, latency on the Mawingu network was 40 to 90 ms, with sub 200 ms for UK/EU and sub 300 ms for USA. We believe this is very acceptable performance worthy of being called ‘broadband.’ Some live demonstrations were to large gatherings of policy makers at events like ITU Telecom World in Bangkok and to the EU Development Days in Brussels. If correct hardware and firmware are used from the latest ‘smart phone’ devices, the overall speed and quality of the TVWS network allows for extremely powerful and very life like conferencing and is a very powerful way to see the much talked of phenomenon of the ‘Digital Divide’ being closed.

57

5. SURVEY OF TRIALISTS

In addition to the technical performance measurements described in the preceding sections, a survey of the trialists was carried out in order to gain insight into their experiences of using the broadband service they have been receiving, and their views on the value and usefulness of broadband for rural communities. We present these surveys in this report, using the language and data directly from the end users.

Four different end user reports are presented:

• Ben’s agent outlet (a rural cyber café) • Laikipia county Red Cross (Kenya) offices • Laikipia county government • Local school ICT labs

5.1. BEN’S AGENT OUTLET

Figure 29 - Benson Maina holds a TV white spaces antenna in front of the Mawingu white spaces broadband pilot container near Nanyuki, Kenya.

The pilot at Ben’s agent outlet offers the public access to Wi-Fi, devices and services free of charge. In this section we feature information and review from Ben, in his own words, and using the data he has collected on the success of the white space project to date.

Mr. Benson Maina is a Mawingu agent – providing Internet access and cell phone and other device charging to his rural community. Opened in November 2013, Mr. Maina’s business is located 17 kilometres from the town of Nanyuki in Kenya, and about 200

58

kilometres from the nation’s capital, Nairobi. Mr. Maina’s shop is part of the Mawingu White Spaces Broadband Project.

The solar-powered container (the lab is actually housed in a shipping container) is the first in what could become a network of containers and other shops across rural Kenya, and hopefully Africa, to provide access to the Internet, electricity to charge devices, and technological expertise and guidance from tech-savvy Mawingu agents such as Mr. Maina. The services are free of charge as the pilot tests the concept. The long-term plan is to charge a very affordable fee for use of all containers throughout the network. Each container will be strategically located to encourage the optimum number of visits from the surrounding community.

“Living here for the last 10 years, I have seen people suffering. If I wanted to know about something in high school, I had to buy a newspaper, but I didn’t have money,” Mr. Maina said. “Mawingu has had a huge impact on the community already. Having access to Internet and technology is life-changing – and it’s the way to alleviate poverty. People in the area will begin having incomes as a result of information obtained from the Internet. In a few years, this area will be different than the rest of the country; we will be icons for what’s to come.”

Figure 30- Students conduct research at the Mawingu white spaces broadband pilot container.

One of Mr. Maina’s first clients, James, is a nearby farmer who wanted to know about government farming regulations and chemical treatments that would benefit his crops. Mr. Maina’s clients include many more farmers and small business owners, like Steven, who came to the Mawingu container to learn about computer software. After researching software solutions online and with help from Mr. Maina, Steven started a prosperous business installing and repairing commercial and residential software in Nanyuki and other towns. Job seekers like Diana, a teacher who was unemployed for a year, use the Mawingu container to research and apply for jobs and communicate with prospective employers. Mr. Maina helped Diana find teaching opportunities online

59

which led to the teaching position she now holds at Doldol Secondary School. Teachers and students are also frequent clients, researching lesson plans and assignments.

“People come to the Mawingu container multiple times a week for a variety of reasons,” said Mr. Maina. “Healthcare is an important use case. One of my clients was a child whose family member had symptoms of malaria. The child researched the possible causes of illness to help determine whether to visit the doctor. Other times, children will learn how to do something on a computer at the container, and then they’ll go home and teach their parents what they learned.”

By offering affordable access to broadband and devices in rural areas without Internet or even electricity, people avoid time-consuming, tiring trips of many kilometers – often travelled by foot – to expensive Internet cafes where they spend their savings checking emails, researching job opportunities and looking for other information.

Mawingu aims to continue to be a centre for trade and a hub for commerce by transforming the traditional, inefficient and expensive marketplace trading routines. Rather than traveling to city centre markets to sell goods, Mawingu offers merchants the opportunity to generate awareness and attract customers to visit the merchants, as well as work with neighboring vendors to form cooperatives. Internet access takes the burden of travel, time and transportation funds off the merchants and allows more time for goods production. Connectivity also enables expansion of product distribution to new markets.

Following are a series of status reports prepared by Ben over recent months:

5.1.1. Agent activity report from September 2013 The following end users are frequent visitors at the Mawingu agent outlet:

For the two month and a half I have received a lot of client but I have done a sample of at least 50 clients at random:

No. AGE OCCUPATION INTEREST HOW OFTEN 1 19 STUDENT WIKIPEDIA ,FACEBOOK DAILY EMAILS, GOOGLE ON FARMING 2 34 WORKING DAILY AND CHEMICAL INFO CASUAL 3 23 NEWS ,FACEBOOK DAILY WORKER 4 26 HOUSE WIFE EMAILS,SEARCHING JOBS 3 TIMES A WEEK 5 35 TEACHER EMAILS, SEARCHING JOBS DAILY EMAILS ,LEARN MORE ON 6 45 TEACHER DAILY NGOS AND PROJECTS EMAILS ,FACEBOOK AND 7 29 WORKING YOUTUBE TO MARKET HIS DAILY ALBUM

60

No. AGE OCCUPATION INTEREST HOW OFTEN LEARNING MORE ON 8 33 WORKING DAILY COMPUTER BASIC 9 15 STUDENT ENCARTA ,WIKIPEDIA 3 TIMES A WEEK GOOGLE TO SEARCHING 10 38 FARMER 2 TIMES A WEEK AGRICULTURE INFORMATION 11 48 WORKING EMAILS ,NEWS 2 TIMES A WEEK CASUAL 12 24 SEARCHING JOB,FACEBOOK 3 TIMES A WEEK WORKER 13 17 STUDENT EMAIL ,FACEBOOK 3 TIMES A WEEK 14 33 WORKING SEARCHING JOB ,NEWS 2 TIMES A WEEK SEARCHING JOB, CASUAL 15 30 NEWS,AGRICULTURE 3 TIMES A WEEK WORKER INFORMATION BUSNESS 16 50 COMPUTER BASIC 1 TIME A WEEK WOMAN 17 13 STUDENT MORE ON EDUCATION 3 TIMES A WEEK 18 16 STUDENT EDUCATION ,FACEBOOK 3 TIMES A WEEK SEARCHING JOB ,NEWS 19 25 WORKING DAILY ,FACEBOOK 20 18 STUDENT FARMING ,FACEBOOK DAILY BUSINESS SEARCHING NEEW FORMULA 21 35 1 TIME A WEEK WOMAN OF MAKING SOAP 22 17 STUDENT FACEBOOK 3 TIMES A WEEK 23 25 WORKING NEWS ,FACEBOOK 3 TIMES A WEEK 24 20 WORKING EMAILS 3 TIMES A WEEK CASUAL 25 19 NEWS FARMING DAILY WORKER 26 17 STUDENT FACEBOOK DAILY 27 26 STUDENT DISTANCE LEARNING DURING HOLIDAY 28 20 WORKING EMAIL,SEARCHING JOBS DAILY 29 28 WORKING NEWS 2 TIMES A WEEK 30 17 STUDENT NEWS FACEBOOK 3 TIMES A WEEK 31 31 WORKING NEWS ,FACEBOOK 3 TIMES A WEEK 32 30 HOUSE WIFE EMAIL,SEARCHING JOB 2 TIMES A WEEK DURING LEAVE AND OFF 33 35 WORKING EMAIL,WORKING ONLINE DAYS 34 25 STUDENT DISTANCE LEARNING 3 TIMES A WEEK 35 32 WORKING EMAILS 3 TIMES A WEEK 36 26 STUDENT WIKIPEDIA ,FACEBOOK 2 TIMES A WEEK 37 27 STUDENT DISTANCE LEARNING DURING HOLIDAY 38 32 WORKING EMAILS 3 TIMES A WEEK CASUAL 39 30 NEWS ,ARTIST FORMULA 2 TIMES A WEEK WORKER 40 26 STUDENT DISTANCE LEARNING DURING HOLIDAY 41 30 WORKING SEARCHING JOB 2 TIMES A WEEK MORE INFORMATION FROM 42 35 WORKING 1 TIME A WEEK INTERNET 43 35 WORKING COMPUTER BASIC 2 TIMES A WEEK 44 34 WORKING INFORMATION FROM NGOs 3 TIMES A WEEK 45 16 STUDENT MORE INFO ON EDUCATION DURING HOLIDAY 46 45 BUSINESS MAN MORE INFO ON INTERNET 1 TIME A WEEK

61

No. AGE OCCUPATION INTEREST HOW OFTEN 47 40 BUSINESS MAN EMAIL SEARCHING JOB 1 TIME A WEEK 48 43 CHIEF MORE INFO ON INTERNET 1 TIME A WEEK 49 38 VILLAGE ELDER FARMING INFORMATION 1 TIME A WEEK 50 30 WORKING EMAIL, FACEBOOK 3 TIMES A WEEK

• TYPICAL ACTIVITIES: Students have come to realize the importance of the Internet in terms of education, they are spending most of their time in searching important things like: past exam papers from www.jesmapublishers.com theme and character. • Students and teachers flock at the Mawingu container especially during the holiday and visit various websites, most notable www.wikipedia.com and www.encarta.com where they can access education information in more detail than currently available to them. • Teachers are trying their best to stay ahead of the students. They regularly visit Encarta program for more and deeper information in various syllabi. • Work and Business –Mawingu container has become the centre of communication for businessmen and working people who are on their off duty or leave. Since County government started working in Kenya there have been a lot of vacancies, so a good number of people are coming regularly to search for job vacancies from various county assembly board positions because they are posted on the net. • Leisure- there is a good number of youth that spend most of their time on-line using Facebook, twitter and Skype - and now communicating and charting with old and new friends from all over the world.

THE BUSY DAYS AT MAWINGU

• From Monday to Thursday the number of people attending Mawingu is constant, but from Friday evening, Saturday and Sunday the number of people rises because many people are off duty / out of school. • 70% of the clients visit Mawingu at least four times a week; i.e., I have 20 regular clients.

TRANSPORT AND GETTING TO THE CONTAINER

Walking is the main means of transport to Mawingu; some clients are known to walk as far as 6kms to get service at Mawingu - and then another 6kms back home! Bicycle is another very common means of transport to Mawingu, and the boda boda (motorcyclist) guys use their motorcycle to get at Mawingu when they don’t have client. A good number of people come from within the village.

HOW MUCH THE CLIENT WOULD LIKE TO PAY

62

• One shilling per minute is the rate that many people are suggesting to be paying since the Internet is delivering to them perfectly. • Some would like to have membership card.

CHARGING

• A lot of the clients come with their phones to be charged when they are on Internet, some of the clients would drop the phone at Mawingu and pick it late during the day when they are fully charged. Charging one phone or Camera would cost ksh15. each

AGENT VIEW

• We should have the printer and scanner to make it easier for those who would like to attach their document to the emails. • It’s proven that Mawingu Internet is the best in the area by receiving different clients from all corners of Burguret.

The most benefited people are as follow:

NO GENDER AGE OCUPATION BENEFITED WITH 1 M 22 Self Works as an freelance I.T man within Nanyuki town , he install employed programs and software to different client. Before Mawingu Internet he was spending a lot of money to travel from Burguret to Nanyuki every day to access Internet. Now with Mawingu Internet he is able to learn more about software and new programs also communicating with his clients through emails without any expenditure .now he is saving more than ksh 400 everyday

2 M 38 Farmer Has been a farmer for the last 15 years, he specialize with perishable produce which require a lot of care, so before he got good information he has been counting loss day by day .But since he visited Mawingu Internet his life has changed, he is able to get the best input e.g. certified seeds, registered chemicals and also good markets.

3 M 37 Police force Works with Kenya police at Kakamega in IT department during his long holiday he can work with Mawingu Internet without any expenses. He can respond to any email at any time which is improving his performance at his career.

4 M 27 Student at - Online literature. JKUAT - Communicating with lectures who is guiding online. - Downloading courses from the Internet. - Research work from the Internet. - Emails. 5 F 30 Teacher Has been working as a teacher in a private school

63

NO GENDER AGE OCUPATION BENEFITED WITH After she started using Mawingu Internet, thing changed she spend most of the time in searching job. She visited site like www.tscjobs.co.ke and finally she got job at government school.

6 M 27 Worker at Working at Tambuzi and also a musician at his own time. Tambuzi Mawingu Internet has help him to market his album through YouTube and also learning more about music.

5.1.2. Activity report for December 2013 holiday period Visitors in the period wishing to identify themselves:

AGE OCCUPATION INTEREST HOW OFTEN 1 18 Form 4 Leaver Mostly comes to Four times a week browse more on universities and colleges plus the courses they offer

2 10 Student Encarta kids to learn Everyday more on the 8-4-4 programme

3 18 Form 4 Leaver Facebook, Youtube 3 times a week and browsing for universities and colleges

4 25 Working Emails Everyday

5 14 Student Completing Holiday Everyday assignments

6 27 Tambuzi Mostly comes to Once or twice a week. Worker learn more about music and different ways of shooting videos since he’s a musician

7 13 Student Mostly comes to Four times a week learn more on the World’s History of various types of gov’ts (Social Studies)

8 22 Jobless Searching for jobs Twice a week

64

AGE OCCUPATION INTEREST HOW OFTEN 9 30 Working Emails Everyday

10 43 Teacher Reading e-mails and Once a week reference for school work

11 14 Face book and Student Everyday YouTube

 During December holiday, we experienced a lot of youth since they were on holiday.  A good number of youth from urban area who visited their rural home enjoyed the services since unlike cyber in town this is free fast Internet and near to them.  The class eight leavers were taking this advantage to know more about the Secondary school they want to join.  The working people were accessing their emails on daily basis from Mawingu Internet despite being in rural area during the holiday.

CLIENT RECOMMENDATION

 To add more computers in the container.  To start a computer basic training program within the plot.

5.1.3. Activity report from 20th May 2014

WHO VISITED MAWINGU

AGE: Below 18 years constitute about 25%, 60% are aged 18-35 years and 15% are aged above 35 years

SEX: 80% of Mawingu clients are male while females constitute 20%.

This is because males have fewer chores to perform especially during afternoon and evening hours.

NUMBER OF CLIENTS: Here at Mawingu cyber we receive average of 20-30 clients a week.

65

DAYS' ATTENDANCE: A few customers visit in the morning but most of the clients stream in in afternoon. We have average of 4-7 persons visiting Mawingu on Monday, Tuesday, Wednesday and Thursday while during the weekends, that is, from Friday to Sunday, clients count increases to even 10 clients. Some of them are even forced to share the computers.

DISTANCE COVERED BY CLIENTS

People travel as far as 6km to access Mawingu cyber services.

MEANS OF TRANSPORT to Mawingu is by walking. Some use bicycles and others hire motorcycles to access our services.

WHAT DO THEY DO

 EDUCATION: Students browse to research on subjects they have learnt in class for better understanding. Former students are visiting the KUCCPS to view the courses offered in the Universities for them to make informed decisions on what to take as their courses. Also, teachers come in to research and gather reference and teaching materials.  BUSINESS & AGRICULTURE: Some people come in to check on current business trends for them to enhance their productivity on the products that they are taking to market. Farmers in the locality stream in to research on new farming methods, chemicals in the market and weather trends for them to make favorable decisions concerning their farming practices.  LEISURE: Some clients come in to connect and communicate with other parties in the entire globe via Facebook and mail services such as Gmail and yahoo mail.  CHARGING: Due to inaccessibility of power in some homes in this locality, some customers come over to Mawingu cyber to charge their mobile phones and cameras.

WHAT ARE CLIENTS WILLING TO PAY?

 Customers are willing to pay about 1 shilling per minute for using the Internet this is fair because the Internet is so fast incomparable with other nets within our towns, and 15 shillings for charging their phones and cameras.

AGENT'S VIEW

 Starting a computer college would do the community good since most of the people around are computer illiterate. It will also be earning the business an extra coin in return.

66

 Setting up other Mawingu container cyber in Burguret and Matanya to reach to people living as far as those areas.

IS THE BUSINESS VIABLE (Yes!)

 It keeps the youths around busy and hence keeping them out of criminal activities and drug abuse especially youth engaging with deadly local beer that has killed hundreds of people in central Kenya.  It is enabling farmers and businesspeople to access crucial information on effective production and marketing.  Students and teachers are able to access necessary information for teaching reference and important leaning data.  Generally in order to eradicate poverty we must fight illiteracy by keeping people informed through very cheap, fast and affordable Internet.

67

5.2. LAIKIPIA COUNTY RED CROSS (KENYA) OFFICES

Figure 31- Laikipia county Red Cross offices, Nanyuki.

As the coordinator for the Laikipia County Red Cross in Kenya, Mr. Anthony Kuria is responsible for liaising between the Laikipia County Red Cross branch outside Nanyuki and its nearby sub-branches, as well as the Kenya Red Cross Society’s headquarters in Nairobi. The Laikipia County Red Cross branch and its four sub-branches provide county residents with disaster relief and community empowerment services such as support to families that have lost homes to fires (one of the most prevalent disasters in the region), flood and drought damage relief, automobile accident response and proactive education on topics such as nutrition and first aid. Blood donation collection, a quintessential Red Cross program, is also conducted by the Laikipia County Red Cross.

The Mawingu White Spaces Broadband Project is providing the Laikipia County Red Cross branch low-cost Internet access, a drastic improvement to the organization’s efficiency and scope of services, and most importantly, its bottom line. Broadband Internet is transmitted via TV white spaces technology from the Mawingu pilot base station in Nanyuki to a receiver on the roof of the Laikipia Red Cross office, which is located outside town. A router then transmits the signal throughout the office creating a Wi-Fi hotspot.

Mr. Kuria says having Internet at the office allows the organization to accomplish more in less time, and saves substantial funds. “Before we had Internet at the office, we would have to shut down the branch to travel to town or to other Red Cross branches so we could exchange information or receive trainings,” said Mr. Kuria. “Now we can

68

accomplish tasks over the Internet that previously required a week of travel. We don’t have to buy expensive service plans for our modems anymore, and we don’t have to travel to costly Internet cafés in town to send reports to other Red Cross branches. The Mawingu White Spaces Broadband Project is a huge upgrade for our Red Cross branch.”

Figure 32 - Home fire disaster, Nanyuki (1 of 2).

Figure 33 - Home fire disaster, Nanyuki (2 of 2).

Now Mr. Kuria and his colleagues use Microsoft Lync and Skype to give presentations to other branches and receive organizational trainings from headquarters. Work plans, reports and other documentation are easily sent via email, and simple matters that previously required in-person meetings or telephone calls can be communicated and resolved by email.

Another hallmark service provided by the Red Cross in developing regions and disaster situations is locating and connecting missing people with their families. The Laikipia County Red Cross’ Family Link Tracing Program has traditionally assisted families in finding their missing loved ones using Red Cross Messages, a paper form mailed via postal service to other Red Cross branches. This was a lengthy process – Red Cross Messages would be sent between multiple branches until the family member was located; then the news would take a few days longer to travel back to the initiating branch by mail. Email has expedited the service tremendously, enabling instant communication between Red Cross branches and families.

69

Figure 34 - Victims receive donations via Laikipia county Red Cross.

There is a hitch, however. Not all Red Cross branches have Internet access – or even electricity. As a result, expensive and time-consuming travel, phone calls and postal mail are still necessary means of communication between many branches. “We are hoping the TV white spaces technology will help improve communication with our sub-branches and other Red Cross offices by providing Internet access and solar power in more locations,” said Mr. Kuria. “The community at large will benefit from the Wi-Fi and electricity available at the Red Cross sub-branches, if they get connected. Having broadband available at additional locations will help us do our jobs more effectively and serve more people in need.”

The vision of increased connectivity delivered by TV white spaces technology for organizations in rural communities like the Red Cross extends throughout Africa, beyond the Mawingu pilot in Kenya.

Providing broadband to rural areas of Kenya is critical to driving prosperity in rural areas, as well as Africa as a whole. It gives students – both children and adults – a new way to experience learning, and rural communities the ability to connect to the world, improving opportunities for economic development via e-commerce and small business growth.

70

5.3. LAIKIPIA COUNTY GOVERNMENT

Figure 35 - Danilo Legei, Director of ICT & Research Services of the County of Assembly of Laikipia.

In November 2013 Danilo Legei was appointed as Director of ICT & Research Services of the County of Assembly of Laikipia. Prior to joining the Assembly, Legei worked at Kenya Commercial Bank (KCB) and has over eight years of experience providing strategic ICT advice.

In 2010, Kenya adopted a new constitution calling for a devolved system of governance structured around two key facets: national and county. The leadership of the County Assembly of Laikipia, which was created in March of 2013, quickly discovered that good governance -- making local laws reflecting the aspirations of the citizenry -- requires direct citizen contact and public consultation.

Danilo Legei, the Assembly’s Director of Information and Communications Technology (ICT) called this “an undeniably pivotal period” for the county’s development.

In February 2014, the Assembly reached out for assistance from Mawingu Networks, Ltd., a telecommunications service provider in Kenya supported by Microsoft and the USAID Global Broadband and Innovations team (GBI), among others, to implement a new high-speed, low-cost network called ‘Mawingu’ (Swahili for cloud).

Thanks to the Mawingu project, Laikipians can now access the Assembly’s newly launched website to read regular news updates, find transcripts from proceedings, communicate directly with Assembly members, respond to calls for participation or submission, and engage via social media.

Assembly ICT Director Legei recently spoke with a representative of NetHope, a principal implementing partner for GBI.

71

“Mawingu is very cost-efficient for us. We’ve never had better or faster connectivity,” shared Legei. “There’s a lot of interest and high expectation on the part of Laikipians for their government.

This connectivity has allowed us to manage the expectation, to be held accountable.

Many people have already come to the Assembly to witness committee proceedings and give input on basic bills passed to kick-start the system.”

The Assembly is just one of several public institutions now relying on the network to improve the lives of rural Kenyans. Utilizing solar power (funded in part through USAID’s Global Climate Change program) in areas with little or no reliable access to the national electricity grid, the latest TV White Space (TVWS) wireless technology is now delivering broadband speeds to primary school computer labs, a community library, and a kiosk staffed by village entrepreneurs. This technology, operating over unused radio frequencies previously reserved for television, has generated significant excitement in the County, and for the Assembly represents an immediate accomplishment delivered to citizens that strengthens their confidence in the importance of local control of government.

Legei hopes one day citizens will be able to watch live streaming of Assembly committee meetings. For now, access to the network continues to stir the engagement and participation needed to ensure effective citizen representation.

5.4. LOCAL SCHOOL ICT LABS

As part of its Laikipia trial, Mawingu Networks has installed ICT labs at four local schools, connected the labs to broadband Internet via TVWS radios and also trained the local teachers so as to enable them to effectively use the ICT labs as part of their teaching activities.

The trial schools were fitted with ultra-low power Microsoft Multipoint servers and workstations, designed, integrated and commissioned in Kenya by GreenBridge Computing Inc. Each lab is fitted with 30 seats. The servers are installed with Microsoft Office 365, the full Khan Academy curriculum and a special school edition of Wikipedia.

In addition the teachers were facilitated to add their own content relevant to their particular teaching needs and styles.

72

The four ICT lab trial schools are:

 Male Primary School,  Male Secondary School,  Gakawa Secondary School, and  Kisima Primary School

The ultimate test for the schools will be to see how access to the ICT labs will affect the students’ results. Mawingu Networks is monitoring the situation and look forward to sharing these findings as they become available.

The following are selected photos from the launch of the ICT labs at Gakawa Secondary School:

73

Figure 36- ICT Lab launch event at local schools.

74

Figure 37 - - RIchard Thanki instructing Charles - school care-taker at Gakawa School - how to use the Internet.

75

6. NEXT STEPS: EXTENDING THE WHITE SPACE NETWORK

Mawingu Networks is presently working closely with Ol Pejeta Conservancy to bring the benefits of affordable electricity and broadband Internet via white space connectivity to communities in the vicinity of the conservancy.

Leveraging a variety of wireless technologies, including TV white spaces, the proposed collaboration would enable broadband delivery in an area of approximately 1,000 km2 and covering approximately 60,000 to 80,000 people.

Ol Pejeta has over ten years of data from community outreach projects that will form a solid starting point for the monitoring and evaluation of the proposed project.

Significant improvements to wild-life conservancy and anti-poaching will be enabled by the provision of a broadband network in and around the perimeter of the conservancy.

Fundraising is currently under-way to raise the necessary capital to proceed with the project.

Through consultation with Ol Pejeta’s local management team, Mawingu Networks has identified 17 primary schools and 11 secondary schools in the communities surrounding the conservancy. In addition there are also two health dispensaries.

Targeted benefits of the Ol Pejeta collaboration include:

 Provide up to 11 rural Kenyan secondary schools with access to broadband Internet equipped teaching facilities through zero-carbon footprint solar- powered ICT labs.  Provide 28 rural Kenyan primary and secondary schools with broadband connected interactive video projection systems.  Enable up to 30 local entrepreneurs to deliver affordable broadband Internet and charging services to the local communities surrounding Ol Pejeta conservancy.  Reduce hydrocarbon usage by local population travelling by motorbike taxi to access Internet services.  Contribute in a meaningful way to the reduction of hydrocarbon based lighting and electricity in the local communities.  Facilitate access to Internet based agricultural input suppliers market pricing information and trading networks.

76

Table of schools in the vicinity of Ol Pejeta:

Name S coordinate E coordinate 1 St. Augustine Secondary School 00 04 095 036 47 978 2 Ol-Taffeta Secondary School 00 04 163 036 048 950 3 Ol-Taffeta Primary School 00 04 282 036 048 825 4 Tharua Secondary school 00 08 028 036 51 678 5 Tharua primary school 00 08 979 036 51 749 6 Withare dispensary 00 02 947 036 45 973 7 Mwituria secondary school 00 01 449 036 43 245 8 Withare secondary school 00 02 228 036 46 639 9 Withare primary school 00 02 228 036 46 639 10 Male secondary school 00 06 187 036 56 795 11 Thome boys secondary 00 04 262 036 36 085 12 Tigithi secondary school 00 03 461 036 56 848 13 Withare Polytechnique 00 03 175 036 43 875 14 Kijabe Primary School 00 03 185 036 43 892 15 Lechugu Primary School 00 04 213 036 56 088 16 Sweetwaters Secondary School 00 00 321 036 58 869 17 Yard DEB primary School 00 00 323 036 58 870 18 Marura dispensary 00 00 269 036 56 776 19 Wathituga primary school 00 01 308 036 58 628 20 Chuma primary school 00 01 715 036 57 810 21 Matanya primary school 00 06 628 036 57 776 22 Sweetwaters primary school 00 00 836 037 01 664

Name N coordinate E coordinate 23 Serat Primary School 00 06 127 036 56 951 24 Tangi Nyeusi Primary School 00 06 995 036 49 566 25 Irura primary school 00 00 153 036 55 434 26 Uasonyiro primary school 00 06 662 036 53 181 27 Endana secondary school 00 09 293 036 54 828 28 Endana primary school 00 09 258 036 54 550 29 Ereri primary school 00 07 376' 036 55 090 30 Njoguini primary school 00 00 468' 036 59 017 31 Loise Nanyuki Girls secondary 00 01 078 037 03 918

77

Figure 38 - Proposed radio mast locations for Ol Pejeta conservancy project.

78

7. CONCLUSIONS

The TV white space broadband trial in Nanyuki has successfully demonstrated the potential of white space spectrum for providing broadband access to remote, hard-to- reach areas in challenging terrain. The trialists from many walks of life, have embraced the network for a wide range of uses related to working life, social life, leisure & recreation, online shopping, and education. Without exception, both the providers and users of the networks speak enthusiastically and positively about the broadband connectivity they have been receiving as part of the trial. In relating their experiences, almost all of the trialists speak of the immediate and significant value that it has brought to their personal lives and their education and business opportunities. Sometimes this value has simply come from being able to do a search on-line to get information.

Key technology results and findings to date for the Kenya TVWS pilot include:

 No interference caused to any Television Service reported to date;  No increase detected in the UHF band’s noise floor;  Co-channel operation is achievable at low power levels;  Solar Power can meet the supply of a typical of TVWS Base Station’s 24 hour ‘base load’ along with their associated Back-Haul radio links;  Cloud Based (and hence remote) management of all network components is possible;  Operations below 4 Watts EIRP have proven to be most successful to date thus allowing for maximum spectrum reuse in small areas;  That local Kenyan systems integrators are very capable of deploying and maintaining the software and hardware with minimal training;  Hardware and radio costs are falling - notably developments of new white space 802.11af chips, along with other technical developments in channel bonding and MIMO techniques, will bring lower costs but also higher connection speeds and efficiencies;  Using available software and terrain data, white space radio frequency (RF) coverage predictions are reliable and straightforward to generate for various local areas of the country;  TVWS is capable of reaching into relatively low population densities in dispersed areas without existing ‘grid power’;  The WS radio technology and supporting hardware is highly reliable, and as a qualitative data point, there have been zero equipment failures over 12 months across all radios used, and in all the ICT Lab’s hardware and solar equipment.

79

 The use of a ‘Hybrid’ MHz/GHz approach (blending TVWS frequencies alongside WiFi and other available frequencies) allows efficient use of the right radios and spectrum meaning that we can demonstrate just 3 base stations covering areas of almost 235 km2.

At the time of completion of this report the project partners have formally requested the new CAK to extend the pilot project for 24 months. Based on the success and outcomes from the first TVWS technology demonstrator phase of this project, the ‘technical’ success of delivering this broadband service and the various social and economic impacts have been clearly demonstrated. Yet much more needs to be done to assess deeper and lasting impacts.

Finally, the next two images help illustrate the longer term aim of this TVWS and other similar projects. When looking at Africa in Figure 39 there is very little ‘light’ evident on NASA’s ‘world light pollution map,’ which is a proxy to showing where electrical grid power is or is not located. The map in Figure 40 depicts Internet address space activity being used at any one time. The two maps show a clear correlation between (grid) electrical power and Internet users/activity. By implication, it might then seem that we do not get high Internet activity until grid electrical power arrives. However this is precisely one of the points we are demonstrating can be circumvented by TVWS. We can bring the Internet to rural areas with renewable and solar power providing enough power to get users on-line with reliable and good bandwidth connections now. So the electrical grid can come later and will of course be welcome when it arrives – but right now we can still get on line with TVWS connectivity.

Figure 39 - NASA's 'world light pollution map'.

80

Figure 40 - 'Heat map' of world's Internet usage.

81

8. FURTHER REFERENCES AND READING

 See www.dynamicspectrumalliance.org/resources.html.  The Broadband Commission for Digital Development: “The State of Broadband 2012: Achieving Digital Inclusion for All”, September 2012.  (http://www.broadbandcommission.org/Documents/bb-annualreport2012.pdf )  ECC Report 159: “Technical and Operational Requirements for the Possible Operation of Cognitive Radio Systems in the ‘White Spaces’ of the Frequency Band 470-790 MHz” (Figure 62, Page 137), Electronic Communications Committee (ECC), January 2011.  BBC R&D White Paper: “Throughput Measurements On A Trial TV White Space Network”, January 2013. (http://www.bbc.co.uk/rd/publications)  BBC R&D White Paper: “Isle of Bute TVWS Trial: Network Architecture and Interference Studies”, November 2012. (http://www.bbc.co.uk/rd/publications )  BBC R&D White Paper: “Analysis of DTT Measurements on the Isle of Bute”, December 2012. (http://www.bbc.co.uk/rd/publications )  C. McGuire, M. R. Brew, F. Darbari, G. Bolton, A. McMahon, D. H. Crawford, S. Weiss, and R. W. Stewart, “Hopscotch – A Low-Power Renewable Energy Base Station Network for Rural Broadband Access.” EURASIP Journal on Wireless Communications and Networking, 2012:112, 16th March 2012.