Report No: ACS9644 . Using Locational Data from Mobile Phones to Enhance the Science of Delivery Ryan Haddad, Tim Kelly, Teemu Leinonen and Vesa Saarinen . June 2014 . TWICT . 1 Standard Disclaimer: . This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. . Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. 2 Table of Contents Acknowledgements ................................................................................................................................... 5 Summary ....................................................................................................................................................... 6 1. Introduction ....................................................................................................................................... 8 1.1 How locational data can be used ................................................................................................... 8 1.2 The evolving locational data toolkit.............................................................................................. 9 1.3 Structure of the report ................................................................................................................. 10 2. How Do Locational Technologies Work?....................................................................................... 12 2.1 A brief history of locational technologies ................................................................................... 12 2.1.1 Cell Tower Triangulation ........................................................................................................... 12 2.1.2 Global Positioning System (GPS) .............................................................................................. 13 2.1.3 WiFi Positioning System (WPS) ................................................................................................ 14 2.2 Current Location Tracking Usage in Mobile Network uses ....................................................... 15 2.2.1 Data from Cell Tower Triangulation .......................................................................................... 15 2.2.2 Collecting GPS embedded Data ................................................................................................. 16 2.2.3 Analyzing GPS embedded data .................................................................................................. 17 2.2.4 WiFi Positioning System and Location-based services .............................................................. 19 2.3: Potential for the future ..................................................................................................................... 20 2.3.1 Smartphone developments .......................................................................................................... 20 2.3.2 GPS developments ...................................................................................................................... 21 2.3.3 Indoor Positioning Systems ........................................................................................................ 21 2.3.4 Other Technologies..................................................................................................................... 23 3. A Conceptual Framework for studying Locational Data Applications ........................................... 25 3.1 Methodological frameworks ....................................................................................................... 25 3.1.1 Ushahidi Haiti Project Review Framework ................................................................................ 25 3.1.2 Systems and software Quality Requirements and Evaluation (SQuaRE)................................... 26 3.1.3 The Challenges of Implementing New Technology in Developing Countries ........................... 26 3.3 Proposed Framework for the Presentation of Cases ................................................................... 27 4. Case studies of locational data ................................................................................................................ 29 4.1. PoiMapper ........................................................................................................................................ 29 4.1.1. Overview ........................................................................................................................................... 29 4.1.2. PoiMapper Use Cases ....................................................................................................................... 30 3 4.2. Ushahidi ........................................................................................................................................... 32 4.2.1. Overview ........................................................................................................................................... 32 4.2.2. Ushahidi Use Cases........................................................................................................................... 34 4.3. Dimagi CommTrack and CommCare .............................................................................................. 37 4.3.1. Overview ........................................................................................................................................... 37 4.3.2. Dimagi Use Cases ............................................................................................................................. 39 4.4. Taarifa .............................................................................................................................................. 41 4.4.1. Overview ........................................................................................................................................... 41 4.4.2. Taarifa Use Cases ............................................................................................................................. 42 4.5. Comparing the applications ............................................................................................................. 43 4.6. Using Big Data to Enhance the Science of Delivery ....................................................................... 45 4.6.1. Using Big Data to Analyze the Mobility of People .......................................................................... 46 4.6.2. Using Big Data to Analyze Social Interaction and Economy ........................................................... 47 4.6.3. Using Big Data to Analyze the Spread of Malaria ............................................................................ 48 5. Conclusions ..................................................................................................................................... 50 5.1 Using Locational Data to Enhance the Science of Delivery ....................................................... 50 5.2 Potential Uses of Locational Data for Addressing Development Challenges ............................. 50 5.3 Recommendations to development practitioners and policy-makers .......................................... 52 5.4 Final reflections .......................................................................................................................... 52 References ................................................................................................................................................... 54 4 Acknowledgements This project was undertaken jointly by the ICT Sector Unit of the World Bank and the Media Lab Helsinki at Aalto University School of Arts, Design and Architecture. The World Bank team was led by Dr Tim Kelly, Lead ICT Policy Specialist, who acted as task team leader for the overall program of work and contributed chapter one, and included Ryan Haddad (consultant), who contributed chapter two. The Aalto team comprised Prof Teemu Leinonen and Vesa Saarinen who contributed chapters three and four. The team collaborated on chapter five and the Executive Summary. The research for the project was undertaken between the project concept note review, which took place virtually, between 8-22 January 2014, and the Decision Meeting held on 4 June 2014, chaired by Randeep Sudan, ICT Sector Unit Manager. The peer reviewers for the program were Ilary Lindy (Innovation Policy Specialist, WBIKE), Siddhartha Raja (ICT Policy Specalist, TWICT) Rosemary Rop (Water Policy Specialist, TWIWP), Saori Imaizumi (Consultant, TWICT) and Samia Melhem (Knowledge and Learning Coordinator, TWICT). In addition, Zaid Safdar (consultant, TWICT) contributed useful comments. The authors would like to thank Suvi Reijonen of the Finnish Ministry of Environment; Petra Hietaniemi, Pertti Lounamaa and Peeter Pruutem of Pajat Inc; and Henna Hakkarainen of Kepa, the umbrella organization for Finnish development aid related CSOs, for their valuable comments and inputs. The research program was funded in part by the Finland Ministry for Foreign Affairs through its “Finland as a Knowledge Economy 2.0” Trust Fund. 5 Summary This report concerns the use of locational data from mobile phone for enhancing the science of delivery, in development programs. The report can be summarized in a series of key numbers: 1 There is one objective of this report, namely to enhance the science of delivery by using locational data from mobile phones. In this context, the “science of delivery” is defined as evidence-based experimentation, and locational data is one of the toolsets that is becoming increasingly available and useful for this purpose. 2 There are two main audiences for this report. The primary focus is on development practitioners, who are looking to use the locational data toolset to provide a more scientific approach to their work. The secondary focus is on policy-makers, particularly in developing countries, who have development challenges they wish to solve, and who may wish to use locational data for that purpose. 3 There are three main ways in which locational data is used: • Locational data recorded by device users (consumers). The main tool would be maps and other navigational aids that combine location-finding services, such as GPS, with data stored on the phone, such as maps • Locational data recorded by service providers (operators). The main tool here is call data records that can be passed on by operators to researchers. • Locational data recorded by third parties (neither the user, nor the operator), as part of a survey, or to report incidents. This is the main focus of the applications presented in chapter four of this report. 4 There are four main technologies used to generate locational data, as discussed in chapter two: • Cell-tower triangulation; the least accurate, but the most widely available; • Global Positioning System (GPS), that use satellites to locate a user; currently the most accurate, but largely restricted in use to smartphones and tablets. • WiFi Positioning Systems (WPS), that use wifi networks to provide additional locational detail. This is potentially even more accurate, but has the smallest geographical range. • Indoor Positioning System (IPS), which marry together data from the three main technologies with additional sensor-derived data, such as air pressure, heat and infrared sensors, to locate individuals within a building or a university campus. This is the most accurate of the technologies, but for the moment is still largely experimental. 5 There are five main survey-type applications that are compared, in chapter four: Poi Mapper, Ushahidi, Dimagi CommTrack, Dimagi CommCare and Taarifa. 8 There are eight major development challenges, or millennium development goals. All of these are addressable to some extent by locational data, as illustrated in Figure 5.1 10 There are mini case studies of locational data in action in development programs covered in the report, as described in chapter four: 6 • Tuberculosis monitoring in Thailand (POImapper) • Oral cancer screening in India (POImapper) • Wildfire monitoring in the Russian Federation (Ushahidi) • Election monitoring in Kenya (Ushahidi) • Supply chain management for community health workers in Malawi (Dimagi CommTrack) • Assistance to home-based care providers in Tanzania (Dimagi CommCare) • School construction in Uganda (Taarifa) • Transport planning in Côte d’Ivoire (Big Data) • Poverty mapping in Côte d’Ivoire (Big Data) • Malaria tracking in Kenya (Big Data) Seven billion There are more than seven billion mobile phone subscriptions around the planet, a number that will soon exceed the human population. Around one billion new smartphones are sold every year, and virtually all of these have GPS and WiFi capabilities, meaning they can make use of all four of the locational data technologies described in chapter two. They have an increasing range of other sensors – for instance for temperature, pressure, fingerprint recognition, camera, compass, gyroscope, pulse measurement, podometer, accelerometer, and so on – that can be used in conjunction with locational data to give a richer range of detail. In the hands of development professionals, these smartphones and tablets provide an unparalleled toolset for bringing scientific measurement to bear on the execution of routine tasks. In the hands of ordinary people, they provide more information, and potential control of that information, than ever before. The era of the “quantified self” is at hand. The locational data capabilities described in this report are part of a wider trend towards quantification and measurement which underlies the emerging science of delivery. Beyond the science of delivery lies predictive analytics, or the ability to predict outcomes, and adapt in a more intelligent way. It is commonplace to say that if you can’t measure something, you can’t understand it, and if you can’t understand it, you can’t control it. Locational data is a small set along the road to measuring, and better understanding, our world. It could lead to an Orwellian future of control, in which nothing happens without being observed. But it can also lead to a utopian future of enhanced understanding, that can release us from the ignorance that leads to poverty. We might not know where we are going, but we can at least try to find out where we are and where we’ve been. 7 1. Introduction More than 70 per cent of the world’s inhabitants now own a small computer, better known as a mobile phone. Those phones typically possess more computer processing power than the Apollo rockets that went to the moon in the late 1960s, yet can easily fit in a pocket. This report is about one specific capability of today’s mobile phones – namely, the ability to locate a phone, and therefore its user, in space and time. For the purposes of this report, this facility is referred to as “locational data” though in practice it relates to a range of capabilities. These include the ability of the user to locate him or herself in geographical space, and to track their progress. The locational data from multiple devices can also be recorded to show, for instance, how individuals move through a city at different times of day and night, sometimes in real-time. 1.1 How locational data can be used In this report, these two specific types of locational data are analyzed: • The active, or intentional, generation of survey data as part of a development intervention. Survey data can be geo-tagged time-stamped, and they can be meshed with other spatial data, allowing, for instance survey data to be shown on a map, with the the time when it was recorded. Five specific examples of active collection of locational data in development project are discussed in the first part of chapter four. • The passive, or ancillary, collection of locational data, for instance through the analysis of mobile call data records (CDRs). Where operators can be persuaded to release anonymized call data for research purposes, this “Big Data” database can then be used, for example, to optimize traffic planning or to study the response to a natural disaster. Three specific examples of the passive collection of location data, and its use for development are illustrated in the second part of chapter four. In general terms, there are three types of location finding used by mobile devices -- using the cellular network of base stations for triangulation, using Global Positioning Systems (GPS), using Wi-Fi positioning (WPS). A fourth broad type of location finding – Indoor Positioning Systems (IPS) – is beginning to emerge, but is not studied in detail in this report. Of these, using the cellular network for "multilateration" is the least accurate but nevertheless has the advantage that it can be used by almost any mobile phone, including basic ones, and can be used to locate a phone independently of the user, as long as it is switched on. It is also the most widely used in Big Data approaches. GPS is more accurate, but usually requires a smartphone, or similar device. It is the most widely used for survey data. The capabilities and WPS and IPS are still to be explored for development purposes, but commercial uses are beginning to emerge, for instance for tracking students within a university campus. There are three generic ways in which locational data from mobile devices can be useful: • Locational data recorded by device users (consumers). The main tool would be maps and other navigational aids that combine location-finding services, such as GPS, with data stored on the phone, such as maps. In popular applications, such as Facebook or Foursquare, the locational data may be combined with social media or mapping data may be overlaid with other data in mash- ups. One example discussed in this report is where users upload reports to a central database, as in the Ushahidi example covered in section 4.2. 8 • Locational data recorded by service providers (operators), or passed on by operators to researchers. The main tool here is call data records and the original motivation for operators was outreach and advertising, for instance to reach people passing a coffee bar with a targeted advert. Location tracking is used by emergency services, and also by law enforcement. With the release of CDRs for research purposes, a number of new services are emerging in this field, for instance in meteorology (measuring the attenuation of mobile signals as a way of measuring humidity) and in "big data" analysis of the transport sector, for instance based on mobile phone traffic. Tracking services are also used in areas like fleet management, congestion charging, and route optimization. • Locational data recorded by third parties (neither the user, nor the operator), as part of a survey. Surveyors may collect non-real time locational data for research and operational purposes. This again allows for mash-ups or repurposing of data from different sources. A good example here would be the use of locational data to verify progress in a vaccination campaign for disease like polio or TB. In such cases, 100 per cent coverage of the target population is essential. Geotagging of photographs of inoculations could be matched with other data (e.g. maps, school records, vaccination cards etc) to build up a visual picture of the progress of a vaccination campaign. This is the main focus of the applications presented in chapter four of this report. 1.2 The evolving locational data toolkit Locational data is a toolset that 2,000 continues to evolve, and developmental applications tend to 1,800 lag behind functional capabilities. 1,600 This is in part because the technical 1,400 capabilities of mobile phones, 1,200 which are explored in chapter two, Featurephone shipments are still evolving. Although 1,000 commercial GPS has been around 800 since the early 2000s, it is only in 600 the last few years that the Smartphone shipments 400 capability has become standard in smartphones and tablets, and even 200 more recently that those - smartphones have become 2009 2010 2011 2012 2013 2014 2015 2016 2017 widespread, overtaking sales of feature phones globally only in the Figure 1.1: Forecast global shipments of feature phones and smartphones, 4th quarter of 2013 (Gartner, 2014: 2009-2017 (millions) see Figure 1.1). Source: Adapted from Business Insider (2013). 9 Because the toolset is still emerging, it is unsurprising that the development community has been relatively slow to absorb these new capabilities of locational data into pilot programs and full scale implementations. There is typically a 5 to 10 year lag in the availability of a particular technical function and its use in applications, as exemplified by the capability to use short message service (SMS). It was included in the technical specifications of mobile phones launched in 1991, but only in Figure 1.2 Mobile user tracking in the wake of the Fukushima Nuclear Accident, March 2011. widespread use ten years Source: Shibasaki, R. (2014). later. In the development community, which is typically not noted for its pioneering use of technology, such delays can be even longer. Nevertheless, as this report demonstrates, locational data holds substantial promise for enhancing the science of delivery, initially within the information and communications technologies for development (ICT4D) community but ultimately in mainstream programs for economic and social development. 1.3 Structure of the report The objective of this report is to examine the potential of locational data for the “science of delivery” in the field of development. The “Science of Delivery” is a term popularized by the World Bank President, Jim Yong Kim, and refers to using evidence-based experimentation to improve development outcomes (Walji, 2013). In this context, locational data is a new tool that is starting to be used in a variety of development fields including health, education, disaster risk management, traffic planning etc. In one illustration, mobile call data records (CDRs) were used to track the evacuation of Japanese citizens from a 30 km zone around the Fukushima Nuclear Power Plant after its failure following the tsunami that hit the coast on 11 March 2011 (see Figure 1.2). These CDRs could then be meshed with health records to optimize the delivery of emergency health treatment. Following this broad introduction to the topic in chapter one, the next chapter explores the technology behind locational data. Chapter three presents the methodology followed in this research and Chapter four, which is the heart of this report, then presents a series of mini case studies of how it is actually being used in a representative sample of different development fields. This is the “evidence-based 10 experimentation” which can be harnessed to improve the “science of delivery”, and examples of both active and passive collection of locational data are presented. Finally, chapter five examines, in broader terms, the longer term potential of locational data as a development tool, once smartphone ownership becomes more widespread. Already, as shown in Figure 1.1 shipments of new smartphones will exceed those of featurephones in 2014, and by 2017, smartphones will account for more than 80 per cent of all new phones sold. During the period from 2009 to 2017, the average selling price of new smartphones will have halved from US$360 to US$180 per unit, while functionality (eg battery life, screen size, memory, operating system etc) will have improved significantly. At some stage during 2014, the number of mobile subscriptions will exceed the number of people on the planet, at just over 7.3billion. With new smartphones arriving on the market at more than one billion a year, it won’t be long before a majority of the installed base of phones in use around the world has GPS, mapping functions and a touchscreen as standard. That becomes a game-changer in terms of their development potential. Thus, while the pilot programs and applications described in this report may appear as experimental in nature, the potential exists for fairly rapid scaling up. But first, their social and economic value for development needs to be proven in action. 11 2. How Do Locational Technologies Work? 2.1 A brief history of locational technologies There are three main location tracking technologies in use worldwide – cell-tower triangulation, global positioning system (GPS) and WiFi Positioning System (WPS), and these are described briefly below. 2.1.1 Cell Tower Triangulation In today’s developing world, not all phones are smartphones (see Figure 1.2); this means that not all mobile phones have the immense technological capacities which many users in the developed economies are growing accustomed to. Such capacities include location-finding technologies, such as GPS, built directly into the handset. Nonetheless, phone companies faced this constraint prior to the advent of GPS- enabled phones and developed other techniques that enable call-tracing. One such method – cell-tower triangulation – was introduced by mobile phone companies aiming to aid emergency response teams in locating callers if communications were lost, as well as for tracking down wrong-doers. Cell-phones operate as two way radios; various towers and base stations are arranged on a grid to form networks. These networks comprise different cells around wireless towers from which radio signals are sent and received to and from cell-phones (see Figure 2.1). This three-way relationship allows phones to communicate with their nearest tower as the base station monitors the signal strength produced by the phone. When moving between cells, base stations in different cells recognize a diminution or a gain in signal strength originating from handset-specific cellular-identification frequencies (cell-IDs). Towers in the cell which the phone is leaving transfer the signal to towers in the cell which the Figure 2.1. Cell-tower triangulation phone is entering (this is called a handover and it is a Source: Park, W. (2008). distinguishing feature of cellular mobile networks). Computers connected to the cell base station can determine locations based on the measurements of signal that the station is recording. These measurements include the angular approach of the signal to the cell towers, the time it takes for the signal to travel between multiple towers, and the strength of the signal when it reaches the tower. Cell tower triangulation typically generates results to an accuracy of 100-200m in digital mobile networks (ie second generation and above) in urban and semi-urban areas. In remote or rural areas, cell phone triangulation is less conclusive as it is often the case that towers are located so far apart that they cannot provide consistent signals, base stations are unable to monitor signal strengths and cells do not overlap as much. Furthermore, physical obstacles such as trees, buildings, mountains, etc. can disrupt signals causing delays or shortages. For instance, cell signal strength in elevators is weak. 12 2.1.2 Global Positioning System (GPS) Given the relatively low level of positional accuracy associated with cell-tower triangulation, mobile phone manufacturers and operators looked to utilize more efficient technologies for enabling location- based services. A particular technological advancement was the incorporation of the mobile phone as a GPS receiver. As with many other functions, such as camera, calculator or web browser, GPS was originally added as a premium feature but is now standard, even in lower-cost feature phones. Although there are a number of GPS platforms in use, or in development, the most popular system is called Navstar; it was developed and introduced by the United States military during the 1960s and early 1970s. It permits anyone using a device with a GPS receiver to pinpoint their location. The GPS receiver uses communication signals (radio-waves) from the 30 or so global positioning satellites to calculate latitude and longitude (see Figure 2.2). The satellites transmit their own positions, times, and pseudo random noise codes (PRN) which receivers use to calculate range. These received signals are converted into position, velocity, and time estimates as the receiver calculates the position of the Figure 2.2. GPS Satellite Positioning Source: Shibasaki, R. (2014). satellite and distance between it and the receiver. Through trilateration (different than triangulation)1 of these signals, the receiver determines its own position. Some receivers permit users to record and save locations (waypoints/points of interest), sequenced locations (mapped routes), as well as tracked directions of the receivers’ movement over time (tracks). With the advent of the handheld GPS device in the late 1980s, new possibilities were uncovered in the sphere of GPS. By the 1990s and early 2000s, commercial carriers such as Garmin and TomTom began to produce GPS devices operating with various degrees of mobility, and without a fixed electric power source. These technologies started gaining traction and have been utilized in numerous fields (e.g. military, urban planning, maritime safety, private navigation, etc.) for collecting locational data. In 1999, a Finnish mobile telecommunications company, Twig Com Ltd., (then called Benefon) launched the first ever mobile phone with GPS integrated capabilities (see Figure 2.3). This started a trend and at the turn of the millennium, various mobile telecom companies began releasing phones with GPS microcontroller chips installed directly into mobile devices, including tablets. Presently, consumers with calling plans and phones equipped with service plans or software that provides navigation can utilize map applications to use the mobile phone for a variety of location-based services. Offline use is also possible if map data is preloaded. What started as simple technology to pinpoint exact locations turned into more sophisticated GPS signal receiving mobiles that could understand various programming languages and offer services like turn-by-turn directions or device tracking. Figure 2.3 1999 Benefon Esc! Source: 13 Webdesignerdepot.com (2009) Initially, the accuracy of the US GPS system was deliberately impaired due to security concerns, but the US government eventually lifted this restriction. Currently, experts perceive that locational data produced by mobile phones and tablets, derived strictly from satellite signals, is accurate to within ten meters or less of actual location. However, GPS-location accuracy can falter due to certain signal-blocking inhibitors. Such inhibitors include dense foliage or buildings – this causes radio signals between GPS satellites and GPS receivers/cell phones to be blocked or distorted. Signals may also be deliberately impaired around sites considered security risks by the US government. This is one reason why other governments are developing their own systems, such as the Galileo system in Europe, Global Navigation Satellite System (Glonass) in Russia), Beidou Navigation Satellite System (BDS) in China, and Quasi Zenith Satellite System (QZSS) in Japan. 2.1.3 WiFi Positioning System (WPS) GPS and cell-tower triangulation technologies have one major flaw: they do not work accurately indoors, or in densely populated areas, due to signal blockage. With the wide-scale proliferation of wireless access points (AP), WiFi Positioning System (WPS) has been used – mostly commercially – to provide location- based services to consumers via Wi-Fi. This may imply that certain limitations will continue to exist in the developing world when it comes to WPS if the commercial potential is considered to be less attractive. WPS was a term coined by Skyhook Wireless - a global location network with a database containing over a billion Wi-Fi APs. Skyhook Wireless utilizes its database of global APs to install new positioning techniques when WiFi equipped devices are connected to the internet. Commercial enterprises such as Google, Apple, and telephone companies (telcos) have compiled their own extensive lists of APs by correlating APs and hotspots with the GPS locations of mobile users. So-called “Hotspots” are public Wi- Fi Aps, but private APs may also be tracked. The two most common WPS approaches to pin- pointing locations are based on signal strength indicators and “radio frequency fingerprinting.” Much like GPS or tower triangulation, the signal strengths between the device and various hotspots are measured; the signal strength indicates the distance from the AP and a geometric calculation against other AP locations is used to locate the device (trilateration). Signal strength positioning is highly dependent on accurate record-keeping of access points. Fingerprinting, on the other hand, Figure 2.4. Smartphone interaction with APs, server, involves the use of previously mapped locations. and database Source: (2014) http://www.mobizen.pe.kr/724. New radio transmissions originating from the mobile each have their own properties including specific frequencies and signal configurations. Therefore, each signal originator has its own specific “fingerprint.” This on-site data is collected and mapped to locations; then it is compared to the locations of previously mapped hotspots, and in turn the transmission can be assigned its own location. The 14 drawback to fingerprinting is that radio frequencies may change quickly, and thus monitoring them effectively is challenging. 2.2 Current Location Tracking Usage in Mobile Network uses 2.2.1 Data from Cell Tower Triangulation In triangulating the signals bouncing off cell towers, as well as recording their time-delays, phone companies are able to pinpoint the locations of cell users to within 100-200 meters of approximate cell-ID handover positions. This is, of course, dependent on the density of towers available in a certain radius (i.e. more towers produce more accurate results). Low levels of accuracy often require that this method be used in conjunction with GPS or WPS. By focusing on call data records (CDRs), tower-triangulation provides the ability to track a cell-phone’s presence over time, which is often critical in solving crimes. It can display roughly where the phone was (in proximity to which tower) when a call was made, or a piece of data (e.g. email, text) was originated or received, as well to which tower the signal was transferred if the phone was moving, and where the cell- phone was when it last received a signal or call. Such CDRs, monitored in coordination with the time stamps from outgoing/incoming calls, allow the network operator to track and plot a phone-user’s mobile path over any period of time (i.e. the route taken plotted by tracking the times when the phone was in use). It is important to note, however, that tracking a cell-phone user’s route does rely a little on logical inference. For instance, not all originated and terminated calls take place in different cells on a map. They may take place in the same cell so all that can be concluded is that specific user’s location on the map at that time. Nonetheless, if time passes and the user’s next call is made from the same cell – for instance, on the next morning – it can be inferred that said user slept at that respective location. “Always on” mobile broadband services provide a higher level of accuracy because the usage of the phone is more regular. That being said, accuracy depends on the density of towers in a given area used to triangulate the location – one tower may display that a signal originated near to that tower, but signal strengths captured by other towers are needed to narrow the area’s range where the cell is located. It also depends on usage and the phone’s battery life. Historically, this method has been used by law enforcement agencies in order to narrow down an area where a suspect might have been at the time that a crime took place, or to track movement of suspects. There are relatively few commercial data-collection applications which depend solely on the cell-tower triangulation as this method typically relies on access to CDRs from major carriers, and this raises privacy concerns. With access to CDRs, millions of pieces of data can be mapped into a program to show movement over time (as displayed from the University of Tokyo Project; see Figure 1.1), but the utility is reduced if the data is anonymised. Software companies generally do not have access to CDRs nor do they have access to a database with respective tower locations. However, some enterprises, like ArcGIS, have analytic tools (ArcGIS 10.2 – Cell Phone Analysis) which allow forensics teams (mostly law enforcement) to analyze cell tower data via CDRs acquired from wireless service providers. In the trend towards “open data” more mobile operators are willing to release generic, anonymized CDRs, but mainly for research, rather than commercial, usage. 15 2.2.2 Collecting GPS embedded Data Recently, due to the evolution of mobile phone/tablet devices, and the widespread adoption of mobile GPS chips, GPS enabled data collection applications have begun to proliferate in the market. These applications (e.g. Ushahidi and POImapper, described in chapter four, and OSMtracker and Cybertracker.org) enable mobile phones, and their users, to become accurate geo-taggers of data with their respective data collection sites.2 They permit users to utilize applications which enable the mobile device to view routes (for instance, when navigating), record/save GPS tracks (for instance, in fitness applications), and store waypoint coordinates (for instance, for locating photographs taken). Mobile applications provide a wide array of data-collection capabilities depending on the phone. In smartphones, and some newer feature phones, these capabilities allow for media files – such as images/raster files and audio – to be collected and geo-tagged with GPS coordinates (latitude/longitude). Almost all phones – including older feature phones – have applications which allow other qualitative data (such as digital questionnaires and forms) to be geo-tagged when populated in the field. Applications (like POImapper from Pajat Solutions Ltd.; see section 4.1) permit forms to be customized in a way that allows users to be very specific in the type of data they are capturing, filtering (i.e. usually through metadata tags), and recording prior to geo-tagging the data. Figure 2.5. Smartphone screenshots from the Taarifa tool. Image (1) shows a page where field-users can select which data-collection form they want from their list of previously customized forms. Image (2) shows a map where GPS location has been pin-pointed on a form, as well a tool to upload media captured at the data collection site. Source: Brar, S. and Relhan, G. (2014). Typically, data collected in the field is updated and edited offline prior to being uploaded into central databases with which the collection tools interact directly. These databases, or content management systems (CMS), can be accessed from the “cloud” (i.e. using remote databases accessible from the internet), on the local premises at a data center, and in some cases, offline, where the device has sufficient memory storage to create a data cache. Some CMS, like GeoNode.org (utilized by a number of World 16 Bank projects) and MapBox’s TileMill, are open source and web-based, permitting a variety of devices to access the same content from a web browser. GeoNode empowers developers to integrate its geo- capabilities into existing platforms and applications – a handy feature for customization of any tool which wants to develop a Geographic Information System (GIS). Lastly, GeoNode allows developers to deploy spatial data infrastructures (SDI) which empower multiple users and tools to interact with stored spatial data in a multitude of ways. Others, like POImapper (proprietary) and CloudGIS’s Mobile Data Collection Tool, allow for existing data, previously collected and stored in private databases, to be edited offline. Such GPS-embedded data can be downloaded, edited, and then uploaded back to the server (Moss, 2012). In some mobile applications like Taarifa or POImapper, real-time geo-tagged data collection is possible. Real-time data collection can make an impact in various sectors. For example, it can help monitor infrastructure installation/maintenance or assist in resource allocation decisions for public utilities like water.3 2.2.3 Analyzing GPS embedded data As researchers and development specialists have discovered, locational data can provide immense insights into the movement patterns of particular communities (e.g. commuters, water seekers) as well as ongoing infrastructure projects, etc. GPS-generated data is basically useless on its own; however, certain computing programs, such as GIS, provide tools that help specialists, through digitized maps, manage and visualize location-enriched data collected in the field. These maps can be manipulated to present the data in comprehensive and telling visualizations (e.g. time-lapse depictions of municipal transport projects). Geo-data, when stored on a GPS-enabled device, can be transferred via a direct connection (e.g. USB, WiFi, Bluetooth or 3G data connection) to a data server or computer with GIS software (assuming, of course, that a direct connection is available). Not all GIS are “smart” enough to interact with GPS data in its raw form. Sometimes, data must be converted (typically through an online conversion tool) into GIS readable languages and Figure 2.6. Vector data points displayed in a formats (e.g. .gpx4, CSV, or shapefile - a vector data GIS map of Arkansas, USA to create a format). Vector data is the geometric data format (points, shapefile, plotting the trajectory of a lines, and polygons) traced onto a map.5 Vector layering tornado’s path over time. Source: 2014, ArcGIS, directly on a map, in conjunction with temporal http://storymaps.arcgis.com/en/gallery/#s=0& information, can, for example, enable time-series n=30&d=1 evaluations, as mentioned above.6 Moreover, vector data can be collected through mobile technology and later uploaded into applications other than GIS such as web-based interactive community maps (e.g. OpenStreetMap). The crowdsourced map can then be used by data collection and visualization tools to provide more accurate templates and reference points when mapping waypoints, routes, or tracks offline. In general, most of the mobile data collection tools discussed do not have customizable maps built into their programming as many feature phones cannot replicate or readily update (due to low bandwidth, 17 processing power, and small screen size) the visual displays of maps on their screens. Instead, they house generic maps on their data-center servers (usually of the region/country in which the application is being utilized in) attained from third parties such as OpenStreetMap, MapQuest, or Google Maps. Newer smartphones can access such maps offline in order to add geo-specific data (Lounamaa, 2012). Certain applications (e.g. GIS Cloud Mobile) allow for waypoints, routes, or areas to be traced through a field office’s desktop server’s map portal; newer mobile devices can sync with the desktop server and be used to gather more information in the field (Holmes, 2014). Older, “basic” phones simply house automatically georeferenced data until it is uploaded to the tool’s congruent in-office platform to be mapped. Some online geospatial CMS (like GeoNode.org and TileMill) are not strictly classified as a GIS, but still have mapping capabilities, as well as other features that create a central hub for managing and visualizing data after it has been uploaded. Whereas GeoNode does not permit map edits directly on a mobile device, those carried out on a computer are saved to the cloud and the populated maps may be published to suitably-equipped mobile devices for field-use during other data mapping exercises. Alternatively, this data can still be extracted directly from a mobile application in the form of a CSV file, a shapefile, or another format like KML7 and imported into a compatible GIS for more advanced mapping and statistical analysis. While certain online technologies and some geospatial CMSs are appropriate visualization tools for basic mapping, GISs like QGIS (open source) and Esri’s ArcGIS (proprietary) provide much more powerful analytic tools. ArcGIS has recently released a cloud version of its software – ArcGIS Online. ArcGIS Online, in conjunction with the ArcGIS mobile and tablet apps (for Android, iOS devices, and Windows) make it accessible and fully functional from any of the above mentioned mobiles or tablets with a data connection (see Figure 2.7. ArcGIS online and mapping online. Source: Esri (2014) “ArcGIS Online Features” Figure 2.7). Furthermore the ArcGIS Runtime SDK (software development kit) enables mobile developers to build mapping applications (for iOS, Android, and Windows) which utilize the mapping, geocoding, and other functions of ArcGIS’s online and desktop tools; this includes a handful of geoprocessing tasks and real-time updating which can be used before, during, and after data has been collected. This gives subscribed users universal access to the locational data and content regardless of where and with what the data was originally collected and uploaded. These GIS utilize data sources called libraries (e.g. the Geospatial Data Abstraction Library - GDAL/OGR) to read and write vector data formats; such libraries can be added onto proprietary technologies via plug-ins in order to provide interoperability for the data files between the GIS and data collection application in the various formats discussed as well as others (e.g. .gpx, KML, PostGIS, GeoTIFF). 18 2.2.4 WiFi Positioning System and Location-based services Currently, but mostly in more developed nations, WPS enables application developers and major companies the chance to offer Location Based Services (LBS) on mobile devices (i.e. phones and tablets) when connected to the internet. This includes accurate positioning and LBS even when indoors, and indeed this generic category is often call “indoor positioning systems” (IPS), though more correctly IPS uses a variety of different sensor techniques, as shown below. As mentioned above, poor Wi-Fi infrastructure hinders the accuracy and ability of WPS altogether. Additionally, even though WPS empowers LBS in devices without GPS capabilities, LBS usually works better when WPS is complemented by GPS. Commercial uses of WPS are typically employed in the areas of marketing (e.g. adverts received when passing a particular store). WPS also enhances GPS in order to strengthen functionality and navigation. From a marketing standpoint, when a smartphone, with its Wi-Fi receiver turned on, is within a certain range of hotspots or APs, some applications (e.g. Foursquare) utilize this data in conjunction with data assigned to geo-referenced physical locations pulled from the application’s server database. The application processes this data relationship in order to push (i.e. serve) location-specific advertisements to the smartphone user for various businesses which are within a certain range of that user’s vicinity. WPS also enables devices to “check-in” to georeferenced locations (this simply means that the physical location has a mapped presence online derived from its AP addresses). Voluntary check-ins are also accessed by some applications to index a user’s preferences in order to serve advertisements of interest in the future. Moreover, voluntary check-ins allow users to geotag social media (e.g. photographs on photo- sharing platform Instagram). Such features are similar to the real-time data updates and data-geotagging capabilities which GPS-enabled data collection tools described above offer. At present, not many data-collection tools similar to those which generate GPS embedded data exist solely for use with WPS. Nonetheless, many developers are exploring this space and potential capabilities as hotspots become more abundant. Some current initiatives include indoor mapping applications which employ the mobile GIS technology discussed. One such application is the Smart UJI Campus; this application is a map-based web service which enables staff, students, and visitors at the Universitat Jaume-I to locate points of interest, review and search geo-tagged information (including the whereabouts of various campus services or employees), as well as other resource management features all empowered by WPS (Vicent et al 2013). The application developer used a mobile device, equipped with the ArcGIS Figure 2.8. Images from campus and SDK and a set of basemaps (two visual maps of the geo- interior map layouts from the Smart UJI spatial information provided by a central directory Campus application. containing campus data), to register different access points Source: Vicent, Joan Pere Avariento (2013) from campus infrastructure as well fingerprints from indoor locations. In essence, the application provides a map of the whole area (i.e. campus) as well as a map for the interiors of buildings – including floor-by- floor schematics. All of the points of interest are stored in a geodatabase which also contains relevant 19 information; moreover, users can use Wi-Fi fingerprinting in order to register new points indoors. A tool such as this can work to enhance the information-communication capabilities of smart buildings as well as to provide navigational functions for large indoor structures. 2.3: Potential for the future 2.3.1 Smartphone developments In developed countries, GPS-enabled smartphones have already reached the mass market. That is, they are typically affordable to the everyday user. General affordability, however, remains a major concern in the developing world. It is perceived that as the costs of smartphone ownership and mobile connectivity go down, user uptake in marginal communities will increase. This was the case with the feature phones which were a luxury in developing countries a decade ago but are now much more affordable. This, of course, is also true for the various proprietary data collection applications, mapping, and analysis tools discussed thus far. So the “trickle down” effect of rapidly falling device costs and mobile broadband prices will democratize the services described here in developing countries in the near future, following a similar model to that followed in developed economies. In parallel, as stakeholders begin to understand the power of harnessing geospatial data, current open- source technologies like OpenDataKit (for data collection) or GeoNode.org and QGIS (for data storage and analysis) may encourage a trend towards open-source and more functional geo-enabled tools (e.g. ArcGIS’s catalogue of tools). Furthermore, there are initiatives which endeavor to promote the use of such tools. One group, the Open Geospatial Consortium (OGC), has developed standards and specifications that make geospatial data interoperable between different OGC compliant interfaces, data stores, and GIS platforms. Standards range from open web mapping services to compatible markup languages (e.g. GML or SQL); however, compliance with few or some of the standards does not ensure compliance with all the standards. Until then, data-based initiatives associated with tower triangulation and GPS-enabled feature phones will be the widely prescribed norm in developing countries. WPS also faces cost-related constraints for the future. As Wi-Fi infrastructures improve in the developing world (i.e. through investments into larger-scale connectivity and an influx of public access points), and as Wi-Fi enabled phones become more common in marginal communities – the likeliness of using WPS to collect and monitor locational data will increase. Instead of relying on field work and georeferencing to tag/capture data from physical locations, these locations will have with their own hotspots and the data mining and analysis procedure will be much easier. In short, we’re moving from a world in which data is actively and consciously collected to a world in which data is passively and unconsciously recorded. That being said, when using smartphones to record tracks or calculate user stay-points, locational accuracy may not be perfect. Data scientists recommend cross-checking the accuracy of tracks, routes, and waypoints collected via mobile technology with other references containing similar data. This could include data reinforcement techniques through tower triangulation, Radio Frequency Identification (RFID) tags built into public transportation infrastructures, or online maps; when dealing with specific data, even images produced by traffic cameras, planes, drones, or satellites (this has been least prescribed considering such images may lack accuracy due to weather) can enhance data analysis techniques. Recently8, the US government has lifted restrictions regarding the accuracy of satellite images; earlier, 20 companies were restricted from using satellite images where visible features were smaller than 50cm. The lifted ban enables the use of more accurate imaging and can allow certain companies to make the highest quality images available to their consumers, data analysts and development specialists included. One proprietary application – Navizon – successfully empowers indoor and outdoor LBS through accurate geographic positioning by utilizing a crowdsourced global database of the geographic locations of Wi-Fi access points and cell-towers, collected by registered users. This can then be triangulated against the GPS location of a phone. Navizon has enhanced healthcare by offering its services on a variety of smartphone platforms and enabling healthcare professionals to deliver better patient care, maintain patient safety (regardless of location), manage assets and resources, as well as optimize workflows. It is unique in the way it utilizes the interaction of all three mentioned location-finding technologies in order to provide accurate positioning on campuses and inside buildings, down to specific rooms. 2.3.2 GPS developments Navizon, and other tools like it, are reliant upon the established (and functioning) infrastructural components that enable tower triangulation and WiFi. That is, it requires more than simple GPS to provide its users full functionality. The American Military’s GPS system – Navstar – has been the sole global, space-based satellite navigation system for decades. This has meant that location based services and location data collection has relied extensively on that satellite system; this can create bottlenecks, for instance when it comes to real-time data collection, in areas where connection is spotty and signal blockage occurs. However, other nations have begun constructing their own global navigation satellite systems. One such nation is Russia which has successfully deployed the “Global Navigation Satellite System” (GLONASS); the 22-satellite system endeavors to complement the traditional GPS system. In 2012, carriers began integrating GLONASS radios into smartphones. This technological inclusion has meant that newer smartphones (e.g. iPhones released after the 4S generation, Blackberry Z10/Q10, and a handful of Samsung Galaxy apparatuses to name a few) are able to be located and capture geospatial data at much more accurate rates – location wise. Qualcomm estimates that the interaction between GPS and GLONASS improves accuracy by 50 percent in “deep urban environments.” As these satellites interact, they are able to trilaterate more accurately as more satellite signals are picked up by the handset. This is expected to continue to improve as more nations develop their own navigation satellite systems; China is working on the Beidou-2 system which is anticipated to be serving customers by 2020 and the E.U. plans to operationalize its satellite system – Galileo – by 2019. 2.3.3 Indoor Positioning Systems Nonetheless, the satellite systems will still falter when it comes to creating accurate location fixes indoors. A handful of indoor-positioning systems (IPS) have been developed but they mostly rely on WPS or some sort of, less accurate, hybrid GPS method (phone captures GPS coordinates then uses sensor data from its digital compass, pedometer, barometer and accelerometer to determine indoor locations). One such initiative was developed by Broadcom which introduced a smartphone chip in 2012 that supports GPS, WiFi, and Bluetooth. The chip makes note of the phone’s entry point (via GPS) then counts steps, directions, and altitude by relying on sensor data produced by the accelerometer, gyroscope, and altimeter, respectively. Given the modest rates of smartphone proliferation in developing countries, adoption of these technologies is lagging relative to some of the commercial, more social, uses (e.g. applications which provide parents real-time monitoring of their children’s indoor-locations, or those 21 which can identify, for users, the number/location of other users, in the same application, with similar interests, or even automate pings and phone calls based on locational positioning) that have spurted in higher-income economies. The potential functionality is limited only by the imagination of the developers. On the other hand, the widespread penetration of such functionality is limited by security and privacy concerns when it comes to data, as well as – again – general connectivity and affordability. Other companies, like Nokia, are experimenting with the use of Bluetooth beacons and sensors to enable off-line, low-energy location based services. With the correct application (e.g. INSITEO for Android or indoo.rs for iOS and Android), the newest iteration of Bluetooth – Bluetooth 4.0 or Bluetooth LE – enables users who have set up, tagged (fingerprinted), and registered a handful of customized Bluetooth beacons to send and receive signals between the handset and the beacons in order to provide real-time localized information and navigation indoors. For developing nations, this is more affordable than establishing entirely new WiFi infrastructures at the macro level, however individual users may find that this is pricey as they will be required to order more beacons due to the fact that accuracy and functionality are greatly heightened when more beacons are being utilized. Some solutions are revolutionary in themselves as they require no extra WiFi or Bluetooth infrastructure whatsoever. One such solution was introduced by IndoorAtlas, a Finnish company. The method relies on the Earth’s innate magnetic field to ascertain a handset’s position. The Earth’s entire surface emits a magnetic pulse, and with a map of these magnetic fields, users can accurately navigate the outdoors and the indoors. IndoorAtlas has developed a smartphone application which relies on the sensitivity of the smartphone’s magnetometer to create magnetic field maps that are accurate up to 10 centimeters. This application populates the map as users walk around picking up new magnetic signals; eventually the application will be able to provide extensive indoor location services where magnetic maps have previously been established. Another company, ByteLight, sells a new form of technology that powers location services through LED lighting. Essentially, ByteLight has created an IPS based on visible light communication; the technology is entirely reliant on LED lighting systems which have ByteLight compatibility previously built in. This is good news for developing nations as most of the buildings in the world have not yet switched to LED lighting and the cost of adding ByteLight compatibility is negligible. On the other hand, it is bad news for any present users of LED lighting systems as ByteLight cannot be built into LED structures after the bulbs have left the factory. When LED lights have ByteLight’s middleware built in, the technology powers the chips in bulbs and handles the applications that work, through smartphone cameras, to identify the unique ID of each lamp. It does not require any beacons or networking equipment. ByteLight controls the pulses of LEDs to generate certain patterns; these patterns are picked up by the camera in smartphones or tablets and, by using the data from the LED transmission, a device can work with an app to perform client-side (as opposed to server-end) calculations and locate the device’s whereabouts indoors. IPS will become very important in the coming years as indoor LBS (e.g. in hospitals or schools) becomes essential for development outcomes. The ability to deliver automated services in the public sphere based on a beneficiary’s indoor location will be paramount in creating stronger feedback loops and engaging citizens in policymaking decisions. Consider the possibility of serving valuable instructions or digital documentation, pertaining to business registration, directly to a citizen’s mobile device via an application or web-service that recognizes when the entrepreneur has entered the building housing the local business 22 registrar’s office. Smart Cities can be made even smarter by incorporating IPS technologies in order to create indoor maps of major commercial, municipal, and government centers. 2.3.4 Other Technologies As costs are driven down and infrastructures improve, a handful of newer technologies will become available in the developing world. For instance, Broadcom has introduced a battery-conserving GPS chip for smartwatches. Such watches will initially permit wearers to collect geo-tagged fitness and health data. In the future, the watches may be used to monitor continuous tracks of locational data and as complementary devices to validate GPS data generated by mobiles – not to mention the convenience of a wearable, as opposed to a handheld, device. Another (relatively expensive) wearable device that has yet to widely penetrate consumer markets is enhanced spectacles, such as Google Glass. The hands-free smart eyewear is classified as a wearable computer with a head-mounted, transparent optical display that reflects images for the user and displays information in smart-phone like format. Through voice commands, users can take pictures and videos, make calls, send messages and emails, request navigation, as well as other features typically associated with hand-held mobiles. Furthermore, like smartphones, applications for collecting geospatial data and geotagged media files exist. BAE Systems recently introduced the GXP Xplorer Snap app which “enables Google Glass users to quickly snap photos and record a report title and brief description using only their voice. The report is automatically geotagged, time-stamped, and uploaded to a GXP Xplorer server, where it is immediately shared and accessible to the rest of the enterprise. The hands-free benefit makes it an ideal technology for use in reconnaissance missions and disaster relief operations” (Ratzer, 2014). Google has been one of many driving forces behind geospatial related initiatives – from its highly adaptable smartphone platforms (Android), Google Glass, to its popular, adaptable mapping software (Google Maps). Perhaps Google’s most intuitive contribution however is the advent of data-collecting vehicles. These cars are equipped with a handful of technologies and sensors which enable 3D locational data to be collected, stored, and utilized to improve Google Maps. The data includes images which endow street-views of georeferenced locations. Google is also working on another type of car – a driverless one. These cars use sensors and a handful of optical devices as well as built in navigation systems to safely pilot through streets; they have been tested in the following U.S. states: California, Nevada, Michigan, and Florida. The prospect of a driverless car, equipped with the data collection tools from the Google Maps cars, presents many opportunities for quickly collecting locational data in the field in hard to reach areas. This is especially true in fragile or conflict ridden states where fieldworkers’ lives may be in danger. Furthermore, like Google Glass’s ability to pull data from paired mobile devices or laptops, innovations in driverless cars could theoretically enable real-time data collection and streaming to paired devices. This could further improve monitoring processes in infrastructure projects, identify structural and transport problem areas in marginal communities, and empower fieldworkers in a handful of industries to remotely observe and track achievements or setbacks for a given project. At the consumer level, Google is apparently poised to introduce a 3D imaging tablet that puts its street view technology into the hands of users9. As might be expected, applications for these new devices lag behind the availability of the technology, but the potential applications are limited only by imagination. 23 The power of the technologies and the data they capture, described in this chapter, will be discussed below in a handful of specific cases, mainly focused on aid programs. It should be noted that the applications currently used in the development community lag someway behind the potential described here, but there is a catching up process underway. This chapter has provided a brief technical overview of how the technologies interact with one another and with the data they generate. It is not meant to be a guidebook for developers or designers who have chosen to incorporate geo-enabled functions into their existing tools and applications. For that purpose, it is recommended that mobile and web developers turn to development kit guides and tutorials provided directly by the originator of the desired platforms, tools, and applications. Endnotes for Chapter Two 1 Trilateration incorporates distances into its mathematical process while triangulation is strictly a gemotetric calculation based on angles 2 See, for instance, ProFor (2014) Information and communication technology for Forest Law Enforcement and Governance; Lessons from a Two-Country project in LAO PDR and Moldova, page 14, 50 3 Pajat Solutions Ltd. (2014) NetHope.org PoiMapper Case Study, http://solutionscenter.nethope.org/case_studies/view/poimapper-real-time-field-data# 4 Wikipedia (2014) “GPS Exchange Format”, http://en.wikipedia.org/wiki/GPS_eXchange_Format 5 Wikipedia (2014) “Shapefile”, http://en.wikipedia.org/wiki/Shapefile 6 QGIS.org (2014) “Vector Data”, https://www.qgis.org/en/docs/gentle_gis_introduction/vector_data.html 7 Wikipedia (2014) “Keyhole Markup Language”, http://en.wikipedia.org/wiki/Keyhole_Markup_Language 8 See BBC (16 June 2014) http://www.bbc.com/news/technology-27868703. 9 See BBC (24 May 2014) http://www.bbc.co.uk/news/technology-27538491. 24 3. A Conceptual Framework for studying Locational Data Applications While the “science of delivery” can be improved using a variety of innovations, technologies and techniques, the focus of this study is on how it can be enhanced by using applications and other software- based solutions that use locational data on mobile phones. The focus is on general purpose mobile phones rather than specialized devices (such as GPS trackers) and on consumer-market which are affordable to the broad mass of users. This question is answered by presenting applications and real-life cases where the applications have been used. The applications presented are related to mobile phones and their abilities to track the phone and its user, using at least one of the technical possibilities presented in the previous chapters. The presentation of applications, and cases where they have been used, is carried out in sections 4.1-4.4. These are then presented and compared with each other in section 4.5. The tables presented in section 4.5 are intended to provide a diagnostic tool to assist potential users in identifying which of the applications reviewed is best to use for different objectives and in different aid delivery programs. In addition to applications and cases presented, location-based “Big Data” can also be used, for example, in development-related research and policy making. In section 4.6, some cases where Big Data has been used successfully in development interventions are presented. When reviewing applications, cases and big data solutions, three frameworks are used to provide a conceptual framework. First, learnings gained from the implementation of Ushahidi following the Haiti earthquake Project (Morrow et al, 2011) are used. Second, when presenting the applications and the case studies, and when making qualitative comparisons, an ISO standard called SQuaRE (Software product Quality Requirements and Evaluation) is used as a loose framework. Third, implementation challenges are briefly discussed. Since these three frameworks consist of a large number of factors, umbrella syntheis framework, consisting of two simple research questions, is built from them and presented in section 3.5. Because of the nature of this research, the frameworks are applied in an agile manner. They provide a backdrop for presenting and benchmarking applications and case studies, but are followed in an interpretative qualitative manner. 3.1 Methodological frameworks 3.1.1 Ushahidi Haiti Project Review Framework In their review paper “Independent Evaluation of the Ushahidi Haiti Project” (2011), Nathan Morrow at al present and evaluate a development aid project conducted between 2010 and 2011 in Haiti. The Ushahidi Haiti Project (UHP) was a large-scale volunteer-driven effort to produce an application-based crisis map after the devastating January 2010 earthquake in Haiti. While the review paper presents and reviews the UHP case in general, and while it presents the use of Ushahidi platform in Haiti case in detail (Ushahidi is one of the cases presented in chapter four), the paper also provides a framework, or a set of research questions, for evaluating the impact of a certain application. This framework is also used in this research, for example as a part of the umbrella framework presented later, and while summarizing the whole research in the final chapter.The UHP framework consists of four factors, or sets of categorized questions. They are: 25 • Relevance: to what extent does the application address the needs of beneficiaries? • Effectiveness: To what extent did responders actually make decisions based on the information the application provided? Why was the information used or not used? • Efficiency: How did the application add value to a project? • Sustainability: What has the application created, for example, has it created new groups, has it been institutionalized, or has it stimulated commitment from donors or other actors? 3.1.2 Systems and software Quality Requirements and Evaluation (SQuaRE) In addition to using Ushahidi Haiti Project review’s framework, the ISO standard SQuaRE (ISO/IEC 25010:2011) is used loosely as a tool for describing both applications and the cases. SQuaRE is an international ISO standard. It is used to describe “the relationship between a quality model, its associated quality characteristics and software product attributes with the corresponding software quality measures, measurement functions, quality measure elements, and measure methods.” (Lepmets et al., 2011.) In SQuaRE, the attributes the standard describes are: • Functional suitability, • reliability, • usability, • performance efficiency, • maintainability, • portability, • security and • compatibility. In a thorough SQuaRE analysis, the quality of a certain software application can be evaluated by measuring its internal attributes (typically by evaluating the application’s static measures), its external attributes (by evaluating the behavior of an applications code) or by the quality of the application’s attributes (by an examination of the use of the application). (Lepmets et al., 2011.) In this research, we will focus on the third: the qualitative attributes of each application. 3.1.3 The Challenges of Implementing New Technology in Developing Countries When applications are implemented in new environments, such as developing countries, they often face a common set of problems. The general challenges of implementing new technology are well-researched and documented. For example, according to Annie Evans (2013), implementation manager for CompassLearning, there are ten general challenges that an implementable technology will face: 1. Avoiding Using Technology for Technology’s Sake: is the application really useful? 2. Creating a Vision: is the vision behind the application really good? 3. Money: where does the funding come from? 4. Professional development: is the development process of an application done properly? 5. Getting everyone onboard: how to win over naysayers with the technology? 6. Scheduling of use: where and when is application used? 7. Good systems and procedures: making functional systems, and taking account of the challenges in environment, such as poor connectivity 26 8. Unlocking motivation: how to keep users motivated? 9. Data and progress monitoring: follow-up and, when necessary, update goals. 10. Maintain enthusiasm: keeping users and developers enthused over the potential outcomes. In addition to general challenges in software implementation, there are some additional challenges in implementing applications in the so-called developing world. For example, Edward F. Hsieh argues in his paper “Investigating Successful Implementation of Technologies in Developing Nations” (2005) that certain common factors can be identified which are important to sustainable technology implementation in developing nations. According to Hsieh (2005, 16), the most important factor is the ability of the community to maintain the product. The next most important factors were the local need for the technology, and the ability to produce the technology locally. In Hsieh’s paper, the other factors which need consideration while implementing technologies to developing countries, are the issues of participation (including the participation of international NGOs or local government) and the question of finance. Hsieh argues (2005, 17), that the financial support given by an NGO or some other partner can work to either help or hinder the success of the project: “The problem can arise is that the community becomes dependent on the subsidy, and abandons the project when the NGO program ends.” While implementing technologies in developing nations, a broader context of successful development aid projects should be taken into consideration. While opinions vary on which methods of development aid are the most beneficial, Kosack (2002) argues that those which seek to improve the quality of people’s aid are often the most effective. He links successful development aid projects to efforts to encourage democratization. This is something that arguably can be advanced using the survey-based applications described in the first part of chapter four, but would not necessarily apply to Big Data applications. For the future of the applications and technologies presented, it is vital for companies, volunteers and activists working with the technologies to identify the challenges of implementation, and to understand the broad context of development aid. Since the technologies presented here have not been used very long, it is also evident that their challenges have not been researched thoroughly. Thus, it remains to be seen, which applications can tackle the challenges they face, and follow-up research on the challenges the technologies face is needed. 3.3 Proposed Framework for the Presentation of Cases The following chapter presents several applications, use cases and big data solutions, which use at least some of the technological possibilities presented in chapter 2, and which might be expected to enhance the “science of delivery. A synthesis between the two frameworks briefly described earlier – UHP review, SQuaRE and the implementation challenges – is applied, using qualitative interpretation. In this synthesis framework, a set of research questions are posed: 1. What are the key features of the applications? This question derives especially from SQuaRE framework, but also from the learnings of implementation challenges and UHP review. 27 2. What are the key learnings and outcomes of cases, where applications have been used? This question derives from the UHP review, but also from the learnings of implementation challenges. This question is answered after each case presentation. All the applications and cases are presented in a similar structure: beginning with an of the application and the vendor and following with a series of use cases and a brief summary of outcomes achieved. Following the presentation of survey data applications and cases in sections 4.1-4.4, and a comparison amongst them in section 4.5, section 4.6 looks at the possibilities of Big Data for enhancing the science of delivery. While Big Data solutions are not directly comparable to the survey data applications, the same synthesis framework questions are posed. The concluding chapter returns to the initial frameworks of UHP, SQuaRE and to implementation challenges, and discusses what has been learned from the analysis of the cases. There are also a further series of questions posed by the research, in particular concerning security and privacy issues, that go beyond the current study but would be suitable for possible follow-up research. 28 4. Case studies of locational data This chapter presents a series of mini case studies of mobile applications that use location-based data for enhancing the delivery of development assistance. It also presents real-world cases where the applications have been used, and analyses the applications presented in a series of comparative tables. The applications are reviewed based on the synthesis framework provided in chapter three. The survey data applications presented are PoiMapper, Ushahidi, CommCare, CommTrack and Taarifa. While there are also other applications that use locational data for development aid purposes, some pre- selection had to be made, for a number of reasons. These five applications were selected, based on the following criteria: • They provide a representative view on the use possibilities of different mobile-based technologies described in the chapter two. • They have both similarities and some differences with each other, and can be relatively easily compared. • The have already proven successful in documented real-life cases, and show good potential for replication and scaling up. • They are among the better known, and well-documented, applications. In addition to the aforementioned survey data applications, there are also cases where Big Data approaches have been used to enhance the science of delivery. While Big Data and its use cases are not directly comparable to the survey data applications presented, they nevertheless provide an interesting possibility to develop research and policies to improve societies. A number of sample cases are explored in section 4.6. 4.1. PoiMapper 4.1.1. Overview PoiMapper is an application for collecting location-based point-of-interest (POI) data, developed by Pajat Solutions Inc, a Finnish software company, in co-operation with Plan International. PoiMapper is a tool for “collecting and utilizing point-of-interest (POI) data with cost-effective mobile technologies”. PoiMapper is a multiplatform application, which is used to collect, view and share data gathered about POIs. PoiMapper is typically used by field workers, experts or volunteers of development aid organizations, but can also be used for other data collection and analysis cases. The targeted users are organizations carrying out point-of-interest related development aid projects, their field workers and management-level employees. According to Pajat, the primary developmental emphasis of PoiMapper has been on its usability. Figure 4.1. PoiMapper in use on different devices The application is designed to be simple to use, even for people with very little experience with Source: poimapper.com mobile phones, and it can be used by a wide variety 29 of mobile phones, ranging from latest smartphones to cheaper feature phones, as long as they are equipped with a camera and a GPS. The secondary developmental emphasis of PoiMapper is that Pajat - in their own words the application has been built for use in “rough” conditions. It can store data in offline mode, and can be efficiently used “Poimapper is developed by Pajat Solutions without a reliable mobile network. Ltd, a mobile software solutions company with personnel in Finland and development partners The key capabilities of PoiMapper allow field workers, for in Kenya and in India. Pajat key personnel have example, to: a long track record of developing new • Define and maintain points of interest, routes and technologies and businesses both in small areas, from which data (text, numbers, single/multiple companies and in large multinationals. We choice alternatives, conditional questions, pictures) is believe innovation will be increasingly driven collected; by major opportunities in Asian, African and • Collect data using a multitude of platforms, along with Latin American growth markets. Our mission is other people using PoiMapper via different mobile to provide affordable, useful and meaningful platforms; mobile solutions that are globally relevant. We • Edit and add to the collected data; also provide consulting services, including • Upload (or download) collected data to/from a ICT4D.” database, using either cellular network or a computer with an internet connection; and • Research, display, maintain and visualize information. The data gathered using PoiMapper can be accessed, administered, edited and visualized using a web application, and accessed via a computer with a service-compatible browser. The main focus of visualization is in maps: data is presented via a map interface – using both Google Maps and OpenSteetMaps. Technologically, PoiMapper is offered as three versions: as a J2ME implementation (introduced in 2010), as an Android application (2011) and as a HTML5 application (spring 2014). It uses a PostgreSQL database, stored in a large commercial cloud service. 4.1.2. PoiMapper Use Cases Tuberculosis Monitoring (with Plan Thailand) Tuberculosis Monitoring is an on-going project where PoiMapper is used in a tuberculosis (TB) monitoring project carried out in the Chiang Rai province of Thailand. The project is coordinated by Plan Thailand with the objective “to strengthen migrants and ethnic minorities to be NGO-organized Community-Based Volunteers (CBV) /Migrant Health Volunteers (MHV) to do Directly Observed Treatment (DOT) provision and/or treatment for TB care, including TB suspect referral.” 30 The Directly Observed Treatment watcher will support patients’ daily tuberculosis treatment, under supervision of community-based volunteers and/or migrant health volunteers (CBV’s and MHV’s) – and Plan Thailand’s staff. PoiMapper is used in this process to collect information relating to tuberculosis patients when they are visited by staff. In 2011, there were approximately 20 PoiMapper users within this project. Key Outcomes and Results Following key outcomes in using PoiMapper in tuberculosis monitoring project have been identified: Figure 4.2. Using PoiMapper in tuberculosis monitoring Source: Pajat and Biocon (2013). • PoiMapper has helped to update the project’s process more quickly. Project staff can update information on patients immediately. • The information collected by using PoiMapper has helped to estimate the prevalence of tuberculosis in a given area. If the area has high prevalence, it can be targeted for appropriate follow-up by researchers and policy makers.. • Tuberculosis patients received follow-up to their status continuously: patients were monitored and visited more systematically than before. • It was possible to change staff during the project, since all relevant data – including the locational data of residence of a patient – were stored and easily accessed. A new MHV could easily pick up and follow-up their colleagues’ work. • Staff could see the history information of a certain area, and could see a whole picture of tuberculosis situation in a certain region. • The design of information collected using PoiMapper facilitates reporting by Plan Thailand to the donor. • PoiMapper “is estimated to save 3 days/month work time for health volunteers and 2 days/month for field officers” within Plan Thailand’s project. Plan Thailand’s tuberculosis programme manager said that monitoring visits and proper treatment programs would not have happened at such a scale if PoiMapper had not been used. Early Intervention Oral Cancer Screening Program (with Biocon Foundation, India) PoiMapper was used in India by the Biocon Foundation, by their field workers and communal health workers, to improve oral healthcare in rural areas. According to Biocon Foundation (Pajat and Biocon, 2013), oral cancer represents around 40 per cent of all cancer cases in India (compared to just 4 per cent in UK). Oral cancer is a major problem in rural areas of India, and thus it is a large-scale problem, since rural India contains over over two third’s of India’s population, with half living below the poverty line. Chewing betel, paan and areca (and other tobacco- 31 related health risks) are known risk factors in developing oral cancer. Early intervention to tobacco- related habits and early diagnosis of oral cancer are critical issues when reducing the number of oral cancer cases in India. According to Biocon Foundation, “community health workers provide primary healthcare in rural areas, but lack skills to perform oral cancer diagnoses.” To address this problem, Biocon equipped community workers with mobile handsets using PoiMapper. Field workers took photos and videos of the mouths of people at risk oforal cancer, and stored the images and locations of patients to a database. Doctors studied the images were able to diagnose the patients remotely, using web applications. Data collected by PoiMapper was implemented to OpenMRS, an open source medical health record data system. Key Outcomes and Results The early intervention project was executed in two sites in Karnataka, India: Mangalgudda village and Anakanur village. A total of 583 people were screened. Of those, 66 had lesions, and 56 needed biopsy. A typical person with lesions was a male of 40+ years. According to Biocon Foundation, early intervention screening can help people to change their tobacco related habits, and thus reduce the possibility of contracting oral cancer. The information collected also: Ushahidi In Their Own Words • gave doctors 24/7 access to oral cancer information, which reduced the time to start the treatment “Ushahidi - a Swahili word which means • gave patients medical assistance right “at their doorstep” “testimony” • made tracking the progress of patients easier We build tools for democratizing • made planning of follow-up treatments possible information, increasing transparency and • made it possible for doctors to help, supervise and lowering the barriers for individuals to research large regions and reduced general healthcare share their stories. costs, thanks to early intervention procedures. We’re a disruptive organization that is willing to take risks in the pursuit of 4.2. Ushahidi changing the traditional way that information flows… 4.2.1. Overview Ushahidi is a not-for-profit technological corporation, based in Since early 2008 we have grown from an Kenya, which specializes in developing free, open source ad hoc group of volunteers to a focused software for data collection, visualization and interactive organization. Our current team is mapping. Its leading product is the Ushahidi Platform, a web comprised of individuals with a wide span application to “easily crowdsource information using multiple of experience ranging from human rights channels, including SMS, email, Twitter and the web”. Other work to software development. We have Ushahidi’s tools include Crowdmap, a tool for “anyone wanting also built a strong team of volunteer to tell a story with a map” and Ping, a simple app which was developers primarily in Africa, but also allows “teams to check-in with each other” (Ushahidi.com). Europe, South America and the U.S.” 32 According to Ushahidi (2014), “the platform is used worldwide by activists, news organizations and everyday citizens”. It is an application used especially for crowdsourcing projects: Ushahidi Platform is used to collect bigger portions of data from larger crowds. Use cases suggested by Ushahidi are: • Monitoring elections. “Using the power of the crowd to monitor and visualize what went right and what went wrong in the election”. • Mapping crisis information. “Whether it’s a natural disaster, epidemic or a political crisis, the tool can be used to handle information coming out of a crisis.” The possible stakeholders in these use examples could be, for example, NGOs or other larger organizations interested in monitoring elections, or even active tech-savvy individual citizens, who are interested in building crowdsourcing-related projects. However, since Ushahidi is a platform as much as an application, many other use cases – and thus possible stakeholders – can be recognized, from commercial uses to needs of individual active citizens. The key features of the Ushahidi platform are: • That is can collect information from several different sources, such as SMS messages, MMS messages (images, video) and smartphone applications. • The information collected is usually shown on an interactive map, accessed over a web interface. • The information can be visualized also using a timelime: reports from a certain location can be seen in a chronological order, allowing for instance to visualize reports at different times of day or week. The primary developmental emphasis of Ushahidi Platform lies in its adaptability: The Ushahidi Platform can be adapted for use in many different scenarios. The platform is designed to be easy to use, from the perspective of the end-user. This is intended, for example, to encourage ordinary citizens to become content providers and to facilitate “citizen journalism”. To start using Ushahidi Platform, the platform needs to be installed to a server, which uses one of the Ushahidi-supported operating systems (OS). The Ushahidi Platform uses a MySQL database, and Google Maps API for geocoding information provided to the platform. Installing, configuring and administering Ushahidi Platform requires high-level knowledge in installing and administering server-side software. Ushahidi also provides a ready-made hosted version of Ushahidi Platform, called Crowdmap “Classic”. It offers some of the functionality of the Ushahidi Platform. Crowdmap “Classic” is not as configurable, as flexible or as secure as a dedicated Ushahidi Platform installation. However, Crowdmap “Classic” might be a sufficient solution for some straightforward cases of location/time based data collection and visualization. . 33 4.2.2. Ushahidi Use Cases Map of Russian Wildfires The Russian Federation experienced a heat wave during the summer of 2010. Wildfires caused by the heat wave were a major challenge which required a rapid response. Many people lost their homes and property, and officials and journalist faced several challenges in addressing the problem. But a volunteer project called “Help Map” was built using Ushahidi Platform. Help Map was used to 1) crowdsource and map unprecedented wildfires and 2) to offer help to individuals who had suffered losses because of the wildfires. With Help Map, people in need were connected to people willing to help. Ushahidi was used to aggregate two kinds of Figure 4.3. Ushahidi used in Help Map. Source: www.russian-fires.ru reports from the general public. Those in need of help answered the question “what is needed?”. Those who had the capacity of helping responded by reporting “I wish to help”. Offers to help included transportation, food, clothing, homes and shelters etc. (Mora, 2011.) The project started as a Ushahidi Crowdmap “Classic” project, but was soon migrated to an independent Ushahidi Platform service running on an independent server, that can still be seen at http://www.russian- fires.ru. According to Global Voices Online (2010) magazine, the project “quickly became an important base for volunteers to provide resources and lend a hand during weeks of crisis” and it is one of the most successful uses of Ushahidi so far. Messages sent to Ushahidi Platform included tweets, emails, SMSs and web form submissions. Messages were analyzed and, as in other Ushahidi implementations, were categorized, geolocated, verified and mapped. When location data could not be automatically determined, users were actively encouraged to indicate their location on a map, for more precise localization. (Mora, 2011.) Key Outcomes and Results According to Mora (2011), outcomes reported from the Help Map project included: • People in need were successfully connected with people willing to provide help during the wildfires. • Implementation of Help Map revealed the altruistic potential of the Russian society, especially in light of the tardy response from government officials. • Online communities and Internet users took on responsibility in helping people in need, and demonstrated that Russian network society is not a passive audience. 34 • The willingness to help of the Russian online community took on new, more creative forms: The Help Map community, for example, helped to develop volunteer firefighter units. Internet provided a platform for 24-hour coordination of help and exchange of help information, and Help Map was one of the core hubs in this. Similar examples include responding to the Haiti earthquake of 2010. Uchaguzi – Communicating with the public on the Kenyan Referendum The initial impetus for creating Ushahidi came in the wake of the post-election violence in 2008. With more time to plan, in 2010 Ushahidi collaborated with several partners to create and utilize the “Uchaguzi-Kenya” platform. The platform, built using Ushahidi Platform, was built to be a channel for Kenyan citizens to communicate openly about the 2010 Kenyan referendum on the new Constitution, where similar violence was feared. The goals of Uchaguzi-Kenya were to • Provide space for information sharing and collaboration, • Amplify the voices of the Kenyan citizens, • Increase the efficiency and transparency of election monitoring system and • Create a mechanism through which citizens could monitor and report incidents they’ve witnessed in election locations and election process in general. Uchaguzi-Kenya was a joint collaborative operation of Ushahidi, CRECO, SODNET, HIVOS and Urala. According to the broad case study (Uchaguzi – A Case Study, 2012) carried out on the Uchaguzi-Kenya project: “Uchaguzi was a new endeavor, not only in technology, but also in partnerships and information flows.” In Uchaguzi-Kenya, information from election locations was collected mainly from SMS messages, but also Twitter and E- mails were sent to report information from voting locations. A broad communications campaign was initiated to encourage people to send SMS messages if they saw incidents worth reporting during voting, not only reports of violence but also evidence of peaceful public participation. In addition to encouraging the general public to send information, a number of CRECO volunteers were trained to use the pilot Uchaguzi platform. The volunteers sent information to Uchaguzi by using a pre-determined code card: using pre-determined codes enabled volunteers to send more structured information in short form. Figure 4.4. Uchaguzi-Kenya screen capture. Source: Uchaguzi – A Case Study (2012) 35 During election day, some 550 CRECO’s volunteer election monitors sent SMS information to the platform. In CRECO offices, 10 staff members analyzed and mapped the data sent and, when necessary, provided user support for volunteers. During the election process, a total number of 2,492 Uzhaguzi messages were received (1,900 SMS, 571 Twitter, 21 E-mail). Some 51 % of these reports were verified. Key Outcomes and Results The “Uchaguzi-Kenya” –initiative was later benchmarked by Ushahidi, Knight Foundation and the Harvard Humanitarian Initiative. They found that the Uchaguzi-Kenya platform was a general success (Uchaguzi – A Case Study, 2012.). The platform was also an example of using Ushahidi platform in conditions where smartphones were not widely available. According to the report, the main successes were 1) collective approach to problem-solving and 2) strong leadership which focused on overall goals of the project. During the voting in the 2010 Referendum: • a relatively high number of reports were sent from a geographically representative range of different voting locations, • over 1500 “actionable” reports were sent from voting locations, and 194 “action taken” –reports were eventually sent, • marketing was (partially) successfully used to advocate active citizenship and use of the Uchaguzi platform, • different organizations collaborated successfully in using and advocating a new technology, • a significant number of volunteers were successfully trained to use and advocate the Uchaguzi platform. Since the first usage of Uchaguzi in 2010, similar initiatives have been successfully implemented in general elections in Tanzania (2010), Zambia (2011) and again in the Kenya (2013). One of the key changes from 2010 to 2013 was the implementation of smartphone applications: during the general election of Kenya in 2013, people could send information from election site using SMSs, Tweets, E-mails, web-based forms – and downloadable apps were available on devices running IOS and Android. 36 4.3. Dimagi CommTrack and CommCare Dimagi In Their Own Words 4.3.1. Overview Dimagi is an American social enterprise, founded in 2002. It offers “Dimagi is a privately held social several applications for improving the science of delivery enterprise founded in 2002 with its including CommCare, and CommTrack. headquarters in Cambridge, Massachusetts, USA. According to Dimagi’s, CommCare is “an easily customizable mobile platform for frontline workers to track and support their We deliver open and innovative interactions with clients”; while CommTrack is “a tool for mobile technology to help underserved logistics and supply chain management”. In addition to these communities everywhere.” applications, Dimagi also provides developer frameworks for building new SMS/web applications (RapidSMS) and Android- “At Dimagi, we believe that many of based survey tools (Open Data Kit), but these tools are not the world’s problems can be presented or discussed here. assuaged with low-cost technological solutions. “ Dimagi CommTrack Dimagi describes CommTrack as “a mobile logistics management “The developing world stands to system for low-resource settings”. It is a multiplatform, open massively benefit from the source application which is used to improve stock tracking, technological advances that have requisition planning and delivery acknowledgement. been made in recent years, and Dimagi wants to bring those benefits CommTrack’s original purpose was “to support health workers and other mobile agents who manage commodities in low- to fruition.” resource settings”. But it can also be used in other field cases, for instance: • to monitor stock activity and capture logistics data, i.e. what is the stock status of certain traced items, and where the items currently are, • to track and approve specific stock-related orders and • to report indicators on various forms non-stock data, such as number of cases treated in a certain medical programme. Stakeholders potentially interested in CommTrack might be, for example, organizations interested in the delivery of development aid in its most narrow sense: for instance, the delivery of goods or other stockable items, such as emergency food supplies or medicines. Using the mobile interface, a field worker can send and receive stock updates over SMS messages, over an offline phone application, using a smartphone application or using a web interface. Location metadata is stored to updates when possible. The web interface of CommTrack can be used to visualize stock data and field worker activity through reports, graphs and map visualizations. If location metadata is included in data sent from the field, stock data can be visualized in real-time over a map layer. 37 CommTrack gives both field workers and managers a real-time information on supply and demand indicators of supplies tracked. This makes streamlining requisitions and delivery possible, and thus can speed the supply chain. Data collected by CommTrack can be shared between different applications via an open API. New applications can also be built, to contribute relevant information to CommTrack’s database. The information is generally stored to Dimagi’s cloud service, but it is possible to install CommTrack to a completely independent server environment. Dimagi offers different “services” and “plans” for user organizations, ranging to free solutions to tailored enterprise options. CommTrack is currently used, for example, in Tanzania, Ghana, Malawi and Uganda. In Tanzania alone, CommTrack technology is used in over 2,300 facilities. Dimagi CommCare Dimagi CommCare is an open source platform which Dimagi describes as a “job-aid for mobile workforces”, a “tool for supervision and evidence-based change” and a “way to capture data in an electronic repository that otherwise sits in thousands of paper notebooks”. (Dimagi 2014.) The goals of CommCare are to 1) enable data-driven management with visibility into performance, and to 2) improve access, quality, experience and accountability of services. CommCare consists of two technology “components”: CommCare Mobile and CommCareHQ. The CommCare Mobile can be used by field workers, to • gather survey data, such as health related information (images, videos, text responses, bar codes etc) about patients or other subject at hand, • record images and videos to add extra content to the data collected, • implement GPS coordinates to the data collected, • educate: the application can show video, audio and image prompts and guides for health services or other services provided. Mobile application sends information to and synchronizes itself with cloud-based database application CommCareHQ. It is used by a compatible web browser to, for example, • analyze information collected using CommCare Mobile (and web forms), • visualize and display information, also using map layers to provide locational information, • build surveys and other methods for collecting data with the mobile application, • manage the field workers and their user permits, as well as other data relevant to usage of mobile application. CommCareHQ analyzes data sent from a mobile application in real time. CommCareHQ synchronizes data with mobile application(s) when network connection is available. CommCare Mobile is a cross-platform product, with support also for low-end Java and Android phones and tablets. As with CommTrack, Dimagi offers various plans for using CommCare, ranging from small- scale, costless options to solutions offered for major corporations or governmental partners. 38 Currently CommCare is used in more than 30 countries and by 50 different organizations. Over 2.5 million reports have been submitted using CommCare solutions. 4.3.2. Dimagi Use Cases Dimagi CommTrack: cStock – Supply Chains for Community Case Management, Malawi CommTrack technology was used in Malawi in a large project called cStock, a collaboration between JSI Research & Training Institute, the Bill & Melinda Gates Foundation and the Malawi Ministry of Health. Health Surveillance Assistants (HSAs) in Malawi carry and prescribe a pre-defined list of medicines such as antibiotics and anti-malarials. cStock software, based on Dimagi’s CommTrack, was used to help with medicine distribution and management. The goals of cStock were to: • improve the resupply process of medication carried and prescribed by HSAs; • provide a mechanism for health centers and district managers to troubleshoot and priorize product availability. This was done by automatically providing them withsufficient information on medicine carried and prescribed by HSAs. HSAs used their personal mobile phones to send geotagged SMS messages on stock information to the cStock database on the medicine they prescribed and carried. The cStock system automatically calculated the need for medicines in the ’as's work and reported stock levels of each HAS automatically to the correspondent health center. This made it possible for the staff of the health center to pre-pack the supplies needed before a HSA arrived. This helped to reduce the wasted trips done by HSAs, since before the implementation of cStock system, health centers often ran out of stock in medicine before the HSAs arrived. For higher-level staff, the amount of medicine supplied provided a possibility for real-time identification of problem areas. The locational data provided from the stock data of HSAs and health centers gave the possibility of monitoring the overall performance of the supply chain in a given area. cStock has been a success: it has been deployed to over 1,500 HSAs in Malawi. Key Outcomes and Results Dimagi reports many positive outcomes and results for cStock projects. According to Dimagi’s study, the application: • Helped community health workers save time: some 99 % of health workers reported that cStock saved them significant time, especially since fruitless trips to health centers were eliminated. • Helped to more than double the product availability after the implementation of cStock. • Improved the health/medicine reporting rates to above 80 % in all districts where cStock was used. 39 CommCare Tanzania According to Dimagi, CommCare is used in over 30 countries, and by over 50 organizations. Key organizations which have used it include World Vision, World Health Organization, Public Health Foundation of India and Save the Children. One of the first projects where CommCare was successfully used was CommCare Tanzania project, which started in 2010 and which is still ongoing. The project is carried out in co-operation with D-Tree International and Pathfinder International, both of whom have a large network of home-based care providers (HBCPs). They have a roster of clients within Tanzania, who HBCPs visit at least once a month. The HBCPs provide social support for their clients, and do screening Figure 4.5. CommCare in use in Tanzania. for clinical symptoms, such as drug related issues or Source: www.commcare.org other problems requiring care. HBCPs collect data using CommCare Mobile. The data collection tool is a mobile application, which prompts the HBCPs to 1) ask pre-determined questions from the clients they are visiting and 2) record the answers given by the patients. The data collection tool also collects other relevant data of the clients, such as basic patient information, and can be used to track GPS coordinates and save images and videos, if the device is capable to doing these. After collecting the data from patients, HBCPs upload the data collected to the CommCareHQ. Pathfinder and the Tanzanian government use it to build monthly health reports. They also plan their follow-up actions according to data gathered by HBCPs. Key Outcomes and Results While a proper research of the outcomes of CommCare Tanzania projects is yet to carried out, some results of the CommCare Tanzania project have been published. For example, according to Dimagi’s case study on CommCare Tanzania (2012), • after implementing an SMS reminder system to CommCare Tanzania’s mobile platform, the timeliness of the scheduled visits of HBCPs improved by 85 %. • Currently over 150 HBCPs use CommCare regularly. HBCPs are trained and monitored by D- Tree staff located on-site in Tanzania. Also constant tracking of client cases provided both NGOs and Tanzanian government information and tools for managing future aid processes. 40 4.4. Taarifa Taarifa In Their Own Words 4.4.1. Overview Taarifa is an open source application for information collection, “Taarifa’s community is diverse visualization and interactive mapping. The platform is created by the and scattered across the world. Taarifa community, which is a group of volunteers interested in They are brought together by the developing the application, but is said to be moving towards to a ideals of Taarifa in collaborating more formal structure. and building tools which help to Taarifa is derived from the Ushahidi project, presented earlier in this shape the world.” chapter. The project started at the World Bank’s Water Hackathon in London in 2012, as a solution to help delivering clean water, removing dirty water safely and separating dirty water from clean. Taarifa’s main extension to Ushahidi was adding problem reporting functionality to the Taarifa client (for example, picture and description of a issue by a mobile phone) and a triaging system to the web application (a report goes through several stages before being marked as “resolved”). (Appcircus: Taarifa, 2014.) Taarifa won the first prize at the London hackathon, and has been developed further subsequently as a platform solution, which allows all kinds of issue reporting, visualizing and resolving. Taarifa is primarily designed to “close citizen feedback loops”. This is done by: • collecting information, especially issue reports from the field, using SMS, HTML5-based web forms, E-mails and tweets; • visualizing and publishing the gathered information using simple map interfaces; and • sending the collected information, the issue reports, to officials. By using the web application of the platform, the reports filed can be followed up and acted upon. For example, 1) a citizen can use his/her smartphone to fill the web form in a service using Taarifa Platform, to report an unlicensed junkyard that has been observed. Locational data is included in the issue report, as well as a photo from the site. 2) The information on the junkyard is displayed on the web service’s map visualization. 3) Information on the issue is automatically sent to officials, who act to close and clean the illegal junkyard. 4) Officials update the web application and report the issue, once it is resolved. This information is, again, saved to the issue report visualized on the map. 5) Citizen(s) can see on the map, that the problem has been resolved. Taarifa helps to engage citizens and broader communities especially in the developing world: Taarifa connects the citizens who send issue reports with officials, who have the ability to fix the reported issues. 41 Taarifa says that it gives citizens “the power to report and address the problems around them” by using spatial information. The issue reports sent using Taarifa contain locational metadata, which is used to place the issue report to a map. With map visualizations, both the citizens and the officials can follow the sent reports, as well as the follow-up actions the reports have generated. The Taarifa Platform is built by an active community, who believe in the ideals of Taarifa in collaborating and building technological solutions to help communities. Therefore the design emphasis of Taarifa has been in building a tool for activating citizens to take interest in local issue by using spatial information and map-based visualizations. Figure 4.6. Example of a Taarifa platform Source: www.taarifa.org As Taarifa is an open source API and not a web service as such, a somewhat high level of expertise is needed to get the service installed, administered and configured. The system requirements of Taarifa are somewhat similar to Ushahidi’s: the installation can be done on a dedicated server or to a cloud service. A MySQL database is needed. The development process is actively followed in Taarifa’s blog site at http://taarifa.wordpress.com, and new developers are actively encouraged to participate in building Taarifa. 4.4.2. Taarifa Use Cases Uganda Pilot Project for school construction After winning the water hackathon competition, Taarifa was used in a pilot project with the Ugandan Ministry of Local Government and the Africa Urban and Water sector of the World Bank. In the pilot project the Ministry wished “to monitor local government project based around improving community cohesion, public services and enterprise.” The pilot project was carried out as a part of the Ministry’s “Improving Systems for the Urban Poor” –programme, and its two sub-programmes “Community Driven Development” (CDD) and “Local Government Management and Service Delivery” (LGMSD). CDD is a match- funding program, while LGMSD is, for example, a program for Figure 4.7. Taarifa used in Uganda pilot project. Source: Taarifa. 42 constructing buildings such as schools. Traditionally the CDD and LGMSD reporting systems were paper-based. The paper forms were filled by pen, then posted to central offices in Kampala, Uganda. The drawbacks of this system were the inreliability of the postal system, the slow speed of the paper-based system, and the workload based upon the reporters. This was something that the Ministry wanted to change. (Ilffe, 2012.) In the Taarifa-based pilot programme, civil servants were trained to use low-end smartphones equipped with Taarifa to upload the forms automatically to the Taarifa application. The intent was to speed up the process, make the process more reliable and decrease the workload of the field workers. Key Results and Learnings The pilot was deemed successful by the Ugandan Ministry. During and after the completion of the pilot: • the platform was rolled out to 111 districts of Uganda, • trained field workers managed to collect CDD and LGMSD information by using the smartphones provided, • Taarifa was thoroughly tested in day-to-day field use, and improvements were made to the Taarifa platform. During the pilot, it was observed that custom forms built for smartphones worked well initially, but when field workers ventured into more distant districts, some problems with the offline mode of Taarifa were discovered. For example, it was not possible to change different survey forms when in offline mode. This problem was later addressed, and was used as a good example of connectivity problems in rural Africa, and a demonstration of the flexibility of Taarifa platform community. 4.5. Comparing the applications The applications presented above – PoiMapper, Ushahidi, CommTrack, CommCare and Taarifa – have successfully been used in a number of real-world cases. While all these applications can be said to enhance the delivery of development aid, at least in some way, the applications, their uses and target groups differ somewhat. From stakeholders’ perspective, it is essential to select the right application for the right need. Table 4.1 presents a functional analysis of the usage of the applications while their technical similarities and differences are presented in Table 4.2. 43 Table 4.1: A functional analysis of the locational data applications PoiMapper Ushahidi CommTrack CommCare Taarifa What it is? Application for A platform to An application to An application to An application collecting, collect, edit and track inventory, gather survey for reporting editing, analyzing visualize crowd- and to track the data, such as location-based and visualizing sourced data on a logistics of supply images, GPS issues and to point-of-interest map. and demand. coordinates, text, track their based data. barcodes, videos, resolution. etc. Typical Use User defines User contributes A field worker A field user User reports an Example what data to to a reports records survey issue that needs collect (eg text, crowdsourcing information on data to a mobile resolving using a photos, videos project by adding stock usage, on- device. Data can mobile phone. etc); gathers data incident reports to site, using a mobile be visualized and The issue is then using a mobile a map which can device. An office- analyzed from visualized on a app and then be analysed based employee the Web map and public visualizes it from to show the spatial can track real-time interface. officials can take the Web build-up of inventory, and act action. interface. incidents over accordingly. time. Typical Field workers and Crowdsource Field workers and Field workers and Citizens targeted volunteers in volunteers and volunteers in volunteers in reporting a local Users NGOs and activists. NGOs and NGOs and issue, and government. government and government and officials with a their managers. their managers. responsibility to fix it. Design Usability: robust Adaptability: can Simplicity: Adaptability: can Simplicity: features and simple to be tailored for use automated SMSs be adjusted to citizen reporting use, in many scenarios. assist with logistics collect many made as easy as management. kinds of data. possible. Key benefits Can be used in Highly flexible and Cost effectiveness: Open source Workflow- (selected) “tough” can run on various can reduce flexibility: can be based: issue environments platforms. unnecessary work adapted for reports can be running on and trips. multiple uses. tracked and various resolved. platforms. Use cases Tuberculosis Mapping of Stock control for Communication Tracking school profiled monitoring in wildfires in Russia; community health with, and construction Thailand; Oral monitoring workers in Malawi tracking, of and cancer screening Kenyan home-based care maintenance in in India. referendum. providers in Uganda Tanzania Comments Location Open source and Usage focused on Can be used as a Developed out information is the widely used for a aid and healthcare teaching aid with of a hackathon “base metadata: variety of different logistics multimedia and collected. purposes. capabilities subsequently volunteer driven. Source: Authors, adapted from websites and case examples. 44 Table 4.2: A technical analysis of the locational data applications PoiMapper Ushahidi CommTrack CommCare Taarifa Developer Pajat Inc., Ushahidi Dimagi, US- Dimagi, US- Taarifa Finland volunteer based social based social community community. enterprise. enterprise. Mobile Android, iOS, J2EE phones, Offline solution J2EE phones and Any SMS- Platforms * J2EE. with apps for for Nokia Android. capable phone. Android, iOS. phones. Apps for HTML5 Android, iOS. interface, apps for smartphones. License Proprietary Open Source. Open source, Open source, Open source (software as a Licensed as GNU and software as and software as service) Lesser General a service a service Public License. User Easy and simple Good admin and Mobile app easy Mobile app easy Good admin and competence to use and installation skills to use but web to use but web installation skills required administer required, but application application required, but data reporting requires requires data reporting easy for users experience experience easy for users Acquisition of Locational data Locational data Locational data Locational data Locational data locational data saved saved can be saved can be saved saved automatically automatically when required. when required. automatically and can be and can be and used in edited. Points of edited. Points of issue tracking & interest shown interest shown report on map. on map. visualization. Website poimapper.com ushahidi.com commtrack.org commcare.org taarifa.org Note: Each application also has a web interface for admin and management. Source: Authors, adapted from websites and case examples. 4.6. Using Big Data to Enhance the Science of Delivery In addition to the applications and case studies presented earlier in this chapter, locational data can also be collected passively in the form of so-called “big data”. This can be defined as very large, complex datasets, which can be researched and, for example, visualized, using appropriate research tools. The results of the research processes can be, in turn, used in both policy making and development aid projects. Big Data can be a powerful research tool, and it can provide an interesting perspective to development aid projects. A good example of the use possibilities of Big Data was demonstrated by Google in predicting flu spread patterns from flu-related search queries. This showed that Big Data, which can be seen as byproduct of citizens’ everyday activities, can be harnessed for making powerful predictions. This, in turn, can steer policy makers’ and citizens’ behavior (Philip, Schuler-Brown and Way, 2013). However, issues such as security or privacy should be considered thoroughly. Some examples of using big data in development aid projects are published in the United Nations Global Pulse “Mobile Phone Network Data for Development” primer (2013). The primer explains how analysis 45 of large quantities call data records (CDRs, sometimes called Call Detail Records), passively collected from mobile phones, can provide information for development purposes, and thus provides case examples of using one type of big data in development aid projects and research. United Nations Global Pulse (2013) defines CDRs as follows: “Whenever a mobile call or transaction is made, a call data record (CDR) is automatically created by a mobile network operator. CDRs are a digital collection of attributes of a certain instance of a telecommunication transaction, such as the start time or a duration of a call. … An additional piece of information that gets recorded in CDRs by a mobile network operator is to which cell towers the caller and recipient’s phones were connected at the time of the call.” The primer presents three types of indicators, which can be extracted by analyzing the big data of call detail records: the analysis of the mobility of people, the analysis of the social interaction and the analysis of economic activity. Both of these are discussed below. A somewhat different example of using Big Data in development aid related research, was a study conducted by a team of researchers from the Harvard School of Public Health and seven other organizations. The team used cell phone data to illustrate the spread of malaria, and to show how human travel patterns contribute to its diffusion. This example is briefly discussed in section 4.6.3. While connecting the frameworks presented in chapter 3 to Big Data research is difficult, at least some observations can be made by using the UHP review framework, where the factors of relevance, effectiveness, efficiency and sustainability are considered. This suggests that Big Data research has great potential in the science of development delivery, for the following reasons: • The availability of “Big” data sets, and the relevance of Big Data research, is increasing. Epidemiology studies, for instance, are impossible, or at least extremely difficult, to produce without large datasets, and information collected from mobile phones is relatively inexpensive, and available in near real-time. • Big Data research can directly inform decisions made by policy makers (evidence-based policy-making), especially the case of mobility analysis presented in section 4.6.1. • From the perspective of research, Big Data can add value to development projects, if only because the technique is still relatively new. • The Big Data approach is inductive, in the sense that researchers look for patterns in the data rather than presupposing, from theory, what they may find. Thus, there is scope for surprises. 4.6.1. Using Big Data to Analyze the Mobility of People By researching call detail records, it is possible to analyze the movement of masses, since the route of calls and messages can be tracked (by “connecting the dots” between different cell towers). This information can be used to analyze and visualize, for example, the daily rhythms of commuting of large quantities of people. 46 One example where CDRs were used to analyze and develop movements of large quantities of people was the in a program for optimizing transport networks in the Côte d’Ivoire’s largest city, Abidjan, conducted by IBM’s AllAboard project’s researchers. According to Mobile Phone Network Data for Development, “rabid urbanization in developing countries has increased pressure on infrastructure such as road networks. Roads and public transportation systems become saturated, and people lose great deal of time traveling from home to work.” This problem was addressed by analyzing 500,000 CDRs over a period of five months: “Mobile phone location data is used to infer origin-destination flows in the city, which are then converted to ridership on the existing transit network. Sequential travel patterns from individual call location data is used to propose new candidate transit routes. An optimization model evaluates how to improve the existing transit network to increase ridership and user satisfaction, both in terms of travel and wait time.” (IBM AllAboard 2013.) According to United Nations Global Pulse Figure 4.8. Example of mapped commuting data. (2013), the research work resulted in a Source: IBM AllAboard (2013). partial solution to the commuting problem: four new bus routes were added to Abidjan’s commuting infrastructure, and one route was extended. This should reduce traveling time by 10 per cent. United Nations Global Pulse sees such a provisioning use of CDRs as “useful for better urban planning and public transportation”. 4.6.2. Using Big Data to Analyze Social Interaction and Economy A second indicator, where CDRs are used to improve the science of delivery, is the analysis of social interaction. This can, in turn, be used to analyze other social patterns, such as poverty. The third indicator, economic development, can also be researched by analyzing CDRs. According to United Nations Global Pulse (2013), the geographic distribution of social connections can be used to build demographic profiles of aggregated call traffic, and understanding changes in behavior: “Studies have shown that men and women tend to use their phones differently, as do different age groups. Frequently making and receiving calls with contacts outside of one’s immediate community is correlated with a higher socio- Figure 4.9. An example of poverty map economic class.” developed using Call Data Records. Source: United Nations Global Pulse (2013) 47 One of the interesting research cases, where CDRs and big data were used to understand social connections and the link between poverty and social interaction, is described in the paper “Ubiquitous Sensing for Mapping Poverty in Developing Countries” (2013) by Christopher Smith, Afra Mashhadi and Licia Capra. In their paper, Smith et al research a large quantity of anonymized CDRs from Côte d’Ivoire. They research the correlation between the CDRs and poverty by combining the CDRs with a Multidimensional Poverty Index created by the University of Oxford, and find that there are “several features of communication patterns among mobile phone users in Côte d’Ivoire that track poverty of regions as defined by the Multidimensional Poverty Index.” Smith et al. see that these research findings have important implications for policymakers and agencies working in societies, where it’s difficult to manually collect large quantities of socioeconomic data. They also claim that the analysis of CDRs “can be used to provide poverty estimates at a spatial resolution finer than previously available.” United Nations Global Pulse (2013) comment that Smith, Mashhadi and Capra in their research “validated the possibility of making poverty maps using call detail records.” It is not a coincidence that both the commuting study and the poverty study use CDRs from the same country, Côte d’Ivoire. That’s because relatively few operators, and relatively few developing countries are currently willing to release call data in the quantities needed to make such analysis work. But data release is becoming more common in developed countries, as the example profiled in Figure 1.2 shows. Big data approaches are, for the moment, much less well developed than the mobile applications profiled in the first part of this chapter. But in the longer term, the potential value of passively collected call data is much higher than actively collected survey data, if only because the unit costs are so much lower. 4.6.3. Using Big Data to Analyze the Spread of Malaria In 2012, researchers at the Harvard School of Public Health (HSPH) and several other institutions published a research paper, where anonymized phone data was combined with information on the regional incidence of malaria, in kenya. The research, showed on largest scale so far, how human travel patterns contribute to the spread of malaria (HSPH News, 2012.) The study, by Amy Wesolowski and her team, published in the October 2012 issue of Science, presents results from a study where researches mapped every call or SMS message made by each of over 14 million mobile phones in Kenya. According to Wesolowski et al (2012), Figure 4.10. Examples of “sources” “Human movements contribute to the transmission of malaria on and “sinks” of malaria. Source: Wesolowski et al. (2012) spatial scales that exceed the limits of mosquito dispersal. Identifying the sources (where malaria comes from) and sinks (where malaria goes) of imported infections due to human travel and locating high-risk sites of parasite importation could greatly improve malaria control programs.” 48 Previous smaller-scale studies had used mobile phones to estimate importation rates of malaria, but lacked resolution on infection risk. In their study, Wesolowski et al used a Big Data approach to identify networks of malaria parasite movements, and pinpoint regions from where disease “sources”, and to where it “sinks”. By analyzing this huge dataset, the researchers came to observe that 1) it was possible to estimate the probability that a particular person is carrying malaria parasites, and 2) it was possible to build a map of parasite movements between areas that mostly emit diseases and areas that mostly receive diseases. The map not only predicted malaria movement, but also showed which locations could be targeted for malaria control and elimination (Medical News Today, 2012). The research team found that a large part of malaria in Kenya comes from the Lake Victoria region and spreads east, mainly towards Nairobi, Kenya’s capital. Using this information, according to the research team, “would yield the biggest benefit nationally”. This is important research, since one million people die from malaria each year globally, and 90 per cent of the deaths are children under five years, living in sub- Saharan Africa. (Medical News Today, 2012.) While Wesolowski et al see some limitations in their approach – the study can only measure mobility among phone owners and in regions where are phone networks, and cannot showcase migration cross borders (Wesolowski et al, 2012) -- nevertheless the research is important and influential. It could be used, for example, by public health officials. They could send geographically targeted text message warnings to the phones of people traveling to high-risk malaria areas, suggesting they use a bednet, and thus help reducing the spread of malaria (HSPH News, 2012.) 49 5. Conclusions 5.1 Using Locational Data to Enhance the Science of Delivery This report has provided insights about the many features of a mobile phone, specifically the technologies which empower users to locate themselves in space and time, to collect and visualize locational data, and to interface with computer databases and maps in “the cloud”. Together, these capabilities serve to enhance the “science of delivery”, when employed in a development context. Specifically, they assist in applying experimentation-based evidence to improve development aid process outcomes. Chapter Two of this report provides the technological underpinning to location tracking, based on three particular tracking technologies: cell-tower triangulation, GPS (Global Positioning System) by satellite and WPS (WiFi Positioning System). The selection of which technology to use for a particular application will depend on the availability of data, the degree of accuracy required, and the location of the user (eg indoors/outdoors, urban/rural etc). It will further depend on the degree of sophistication of the mobile devices used. In addition to the technical solutions which cell-tower triangulation, GPS and WPS can provide, simple innovations in mobile phone use can also be used to pinpoint locations. For example, SMS text messages sent by a user from a certain location can contain a code, or other piece of geo-specific information. This code can be tied to a certain known location – an election site, for example, as presented in Uchaguzi- Kenya referendum monitoring project presented in section 4.2.2. Chapter Three presented a methodological and conceptual framework for the benchmarking of specific cases carried out in Chapter Four. Several mobile applications and real-life use cases were highlighted that have recently been used to enhance the delivery of development assistance. The applications presented were PoiMapper, Ushahidi, CommTrack, CommCare and Taarifa. The similarities and differences between the different applications are summarised in the benchmark crosstabs presented in section 4.5. In addition to the five benchmarked applications, some potential uses of location-related Big Data – drawing upon very large samples sizes of data – were explored. Section 4.6 focused on three cases where Big Data was used in research related to development interventions. In the first two cases, the analysis of call detail records (CDRs) from mobile phones can be used in development projects and research: in a project carried out in Côte d’Ivoire, the analysis of CDRs led to improvements in public transport. In a research project case presented, it was shown how CDRs can be used to map poverty in developing countries. In addition to these cases, a research case where sources and sinks of malaria were mapped from phone analysis data, was presented. 5.2 Potential Uses of Locational Data for Addressing Development Challenges This report has established that locational data applications can, and are, being used by development practitioners. Although their use is still largely small scale and experimental, and the technology continues to evolve, there is sufficient evidence of beneficial outcomes to move to deepen and broaden the experiment. In other words, pilot programs show sufficient potential for scaling up and extending to additional development challenges. One way of conceptualizing the potential of locational data is by examining, in broad terms, what development challenges are facing today’s world and then considering how locational data could be used to respond to them. Table 5.1 presents a first cut at this. The eight development challenges presented here are the Millennium Development Goals, as defined by the UN Millennium Summit in 2000. 50 Millennium Development Goal Possible use of mobile survey data Possible use of “Big Data” application 1. Eradicate extreme poverty and Potential: High Potential: High hunger Mobile tablets can be used for conducting household surveys Big Data from Mobile Call Records can be used as a back up, or sanity for poverty assessments. The World Bank has used mobile test, for other sources of data, for instance by showing the ratio of tablets, equipped with geotagging capabilities, to replace paper- incoming to outgoing calls in different parts of a country (with a high based surveys for rapid assessments in South Sudan. ratio of incoming calls suggesting lower income levels). 2. Achieve Universal Primary Potential: High Potential: Moderate Education School teachers may submit attendance statistics for their Mobile Call Records from the phones of teachers could be used to track classes, which can be geo-tagged and time-stamped to identify whether they are in the same location as their school, to track absentee regions where attendance is low, and can track trends over time. or “ghost” teachers. 3. Promote gender equality and Potential: Moderate Potential: Low empower women Data on school attendance (see above) can be tracked against In general, anonymized call records do not allow to distinguish between geotagged photos on school maintenance, showing which male and female users. But Big Data can be combined with survey data schools have working toilets. Anecdotal evidence suggests to track, for instance, the routes followed by women when gathering girl’s participation in class is lower in schools than lack good firewood or collecting water, typically female roles in many countries. sanitation facilities. 4. Reduce child mortality Potential: Moderate Potential: Low Vaccination campaigns for infectious childhood diseases, such In general, anonymized call records do not allow to distinguish between as measles, can be followed using geotagged and timestamped child and adult users. However Big Data can be combined with survey survey data, to ensure full coverage in a particular area. data, for instance, to track children’s movements in areas close to landmines. 5. Improve maternal health Potential: Moderate Potential: Low Doctors or midwives making home visits to expectant mothers SMS messages, send to expectant mothers with information on could make geotagged emergency calls requesting urgent maternal healthcare can be targeted, using triangulation techniques, to medical assistance cells where maternal mortality is high 6. Combat HIV/AIDS, Malaria Potential: High Potential: Moderate and other diseases Inventory tracking for community health workers allows more Tracking of mobile CDRs in an area around an outbreak of an efficient restocking. infectious disease allows for estimations of risk of contagion and epidemiological modelling. 7. Ensure environmental Potential: High Potential: Moderate sustainability Camera-equipped mobile phones can be used to track the state Call data records can be used to estimate the population density of rural of coastal sea defences during high tide. areas, which can be meshed with satellite imagery, to better plan for anti-deforestation campaigns, for instance by providing alternative cooking fuel. 8. Develop a global partnership Potential: Moderate Potential: Low for development Mapping of donor initiatives, for instance by geo-tagging field Sharing of call data records around the time of disaster events, such as reports, can be used to better coordinate development earthquakes or tsunamis, can help in disaster risk planning. interventions. Figure 5.1: Examples of the potential uses of locational data in responding to global development challenges Source: Authors. 51 In general, mobile survey data applications have the highest potential, notably for the development challenges relating to poverty and hunger (#1), education (#2), control of infectious diseases (#6) and environmental sustainability (#7). The applications of Big Data are less obvious, but perhaps that is constrained by data availability. Big Data has the most promise in carrying out rapid, or real-time poverty assessments, particularly at times of stress, for instance associated with famine or civil unrest. Big Data approaches generally need to be combined with other data sources, for instance associated with satellite imagery, to be effective. 5.3 Recommendations to development practitioners and policy-makers Some observations can be made on the uses of the applications and big data. First, all the applications presented in this report have been used to gather, edit and display information that is relevant for development purposes. Although the types of information collected can vary – for instance, survey data, images, videos, or inventory status information -- all the showcased applications have proven successful in enhancing the science of delivery. Second, there is a wide variety of potential applications that respond to range of different development challenges. The application chosen will vary according to the function of the end-user: for instance field workers may be more likely to use mobile device while office-based staff may be more likely to use web- based applications. Third, the design emphasis of the applications, their key benefits, and intended uses also vary. It’s up to the project management and, ultimately, to those implementing the project, to choose which platform or application is the best choice for a given situation or delivery outcome . Fourth, in addition to the use of locational data for evidence-based policy-making, it can also be used to improve organizational efficiency, particularly for development functions that rely on timely updating of inventory (eg community health workers or emergency food supplies). The tools can also be used to improve project management and donor coordination, for instance through easy visualization of areas being targeted by different organisations. Fifth, further research is required to address concerns over information security and privacy. In practice, these concerns appear to have been overstated in that it is relatively easy to anonymize data, for instance for call data records. But nevertheless, guidelines for best practice use should be set. Sixth, policy-makers may need to take a lead in encouraging the release of call data records that privately-owned operators may consider too commercially confidential to release. Policy-makers can raise awareness by showing the beneficial outcomes from Big Data projects, and by setting up neutral clearinghouses that will apply basic privacy and security standards. 5.4 Final reflections After presenting and comparing the technologies which underlie locational data, and having reviewed a series of mini case studies of applications and Big Data solutions, the following final reflections can be offered: • Locational data can be generated from three main sources: cell-tower triangulation, GPS- equipped mobile devices, and WPS. • Locational data can be used to identify key waypoints (points of interests) for monitoring outcomes, tracking user movement, or displaying certain routes. • Locational data, when complemented with temporal data, can provide a measurable depiction of mobility and progress, for instance, in the case of geo-tagged and time-stamped data such as SMS-survey responses, media files, or user-generated reports. 52 • The technologies have their limitations. Cell-tower triangulation relies on the density of cell- networks and towers in a given geographical area. GPS is limited by dense foliage, crowded urban areas, and other signal blocking inhibitors. WPS is entirely dependent on dense WiFi connectivity. • Locational data provided by mobile phones and applications related to them can be used – and are being used – to enhance the science of delivery. • Case examples presented in the report show that development interventions, such as providing healthcare in difficult environments or monitoring elections, can be made easier by using technologies and applications that generate, and make use of, location-based data. • The Big Data related cases presented here show its potential in research and evidence- basedpolicy-making. • Technologies and applications used to enhance the delivery of development aid vary. So do the uses, target groups and properties of applications and cases presented here. The applications don’t directly compete with each other, but complement one another. Therefore, the appropriate application must be chosen for the specific development challenge. 53 References Alexandrova, Alexandra (2014) “Bluetooth Low Energy Solution Released”, an Indoo.rs published blog post, dated 27 November, 2013 and available at: http://indoo.rs/bluetooth-low-energy-solution-released/ . Anthony, Sebastian (2012) “Indoor Navigation on your Smartphone, using the Earth’s Magnetic Field – just like a pigeon”, an Extreme Tech digital news article dated 9 July, 2012, and published by Ziff Davis, LLC., available at: http://www.extremetech.com/computing/132484-indoor-navigation-on-your- smartphone-using-the-earths-magnetic-field-just-like-a-homing-pigeon Anthony, Sebastian (2012) “Think GPS is cool? IPS will blow your mind”, an Extreme Tech digital news article dated 24 April, 2012, and published by Ziff Davis, LLC., available at: http://www.extremetech.com/extreme/126843-think-gps-is-cool-ips-will-blow-your-mind/2 . Appcircus: Taarifa (2014): Taarifa Description at Appcircus. Broadcom.com (2014) “Broadcom Delivers First Global Location Chip for Wearables: Enables Consumers to More Accurately Manage Fitness Data On-the-Go”, press release dated 20 February, 2014, available at: https://www.broadcom.com/press/release.php?id=s827044 Burton, Marcus (2013) “Location-based Wi-Fi services can add immediate value to Wi-Fi deployments”, A web articled dated 10 October 2013, available at: http://www.networkworld.com/news/tech/2013/101013-wifi-location-services-274734.html Business Insider (2013) “Smartphone Market Forecast”, available at: http://www.bullfax.com/?q=node- smartphone-market-forecast-how-price-sensitive-global-c. Cangeloso, Sal (2013) “Forget WiFiSlam – ByteLight uses LEDs for indoor positioning,” an Extreme Tech digital news article dated 25 March, 2013, and published by Ziff Davis, LLC., available at: http://www.extremetech.com/extreme/151068-forget-wifislam-bytelight-uses-leds-for-indoor- positioning Carothers, Thomas (2010): The Elusive Synthesis. Journal of Democracy 21.4, pp. 12-26. ESRI (2014) “ArcGIS Online Features”, a product description, available at: http://www.esri.com/software/arcgis/arcgisonline/features/apps . Evans, Annie (2013): The Top Ten Challenges of Implementing New Technology. Retrieved from http://gettingsmart.com/2013/08/the-top-ten-challenges-of-implementing-new-technology/ Fend, Jiangfan & Liu, Yanhong “Wifi-based Indoor Navigation with Mobile GIS and Speech Recognition”, a journal article from IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 6, No 2, dated November 2012, available at: http://www.ijcsi.org/papers/IJCSI-9-6-2-256-263.pdf Gartner (2014) “Market share analysis: Mobile Phones worldwide, 4Q2013 and 2013”, available in summary form at: http://www.gartner.com/newsroom/id/2665715. Global Voices Online: Russia “Help Map” Wins Runet Award (2010): http://globalvoicesonline.org/2010/12/03/russia-help-map-wins-runet-award/ Guest, Greg; Macqueen, Kathleen M and Namey, Emily E (2012): Applied Thematic Analysis. SAGE. 54 Holmes, Chris (2014) “Distributed Versioning for Geospatial Data (Part 1): Distributed Versioning Implementation’, A white paper available at: http://boundlessgeo.com/whitepaper/new-approach- working-geospatial-data-part-1/ . Hsieh, Edward F. (2005): Investigating Successful Implementation of Technologies in Developing Nations. Available at http://dspace.mit.edu/bitstream/handle/1721.1/32887/62616466.pdf?sequence=1. HSPH News (2012): Using Cell Phone Data to Curb the Spread of Malaria. Available at http://www.hsph.harvard.edu/news/press-releases/cell-phone-data-malaria/. IBM AllAboard (2013): Smarter Urban Dynamics. http://researcher.watson.ibm.com/researcher/view_group_subpage.php?id=4746 Iliffe, Marc (2012): Taarifa: Public Service Delivery. http://www.agi.org.uk/storage/GeoCommunity/AGI12/Papers/MarkIliffe.pdf%20 Insiteo (2014) “INSITEO first to announce Bluetooth 4.0 indoor location on Android”, a press release on Webpage available at: http://www.indoor-gps.com/joomla/index.php/en/latest-news/185-insiteo-first-to- announce-bluetooth-40-indoor-location-on-android-2 ISO/IEC 25010:2011 (2011): Systems and software engineering -- Systems and software Quality Requirements and Evaluation (SQuaRE) -- System and software quality models. Kosack, Stephen (2002): Effective Aid: How Democracy Allows Development Aid to Improve the Quality of Life. World Development vol. 31 n:o 1, pp. 1-22. Lepmets, Marion; Ras, Eric and Renault, Alain (2011): A Quality Measurement Framework for IT Services. Service Science and Innovation Public Research Centre Henri Tudor. Lounamaa, Pertti (2012) “PoiMapper: A Mobile GIS Cloud Service to improve field work efficiency, monitoring, and planning,” A Pajat Solutions Ltd power point presentation given on 15 May 2012, available at: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=7&cad=rja&uact=8&ved=0CF8 QFjAG&url=http%3A%2F%2Fwww.bond.org.uk%2Fdata%2Ffiles%2FEffectiveness_Programme%2FP oiMapper_BOND_100512.ppt&ei=DnNhU9DLGOKQyAHozYHQAw&usg=AFQjCNFlz3sTFcGvwICY QZZgVQbDEgIEGg&sig2=x5_pz0iBEeU5fjxWlwGJdw&bvm=bv.65636070,d.aWc . Medical News Today (2012) “Curbing Malaria Spread With Cell Phone Data”. A news article available at http://www.medicalnewstoday.com/articles/251425.php. Mora, Fernando (2011): Innovating in the Midst of Crisis. A Case Study of Ushahidi. http://www.researchgate.net/publication/231537244_Innovating_in_the_midst_of_crisis_A_case_study_o f_Ushahidi Morrow, Nathan; Mock, Nancy; Papendieck, Adam; Kocmich, Nicholas (2011): Independent Evaluation of the Ushahidi Haiti Project. Retrieved from http://ggs684.pbworks.com/w/file/fetch/60819963/1282.pdf Moss, Magdalene (2012) “Collecting and Utilizing Point-of-Interest (POI) Data Cost-Effectively with Mobile Technologies,” Journal Article from Vol. 4 of the Sustainable Research and Innovation Conference Proceedings held May 3rd-4th at the African Institute for Capacity Development, available at: http://elearning.jkuat.ac.ke/journals/ojs/index.php/sri/article/view/238 Navizon|Accurate Positioning Anywhere (2014) “Applications in Health Care,” Webpage available at: http://www.navizon.com/health-care . 55 Online Dictionary of the Social Science (n.d.): Qualitative research. Retrieved from http://bitbucket.icaap.org/dict.pl?alpha=Q. OpenGeospatial.org (2014) “Compliant Products,” A webpage offering information on the difference between OGC compliant and implementing technologies, available at: http://www.opengeospatial.org/resource/products/compliant Pajat and Biocon (2013): Case Study on Oral Cancer Prevention Pajat Solutions Ltd (2014) “PoiMapper: Real-Time Field Data,” case study available at: http://solutionscenter.nethope.org/case_studies/view/poimapper-real-time-field-data# Pajat Solutions (2014). Tuberculosis Monitoring by Plan Thailand – An Abstract Park, Will (2008) “Federal judge sets cellphone privacy precedent – Requires warrant for cellphone location data”, web article dated 13 September 2008 from intomobile.com via the Washington Post, available at: http://www.intomobile.com/2008/09/13/federal-judge-sets-cellphone-privacy-precedent- requires-warrant-for-cellphone-location-data/ Philip, Thomas M.; Schuler-Brown, Sarah; Way, Winmar (2013) “A Framework for Learning About Big Data with Mobile Technologies for Democratic Participation: Possibilities, Limitations, and Unanticipated Obstacles”, available at Tech Know Learn 18:103-120 QGIS.org (2014) “QGIS as OGC Data Client,” A chapter from the QGIS 2.0 guidebook, available at:https://www.qgis.org/en/docs/user_manual/working_with_ogc/ogc_client_support.html QGIS.org (2014) “Supported Data Formats,” A chapter from the QGIS 2.0 guidebook, available at: https://www.qgis.org/en/docs/user_manual/working_with_vector/supported_data.html QGIS.org (2014) “Vector Data,” A chapter from the QGIS 2.0 guidebook, available at: https://www.qgis.org/en/docs/gentle_gis_introduction/vector_data.html Ratzer, Charles (2014) “BAE Systems Unveils Geospatial Intelligence Mobile App for Google Glass”, A news article dated 15 April, 2014, available at: http://www.baesystems.com/article/BAES_166261/bae- systems-unveils-geospatial-intelligence-mobile-app-for-google- glass%E2%84%A2?_afrLoop=570592627905000&_afrWindowMode=0&_afrWindowId=null&baeSessi onId=qRvhTz9b7RDDMnbvfVjfhKNkQJD9mXtnGj1PHdcXRl6yLFQgc5TW!425070103#%40%3Fbae SessionId%3DqRvhTz9b7RDDMnbvfVjfhKNkQJD9mXtnGj1PHdcXRl6yLFQgc5TW%2521425070103 %26_afrWindowId%3Dnull%26_afrLoop%3D570592627905000%26_afrWindowMode%3D0%26_adf.c trl-state%3D5an5qm1ff_4 Segan, Sascha (2011) “Smartphone GPS Just Got a Whole Lot Better, and Russian”, A PCMag Digital news article dated 7 September, 2011 and published by Ziff Davis, available at: http://www.pcmag.com/article2/0,2817,2392569,00.asp Shibisaki, R. (2014) “The flows in cities: Tracking movements of cell phone subscribers for infrastructure planning and disaster risk management”, Presentation based on written research available at: http://pflow.csis.u-tokyo.ac.jp/paper_e.html. Smith, Christopher; Mashhadi, Afra; Capra, Licia (2013): Ubiquitous Sensing for Mapping Poverty in Developing Countries. http://www0.cs.ucl.ac.uk/staff/l.capra/publications/d4d.pdf Spirent Communications (2014) “What is Wi-Fi Positioning” A webpage available at: http://www.spirent.com/Positioning-and-Navigation/What_is_Wi-Fi_Positioning 56 “The Evolution of Cell Phone Design Between 1983-2009”, A webdesignerdepot.com report dated May 22nd, 2009, available at: http://www.webdesignerdepot.com/2009/05/the-evolution-of-cell-phone-design- between-1983-2009/ TipSense, LLC (2014) “Poimapper”, An App Discovery Engine and Application review web page available at: http://appcrawlr.com/android/poimapper. United Nations Global Pulse (2013): Mobile Phone Network Data for Development. http://www.slideshare.net/unglobalpulse/mobile-data-for-development-primer-october-2013 Ushahidi (2012) Uchaguzi – A Case Study: http://blog.ushahidi.com/2012/10/24/ushahidi-kenya- evaluation-case-studies/ Vicent, J. P. A (2013) “WIFI Indoor Positioning for Mobile Devices, An Application for the UJI Smart Campus” A Dissertation for the Degree of Master of Science in Geospatial Technologies at the Jaume-I University dated March 2013, available at: http://run.unl.pt/handle/10362/9193 Walji, Aleem (2013) “What is science and what is delivery”, blog article, dated 19 May 2013, available at: http://blogs.worldbank.org/voices/what-science-and-what-delivery. Wesolowski, Amy; Eagle, Nathan; Tatern, Andrew J; Smith, David L; Noor, Abdisalan M; Snow, Robert W; Buckee, Caroline O (2012): Quantifying the Impact of Human Mobility on Malaria. Author manuscript available at http://europepmc.org/articles/PMC3675794. Full article available at Science, Oct 2012. Wikipedia (2014) “GPS Exchange Format”, last modified 30 April 2014, available at: http://en.wikipedia.org/wiki/GPS_eXchange_Format Wikipedia (2014) “Keyhole Markup Language”, last modified 17 April 2014, available at: http://en.wikipedia.org/wiki/Keyhole_Markup_Language Wikipedia (2014) “Shapefile”, last modified 27 March 2014, available at: http://en.wikipedia.org/wiki/Shapefile Wilson, Tracy V. "How GPS Phones Work" 24 October 2005. HowStuffWorks.com article available at: http://electronics.howstuffworks.com/gps-phone.htm The World Bank (2014) “Information and communication technology for Forest Law Enforcement and Governance; Lessons from a Two-Country project in LAO PDR and Moldova,” report. 57