Rural areas are defined, in part, by their lack of infrastructure. In many parts of this column author Kurtis Heimerl's home state of Alaska, communities lack power infrastructure and learn to set up and use generators as a solution; in column author Kashif Ali's home country of Pakistan, filters are deployed to provide clean water in areas without a generalized potable source. Building sustainable infrastructure solutions for these types of placeswhich exist in some way in all countries with substantive rural areasis a complex problem. While many (if not most) issues are not technical in nature (instead involving things like local buy-in), research has shown designing the technical portion of an intervention with a deep understanding of the local context is important to the success of the project.2 This understanding includes the capabilities, affordances, knowledge, and infrastructure available within a community; as well as the ability to leverage that understanding to build technologies that are inexpensive, robust, and understandable by rural users.
OpenCellular (OC),8 an initiative of the Telecom Infrastructure Project (TIP),10 is an open source hardware and software infrastructure platform that implements a cellular access point, either GSM or LTE. It was created to provide coverage to the hundreds of millions of people living in areas currently without cellular coverage and includes a variety of optimizations across hardware, software, and business models to better fit the needs and capabilities of these rural communities. OC's design leverages the extensive research literature on operating in rural areas as well as our personal experiences in places like Kenya, Indonesia,5 and the Philippines.1 In this column, we describe four important design choices in the context of the OpenCellular project: platform, power and networks, business models, and customizability. While the OC developers are still conducting their initial deployments and the eventual impact of the project is not known, we believe these designs are crucial to the long-term success of the platform.
The first design choice is what device to leverage for your intervention. A common viewpoint is that we live in a world of ubiquitous access; everyone has access to at least basic cellular connectivity. Unfortunately, this is not the case. Recent estimates have GSM cellular coverage at 80% of the world.11 These networks have stopped expanding as operators instead invest in higher-revenue urban 3G or LTE installations. Ironically, while ubiquitous connectivity may be a myth, the ubiquity of the cellular phone itself is not. In our travels, we have yet to come to a location where these devices were not already woven into the fabric of the community. For example, when we first set up our network in Papua, Indonesia, the network recorded over 3,000 unique mobile phones despite being a four-hour drive from the closest cellular network.9 It may seem counterintuitive for there to be cellular phones where there is no network, but the devices are more than just phones. They are also rugged rural entertainment consoles, with built-in battery power, speakers, and headphone jacks. Where there is no cell network there is often no radio and instead, people bring their entertainment with them. All the while, many members of the rural community regularly travel to dense urban areas where there is coverage and use their phones there.
This situationa large installed base of mobile phones in areas without any cellular accessprovides an opportunity for novel connectivity solutions. OC is designed to meet the current capabilities and needs of users in these communities. The first revision, OC-SDR (OpenCellular Software Defined Radio), is a GSM cellular base station with support for basic GPRS/Edge data connectivity. The second revision, OC-LTE is an LTE-based extension of the platform. These two access points allow locals to get on the network using their existing phones. Then the operator can determine various services the community would like (for example, using IVR for low-literate populations) and eventually upgrade to broadband as economics and availability of other associated infrastructure (such as power for smartphones) matures.
Power and Backhaul
To keep costs low, one must leverage what infrastructure is available in the remote communities. Surana et al.9 found that grid power was unreliable in rural India, producing both brown-outs and voltage spikes capable of destroying equipment. They also learned that the lack of general Internet access was a huge issue in diagnosing failures. OpenCellular mitigates these problems by building backhaul and power solutions into the system. The platform includes power cleaning, variable input voltages, and support for Power-over-Ethernet (PoE). The system also supports PoE's power sourcing equipment (PSE) standard, allowing the OpenCellular access point to "daisy-chain" power to phone chargers or even another OC instance. To better support renewable energy sources, OC also features two internal solar charge controllers for external sealed lead acid and internal lithium ion batteries. The internal lithium-ion battery, with a built-in UPS system, works as a backup to allow the system to fail gracefully when the local grid fails. This suite of power cleaning and support systems allows OpenCellular systems to coexist in the chaotic reality of rural power.
While ubiquitous connectivity may be a myth, the ubiquity of the cellular phone itself is not.
For network, OpenCellular includes a built-in out-of-band satellite back-haul. This is not designed to be used for daily communications (as the cost could be prohibitive) but instead to reduce the cost of debugging a failed network. Operators will not need to send an engineer out to the system (see Figures 1 and 2) if the backhaul has failed and instead can use the satellite link to gather critical data about the issues with the system, such as the status of individual hardware components or stability of power or backhaul subsystems. This collected information can also be relayed to the local maintainer (potentially by the access point itself) who can then assist in maintenance (for example, cleaning the solar panels) even when the main power and backhaul is down.
The highly variable power and network situation in rural areas invites another design imperative: OC optimizes for low cost over reliability. In rural areas, the vast majority of downtime will be due to failures in the related infrastructure, including power and network. Increasing the reliability of the OC system itself will only marginally increase the overall uptime of the network. Instead, it is better to be cheap and easy to replace or repair.
Thirdly, you must design your intervention to sustain. That involves creating business models that encourage local participation and support. While existing business models for cellular exist (and are quite lucrative), many rural areas remain underserved. Even with the power and network advances mentioned here, it is always going to be expensive to send engineers and equipment to remote parts of the country for installations and maintenance. For coverage to reach the entire world in a sustainable manner, OC must support a variety of different business models that can cover the diversity of the rural world. OC enables two key business models: community-focused and traditional (see Figures 3, 4, 5).
In many remote rural areas, much of the infrastructure is owned and operated by local agents. Co-production7 is one model that makes use of this fact, with core infrastructure such as power and water built and operated in close collaboration with the populations served. Galperin4 suggested extending local ownership to cellular, with smaller local telecoms providing service. OpenCellular supports these local business models. CommunityCellularManager (CCM)3 is one OC-supported software suite that allows small communities to operate their own small OpenCellular-based networks. It provides both client and cloud support for management, routing, and interconnect. With CCM, the local rural community can then personally maintain and operate the network.
OC can also operate as a traditional cellular access point, supporting a variety of open and closed-source base-bands and cellular stacks that when configured can connect to traditional core networks (EPC, in case of OC-LTE). This allows existing incumbents to utilize OC to decrease the cost of their rural installations while requiring minimal changes to the rest of their infrastructure.
While OpenCellular has been designed with our own rural experiences in mind, the appropriateness of specific technologies will vary widely across areas. For example, OC's built-in satellite backhaul may be appropriate where wireless is used for backhaul6 but overly expensive if the installation is backhauled over a more robust medium such as fiber. Similarly, new technologies such as 5G or LoRa6 may see rapid uptake in the next few years and overtake LTE. For this reason, OpenCellular needs to be extensible and customizable to enable new access models and new technologies. Enabling this customizability in Open-Cellular consists of two distinct design choices: modularity and open source.
The OpenCellular hardware is designed in a modular fashion, with individual components like the power subsystem separated from the rest of the system. This allows organizations building OC devices to "pick and choose" the features that are relevant to their context. For instance, daughter-boards allow for the base system to expand to radio technologies outside of GSM and LTE and into future technologies like LoRA. Similarly, the components listed here, such as the built-in battery backup, can be removed from the board during manufacturing to save cost if the grid power is expected to be clean. Lastly, new subsystem modules, such as an inexpensive WiFi hotspot, could be added before manufacturing.
OpenCellular enables two key business models: community-focused and traditional.
Lastly, we need to enable these new designs to scale. To do this, OC has been released as open source hardware, including all the schematics, layout, CAD, BoM, and firmware needed to enable large-scale industrial manufacturing. Additionally, all testing software is also open source, so anyone (either an OEM/CM or university students) can replicate and produce OC hardware at the same quality level as current industrial partners; the software is available at https://bit.ly/2xREpUD. Because of the rich suite of software and hardware supporting the platform, motivated organizations can extend OpenCellular to meet their needs and then manufacture the equipment at scale locally in the country, increasing local capacity, reducing costs, and stimulating the local economy.
Designing infrastructure for rural areas that can leverage the local contextthe skills, knowledge, and affordances of the communities that live thereis a difficult task. With OC we chose to focus on four key elements. The first is the user platform, ensuring the intervention uses technologies that are common and available. The next is ensuring we support a diverse range of power and network technologies as well as business modelssome of the key differentiators between rural communities. Lastly, we recognize the limitations of our own designs and capabilities by releasing OC as open source hardware, complete with all of the designs necessary to modify and manufacture the solution. It is our hope that, through the lens of OpenCellular, readers can see how to similarly design their own interventions with these concerns in mind. While we are hopeful that OpenCellular itself brings connectivity to the world, rural access problems always require holistic solutions that are driven by the needs, abilities, and limitations of the communities themselves. As such, we also aspire to allow motivated individuals and organizations take OpenCellular, expand it to fit their needs, and create a diverse ecosystem of rural access solutions. Join us at https://bit.ly/2JoFa8Y to participate in this process.
1. Barela, M.C. Towards building a community cellular network in the Philippines: Initial site survey observations. In Proceedings of the International Conference on Information and Communication Technologies and Development, 2016.
2. Chetty, M., Tucker, W., and Blake, E. Developing locally relevant software applications for rural areas: A South African example. In Proceedings of the 2004 Annual Research Conference of the South African Institute of Computer Scientists and Information Technologists on IT Research in Developing Countries, 2004.
3. CommunityCellularManager; https://bit.ly/2Lvgxs5.
5. Heimerl, K. et al. Local, Sustainable, small-scale cellular networks. In Proceedings of the Sixth International Conference on Information and Communication Technologies and Development: Full PapersVolume 1, New York, NY, 2013.
8. OpenCellular Telecom Infra Project; https://bit.ly/2hJE0u2
10. Telecom Infrastructure Project; https://bit.ly/2sF1hlj
The Telecom Infra Project is offering grant opportunities for organizations looking to use OpenCellular; see https://oc.telecominfraproject.com/opencellular-grant-program/ for more information. Applications will be due in Fall 2018.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2018 ACM, Inc.