Scaling Community Cellular Networks

Scaling Community Cellular Networks with CommunityCellularManager Shaddi Hasan, UC Berkeley; Mary Claire Barela, University of the Philippines, Diliman; Matthew Johnson, University of Washington; Eric Brewer, UC Berkeley; Kurtis Heimerl, University of Washington https://www.usenix.org/conference/nsdi19/presentation/hasan This paper is included in the Proceedings of the 16th USENIX Symposium on Networked Systems Design and Implementation (NSDI ’19). February 26–28, 2019 • Boston, MA, USA ISBN 978-1-931971-49-2 Open access to the Proceedings of the 16th USENIX Symposium on Networked Systems Design and Implementation (NSDI ’19) is sponsored by Scaling Community Cellular Networks with CommunityCellularManager Shaddi Hasan Mary Claire Barela Matthew Johnson UC Berkeley University of the Philippines, Diliman University of Washington Eric Brewer Kurtis Heimerl UC Berkeley University of Washington Hundreds of millions of people still live beyond the cov- tion infrastructure, to provide commercial mobile service. erage of basic mobile connectivity, primarily in rural ar- In spite of their size, incumbent mobile network operators eas with low population density. Mobile network opera- (MNOs) are capital constrained and struggle to justify in- tors (MNOs) traditionally struggle to justify expansion into vestment in rural areas with marginal business cases when these rural areas due to the high infrastructure costs neces- compared to more profitable and lower-risk urban markets. sary to provide service. Community cellular networks, net- Community cellular networking provides one alternative works built “by and for” the people they serve, represent approach for bringing mobile connectivity to these under- an alternative model that, to an extent, bypasses these busi- served populations [2, 28]. Community networks, networks ness case limitations and enables sustainable rural cover- built “by and for” the people they serve in a community- age. Yet despite aligned economic incentives, real deploy- centric, often cooperative, fashion [8, 52, 58], allow for cre- ments of community cellular networks still face significant ative localized schemes for sustainable operation that work regulatory, commercial and technical challenges. even when covering low-income, remote, or sparse commu- In this paper, we present CommunityCellularManager nities. The strength of the community network model is its (CCM), a system for operating community cellular networks ability to flexibly adapt to local needs using local resources at scale. CCM enables multiple community networks to op- and local insights. A network serving one community can erate under the control of a single, multi-tenant controller provide services suited to that community, and can institute and in partnership with a traditional MNO. CCM preserves their own policies that would not make sense in a nation- flexibility for each community network to operate indepen- scale network (for example, free local SMS mailing lists). dently, while allowing the mobile network operator to safely They can also tailor their network deployments to leverage make critical resources such as spectrum and phone numbers local capacity, such as existing towers and power systems, available to these networks. We evaluate CCM through a lowering deployment costs. Yet even where communities multi-year, large-scale community cellular network deploy- could run profitable networks, they face high barriers to en- ment in the Philippines in partnership with Globe, the largest try in acquiring expensive and exclusive spectrum licenses, MNO in the country, providing basic communication ser- as well as operating licenses necessary to secure phone num- vices to over 2,800 people in 17 communities without requir- bers and interconnection with the global phone network. ing changes to the existing regulatory framework, and using We are left with an unfortunate market failure: incum- existing handsets. We demonstrate that CCM can support bent MNOs hold resources and capabilities, in the form of li- independent community networks with unique service offer- censes and interconnect, necessary to provide rural coverage, ings and operating models while providing a basic level of but can’t justify investment in underserved areas. Concur- MNO-defined service. To our knowledge, this represents the rently, community networks could provide sustainable ser- largest deployment of community cellular networks to date. vice, but lack a combination of the commercial, legal, and technological capacity to operate in a telecommunications 1 Introduction regime designed for large carriers. We argue that rather than standing in opposition to each other, incumbent operators Despite the global expansion of mobile networks, millions and community networks have an opportunity for cooper- still live without access to basic voice and SMS connectivity. ative interaction, leveraging the strengths of each to fulfill A key reason for this is that most countries effectively al- mutually beneficial business and social goals. Fundamen- low only a small set of nation-scale actors to hold the neces- tally, such a partnership requires a technology stack enabling sary operating and spectrum licenses, as well as interconnec- collaborative resource management and provisioning of com- USENIX Association 16th USENIX Symposium on Networked Systems Design and Implementation 735 mon services between MNOs and community networks. Community Networks. Researchers have explored wire- In this paper, we present CommunityCellularManager less community networks [31] for years with contributions (CCM), a solution for operating community cellular net- such as specific wireless technologies [18], topologies [59, works at scale within the existing telecom ecosystem. CCM 58], and mesh protocols [5] more appropriate for their unique enables multiple community networks to operate under the constraints. The above systems all use WiFi for access, pri- control of a single, multi-tenant controller, preserving flexi- marily because of low cost and use of unlicensed spectrum. bility for each community network to implement their own While these networks have shown to have the potential to unique services while allowing them to share MNO re- empower local communities [14] and create more resilient sources, such as spectrum and phone numbers. For MNOs, networks [8], this limits these networks to higher-end de- CCM resolves the administrative problems they face grant- vices and small coverage areas without mobility. ing many small third parties access to their core network by More recently, researchers and practitioners have built aggregating their traffic behind one logical point of intercon- community cellular networks [2, 28, 51, 12]. Ideologically nection, while providing a point of control to mitigate the aligned with the goals of community networking, these net- risk they take on by sharing their resources and licenses. We works use cellular protocols to provide wide-area coverage evaluate CCM in a deployment to expand the service of an to basic phones at lower cost. Unfortunately, they remain MNO into unserved areas via community networks using the limited by the technical affordances of cellular, including the MNO’s spectrum and interconnect (Section 5). need for licensed spectrum, phone numbers, and intercon- The contributions of this paper are as follows. First, we nect with the global phone network. CCM provides a plat- describe the design and implementation of CommunityCel- form for community cellular that resolves these concerns. lularManager, a system for deploying rural community cellu- Mobile cores. Cellular core network standards are devel- lar networks at scale by facilitating cooperation and resource oped by the 3GPP [46]; recent work focused on improving sharing with traditional mobile network operators. Second, scalability and reliability of core networks [50, 6, 49]. Sim- we present a multi-year, large-scale community network de- ilar to SoftCell [33], CCM is a clean-slate architecture, and ployment in the Philippines that provides basic communica- was designed for public cloud environments like ECHO [42]. tions service to 17 communities and over 2,800 monthly ac- CCM differs in its focus on shifting administrative bound- tive users, powered by CCM and in partnership with Globe, aries of the mobile core to enable cooperation between MNO the largest MNO in the country. We further demonstrate and community networks. CCM’s ability to support key use cases necessary for rural community cellular networks, such as offline operation, as 3 Design Goals well as its ability to support independent local services on a subset of the community networks in our deployment. CCM is targeted to a specific but important use case: extend- ing mobile network service provided by an incumbent MNO in a cooperative partnership with community cellular net- 2 Related Work works. Prior work [28, 51] has shown that community cellular networks can effectively serve areas unserved by MNOs. Rural Access. Access to connectivity supports numerous Such cooperative arrangements [11, 39] promise to unblock services, such as health [37], education [35], and finance [9]. regulatory and commercial barriers faced by community cel- However, access remains unequal, with connectivity signif- lular networks, such as access to spectrum. The architectural icantly better in dense, urban, and developed environments rigidity of mobile networks, however, makes achieving our than in sparse, rural, and developing ones. In attempts to re- federated

Load more