Supporting a Diversely Connected World via Community Cellular Networking

Abstract

Community cellular networks (CCNs), small-scale cellular networks owned and operated by members of the communities they serve, have the potential to provide sustainable wide-area coverage across a diverse range of contexts on the margins of the Internet. This dissertation first characterizes some of the operational challenges observed in an existing CCN, and then develops new designs for future community cellular systems to address the identified challenges.To ground the research, I gather a unique dataset characterizing the operation of a remote, expensive, data-only community LTE network. Through tight integration with the operator's infrastructure, the dataset has visibility to drill down to individual user behaviors in terms of both network traffic and prepaid credit purchases and transfers. I find that despite the network's constrained capacity, use is still highly unbalanced, with a handful of heavy consumers providing an outsized portion of network revenue. A large portion of the userbase only has intermittent access in this model. 45% of users are offline more days than online, and the median user consumes only 77 MB per day online and 36 MB per day on average, limiting consumption by frequently "topping up" in small amounts. The network is also only somewhat reliable, with downtime caused by software failures, network failures, and the generally harsh conditions on the rural edge. To address the reliability challenge, I observe that multiple hotspot providers were already operating in the area, and propose a new authentication architecture to allow fallback to an overlapping provider even when the user's home network is offline. This system, dAuth, is fully backwards compatible with standard off-the-shelf devices, making it feasible for immediate deployment. I design, prototype, and evaluate dAuth against existing private edge-core and cloud-core architectures, and find it comparable to the status quo while providing robust operation in the presence of failures and good scaling properties as the number of small operators increases. To address the challenge of providing sustainable access while avoiding intermittent availability for low-income users, I explore the design of community-based congestion-management policies and network management mechanisms through the lens of network capacity as a common pool resource. I present qualitative insights from a series of workshops and interviews exploring designs for community-scale networks with resource sharing. Participants expressed conflicting desires for preserving individual privacy while collecting longitudinal data to track the network's impact on the community, prioritization of high-value applications, equal link allocation between users, and human-mediated congestion management in lieu of automated enforcement. Overall my work provides new directions for the design of next-generation community cellular networks that can be more reliable and operate with a wider range of governance structures to sustainably serve a wider range of communities.