Both the Internet and phone system exhibit deep assumptions about continuous connectivity that are simply not true in much of the world, including rural areas and developing regions in particular. Intermittent network connectivity is a fact of life in most developing countries due to a variety of factors including high usage costs, frequent power outages, network failures and the use of delay tolerant networks. Traditional networked applications are not designed to work in intermittent environments. Although work on delay-tolerant networking (DTN) has created viable low-level networking protocols that support intermittency, it remains difficult to write applications that tolerate disconnections well.
This project proposes Intermittent Aware Network Architecture (IANA), a new network platform that can enable a spectrum of intermittent-aware applications. This work will develop several novel applications that show the power of intermittent networking in developing regions and that test and validate the IANA platform. The proposed applications include intermittent versions of web search, collaborative wiki software, automatic teller machine (ATM) support, and voice-messaging cellular phones. These are challenging applications that cover a range of scalability, data sharing, and security issues. Pilot versions of all the four applications will be deployed in India or Africa to understand how well the platform enables real applications in challenging environments.
An overarching goal in the design of IANA is to determine the underlying principles in building a generic network architecture that can be both adopted across different types of intermittent networks and be used as a common platform across several intermittent applications. An important related challenge is to determine an appropriate and generic Application Programming Interface (API) for intermittent applications. In pursuit of this larger goal, this work will 1) classify the kinds of intermittent networks, which facilitates appropriate responses to disconnection; 2) develop an intermittent-aware overlay network above DTN that enables application-specific in network optimizations, which in turn enables both better performance and better data distribution and sharing; and 3) develop a range of new APIs that move beyond low-level connection APIs (sockets) to support intermittent sessions, data collection and distribution, data replication and consistency, and mobility.
Intellectual Merit: This work will result in a substantial increase in the ability to create applications that work well in intermittent environments. The key contributions include an overlay network to help manage replication, mobility, aggregation and sharing, new higher-level APIs validated by real applications, and an evaluation of the effectiveness of the new platform via large-scale deployments in developing regions.
Broader Impact: The target applications have direct value in bridging the digital divide. The collaboration software empowers rural content creation and sharing and facilitates content distribution for education, health care, and emergency response. ATM software can help address the economic inefficiencies due to limited cash flow and limited access to banking services. The voice-messaging phone can extend the impact of the cellular revolution to rural areas and to low-literacy users. The PIs have a history of impactful applications in developing regions and expect similar impact from the proposed deployments and from the free, open-source release of the apps. Field work and new curriculum will also promote development-aware graduate students with strong international exposure. The applications and deployments facilitate inspiring internships, which will encourage diversity and multi-disciplinary research.
