The ``last-mile'' broadband problem is arguably the most important long-standing techno-economic challenge faced by the telecommunications industry. Grassroots community wireless networks are emerging in various parts of the world as an interesting paradigm to connect a community of users to the nearest broadband wire. Community wireless networks (CWNs) have a number of characteristics distinctly different from both mobile ad-hoc wireless networks (MANETs) and traditional IP-based internetworks (eg: OSPF/IS-IS/BGP routed networks). They can also leverage a variety of low-cost capabilities available in residential areas to dramatically simplify configuration and boost effective network capacity. Our project focuses primarily on leveraging the design freedom made possible due to these differences.
Technical Merit: The PIs propose an broad experimental research program into the community wireless network (CWN) concept. The research will be conducted in the context of an operational medium-scale community network to be built in Troy, NY (upto 100 nodes in a 1 mile radius around RPI campus) and in collaboration with a commercial 802.11 access network provider in Harrisburg, PA (see attached letter of support). PIs divide our proposed investigation into the following parts:
First, at the physical/link layer we focus on the problem of auto-configuration and network capacity maximization. PIs propose to use a mix of cheap 802.11b, 802.11a and 802.11g links, leverage the directionality of streets, the GPS (latitude/longitude) location of homes, and the availability of backup channels (eg: telephone line or 915 Mhz ISM band radios for remote auto-configuration) to simplify the setup of nodes. The proposed hardware/software design combined with real-world deployment is a unique feature of our proposal.
Second, at the routing layer, The PIs will develop a broad framework and mechanisms for efficient explicit/multi-path routing using existing routing protocols (eg: OSPF, BGP-4, MANET protocols like DSR etc). Within this framework, PIs will investigate a number of hybrid geographic and topologically-informed routing to seamlessly interconnect organically evolving community wireless networks.
Third, PIs propose to model the growth and evolution of the network; develop strategies to characterize and improve the robustness against node failures. Finally, PIs will measure and characterize the real traffic and usage patterns of the operational community wireless network in the Troy, NY area from its earliest stages of deployment.
Broader Impact: The proposed research will provide real-world experimental validation of the nascent community wireless network (CWN) concept, and make significant contributions to the architecture of last-mile wireless networks, routing, traffic engineering and growth modeling of wireless networks. The project will involve the RPI ACM undergraduate chapter (see attached letter) and a large number of undergraduate students in the research. It will also be used as a context for term-projects in the RPI networking curriculum. We already see tremendous excitement in campus during our pilot activities this semester.
