Abstract
We propose a novel architecture, Grassroots, for collecting, disseminating, and querying information about the physical world. The Grassroots architecture consists of a massive number of individual agents called dataflies, which move around, collect, summarize, and query information about their immediate physical environments. We make use of this architecture in developing a system for collecting and disseminating real-time traffic information. Such a system would prevent traffic congestion from building up. Today a number of commercial solutions exist for disseminating traffic information (e.g., Traffic.com, Metrocommute, EtakTraffic). However, these solutions are plagued by prohibitive deployment and maintenance cost that prevents widespread deployment. As a result, these solutions cover select highways while leaving out major fractions of the highways. In this paper we present our vision of a low cost, highly scalable zero-infrastructure system for collecting and disseminating information about traffic conditions. Our goal is to devise a system that can cover the entire road network of US. We describe the proposed architecture for the system, and evaluate it using realistic model of transportation network and realistic traffic conditions.