The recent loss of lives, and traffic jams stretching for tens of miles as hurricanes Rita and Katrina approached the Gulf Coast demonstrate the difficulty of evacuating urban areas. Mass evacuations are among the most difficult problem areas in Transportation Science because they violate key assumptions underlying traditional theories, e.g., Wardrop equilibrium among selfish commuters. A key challenge in this domain is to develop an understanding of non-equilibrium traffic dynamics over transportation networks to aid in the design of emergency traffic management techniques. This is a formidable task due to the data-intensive nature of the problem, and the semantic gap between current database management systems and transportation science. The goal of this project is to research novel and scalable data management concepts in partnership with the development of novel transportation science models and theories to understand emergency traffic. New collaborative computer science research is proposed to probe innovative database concepts underlying network non-equilibrium dynamics data and queries.
Intellectual Merit These approaches to time-variant graphs and spatio-temporal database support for flow networks significantly differ from the traditional approaches in the database literature. Th project is expected to create innovations in the following areas: 1) graph-aggregates, a novel representation of time-varying graphs, 2) database support for flow network operations, e.g. min-cut, and max-flow, 3) the proposed database concepts will be designed and evaluated in collaboration with domain scientists and professionals using grand challenge problems (e.g., emergency traffic management) and datasets (e.g. large urban evacuation scenarios, population distributions, and flow networks). The hope is to significantly enhance scientists' ability to understand and manage non-equilibrium network behavior, not only in Transportation Science, but also in many other important domains including logistics, telecommunication networks, electric power grids, and distribution networks for gas, water, etc.
Broader Impact Teaching materials (e.g., slides, software prototypes) to facilitate incorporation of research results in courses and classroom activities will be prepared. This project will broaden the participation of underrepresented groups at many levels: one PI has a track record of participation in summer institutes involving undergraduate students from historically black colleges and universities. If successful, the results wil benefit society by reducing evacuation time, which may save lives and reduce exposure of vulnerable populations in the face of man-made and/or natural disasters.
