COMET: An Efficient and Scalable Trajectory Data Management System

University Of Wisconsin-Madison, Madison WI

Investigators

Abstract

The use of location-aware devices, such as cell phones with GPS or objects with RFID (Radio Frequency Identification) tags, is exploding in a number of emerging spatio-temporal applications. Traditional database management systems (DBMS) are not designed to handle such applications, especially if the application requires managing a large number of moving objects. The goal of the COMET (Continuous Management of Evolving Trajectories) project is to design, implement, and build a database management system for managing large repositories of continuously evolving trajectories data sets. Since in these environments location updates are issued continually, the DBMS must support extremely efficient methods for dealing with updates. In addition, to allow querying on previous locations of the moving objects, the DBMS must keep track of past trajectories. As time passes these trajectories continue to increase in length, and with large and often increasing number of moving objects, the database size can increase dramatically. Consequently, the backend DBMS must deploy scalable techniques to deal with increasing data sizes, and increasing number of mobile objects. The key focus of the COMET project is on developing efficient and scalable methods for querying on past, present, and future locations of moving objects, and on scalable trigger mechanism in this environment. The expected results of this project may have a strong impact on emerging notification-based applications, such as emergency response systems, in which critical data needs to be disseminated to a physical mobile user based on the user's current and changing spatial location. The project will also train a number of graduate and undergraduate students. The project Web site (http://www.eecs.umich.edu/~jignesh/comet) will be used for making the COMET software, all developed applications, and real user movement data freely available to a broad research community.