Efficient Strategies for Continuous Distributed Tracking Tasks

While traditional databases have focused on single query evaluation in a centralized setting, emerging applications require continuous tracking of queries on data that is widely distributed and constantly updated. We describe such scenarios, and describe the challenges involved in designing communication-efficient protocols for the tracking tasks we define. We outline some solutions to these problems, by abstracting a model of the communication system, defining the tracking tasks of interest, and building query-tracking schemes based on three guiding principles of minimizing global information, using summaries to capture whole data streams, and seeking stability of the protocols.