Regresar

D-FOPA: A dynamic final object pruning algorithm to efficiently produce skyline points over data streams

Abstract:

Emerging technologies that support sensor networks are making available large volumes of sensed data. Commonly, sensed data entries are characterized by several sensed attributes and can be related to static attributes; both sensed and static attributes can be represented using Vertically Partitioned Tables (VPTs). Skyline-based ranking techniques provide the basis to distinguish the entries that best meet a user condition, and allow for the pruning of the space of potential answers. We tackle the problem of efficiently computing the skyline over sensed and static data represented as Vertically Partitioned Tables (VPTs). We propose an algorithm named D-FOPA (Dynamic Final Object Pruning Algorithm), a rank-based approach able to dynamically adjust the skyline by processing changes on values of sensed attributes. We conducted an empirical study on datasets of synthetic sensed data, and the results suggest that D-FOPA is not only able to scale up to large datasets, but reduces average execution time and number of comparisons of state-of the-art approaches by up to one order of magnitude.