Continuous monitoring of skycube queries over sliding windows in data stream environment

Given a dataset S of d-dimensional points, skycube is a union of skyline results of all the non-empty subsets of d dimensions. The computation of skycube has received considerable attention in conventional databases, but the existing methods are inapplicable to stream applications involving numerous long-standing queries because they assume static data rather than dynamic data that continuously arriving or expiring, and they focus on one-time execution that returns a single skycube in contrast to constantly tracking skycube changes. In this paper, we study continuous monitoring of skycube queries over a fixed-size window W of the most recent data points. The window size can be expressed as time units. We propose the architecture of skycube queries in stream environment, and present three techniques of processing modules to accomplish pre-processing, updating and computation respectively. Extensive experiments show that our techniques can monitor the arrival of data and track skycube changes in stream environment constantly and efficiently, guarantee the updating of skycube real-time, and keep the results exact and effective.