Efficient computation of multiple sliding window skylines on data streams

Abstract

Given a set of objects, the skyline query returns those objects which are not dominated by other objects in the same dataset. An object o dominates another object o' if and only if o is strictly better than o' on at least one dimension and o is not worse than o' on the other dimensions. Although the skyline computation has received considerable attention recently, most techniques are designed for static datasets. However, in many applications, skyline computation over data streams is highly required and techniques for static datasets are inefficient or useless in data streams. Since data streams are unbounded, queries on them generally have sliding window specifications. When many concurrent users ask queries over a data stream, the sliding windows that different users are interested in can vary widely. In this paper, we propose skyline computation techniques for processing multiple queries against sliding windows efficiently. We first present two naive techniques called MSO and SSO, then propose a hybrid method called SMO which exploits the advantages of both MSO and SSO. The experimental results show that SMO processes skyline queries efficiently.