Feedback based continuous skyline queries over a distributed framework

Abstract

Continuous skyline query processing is becoming wide spread. Most of the work done in this field is focused to process skyline queries on a single machine. Our focus is to process continuous skyline queries over data streams, where data is arriving at server in the form of continuous updates from multiple distributed input sources. A single machine solution to run continuous skyline queries over streaming data is not very scalable. Moreover, streaming data arriving from multiple sources can overwhelm server’s computing power, specially if the skyline queries are involved to compute high quality multidimensional skyline points. We propose a three layer solution to compute continuous skyline points. A bottom layer in our approach sends the local skyline points to the middle layer, which after receiving feedback from the server filters the false-positives, and produces the semi-global skyline points to be sent to the server for global skyline. Our approach being scalable distributes the workloads across the network on multiple machines and reduces the number of unnecessary data points to be sent to the server, allowing it to produce qualitative skyline points.