WOLAP: Wavelet-Based Range Aggregate Query Processing

The Discrete Wavelet Transform has emerged as an elegant tool for data analysis queries. It was not until the time we proposed a new wavelet technique, ProPolyne, for fast exact, approximate, or progressive polynomial aggregate query processing that data did not have to be compressed, unlike most of the prior studies in this area. In this paper, after reviewing our ProPolyne technique in more depth with more intuitive and practical discussions, we address its inefficiency in dealing with scientific datasets due to the cube sparseness, subsequently, we propose a new cube model, CFM, to enhance ProPolyne’s both space and query efficiency. While ProPolyne assumed storing the data as large data frequency distribution cubes, CFM organizes the data as a collection of smaller fixed measure cubes to reduce the overall query and storage costs. We combine both cube models in an integrated framework, called WOLAP, for efficient polynomial aggregate query processing. We further enhance WOLAP by proposing practical solutions for real-world deployment in scientific applications. In particular, we show how to incorporate data approximation, how to improve wavelet filter selection, and how to work on datacubes with arbitrary domain sizes.