Roogle: Supporting Efficient High-Dimensional Range Queries in P2P Systems
Multi-dimensional range query is an important query type and especially useful when the user doesn’t know exactly what he is looking for. However, due to improper indexing method and high routing latency, existing schemes cannot perform well under high-dimensional situations. In this paper, we propose Roogle, a decentralized non-flooding P2P search engine that can efficiently support high-dimensional range queries in P2P systems. Roogle makes improvements on both indexing and routing. The high-dimensional data is indexed based on the maximum or minimum value among all dimensions. This simple indexing method performs rather well under high-dimensional situations and tolerates data points with missing values or different dimensionality. To speed query routing, Roogle is built on top of our proposed structured overlay – Aurelia, which has better routing performance by exploiting node heterogeneity. Aurelia also guarantees the data locality and efficiently support range queries. Experimental results from simulation validate the scalability and efficiency of Roogle.
Unable to display preview. Download preview PDF.