J.-W. Chang and H.-M. Kang (Korea)
Parallel CBF method, cluster-based parallel architecture, shared-nothing architecture
To solve the 'dimensional curse' problem, the cell-based filtering scheme has been proposed, but it shows a linear decreasing on performance as the dimensionality. To cope with it, we propose a parallel cell-based filtering scheme using a horizontally-partitioned technique. In order to maximize its retrieval performance, we construct our parallel CBF scheme under a SN(Shared Nothing) cluster based architecture. In addition, we present data insertion algorithm as well as range and k-NN query processing ones. Finally, we show that our parallel CBF scheme achieves good retrieval performance in proportion to the number of servers and that it outperforms a parallel version of the VA-File when the dimensionality is over 10.
Important Links:
Go Back
