J.-W. Chang and Y.-K. Kim (Korea)
Parallel CBF method, cluster-based parallel architecture, shared-nothing architecture
To efficiently retrieve high-dimensional data in multimedia database and data warehousing applications, the VA-file and the CBF (Cell-based filtering) method have been proposed. But they show a linear decreasing on performance as the dimensionality. To cope with the problem, we, in this paper, propose a new parallel CBF method for shared-nothing (SN) cluster-based parallel architecture. In addition, we devise data insertion, range query processing, and k-NN query processing algorithms that are suitable for the SN parallel architecture. Finally, we show that our parallel CBF method achieves good retrieval performance in proportion to the number of servers in the SN parallel architecture, compared with the conventional CBF method.
Important Links:
Go Back
