A New Multidimensional Point Access Method for Efficient Sequential Processing of Region Queries - Advances in Computational Science and Engineering - page 217

MGSFC order value considers all the attributes. This method uses the MGSFC
order value as the clustering attribute.

5 Conclusions

The very famous one dimensional index structure B-tree was extended to the
B + -tree for supporting sequential processing [1]. But, there are no studies for
supporting of sequential processing in multidimensional PAMs.
To support query processing for region queries sequentially in multidimen-
sional space, the cells in multilevel and multidimensional space must be ordered
linearly. To do this, we can use the SFC with some extension. We explain that
the SFC has properties of locality and recursion basically. There are the Z-order
and Hilbert order as the very famous SFC. These orders assume the full grid
pattern for ordering of cells. But the full grid pattern yields a very large index
structure.
To overcome the disadvantage of the full grid pattern, there are many studies
of tree structured multidimensional PAMs. These PAMs integrate cells pointing
to same disk blocks as one cell [2,10]. We restrict the case that the PAMs use
the half splitting strategy and have the square like cells having the length of an
axis with equal or twice of the other axes. We define this type of a grid pattern
as the MGG. We also discussed that an upper level MGG is an abstraction of
a lower level MGG. We define the MGSFC, an extension of the SFC, for linear
ordering of cells for the some case of the MGG. We also discussed that the split-
ting strategy used for a multidimensional PAM must reflects the characteristics
of the MGSFC. This method is used for ordering of cells in a multilevel and mul-
tidimensional space to support sequential processing of region query. We present
sequential query processing algorithms using this approach. The approach dis-
cussed here can be applied not only to the MLGF but also to all PAMs managing
multilevel and multidimensional space with the MGG style.
We also proposed the detailed method for applying this approach to the
MLGF. We call the new structure the MLGF-Plus. After implementing this
approach, we will study the advantages and disadvantage of our approach via
extensive experiments.

Acknowledgements

This work was supported by the Korea Research Foundation Grant funded by the
Korean Government (MOEHRD, Basic Research Promotion Fund) (KRF-2007-
314-D00221), and was partly supported by the MKE(Ministry of Knowledge
Economy), Korea, under the ITRC(Information Technology Research Center)
support program supervised by the IITA(Institute of Information Technology
Advancement) (IITA-2008-C1090-0801-0040).