Database Reference
In-Depth Information
investigate query optimization in the context of distributed query processing in
a Sensor Grid Database.
Caching is a promising technique, widely used for query optimization. However,
as the query cost model in sensor grids depends on different parameters than in the
centralized setting, the cache management and the cache replacementpolicy needs
to beadapted.Inthis paper,weproposetwotechniques:First,weproposea seman-
tic cache, which is an optimized data structure for improving the query eciency.
Generally speaking, semantic cache stores the query result and query semantic in
the cache, for responding the future query; Second, we propose a cache replace-
ment policy, accounts for the location dependent queries cost model, keeping in
the cache the most relevant data for better eciency [1].
However we find that traditional approaches have a limitation. For example,
the paper in [4] proposes the approach of using a Clustering Semantic Cache
(
CSC
), which considers semantic and time factors, however there is only quali-
tative analysis and no quantitative description. In a previous paper [5], the cost
model of the branch and bound and Greedy Dual-Size Frequency (
B
&
B
GDSF
)
considers semantic and time factors, which provides the advantage of quantita-
tive analysis, but , since the geographical context where rather static, the model
did not consider any dynamic location factor. In another paper [2], the cost model
of Collaborative Spatial Data Sharing (
CSDS
) considered a dynamic location
factor, but did not consider semantic and time factors. An advanced application
needs the new approach of semantic cache technology to synthetically consider
the multiple factors that are present.
How to organize the data in a mobile sensor device for access and query;
how to calculate the cost of a cache item for replacement and storage; and
how to synthetically consider semantic, time and location factors, and attain an
optimum cost model of a semantic cache are very challenging issues.
Our contribution is summarized as follows. We propose an approach that has
the following characteristics: (i) Based on the Nash Equilibrium scheme, an appli-
cation of three scalar coecients is proposed in term of the analysis of the relation-
ship among semantic, time and location factors. (ii) After that, we specialize and
summarize the general Nash Equilibrium scheme utilizing the equal correlation
coecient among the new and old vectors to find the point of Nash equilibrium
and resolves the scalar coecients. (iii) It uses these scalar coecient to attain an
optimum cost model of the semantic cache, for query optimization in the context
of distributed query processing. (iv) We emphasize that this method can extend
to any finite-dimension, which are other schemes cannot do.
2 Related Work
We choose three different ways for choosing and calculating a cost model for the
replacement of semantic cache segments.
1) Clustering Based Semantic Cache
CSC
:
The paper [4] proposes the
approach of a Clustering Semantic Cache (
CSC
), which considers semantic and
time factors but only performs qualitative analysis, with no quantitative descrip-
tion. Early in the algorithm the technique of Clustering Semantic Cache divides
Search WWH ::
Custom Search