Database Reference
In-Depth Information
ted. Maps are updated incrementally at the base station, which matches
the maps against the predefined patterns and reports the events.
6.3 Approximation Queries
Approximation is a common approach to achieve data reduction if
no exact answers are required. In applications such as military surveil-
lance, in order to achieve high responsiveness, it is desirable that the
system quickly returns a coarse result for the overall battlefield, and
then refines the result on specific areas iteration by iteration, similar to
zoom in operation on image viewing. [42] proposes data shuing algo-
rithm to tackle this problem. The purpose of data shuing algorithm
is to determine the reporting order of sensor readings so that the al-
gorithm is able to compute the approximated result from the first few
reports and then refine the result on receiving the additional reports.
[43] proposes another approach, called the multi-resolution compression
and query (MRCQ). In MRCQ, sensors are conceptually organized in a
layered structure where a node may work in multiple layers. The raw
readings are stored at the bottom layer in the form of matrices. A bot-
tom layer node applies the discrete cosine transform on its matrix and
gets i) the compressed representation of its matrix and ii) the stored
data. The major characteristics of the matrix are stored in i), which
is transferred to the upper layer node as the approximation, i.e., the
low resolution result. In order to recover the original data, both i) and
ii) are necessary. In case refinements are required later, a bottom node
caches the stored data locally. The same procedures are repeated in the
intermediate nodes, until the compressed data reach the base station.
In order to refine the query result, the base station requests the stored
data cascading down. MRCQ achieves different resolutions of the result
by stopping the cascading at different levels of the layered structure.
7. Conclusion
In this chapter, we survey the recent works in WSNs under a database
point of view. Starting by investigating the characteristics of sensors,
it addresses the common issues in designing algorithms in the WSNs.
It follows to introduce the network topologies of WSNs, which are the
foundation of data transmission. Then we investigate the general data
acquisition and aggregation frameworks and shows the common query
languages and query types in WSNs. More specific applications, query
types, and various approaches are also discussed.
