3 QUERYING SENSOR FIELDS BY
USING QUADTREE BASED DYNAMIC CLUSTERS AND TASK SETS Çağhan ÇİMEN, 2003.
Keywords : Sensor networks, Data aggregation,
Data querying, Data dissemination, Task assignment, Quadtree addressing.
Recent
scientific improvements have enabled the development of new wireless devices
which are named sensor nodes. Since they are developed for independent usage
concept, such devices have different architecture from traditional wireless
network devices. Low cost, low power consumption, small size, data gathering
and conveyance functions are typical differences. In sensor networks all
network elements gather data from their surroundings around and send these data
to any end point (named sink) by using wireless medium with a collaborated work
by using limited battery power. Because of these properties of sensor networks,
knowing location of sensor nodes is an important design issue. As a result of
these, power aware location schemes which are also effective in tasking and
querying, must be developed for sensor networks.
In
sensor networks, it is more important to know the area where the data belong
to; instead of the identification of the sensor node which sends the data. A
coordinate or grid system can be used to address the nodes based on their
location. Instead of these systems, we proposed more efficient and advantageous
scheme in this thesis. This is quadtree based addressing.
Quadtree
addressing is the process of dividing the sensor field into four equally sized
sub areas until at most one sensor node remains in every sub area. We can
uniquely address a specific sensor node by addressing the quadtree sub area
where the sensor node is. A sensor node responds to a query if it is in the
rectangle that represents the quadtree sub area.
In
this study also a dynamic clustering scheme is introduced where a task set of
sensors is formed for each task by using quadtree based addressing. The number
of nodes in a task set depends on the task specifications. Hence, the sensed
data is retrieved from a sensor field in the level of detail specified by
users. Our performance evaluations show that quadtree based dynamic clustering
scheme reduces the number of sensors involved in a query. This supply an energy
conservation during querying.