To determine an event region, it su?±ces to detect the sensor nodes near
or right on the boundary of the event. These edge sensor nodes, if not faulty,
usually have abrupt changes in readings compared to other sensors not in the
region. Thus, the problem of boundary detection for sensor networks is to
find the location of these edge sensors. The boundary detection algorithms
for sensor networks can be categorized into four classes:
??? Localized edge detection: Based on a localized algorithm or the data collected
from its neighbors, a sensor node determines whether it is positioned
on or near a boundary. Designing a localized edge detection algorithm is
challenging, but it results in low energy consumption compared to other
approaches due to localized communication.
149
Ren-Shiou Liu, Lifeng Sang, and Prasun Sinha
??? Centralized edge determination: A powerful node or the sink gathers data
from all sensor nodes and, through examining these data, edges are accurately
determined. Nevertheless, if the sensor field spans across a large
geographical area, the cost of collecting data from all sensor nodes would
be very high.
??? Hierarchical edge estimation: To reach a balance between energy consumption
and accuracy, several hierarchical edge estimation schemes have been
proposed. Sensor nodes are organized into multiple hierarchies, and a parent
sensor in a higher level of hierarchy tries to estimate the event boundary
based on the data gathered from its child sensor nodes.
Pages:
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264