This definition not only provides a definite guideline for discriminating edge
sensors from normal sensors, but also o?®ers an in-depth view for possible
performance evaluation metrics.
Though the design of a localized edge detection scheme is challenging, it
is still a topic of importance for sensor networks. If each node can detect
whether it is an edge sensor locally and reliably, the boundary can be easily
determined by traversing all the edge sensors of the sensor field of interest
and power consumption can, thus, be minimized.
Edge detection has been extensively studied in other fields such as image
processing literature and pattern recognition literature. These techniques can
section begins with three classical localized edge detection algorithms[1] and
a localized faulty sensor detection algorithm[2].
2.1 Statistical approach
In a sensor network, each node is equipped with a sensing device which can
be used to monitor an interesting phenomenon or event. The function which
determines whether an event or a phenomenon is detected by a sensor node is
denoted as the event predicate for that sensor node. Thus, a general statistical
neighborhood, perform statistical analysis, and feed the analysis result to a
boolean decision function to decide whether it is an edge sensor. The boolean
threshold, the more likely a node will be claimed as an edge sensor, and the
As illustrated in Figure 2, a sensor node S0 probes the values of event
predicates for those sensor nodes whose distances are within a predefined
probing radius.
Pages:
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266