Assume the total number
of sensor nodes in the field is N, and the width of the field is b. The expected
number of sensor nodes falling into the closed disk would be m = ??c2N
b2 . Given
a preselected positive number m, the radius of the disk can be determined as
c = qmb2
??N .
The procedure of finding C2 and C3 is summarized as follows:
1. Construct {N} and {N?¤}. Apply the faulty sensor detection algorithm to
produce the set C1 (?µ = ?µ1).
2. For each sensor Si 2 (S ??’ C1), obtain the di from NN(Si) to replace the
di from step 1 while keeping other d values unchanged.
3. Use equation (21) to recompute yi. If |yi| ?? ?µ, assign Si to set C2 (?µ = ?µ2).
4. Obtain C3(m), where m is a predetermined positive number representing
the expected number of sensor nodes in the closed disk used to detect C3
sensors.
2.6 The Bayesian algorithm
Unlike other localized methods, the Bayesian algorithm devotes most of its
e?®orts to trying to disambiguate faults from events and relies on other distributed
region growing algorithms [10] or centralized methods such as those
introduced in Section 3 to estimate the extent of an event region.
The first step in event boundary detection is for each sensor node to determine
whether its reading particularly corresponds to any events of interest.
Nevertheless, due to the possibility of faulty measurements, there could be
unwanted detection or missed detection which makes this problem even more
challenging.
Pages:
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278