The home node
corresponding to this location is the root of the hierarchy, called the apex.
Then the sub-region in a 0-level are divided into four equal sized quadrants,
which constitute the 1-level in the hierarchical structure. One node is selected
in each quadrant of the sensor network??™s geographic region as a cluster head.
All the cluster heads are sub-nodes of the apex in 0-level. To construct the
2-level from the 1-level, each cluster head in 1-level is regarded as the apex of
that sub-region. Then each sub-region in a 1-level are divided into four equal
sized quadrants, which constitute the 2-level in the hierarchical structure. One
node is selected in each sub-region in 2-level as a cluster head. All the cluster
heads in 2-level are sub-nodes of the apex in 1-level. This process continues recursively
until d levels are all constructed in the hierarchical structure. Figure
8 shows the construction of the hierarchical structure.
4.5 Distributed Indices on a Single Key
Besides spatial-temporal aggregates and exact match queries, sensor network
applications will need range queries of the form: ???List all the events for which
the temperature values were between 50?—¦C and 60?—¦C??™. Equally useful are
geographically constrained range queries: ???List all the events that occurred in
region A of the sensor network for which the temperature values were between
50?—¦C and 60?—¦C??™.
The Distributed Index for Features in Sensor networks (DIFS) [16] system
takes a slightly di?®erent approach that increases the e?±ciency of range queries.
Pages:
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491