By GPSR??™s forwarding rules, this
node will initiate a new perimeter traversal that will pass through the previous
home node. The previous home node will detect this condition and remove
its association as a home node for that event. When the perimeter traversal
returns to the new home node, the association between the event and that
node will have been completed. In this manner, GHTs are able to detect and
adjust for the arrival of new nodes into the system.
GHT Scaling: Structured Replication
If many events are hashed to the same location, the home node can become
a hot-spot which will a?®ect the performance and the network lifetime.
302
Chapter 12 Data Management in Sensor Networks
To avoid this, structured replication hierarchically decomposes the geographical
region enclosing the sensor network in a manner shown in Figure 7.
In structured replication, the rectangular or square boundary encompassing
the sensor network is split into 2D equally sized sub-regions, where D is the
depth of the replication. Consider an event whose home node is x; x is called
the root node for the event. Clearly x is located in one of the sub-regions
defined by the spatial decomposition. In each of the remaining sub-regions,
one can then compute the location of a mirror of x; this mirror has the same
coordinates in its sub-region as x has in its own.
This spatial decomposition can be used to define a hierarchy of mirrors, as
shown in Figure 7.
Pages:
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486