We measure the cluster structure every 5s and count the total number
of nodes that have changed clusters during the first 100s period, so there
are a total of 2000 measures for 100 nodes. As shown in Table 1, nodes in
LEACH change their clusters all the time, while the other two schemes are
more stable. We also observed that Max-Min greatly depends on the ID of
the moving node. A large ID always causes more changes, while Forest is not
sensitive to node ID at all. When node 50 moves, 23 of them reported changes
in Max-Min, and 47 of them reported changes in Forest, which is 1.15 % and
2.35 % respectively. When node 98 moves, the reported changes are 84 in
Max-Min and 48 in Forest.
360
0
15
30
45
60
75
90
105
120
135
150
-75 -60 -45 -30 -15 0 15 30 45 60 75
Y-Coordinate
X-Coordinate
Motion of a node in a 100-node random network
Chpater 14 Performance Comparison of Clustering Schemes
Table 1. Total number of changes when a node moves
mobile node ID=50
Total changes Percentage(%)
LEACH 2000 100
Max-Min 23 1.15
Forest 47 2.35
mobile node ID=98
Total changes Percentage(%)
LEACH 2000 100
Max-Min 84 4.2
Forest 48 2.4
5 Conclusion
The area of sensor networks has been receiving increasing attention among
researchers in recent years, as available microelectro-mechanical systems and
wireless communications are now capable of supporting this technology. Over
the past few years, a variety of clustering schemes targeted specifically at the
sensor network have been proposed.
Pages:
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576