The database is a centralized one. The sensors
in this example are temperature sensors.
AVG
D D D
DataBase
PlanB
AVG AVG AVG
AVG
D D D
DataBase
PlanA
Fig. 12. Comparison of two query processing techniques.
311
Jinbao Li, Zhipeng Cai, and Jianzhong Li
PlanA in Figure 12 shows the centralized query processing technique. Every
sensor returns the current temperature at a user defined sampling rate.
When the sampling data arrives at the database server, the values which are
lower than 30?—¦C are filtered out, and the arithmetic mean of the remained data
is calculated. Apparently, the query only concerns sensors that have readings
greater than 30?—¦C, whereas those values lower than the threshold are useless.
However, since all the sensors must return their values according to planA, it
wastes the limited energy, increases the load of the sensor network, produces
unnecessary transmissions and causes congestion in the network.
PlanB is the distributed query processing technique. During query processing,
the select and aggregate operations are performed on the sensor nodes
separately. In PlanB, only qualified data are involved in aggregation and only
partial aggregations are sent to the central database to derive the final results.
Consequently, this reduces the communication tra?±c in sensor networks and
saves bandwidth resource.
5.2 Aggregation Processing in Queries
Aggregation is a common operation for queries.
Pages:
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501