Subsequently, query processing takes place on the
centralized database. This approach is simple. However, the central server is
the performance bottleneck and single point of failure. In addition, all sensors
are required to send data to the central server, which incurs large communication
cost.
2.2 Semi-distributed Model
Sensors nowadays have improved computation and storage capabilities. Therefore,
certain computations can be performed on the raw data at each sensor
node. Most recent research focuses on the semi-distributed model. We present
two representative systems for this model.
(1) Fjord
Fjord, part of Telegraph (a developing project at UC Berkeley) [13], is
an adaptive dataflow system. Fjord has two major components: adaptive
query processing engine and sensor proxy. Fjord processes queries based on a
dataflow computation model. In Fjord, data streams are pushed to the query
processing engine instead of being pulled as in traditional database systems.
At the same time, non-sensor data is pulled by the query processing engine.
Therefore, Fjord is a query processing engine combining push and pull mechanisms.
In addition, Fjord integrates Eddy to adaptively change the execution
plans according to computing environments on a tuple-per-tuple basis. The
second major component of Fjord is sensor proxy, an interface between a single
sensor and query processor as shown in Figure 1. A sensor node only needs
to deliver data to its sensor proxy.
Pages:
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469