| Summary: | 碩士 === 國立成功大學 === 電機工程學系專班 === 96 === The wireless sensor network is one of the new important applications recently and it has been used in military, environmental monitoring and health-care. Sensor nodes collect the data and disseminate the data to server. There are lots of researches in those topics. We proposed a novel application which constructs a distributed database in wireless sensor network. It is different from general distributed database systems such as TinyDB. The distributed database system in wireless sensor networks can deal with searching the unknown data captured by sensor nodes. For example, we can dispatch the image database into the sensor nodes, and randomly deploy the nodes in the forest or grassland. It is an interesting problem for searching the unknown image captured by sensor in the network.
Each sensor node has limited processing capability and storage space and it has to dispatch the database to the different nodes. After randomly deploying the nodes in the area, there are two important designed issues for our system. The first is to find the object within a limited TTL. The second is to decrease the search overhead.
In this thesis, we do not consider the processing of images. Further, we proposed four mechanisms to dispatch the database. The most important one is BIBD-based dispatch mechanism. Furthermore, we propose a selective flooding mechanism to improve the system performance, and it can hold the successful ratio of search. From our experiment results, we know that the BIBD-based mechanism has the best successful ratio in finding the object and performance than the others.
|
|---|
