| Summary: | 碩士 === 國立臺北科技大學 === 資訊工程系研究所 === 101 === Using data broadcasting to provide services has attracted much attention for years. Many types of location based services using data broadcasting have been studied, including point query, range query, and nearest neighbors search. In this paper, we consider the skyline query [4] where the location is an importation factor. For example, a user may want to find the nearest hotel having the lowest price. In such a query, there are two kinds of attributes, spatial and non-spatial attributes. The values of non-spatial attributes are static to all the users but the values of the spatial attributes sometimes are different to different users due to the locations of users. The results thus are different for different uses at different places. We consider two indexing structures based on R*-trees, named R*-tree and AR*-tree, for broadcasting data and propose effective approaches that allow the users to execute the location-dependent skyline query any time at any location. R*-tree indexes spatial and non-spatial attributes direct, and AR*-tree indexes spatial attributes and each node has additional non-spatial attributes. Different broadcast schedules on the broadcast index structures are considered. We last evaluate the proposed broadcasting protocols by performing the experiments and compare the performance in terms of the latency, tuning time, and memory usage when executing a query.
|
|---|
