关键词:
Mobile computing
Mobile communication systems
Wireless communication systems
Electronic data processing
摘要:
In the approaching era of pervasive computing, we can expect to see a huge population of mobile clients, including people and devices, which require access to information at any place. Hence, the locations of clients and data objects have become important parameters for many information services - what we call location-dependent information services (LDISs). This thesis focuses on the provision of LDISs via wireless broadcasting. Compared to point-to-point connection, wireless broadcast systems provide a more efficient data delivery method that allows simultaneous retrieval of information by numerous clients. However, they introduce many limitations that have not been considered in traditional computing environments. This thesis conducts research on advanced indexing techniques to improve the performance of LDISs;For location-dependent data (LDD). the validity of a data value is dependent on the location of the client retrieving the data value. The retrieval of LDD requires the identification of the region that the client is residing in, given a partition of the geographical space. For example, finding the name of the city that the client is visiting requires identifying which city the client is located in given the city boundaries of a country. D-tree is proposed to efficiently determine the target region by indexing the division of adjacent regions. Compared to the existing approximation methods, such as Minimal Bounding Rectangles (MBRs) in R-tree, D-tree does not index any overlapping region and therefore avoids back-tracking. Compared to decomposition schemes, such as the trapezoidal map, D-tree represents the divisions in the search space and does not introduce any auxiliary line segments to the space. Therefore, the storage cost of D-tree is very low. Simulation results of both synthetical and real datasets show that D-tree outperforms existing indexes significantly in terms of search time;Next, we focus on nearest-neighbor search, which can be answered based