| Summary: | 碩士 === 國立清華大學 === 資訊工程學系 === 94 === With the advances in wireless communication and global positioning systems, today’s moving objects such as moving cars have the ability to update their location and velocity information to a central server. Range queries for querying the current and future positions of the moving objects are becoming increasingly necessary. Existing methods have been developed to support range queries, but they unreasonably assume that an object only moves according to its predicted single path. In general, the certain location of an object is unknown until the object updates its location information to the server. After the update, the uncertainty of the object’s location starts increasing until its next update. Although querying these uncertain data results in imprecise answers, these answers can be possibly estimated with probability guarantees by uncertainty models. In this paper, we study an uncertainty model, which is a function that expresses the possible movements with corresponding probabilities of the moving objects. Unfortunately, due to the complexity of the probability evaluation and the large number of objects to examine, the process of querying with probabilities is very costly. To overcome this problem, we map the uncertain movements of all objects to another space for an easy indexing. Our proposed method first eliminates infeasible answers by querying on the index. Then, for evaluating the remaining objects, an approximate examination with an error bound is employed to lower the overhead of the probability evaluation. The experimental study shows that our technique reduces the number of object examinations and the cost of the probability evaluation.
|
|---|
