الفهرس | Only 14 pages are availabe for public view |
Abstract Emerging data stream management systems require the underlying processing server to be able to efficiently process the infinite stream and to support large num- ber of clients. Exisiting systems either can not sustain the large number of connected clients or do not support all the required types of queries. Three methods for data streams processing are proposed in this thesis, which add to the existing literature. 1. A single-server algorithm that is based on a state-of-the-art algorithm for k-NN queries continuous monitoring. This algorithm is extended to support range queries. The effect of pre-computing and storing the required data on the processing time of the algorithms is stud- ied and shown to enhance the performance. 2. A distributed system for processing the same type of stream based on an existing distributed system. The proposed system adds to the existing algorithms another model for the motion of clients, which considers the underlying road map. 3. A combination between the previous 2 methods to more accurately find the query evaluation region. The algorithm of (1) above is extended to operate in a distributed environ- ment and then employed in the distributed system in (2) above to improve its performance. The complexity analysis of the proposed algorithms is studied. A simulation model is developed to assess the performance of the algorithms with comparison to other existing algorithms. The simulation model is validated and verified. Simulation results demonstrate that the proposed algorithms are superior to exisiting algorithms in terms of either functionality or response time. |