Research On Fast Indexing Mechanism Of P2P Based Spatial Vector Data

With the extensive application of geographic information systems, it is gradually integrated into people's daily lives and used by more and more people. Thus spatial information service system is facing major challenges. The performance of the server was reduced by large number of concurrent users'online requests. Additionally, the transmission of massive spatial data in the network led low transmission efficiency. The emergence of P2P (Peer-to-Peer) network solved the two major problems which were faced by traditional systems. In P2P networks, fully use of clients'computing and service capabilities could improve the efficiency of spatial data indexing and achieve the mutual collaboration idea of"All for one, one for all".In this thesis, indexing structure and cache mechanism of spatial vector data based on P2P network were studied and disscussed in-depth. Based on the model of hybrid structure network, an indexing network architecture which integrated multi-scale and multi-layer was proposed. The architecture, which integrated spatial data of multi-scale and multi-layer with distributed Quad-Tree by pyramid model, reduced the network load and improved the efficiency of index. Based on this, a query algorithm for spatial vector data of multi-scale and multi-layer was proposed. The caching mechanism of this algorithm was focused on, and a client cache structure which supported multi-layer was designed. The simulation results of the three classical cache replacement algorithms (FIFO, LRU, LFU), showed that LFU optimized the system best and it could be used in the prototype system. Finally, a prototype system which based on fast indexing mechanism of spatial vector data in P2P network was achieved on the JXTA platform. Tests showed the effectiveness of the system. The research results of this thesis had high theoretic value for spatial vector data online service system, and also provided a new way for other networks with massive data.