Abstract:

Geographic Information System (GIS) is increasingly managing very large sets of data, hence a centralized data index may not always provide the most scalable solution. Recently, the P2P networks have become very popular for sharing information in a totally decentralized manner. In this paper, a new Hybrid P2P Spatial Indexing Network (HPSIN) is proposed, which combines distributed Quad-Tree with DHT-based Chord network to maintain both query efficiency and system load balance. In addition, a simple theoretical model based on opened queueing network for HPSIN is established, which captures the fundamental characteristics of the system. The theoretical analysis and numerical computing results show that there exists an optimum point of tradeoff between efficiency and load balance, which can achieve the minimum overall query delay.

Key words:

GIS, P2P networks, spatial indexing, queueing network, theory