The Complexity Of The Structured P2p Network Query Mechanism

The emergence and popularity of Peer-to-Peer (P2P) networks have reversed the traditional C/S network. The extensive use in resource sharing, distributed computing, and real-time communications have aroused the huge interest in P2P.Nowadays, the ubiquitous networks provide us with all aspects of information resources, and how to search the distributed data of network resources effectively is also becoming a research hotspot. In P2P networks, routing algorithm is the core and directly affects the performance and scalability. Compared with several popular P2P search mechanisms, we think that DHT-based search mechanism is the future direction. Howerver, in large-scale networks, as there are a large number of nodes joining and leaving P2P networks optionally, it's difficult to maintain the DHT in the global range, and hard to achieve complex queries. As the appearance of more and more complex data types, effective complex search for distributed data becomes increasingly important. Therefore, complex query mechanisms of P2P networks have important theoretical and practical research value.The thesis brings a locality-preserving hashing mechanism and kd-tree technology into the multi-dimensional region queries, and proposes an efficient index architecture. It also proposes an advanced routing algorithm to improve query performance. But it needs more practices and improvements to apply these techniques into practice. The thesis first describes the development of P2P networks and characteristics of P2P computing, and compares it with the traditional C/S model. It also studies three types of P2P networks topology models. The thesis mainly researches the following issues:(1) Discusses the research focus of P2P networks. Besides discussing the DHT technology in detail, the thesis analyzes and compares some typical P2P networks, and also discusses some research achievements of a variety of P2P search methods.(2) To solve the problems of structured P2P networks for complex information retrieval, the thesis adopts a uniform locality-preserving hash mechanism and kd-tree to partition and map the multi-dimensional data space to a one-dimensional key space, and has designed an index architecture for multi-dimensional range queries. The experimental results show that the method can efficiently and reliably support range queries.(3) The thesis has proposed an efficient query routing algorithm to progressively refine and deliver queries to the corresponding index nodes. And based on the Pastry protocal, it achieves range queries on multi-dimensional data of structured P2P networks.(4) According to the proposed index architecture, the thesis developes a light-weighted load-balancing mechanism to adjust the load among nodes and ensure that no node is unduly loaded.The simulation results show that the proposed architecture is efficient and scalable compared with other range query methods.