A Study on Improving the Search Mechanism for Unstructured Peer-to-Peer Network Systems

• Main Authors: Chia-HungLin, 林家弘
• Other Authors: Sun-Yuan Hsieh
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/70813462537686090509

博士 === 國立成功大學 === 資訊工程學系 === 103 === 　　Peer-to-Peer (P2P) network systems are convenient and widely used because they enable data sharing, enable co-processing, and promote communication. This is especially true for unstructured P2P network systems that are easy to maintain, optimize parallel processing ability, and greatly increase data exchange all without requiring a customized server or specialized equipment. These qualities have made unstructured P2P networks common and widely used. 　　However, unstructured P2P network systems search data by using the “flooding” mechanism, and, even though the flooding mechanism is easy to use and fast, it requires a larger scale network, causes enormous traffic overhead, and wastes bandwidth. Recent improvements made to the flooding mechanism have provided limited advantages, but still cannot save bandwidth while maintaining good search performance. 　　The purpose of this study is to address these problems of the unstructured P2P network system search mechanism by providing an improved search mechanism which overcomes limitations, builds data sharing network convenience, and increases network communicating. 　　This dissertation is organized into 4 sections: (1) observation and analysis of the severity of current problems of unstructured P2P network systems with varying sizes of network topologies, (2) building an unstructured P2P network search mechanism focused on alleviating current search mechanism problems by providing increased search efficiency while saving bandwidth usage and keeping optimal performance, (3) evaluating efficiency by comparing mathematical models of the new search mechanism to various authentic network systems, and (4) performing a system simulation in various sizes of authentic networks to measure the efficiency improvements. 　　Our research indicates that our new search mechanism had significant improvements for various sizes of networks in every ranking compared to the traditional flooding search mechanism. In fact, our findings indicate our search mechanism reduced 80 percent of bandwidth waste. The findings give evidence that in both theoretical and actual testing results, our new search mechanism showed improvements in efficiency and practical application.