Research And Improvement Of File Sharing In BitTorrent System

With the rapid development of network technology, the traditional B/S structure and C/ S structure has a bottleneck, Internet users have a higher demand of many application areas. The appearance of P2P concept subverts the traditional structure of the network topology, which causes a wave of research and application in the Internet industry.Functions and statuses of all individuals in the entire P2P network is the same, so that no primary and secondary peers. Every individual in P2P calculates and transmits of information without the need of the existence of a similar traditional server, which characterized by rapid convenience, scalability and stability. So far, P2P applications have been used widely in a number of areas.As the most representative of P2P applications, resource sharing technology is the focal point of the research staff. In this paper, for which the BitTorrent system as the research object, we take its Tracker algorithm and the peer selection algorithm as a research priority, noting that the inadequacy of the original algorithm, to do the research and improvement to optimize the efficiency and overall performance of BitTorrent system.In this paper, based on the detail analysis of BitTorrent principles, key algorithms and communication protocols, we do the analysis and research of the Tracker algorithm and the peer selection algorithm. We mainly to optimize and improve the Tracker algorithm for the original random return neighbor selection and choking algorithm issues.Firstly, we qualitatively define process download the node into a different phase of treatment for different stages. Secondly, we introduce the important concepts in P2P systems-incentive mechanism, the definition of "contribution" as an assessment of the credibility of the standard nodes, which as the network thought to contribute to the degree of coverage for the measure to provide different quality of services. Finally, according to the request node, the different roles and situations and download status, we carry out a targeted treatment, in accordance with the different roles in the overlay network, transmission speed, the contribution of group sorting, the node for the request different situations, from different queue to find the most suitable node to transmission speed as the main criterion, followed by the introduction of the contribution as a secondary criterion. If there is little difference in speed with the contribution of the case as Select metrics, to some extent to achieve the implementation of incentive mechanism, as award and punishment mechanism is to punish the "free-rider", to improve the fairness of the whole system, robustness and overall performance.In order to verify the improvement theory of the original algorithm, we prepare a BT software written by C++ language as a network downloading node client, for improvement has been re-written into the text's contribution, the queue concept, build Tracker server as a file sharing platform, the dispatch center, and then on the inadequacies of the original algorithm validation is given, the last of the improved algorithm is tested on the experimental.The main focus of this paper is the analysis of incentives in peer to peer network. The incentive mechanism, node search, fault-tolerance research and security issues have always been the focus of the research. How to design a fair and reasonable incentive standard and to improve the robustness of such networks is one of the most important issues in BitTorrent. Specific to the BitTorrent system, or a central server unstructured peer network, how to measure the contribution of the node is the key to the node's contribution, in this article we set upload bandwidth and download bandwidth utilization ratio as a measure of utilization, while take into account the different nodes of the download status of the request, return the node of the request, taken together, made the original Tracker algorithm and the choking algorithm. After a experiment, we found that the improved algorithm of played a incentives role to some extent.