多人语音系统中多播树的设计与实现_学位论文.doc

第 PAGE \* MERGEFORMAT 35页 毕业设计论文 多人语音系统中多播树的设计与实现 摘 要 近年来，随着计算机网络和通信技术的飞速发展，对网络带宽要求较高的流媒体技术也迅猛地发展起来，多人语音系统的研究和应用已成为当前热点之一。在传统的多人语音系统中，通常采取C/S(Client/Server)模式，即用户从语音服务器点击收听，然后语音服务器以单播的方式把音频流推送给用户。它的弊端在于：1）大量数据集中在服务器上存储和传输，不仅对中央服务器的带宽和处理能力要求很高，而且使得多人语音系统严重依赖于中央服务器，导致系统可扩展性差、可靠性差；2）中央服务器到与会节点时延过大且不均衡，造成音频质量不理想和语音包接收滞后等问题。 而引入P2P和应用层多播技术，作为多人语音系统的支撑技术，则可以突破这些限制，有效地解决上述问题。它通过建立和动态维??基于P2P的应用层多播树来传输音频数据，给用户提供高质量的音频收听服务。在深入分析P2P技术和应用层多播技术后，设计了应用层多播树的构建和动态维护算法，然后根据划分的各功能模块，在仿真实验平台上实现了基于P2P的多人语音系统。 测试表明，多人语音系统的功能实现基本达到目标。在一般情况下系统能够向用户提供较高质量的音频服务，具有良好的可扩展性，但由于均衡负载的能力不够，性能的稳定性有待提高。 关键词：多人语音； P2P； 应用层多播； 多播树 Abstract In recent years, as the rapid development of computer networks and communication technology, high network bandwidth requirements of streaming media technology is also advancing fast, and the research and application of multi-person phonetic system have become one of current hot spots. In traditional multi-person phonetic system, usually adopts C/S(Client/Server) pattern, namely the user clicks the pronunciation server to listen, then the pronunciation server gives the audio frequency class to the user by the unicast way. It has several malpractices: 1) Large amounts of data stores and transports on the sever, which not only has a very high request to the central servers band width and the handling ability, also causes the multi-person phonetic system to rely on the central server seriously, and leads to the bad extendibility and reliability of the system; 2) The latency between the central control server and the node is oversized and imbalanced, which causes some problems, such as not ideal of the audio frequency quality, and the lag of the receive of pronunciation package. However, the introduction of P2P and the technology of the application layer broadcasts, as the support technology of multi-person phonetic system, can breakthrough those limits, and solve the above problems effectively. It provides the user high grade audio frequency service, which trans