基于源根组播的多点视频会议系统模型及其实现

A Model and Its Realization of Multipoint Videoconferencing Based on Source-Specific Multicast

视频会议系统的传输结构对于系统的整体性能起着关键作用 .传统的面向连接的传输结构存在网络带宽开销过大、终端负担重、伸缩性差以及控制复杂等缺点 .本文定量分析了传统传输结构的性能 ,提出了一个使会议系统结构得到简化、可靠性和规模伸缩性得到提高的设计原则 .遵循这一原则 ,本文给出并实现了一个基于源根组播的会议系统模型 ,它强调网络组播的必要性 ,通过媒体传输与会议应用的分离 ,大大减少了网络传输开销 ,降低了系统复杂度 ,并支持更多用户的全交互式应用 .

A multipoint videoconferencing system is composed of four main components that are user interface,codecs,conferencing control and media transmission architecture.Among them,the media transmission architecture is most critical for a videoconferencing system.Traditional methods have many drawbacks such as high network resource overhead,heavy burden on terminals,low scalability and complicated control.This paper proposes a principle for design of videoconferencing,which makes the system simple,robust,feasible and scalable.According to the principle,a new model of multipoint videoconferencing based on multiple source-specific multicast trees is given.This model leads to simplifying design of the videoconferencing,reducing the overhead of networking resources and rising scalability.The paper introduces the evaluation method for transmission architectures,comparison of three types of traditional structures,the new transmission model and a practical videoconferencing developed by the anthers.