【摘要】 与集成服务(InteServ)网络模型不同,区别服务(DiffServ)网络模型中的中心路由器是根据几个有限的服务类别而不是会话流的具体情况来转发调度数据包,这就需要一个重要的服务质量映射的过程。依据DiffServ实现方式的不同,服务质量映射的物理框架也有所不同,但一个关键的映射环节总是存在,即依据数据包的相对优先级系数RPI(RelativePriorityIndex)将其映射入相应的服务类别。该文运用数学工具对这一典型环节进行建模,从视频质量入手分析映射函数,最后在一定条件下得到保证视频质量最优的一种映射函数表达式,并从数学证明和实例分析两个方面论证了这种映射函数表达式的最优性。更多还原

【Abstract】 The bottleneck problem in InteServ networks was resolved by providing services on a perclass basis rather than a perflow basis by central routers in DiffServ networks. The physical framework of the quality of service (QoS) mapping differs according to different DiffServ realization methods. Analysis of different frameworks shows that an important block always exists when mapping individual packets into a relative class based on their relative priority index. A mathematical model of this block was used to analyze the impact of the mapping function on video quality. For some conditions, an optimum mapping function solution can be calculated mathematically and by an example.更多还原