一个NGN中骨干网的域内QoS估算和域间QoS通告模型

NGN是一个面向服务的网络, 整个网络被划分为不同的管理域（也可称之为自治系统）. 以前对网络QoS的研究的目标主要是网络内部报文流的处理及其相应的网络资源分配, 如分类/整形/调度, QoS路由选择, 带宽分配等, 但在NGN环境中, 有更多的因素需要考虑. 本文主要提出并分析了三个问题： （1）如何在网络内部计算最大的带宽瓶颈路径;（2） 如何至少在一定程度上将QoS的三个基本参数（带宽,时延,丢包率）划归为一个单一单数（带宽）, 从而简化域内QoS的估算; （3）如何进行域间的QoS协商和通告.本文分析上述问题更多的是按＂每服务＂而不是＂每流＂或＂每报文＂, 因为按＂每服务＂的QoS粒度符合下一代网络面向服务的思想; 其次,在＂域间＂这个层面上,按＂每服务＂的视角能够建立一个较宏观的QoS模型.

Model of Intradomain QoS Estimation and Interdomain QoS Advertisement in NGN

The NGN is service-oriented and a given service is bound to certain QoS parameters. The whole networks are divided into different administrative domains (also referred to as ASs). The objectives of previous research mainly focus on the QoS inside of a network, namely the process of packet flows and the corresponding resources allocation, yet more factors have to be taken into consideration in NGN. This paper raises and analyzes three problems: (1) how to select the route with max bottleneck in a network with respect to bandwidth, (2) how to, at lease to some extend, unify the three basic QoS parameters (bandwidth, delay, packet loss rate) into a single parameter(in this paper, bandwidth) and thus simplify the estimation of intradomain QoS, and (3) how to perform interdomain negotiation and advertisement. This paper puts forward a model to try to answer the three problems and the perspective is based upon "per service"instead of "per flow"or "per packet". Our philosophy stems from the following two considerations: firstly at the level of "interdomain", the granular of "per service" conforms better to NGN's service-oriented paradigm; secondly from the angel of "per service" , one can easily build a macro model.