基于流量开销模型优化互联和路由策略的理论研究和实际应用

项目名称： 基于流量开销模型优化互联和路由策略的理论研究和实际应用

项目编号： No.61202356

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 王会

作者单位： 清华大学

项目金额： 24万元

中文摘要： 互联决策和路由决策是ISP运营中最重要的两个问题。之前该领域的研究主要集中在通过路由选择实现特定的流量分布目标，忽视了互联决策对路由决策和ISP业务的影响。而对互联决策的少量研究则侧重在互联协商的过程和互联建立的条件上，缺少对通过互联决策优化ISP业务的数学建模研究。本课题在对网络进行合理抽象的基础上，对ISP通过互联和路由决策优化网络的行为进行数学建模。基于流量开销分配模型，研究实际网络可用的互联和路由策略优化算法。并分析在更新网络配置实现优化策略的过程中的性能稳定性和自动化问题，研究其解决方案。在这些研究的基础上，实现支持预测分析和交互调整的互联和路由策略优化的决策支持和自动化系统。该系统的预测分析和决策支持功能，是对目前网络管理系统仅关注网络历史和当前状态监控的必要补充。

中文关键词： 路由策略；互联策略；网络性能；网络配置；网络管理

英文摘要： Peering strategy and routing strategy have significant implications on ISPs' business. However, previous research efforts in this area focused only on how to accomplish particular traffic distribution goals through routing selection, without any considerations on the influence of peering strategy on routing strategy and ISPs' business. Almost all of the few works on peering strategy are about peering negotiation techniques or peering conditions. There lacks mathematical formulation and analysis of optimizing ISPs networks by making proper peering decisions and routing decisions. In this research effort, we formulate ISPs' peering and routing strategy as an optimization problem based on a reasonable abstraction of ISP networks. We propose a traffic cost distribution model to solve the optimization problem using greedy algorithms that can be applied in real networks. We further study the challenges, such as forwarding loops and configuration automation, during updating network configurations to implement the optimal strategy derived by greedy algorithms. Finally, based on these research results, we design and implement a prototype to help network operators make decisions and configure networks. The decision support and result prediction functions of our prototype are necessary complement to current network manage

英文关键词： Routing Strategy；Peering Strategy；Network Performance；Network Configuration；Network Management