项目名称: 多业务网络流量矩阵估计方法研究
项目编号: No.61201127
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 电子学与信息系统
项目作者: 钱峰
作者单位: 电子科技大学
项目金额: 25万元
中文摘要: 网络流量矩阵估计是网路规划和优化、网路故障分析、流量分析等各种网络管理工作的基础,是目前备受国内外学术界和工业界关注的前沿研究课题之一。本研究针对目前网络流量矩阵估计方法不能区分业务的不足,提出了多业务流量矩阵的这样新的问题并建立了系统模型,并围绕多业务流量矩阵求解先验模型、适定性问题、求解实时性和非线性问题展开研究,通过解决相应理论问题,突破现有方法的局限,形成能获得节点对之间各种主要业务流量大小的多业务流量矩阵估计方法,进一步满足通信网络管理和优化需求。主要创新工作有:提出多业务智能先验模型和基于广义分布的概率统计模型作为多业务流量矩阵估计的先验模型;使用通信网络固有的特征信息进行约束求解、采用约束多目标最优化等数学方法,提高方法的稳定性,降低多解性;提出基于非线性最佳滤波理论和非线性混合优化方法的多业务流量矩阵求解方法;提出两种提高计算速度的方法来增强方法的实时性。
中文关键词: 先验模型;适定性;实时性;Hampel 范数;多业务流量矩阵
英文摘要: Network traffic matrics estimation is the foundation of a variety of network management activities including: network planning and optimization, network fault analysis, traffic analysis, etc, which has become one of the international academic and industrial circles-focused new network technologies. According to the defeet of the current network traffic matrix estimation method that can not handle multi-service, this research firstly puts forward a Multi-Service Network Traffic Matrics Estimation(MSNTME) problem and builds corresponding system model.Then this research carries out study on MSNTME priori model, well-posedness, real-time and nonlinear solution problems. By addressing the related key theoretical issues to breakthrough the limitations of current methods, this research strives to design a novel multi-service network traffic matrics estimation method which can obtain accurate flow traffic of different services between a source-destination pair, thus satisfies the requirement of network management and optimization. The main innovations include: propose utilizing multi-service intelligent proir model and generalized distribution based probability statistics model as MSNTME prior model; utilize the inherent characteristics of the communications network for constraint solving; utilize some mathematic method
英文关键词: priori model;well-posedness;real-time;Hampel norm;multi-service