一种基于压缩感知的重叠网络路径性能评价方法  

Compressive Sensing Based Method to Evaluate Path Performance in Overlay Network

作　　者：李广野[1,2] 雷为民[1] 胡丹丹[1] 张伟 

机构地区：[1]东北大学计算机科学与工程学院,沈阳110819 [2]国网辽宁省电力有限公司,沈阳110004

出　　处：《小型微型计算机系统》2017年第1期67-71,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61401081)资助;教育部中国移动科研基金研发项目(MCM20150103)资助;中央高校基本科研业务费专项资金项目(N150404005)资助

摘　　要：路径性能直接反映了网络提供传输服务的能力,准确地评价路径性能是提供高质量网络服务和用户体验的基础.在面向重叠网络架构场景中,由于网络异构和虚拟化特点,现有的普通网络路径评价方法已无法适用,是制约重叠网络技术深入发展的严峻挑战.利用路径性能在一定时间内具有自相似的特性,创新性地提出一种基于压缩感知的重叠网络路径性能评价方法.在对重叠网络路径性能参数研究的基础上,该方法通过建立信号的稀疏表示、设计观测矩阵、重构信号灯3个步骤,对重叠网络路径性能估计和预测.通过OMNeT++仿真及与传统算法对比,验证了该方法的可行性和有效性.Path performance directly reflects the ability of the network to provide transport services,and accurately evaluate the path performance is the basis of providing high quality network services and user experience. Due to the characteristics of network heteroge- neity and virtualization, the existing common network path evaluation method is not suitable for the overlay network architecture. It is a serious challenge to restrict the development of overlapping network technology. In this paper, we propose a performance evaluation method of overlay network path based on compressive sensing by using the characteristic of path performance in a certain period of lime. On the basis of the research on the performance parameters of overlapping network path, this method is based on the sparse repre- sentation of the signal,the design of the observation matrix and the three steps of the reconstructed signal. The feasibility and effective- ness of the proposed method are verified by OMNeT＋＋ simulation and compared with the traditional algorithm.

关 键 词：重叠网络 网络路径性能评价 压缩感知 服务质量 用户体验质量 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]