期刊文献+

Internet链路时延分布的推测 被引量:1

Inference of Internet link delay distributions
下载PDF
导出
摘要 由于因特网越变越大,越来越复杂,但网络的监控管理又不能完全依靠路由节点监测链路的性能参数,QoS推测成为一个新的研究热点。在MFMM模型的基础上提出了一种方法,能够只根据测得的端-端的时延数据,推测单一测量路径上各链路的时延密度函数。文中给出了仿真结果,并分析了算法的有效性和进一步的研究方向。 Intemet is becoming more complex than ever. It results in the difficulty of managements based on the parameters monitored by the routers in existence. Intemet inference is a new research field. Based on the MFMM (Mixed Finite Mixture Models), this paper proposes a method that can estimate intemal link delay distributions along a single measurement path from end-to-end measurements. Simulation results are given. This paper also analyzes the efficiency of the method and further researches.
出处 《信息技术》 2005年第9期114-116,共3页 Information Technology
关键词 网络测量 网络推测 有限混合模型 network measurement network inference finite mixture model
  • 相关文献

参考文献4

  • 1Coates M, Hero A O, Nowak R, et al. Internet Tomography[ J]. IEEE Signal Processing Magazine, 2002, (5): 47 - 65.
  • 2Shih M F, Hero A O. Unicast - Based Inference of Network link Delay Distributions Using Mixed Finite Mixture Models[J]. IEEE Trans. On Signal Processing, Special Issue on Signal Processing in Networking,2003,51 (9): 2219 - 2228.
  • 3朱周华.期望最大(EM)算法及其在混合高斯模型中的应用[J].现代电子技术,2003,26(24):88-90. 被引量:21
  • 4Mario A T Figueiredo, Anil K Jain. Unsupervised Learning of Finite Mixture Models [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(3) :381 - 395.

二级参考文献3

  • 1Feser M, Weinstein E. Parameter estmation of superimposed signalsusing the EM algorithm [J] . IEEE Trans on Acoustics, Speech, and Signal Proceesing, 1988, 36, (4). 477-489.
  • 2Demrster A P, Larid N M, Rubin D B, et al. Maximum likelihood from Incomplete data via the EM algorithm [J]. Royal Statistical Society, 1977, 39, (11): 1-17.
  • 3Moon T K. The expectation - maximization algorithm [J] . IEEE Signal Processing Magazine, 1996,(11).

共引文献20

引证文献1

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部