Dissertation > Excellent graduate degree dissertation topics show

Loss Performance Inference Algorithm Based on Network Tomography

Author: LinWen
Tutor: ZhangJianZhong
School: Xiamen University
Course: Signal and Information Processing
Keywords: Network chromatography Packet loss rate Multicast
CLC: TP393.01
Type: Master's thesis
Year: 2009
Downloads: 151
Quote: 3
Read: Download Dissertation

Abstract

With the expansion of the scale of Internet increasingly complex network structure optimization services , the abnormal link monitoring network major issue facing unprecedented challenges . However , the organizations or departments do not want to share their network status , exacerbated by the complexity of the problem a lack of mutual collaboration . Usually no direct access to all internal network performance parameters or topology , the distribution of this Internet do not collaborate , heterogeneity and other features make the the Internet measurement of extremely challenging . Network chromatography (Network Tomography) is an emerging technology . It is the use of end-to-end measurement to infer the network internal performance parameters ( such as : packet loss rate , delay , etc.) and topology . Chromatography network does not require the participation of all network nodes do not need to deploy measurement equipment to ensure the security of user information , while also reducing the measurement information transmission network . The network tomography technique has become one of the focus of the domestic and international network measurement . The study focused on the packet loss rate inference algorithm . The analysis of the existing inference algorithm based on link-level multicast packet loss rate characteristics , focusing on analysis MLE algorithm and PMLE algorithm . EM-MLE algorithm and the shortcomings of the algorithm of PMLE an improved EM-PMLE algorithm This algorithm combines the advantages of the MLE algorithm and PMLE algorithm , in the case to maintain a high inference accuracy , to reduce the amount of computation . Disadvantage against the PMLE algorithm estimation accuracy is not high , there is proposed a correction factor the μ-PMLE algorithm , by introducing a correction factor , to reduce the PMLE Algorithm estimation error when the packet loss rate , thereby improving the estimation accuracy of the algorithm . In addition , multicast for multi-point measurement technique to achieve the use of single-point multicast packet loss rate inference algorithm of multi- point multicast packet loss rate inferred weighted average fast algorithm . For the network does not support multicast , a variable package unicast packet loss rate inference algorithm . Research algorithm based on constructed chromatography simulation platform NS2 - based packet loss rate , using the platform to simulate a different network simulation environment , the aforementioned network packet loss rate inference algorithm simulation and through the analysis and comparison of the simulation results , summarizes the characteristics of the performance of these algorithms .

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Computer network > General issues > Theory of computer networks
© 2012 www.DissertationTopic.Net Mobile