Use of End-to-End Multicast Measurements for Characterizing Internal Network Behavior

摘要

This is the Final Technical Report for MINC: Multicast-based Inference of Network-internal Characteristics. The primary aim of the MINC project was to develop a statistical foundation on which to develop and evaluate end-to-end based tools for inferring internal network behavior. The project accomplished two goals. First, a series of techniques based on maximum likelihood estimation was developed to infer link-level loss and delay behavior, and internal topological structure based on end-to-end multicast and unicast observations. Briefly, these techniques exploit the correlation inherent in multicast-based measurements to produce accurate estimates. Second, measurement software required to produce the necessary observations and a web-based analysis and visualization tool were developed and made available to the community. The measurement software can be included in any multicast application as well as used on the National Internet Measurement Infrastructure. The analysis/ visualization tool allows the user to obtain and visualize time varying loss rate estimates.