Recent News
Computer Science Colloquium will discuss strategies for sustainable AI data centers
March 10, 2025
Dissertation defense: Alyshia Bustos
March 5, 2025
Computer Science undergraduate honored for cybersecurity research
February 20, 2025
Associate Professor Matt Lakin wins PECASE Award
January 31, 2025
News Archives
Using Network Tomography to Determine Link Delays in a Tunneled Network
October 5, 2004
Date: Tuesday October 5, 2004
Time: 11am-12:15pm
Location: Woodward 149
Hal Burch <hburch+@cs.cmu.edu>
School of Computer Science Carnegie Mellon University
Abstract: Network tomography is analyzing measurements of the combination of factors to separate out the individual components. An example of network tomography is using end-to-end delay measurements, which are the sum of delays over multiple links, to determining the delay across individual links. This type of network tomography is of interest to network operators, as it obviates the need to deploy measurement systems at all nodes within a network. Previous network tomography techniques require deploying monitors on a large fraction of the nodes of the network, which can be expensive and intrusive on large commercial networks. Moreover, past network tomography algorithms were based on assumptions such as symmetric delays, tree-like topologies, and the ability to send packets to internal nodes. This talks will present a system to determine link delays within a tunneled network that requires only one measurement host. Using linear programming techniques, our algorithm does not require packets to be sent to internal nodes or assumptions of symmetry or a tree-like topology. This talk will also present the results from the deployment of our system on one of AT&T's networks. This is joint work with Chris Chase and Albert Greenberg of AT&T Labs.