|Stochastic Systems Group|
Professor Robert Nowak
Dept. of Electrical and Computer Engineering
The explosive growth of communication networks, combined with rapid and unpredictable developments in applications and workloads, has rendered network modeling, control and performance prediction increasingly demanding tasks. Optimizing network performance requires that end-systems have knowledge of internal network conditions. However, it is impractical to directly monitor traffic at each and every router. Measurements at the edge of the network (at hosts and/or edge routers) are relatively easy and inexpensive in comparison. The INCITE (InterNet Control and Inference Tools at the Edge) project at Rice University focuses experts from the fields of networking, digital signal processing, and applied mathematics towards the goal of characterizing network service using only edge-based measurement. This talk will overview the INCITE project and describe the "network tomography" problem in detail. Optimizing communication network performance and detecting attacks and intrusions requires knowledge of loss rates and queueing delays at different points in the network. Can we infer the loss rates and delays experienced at internal points in the network from edge-based measurements alone? This is the network tomography problem. A unified approach to internal loss and delay estimation will be presented, along with an expectation-maximization algorithm for computing maximum likelihood estimates. Analysis and simulation experiments will demonstrate the efficacy of the approach. An outlook on key open problems and future goals of the INCITE project will be discussed as well.
Problems with this site should be emailed to firstname.lastname@example.org