Efforts have been made to determine traffic matrices based on per-link
utilization metrics and routing information. Difficulties with this
approach arise from the need to break the traffic on links into its
source-destination pairs. The breaking has to be done in such a way that
the network-wide link utilizations are achieved. Currently I am working
for Sprint Labs as an Intern and I am studying approaches to attack this
problem.

In the talk I will describe the research environment Sprint is offering
me, briefly describe the research topics being pursued in the lab and
discuss the general problem I am working on, emphasizing some of the
approaches previously taken. The nature of the talk is
"research-oriented" which means that I expect a discussion to take
place.