* Which links in fact experience congestion in the current Internet (and what kinds of links can be expected to experience congestion in the future Internet). Where does the congestion occur? In edge networks? Peering points? Transoceanic links? Links outside of North America? How can the congested links be characterized, in terms of bandwidth, propagation delay, level of statistical multiplexing, and the like.
* Jim Gast at Wisconsin is looking at some of these questions.
* Barford and Plonka in [BP01] have begun work on characterizing network anomolies such as network operation anomolies, flash crowd anomolies, and network abuse anomolies from flow-based measurements. [BP01] also surveys the existing literature in detecting DDoS attacks.
[BP01] Paul Barford and David Plonka, Characteristics of Network Traffic Flow Anomalies, ACM SIGCOMM Internet Measurement Workshop 2001.