Our goals and objectives for this project were to explore models and software for support of simulation of large complex networks. We focused on routing instabilities caused by adverse traffic, and on developing the simulation infrastructure needed to support that. In the out-of-core context we studied the relationship between pre-fetching and scheduling, extending previous work in the area developed in the context of caching. In the course of this we've done a good deal of code development, and with experimentation with models to study behavior.
On the educational end, we trained Anna Poplawski, Jason Liu, Srdjan Petrovic, Yougu Yuan, and Guanhua Yan to develop problem identification skills, as well as skills in analysis of literature and presentation of research materials. Liu obtained a Ph.D.
Our essential findings are that caching effects in routers, caused by worm traffic, can explain the routing instability observed in the real internet in the wake of the Code Red II and nimda worm attacks. Future work is focused on studying defensive measures that might be employed by routers, and modifications that might be made to BGP.