This proposal deals with the dynamic control of large networks that integrate real-time and non-real-time applications with heterogeneous Quality-of-Service (QoS) requirements. To meet these requirements, adaptive algorithms are needed for the routing, scheduling and admission control components. Routing provides a selection of routes, which should have sufficient resources to satisfy the applications' QoS. Scheduling allocates link resources (bandwidth, buffers, etc.) among the different service classes. Admission accepts or rejects a new incoming application, based on the requested QoS and the available resources. Because each control component causes and reacts to changes in the network state, there is a strong time-dependent interaction among them. By taking this interaction into account, the proposed work will develop QoS routing algorithms that will provide significant improvements in performance. To this end, we will develop dynamic integrated models. A novel numerical-analytical technique, called "Z-iteration", will be used to solve these models. The Z-iteration promises to provide rapid and accurate solutions to transient performance of integrated networks consisting of both wired and wireless links. Networks based on ATM and integrated services IP will be considered. The proposed work will develop software educational tools and projects for integrated networks, and will experiment with providing QoS support to educational multimedia applications. More information about this project can be found in www.ccs.neu.edu/home/matta/proj/project.html
