Shortcutting IP flows over large ATM networks

In this paper we propose approaches for shortcutting of IP flows over large ATM networks. With large ATM networks we mean that the single physical ATM network is logically structured into multiple logical ATM subnetworks. Shortcutting across such large, logically structured ATM networks is a technique to avoid network layer processing as much as possible by maximizing the switched path across the ATM network. Existing schemes for shortcutting only provide mechanisms for constrained situations, as e.g. being solely applicable to unicast best-effort transmissions. Hence we try in particular to approach the currently unsolved respectively untreated cases of QoS and multicast transmissions.