Design and evaluation of router-supported and end-to-end multicast receiver-based scoping protocols

IP multicasting allows a source to deene a multicast group address and receivers can dynamically join and leave this group. Currently the propagation of multicast packets is controlled by two scoping methods: TTL scoping and administrative scoping. Both of these approaches require the source to control the scope of the multicast. In this paper we explore two receiver-based scoping protocols. Receiver-based scoping allows a receiver to place conditions on its multicast join that must be met in order to join successfully and remain a member of the multicast group. Such a scoping mechanism would be useful in environments where the receiver incurs a cost for its membership of the multicast group. In this paper we discuss receiver-based scoping and its design goals. In order to perform receiver-based scoping a receiver needs information about the shape of the multicast tree. We describe two receiver-based scoping approaches to obtaining this information: router-supported and end-to-end. The router-supported approaches requires state to be maintained in routers while the end-to-end approach uses the multicast traceroute tool and does not require state maintenance by the routers. We evaluate the performance of these approaches in terms of their accuracy, bandwidth overhead and state requirements.