PRMP A Hierarchical Polling Based Reliable Multicast Protocol

Traditional sender initiated unicast protocols do not scale well for one to many reliable multicast due mainly to implosion caused by excessive rate of feedback packets arriving from receivers In our previous work we showed that such problem can be circumvented by making the sender poll the receivers at carefully planned timing instants so that the arrival rate of feedback packets is not large enough to cause implosion However there are two compelling reasons for extending such at protocol into a hierarchic version First given the amount of state per receiver necessary at the source the at scheme cannot scale to thousands of receivers Secondly when the receivers of a group are spread over a wide area network local error control and recovery may bring substantial gains in terms of throughput reduced latency if recovery can be done by nearby receiver and network cost recovery isolation In this paper we present prmp the extended hierarchic version of our polling based reliable multicast In redesigning the protocol we take advantage of the polling based implosion avoidance mechanism adapt the error and ow control mechanisms for the hierarchic case and nally add both session and congestion controls We propose two congestion control mechanisms window and rate based and show their e cacy through simulation We comment on related protocols compare but not evaluate As all prmp mechanisms are designed having network cost and throughput in mind prmp can achieve reliable multicast with the same kind of reliability guarantees provided by tcp but without incurring prohibitive costs in terms of network cost or recovery latency found in other protocols