Accommodating heterogeneity and scalability for multicast communication

Multicast communication targets applications potentially involving a large number of receivers with heterogeneous data reception capabilities. It accommodates receivers' heterogeneity in one of two ways: a single-rate mechanism in which the multicast source adjusts its (exclusive) transmission rate based on feedback it collects from the network and/or the receivers, or a multi-rate mechanism in which the multicast sender can transmit at different rates. A rate-based multicast mechanism generally requires criteria for the selection of the session rates and the allocation of receivers to the rates. Although current single-rate and multi-rate paradigms do provide mechanisms to meet these requirements, they are ad hoc and are not grounded in a formal optimization criteria.; To address the above issues, we propose to define a formal fairness criteria, the inter-receiver fairness (IRF) under the single-rate scenario, to guide the choice of session rates and receiver allocation and then investigate possible problems associated with the deployment of the criteria under different network environments. The session rate of a multicast connection can be derived optimally by maximizing the inter-receiver fairness of the whole multicast group. The framework is first implemented within the ATM ABR service, with the enhancement of novel multicast ABR consolidation control algorithms.; To further investigate the inter-receiver fair multicast scheme, we extend the scheme from the (switch-assisted) ATM environment to the more generalized (end-to-end) Internet environment by using a special multi-rate model. We find that the inter-receiver fair mechanism can efficiently guide the settings of the rates for different multicast groups as well as the allocation of receivers to different groups. We also briefly investigate the TCP-friendliness when IRF-based multicast sessions share bandwidth with TCP connections.; Both the single-rate and special multi-rate models are limited in their abilities to improve the inter-receiver fairness for a multicast session when receivers in the session have significantly different capabilities. Therefore, we generalize the inter-receiver fair multicast to the multi-rate model. Moreover, we investigate the influence of the multi-rate inter-receiver fair multicast on other connections in the same network configuration.; We believe that the results of the dissertation can form the basis for a variety of additional work, such as investigating the possibility of applying the inter-receiver fair mechanism to real-time audio/video transmissions and to improve multicast bandwidth allocation and congestion control algorithms.