Multicast Rate Control under Discrete Bandwidth Constraints

In this paper, we address the rate control problem for layered multicast sessions, with the objective of solving a generalized throughput/fairness objective. Our approach is based on a combination of Lagrangian relaxation and dynamic programming. Unlike previously proposed dual-based approaches, the algorithm presented in this paper scales well as the number of multicast sessions in the network increases. Moreover, unlike all existing approaches, our approach takes into account the discreteness of the receiver rates that is inherent in layered multicasting. We show analytically that our algorithm converges to rates that are approximately optimal. Simulations carried out in an asynchronous network environment also demonstrate that the algorithm quickly converges to rates that are close-to-optimal.