Revisiting Optimization Based Rate Allocation for Application Layer Multicast: Distributed Model and Approaches

Abstract

Multicast is an efficient method to deliver data to a large number of receivers. In this paper, we revisit the topic of distributed and optimal rate allocation in application layer multicast. First, we propose a fully distributed network model based on our observation and assumptions, and form an optimization problem to maximize the aggregate utilities of all receivers. Second, we propose a novel primal approach and a typical dual approach as well as the accordingly designed algorithms that solve the optimization problem. Third, we extend the algorithms to an asynchronous setting to match the reality of large networks. Finally, we evaluate the performance of the proposed two algorithms in terms of aggregate utility, time complexity (convergence rate), and messaging complexity. Extensive experiments show that both proposed algorithms generate minimal messaging overhead, and that they are optimal in terms of overall utility for multicast applications.