Layered multicast congestion control by particle swarm optimization in heterogeneous environment

To enhance multicast throughput in heterogeneous environment, a new layered multicast congestion control scheme is proposed. With the goal of maximizing global satisfaction of the whole group, allocating sending rate in each layer is formulated to an optimization problem. Since the problem is noncovexity, the sender uses particle swarm optimization to search a set of optimal layers rates. The new scheme also eliminates ‘lowest-first’ phenomenon by proposing a feedbacks suppression algorithm named equal-probability sampling (EPS). Upon EPS all the receivers send feedbacks at equal probability without bias. Simulation results prove that the new scheme can enhance global satisfaction and multicast throughput efficiently, compared with the traditional layered multicast congestion control scheme based on representatives.