Scheduling for fair allocation of rates in multirate multicast networks

We study fair allocation of service rates for real time loss tolerant traffic in arbitrary networks with multicast capabilities. Multicasting poses some specific fairness challenges. The fairness objective is that every receiver receives service at a rate commensurate with its capability and the capacity of the path from the source. Hence, different receivers should receive information at different rates. The source encodes the signal into several layers that can be incrementally combined to provide progressive refinement. Every receiver must receive the most significant layer (layer 1) for basic information. If a receiver has additional bandwidth, it can subscribe to other layers for better reception quality. A layer carries meaningful information, only when all the more significant layers have been successfully decoded. Thus the objective is to provide fair rates of service to the receivers, and to limit the packet losses to the less significant layers. Once the fair rates are known, many congestion control policies such as fair queueing can be used to attain the computed rates. However, rate computation requires the exact knowledge of system parameters, such as link bandwidths. We propose a scheduling policy which attains the maxmin fair rates without computing them beforehand. In addition to guaranteeing fairness, this policy confines packet losses to less significant layers, and protects the more important layers, when there is shortage of bandwidth. Furthermore, this policy does not require any knowledge of traffic statistics, is computationally simple, and is essentially local information based