Proportional fairness scheduling on tandem network

We study packet scheduling algorithms that satisfy multiple performance objectives simultaneously. We have realized the proportional fairness principle based QoS model, which defines both delay and loss rate requirements of a class, to include fairness. The resulting proportional fairness scheduling algorithms on a tandem network formalize the goals of the network performance, user's QoS requirements and system fairness, and expose the fundamental tradeoffs between these goals. In particular, it is difficult to provide these objects simultaneously. By performing simulation and measurement experiments, we evaluate the proportional fairness of the algorithm.