Dynamic Revenue Management for Flows in Packet Networks

In a packet network with several classes of service (CoS), classes of packet flows are characterised by their quality of service (QoS) and their prices. Several studies exist where the QoS of the classes is studied when the prices are given, as well as studies where the QoS is given and the optimal prices are calculated. In this paper, we propose an approach where the prices and the QoS offered to each class depend on the current congestion level. We calculate the optimal decisions in terms of income maximisation using dynamic programming. Simulations results show that our approach has up to approximately 24% better results than the existing ones in terms of income for the provider.