Resource Adaptation for Continuous Profit Optimization in Overlay and Virtual Networks

Abstract

Virtual networks and service overlay networks lease bandwidth from Internet autonomous systems to offer end to end quality of service at competitive costs. In this paper, we propose a novel network resource management approach, based on an economic model, which allows continuous optimizing of the network profit, while keeping acceptable grade of service. The proposal integrates leased link capacity adaptation and connection reward adaptation with connection admission control and routing policies. In our approach, these policies are based on a decomposition of the Markov decision process. The approach implementation is based on link traffic measurements. Numerical analysis validates the approach and shows the convergence and stability of the adaptation as a function of measurement parameters.