Dynamic admission control and path allocation for SLAs in DiffServ networks

Today's converged networks are mainly characterized by their support of real-time and high priority traffic requiring a certain level of quality of service (QoS). In this context, traffic classification and prioritization are key features in providing preferential treatments of the traffic in the core of the network. In this paper, we address the joint problem of path allocation and admission control (JPAC) of new Service Level Agreements (SLA) in a DiffServ domain. In order to maximize the resources utilization and the number of admitted SLAs in the network, we consider a statistical bandwidth constraints allowing for a certain overbooking over the network's links. SLAs' admissibility decisions are based on solving to optimality an integer linear programming (ILP) model. When tested by simulations, numerical results confirm that the proposed model can be solved to optimality for real-sized instances within acceptable computation times and substantially reduces the SLAs blocking probability, compared to a the Greedy mechanism proposed in the literature.