Joint routing and admission control problem under statistical delay and jitter constraints in MPLS networks

Abstract

Admission control algorithms are crucial for ensuring a near optimal resource allocation while guaranteeing the required Quality of Service (QoS) constraints. In this paper, a new routing and Admission Control mechanism in multi-protocol label switching (MPLS) based networks is proposed. The proposed mechanism relies on the packets’ delay distributions that are empirically built on every link, to make adequate admission decisions. As a new connection request arrives, the admission control mechanism computes a feasible path, satisfying the connection’s statistical QoS constraints (the maximum end-to-end delay, jitter and the QoS violation probability). For that, we formulate this problem as an Integer Linear Programming (ILP) model that is solved to optimality using the CPLEX mathematical programming tool for real-size networks. Simulation results show that the new admission control mechanism allows for low blocking rates while statistically satisfying the QoS constraints for both the new and the ongoing connections.