Resource allocation and admission control based on flow congestion probability in MPLS networks

Multi-Protocol Label Switching (MPLS) technology can provide service differentiation and fast packet forwarding through classifying packets into forwarding equivalence classes (FECs) at edge nodes and label switching in the core. By reserving resources for each FEC in the network and implementing admission control at the ingress nodes, the Quality of Service (QoS) requirements of flows belonging to a FEC could be met. User traffic in each FEC could be classified according to its traffic characteristics, thus we can calculate the capacity region of each FEC. Furthermore, the probability of flow congestion as the flow number combination has exceeded the capacity region can be found according to the arrival process and duration time of traffic classes. The main contribution of this paper is to present an optimized resource allocation algorithm and an admission control method based on the flow congestion probabilities of an FEC. Such that we can both provide QoS guarantees and improve the efficiency of network resource utilization in MPLS networks.