Efficient end-to-end QoS mechanism using egress node resource prediction in NGN network

This paper proposes an efficient end-to-end QoS mechanism using egress node resource prediction via probe method in next generation network (NGN). As we want more smart and intelligent network, NGN is the most important issue to provide users converged and guaranteed services. To provide these services, NGN should support proper end-to-end mechanism to support a heterogeneous QoS environment in packet networks. We have studied in IPv6 DiffServ, MPLS DiffServ and other end-to-end QoS with differential service and, there are four ways to provide end-to-end QoS from current best-effort network to NGN; best-effort QoS in traditional IP network, QoS classification with priority in DiffServ network, QoS aggregation/TE in MPLS network and individual QoS/TE. One of the current problems to evolve NGN are there are many legacy equipments which we can not replace at once and various QoS differential service among networks. To address this, there should be proper admission control mechanisms to protect the core network and support end-to-end QoS in the access network. While planning-based admission is simple but not efficient, and probed-based admission control is efficient but has overhead in the network, egress resource prediction-based admission control is efficient and less overhead compared to the previous mechanisms. Egress resource prediction-based admission control mechanism has two parts. First, it checks the utilization of the egress node by probing. But not like probe-based admission control, it does not send probes as always as there are QoS requests. It handles a bundle of requests with probing, predicts the state of the egress node and sends the probe to handle the next bundle of requests. In this mechanism, how to measure the current state of egress node and predict its future state