A new practical packet loss estimator for MPLS VPN services

For provisioning QoS guaranteed VPN services over packet-switching networks, the service controlling system must maintain the subscribed values of QoS parameters, especially the packet loss probability, to be below a preset number. Thus one of the main issues to be solved is to estimate the packet loss accurately and effectively based on the input stochastic traffic process. Inspired by the large deviation theory (LDT), two types of asymptotes loss estimator have been studied in the practical MPLS VPN networks: the large buffer asymptotic estimator (LBE) and the aggregate traffic approximation estimator (ATE). However, both estimators exhibit a large error from the actual loss ratio. A simple reactive estimator is proposed, which can adapt to the different contexts. The basic idea is to adjust the original estimator with one dynamic item that is based on the feedback of loss ratio measurement and adapt it to the changing of traffic model and buffer size. A series of experiments were devised to evaluate the performance of the new estimator under different traffic arrival models and different buffer sizes. The results show that the new practical estimator can calculate the loss probability more accurately.