A Fast and Efficient Node Loss-Rate Estimator for Wireless Networks

This paper presents the Fast Loss Rate Inference (FLRI), an efficient end-to-end method with explicit estimation for inferring the per-node loss rates for wireless networks. The proposed estimator in this algorithm is proved to be consistent and able to correctly converge to the real node loss rate. The accuracy of this approach is evaluated, including analyzing the dependency of accuracy against different parameters for different network loss models. The results, including convergence time, show the ability of FLRI to be used to support real time content delivery in wireless networks.