An End-to-End Performance Inference Technique for Peer-to-Peer Networks

Abstract

For voice/video applications that are based on Peer-to-Peer (P2P) models, ensuring the end-to-end Quality of Service (QoS) is crucial, especially if users are paying fees. In this paper we propose an End-to-end Performance Inference Technique (EPIT) that uses a prediction-based approach to map the ingress traffic levels of the P2P network to the end-to-end QoS in the network. Furthermore, by coupling Simulated Annealing (SA) with EPIT, we describe a traffic engineering solution in such a way that the QoS constraints are met while traffic flows into the network are maximized.