Research Project: VoIP: Measurements of End-to-End Quality of Experience on PlanetLab


William Liu, Krys Pawlikowski, Harsha Sirisena


As the Internet grows and routing complexity increases, network-level instabilities are becoming more common. Among the problems causing end-to-end path failures are router mis-configurations, maintenance and power outages, fiber cuts etc. Inter-domain routers may take tens of minutes to reach a consistent view of the network topology after network failures, and then end-to-end paths may experience outages, packet losses and delays. These routing problems can affect service performance and availability, which are negatively perceived by the end customers and degrade their evaluation on service quality of the Internet provider.

The primary target of this end-to-end measurement work is to design a measurement procedure which would translate QoS parameters to a new metric that can be used as Quality of Experience (QoE) in service level agreements (SLAs). Delay and loss of data are commonly used QoS metrics reported as a part of the SLAs. These metrics may have little direct meaning to the end-user because knowledge of specific coding and/or adaptive techniques is required to translate delay and loss to the user-perceived performance. The failure events will provide a new set of QoS metrics for real-time multimedia applications. Failure events are observable by the end-users independently of coding, adaptive playout or packet loss concealment techniques employed by their multimedia applications. Internet service providers can use the procedures developed in this work to explore the failure events within their networks. The time between failures and duration of failures are easy to understand QoS metrics that can be reported to the customers as a part of the SLAs. Customers can use the same procedures to verify the statistics reported in the SLAs. We will propose to use client-server model, active probe and passive measurement techniques to monitor VoI traffic flows on PlanetLab testbed, to understand network failure behavior in more detail and develop a new QoE metric.