Assessing the service quality of an Internet path through end-to-end measurement

摘要

Understanding the service quality of Internet paths is of great importance for both the service provider and the user. Motivated by this and particularly the aim of extracting such information from a given dataset collected through end-to-end measurement, this paper presents a novel approach, called SCI (Service-quality Characterization of Internet-path). Specifically, SCI utilizes delay and loss measurements collected from vantage points at the two ends of an Internet path. It transforms collected probes into performance signals, namely aggregate delay, average delay and aggregate loss signals. On the aggregate delay and average delay signals, abrupt changes are detected using Principal Component Pursuit (PCP). The detected abrupt changes in aggregate delay and average delay, together with the loss information, are further mapped as different types of events causing degradations and failures in the service within the path. Predicated on the set of identified such service-level events, SCI characterizes the service quality of the measured Internet path using three metrics, namely availability, stability and fatigue. The proposed approach has been systematically evaluated on the dataset collected from a real operational environment. The results show that SCI achieves good accuracy in detecting abrupt changes in aggregate delay and average delay and identifying service-level events, boosting the feasibility of service quality assessment of an Internet path based exclusively on end-to-end measurement.