Real-Time QoE Estimation for DASH Video Using Active Network Probing

Abstract

Video on Demand (VoD) accounts for a significant amount of traffic on IP networks. To meet users' expectations, network operators need means to monitor and to identify when service quality is degraded in order to take actions to avoid customer churn. Most solutions cannot monitor end-to-end conditions without modification on video player applications or require deep packet inspection techniques, which may raise privacy issues. In this demonstration, we use active network probing to measure end-to-end network Quality of Service (QoS) conditions and use a Machine Learning model to infer users' Quality of Experience (QoE) in real-time. The results show that the method allows us to identify whether the network conditions allow video sessions with high QoE, or situations in which the user's QoE is degraded.