In-Network QoE and KPI Monitoring of Mobile YouTube Traffic: Insights for Encrypted iOS Flows

Solutions for in-network monitoring of QoE-related KPIs are a necessary prerequisite to detecting potential impairments, identifying their root cause, and consequently invoking QoE-aware management actions. We leverage a machine learning approach to train QoE and KPI classifiers for mobile YouTube video streaming sessions using features extracted from encrypted QUIC traffic. With previous studies having shown different service behavior across different access networks and different OSs, we go beyond related work and specifically address iOS measurements and models. We assess the performance of models trained on data from a lab WiFi environment and an iOS device through cross-validation, achieving promising results. Using the dataset collected in a lab WiFi network, and two additional datasets collected in operational mobile networks, we further report on the promising applicability of classifiers trained using the WiFi dataset when applied to traffic collected using mobile network probes. The implications of such findings show the potential to use the same classifiers for multiple usage scenarios, thus reducing efforts needed for data collection and training. Finally, we discuss the extent to which models previously trained for Android usage scenarios are applicable for the iOS platform.