An Educated Guess on QoE in Operational Networks through Large-Scale Measurements

Network monitoring and reporting systems as well as network quality benchmarking campaigns use the Average Downlink Throughput (ADT) as the main Key Performance Indicators (KPIs) reflecting the health of the network. In this paper we address the problem of network performance monitoring and assessment in operational networks from a user-centric, Quality of Experience (QoE) perspective. While accurate QoE estimation requires measurements and KPIs collected at multiple levels of the communications stack -- including network, transport, application and end-user layers, we take a practical approach and provide an educated guess on QoE using only a standard ADT-based KPI as input. Armed with QoE models mapping downlink bandwidth to user experience, we estimate the QoE undergone by customers of both cellular and fixed-line networks, using large-scale passive traffic measurements. In particular, we study the performance of three highly popular end-customer services: YouTube, Facebook and WhatsApp. Results suggest that up to 33\% of the observed traffic flows might result in sub-optimal -- or even poor, end-customer experience in both types of network.