Machine Learning-based End-to-End QoE Monitoring Using Active Network Probing

Video on Demand (VoD) is responsible 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. Many proposals in the literature correlate network Quality of Service (QoS) metrics with indicators of user Quality of Experience (QoE). However, 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 previous work, we proposed a method to estimate QoE using active ICMP probing, which is widely supported by network devices and can be used for end-to-end network measurements. In this work, we improve our previous method by adding a secondary model that operates over the first step of QoE inferences. We also extend the evaluation of our approach by using two wireless and wired testbeds, reporting our results for different end-to-end setups subject to distinct connectivity conditions. Finally, we identify and discuss the advantages and limitations of our methods and assess their suitability in real-world production deployments.