Improving the Transfer of Machine Learning-Based Video QoE Estimation Across Diverse Networks

With video streaming traffic generally being encrypted end-to-end, there is a lot of interest from network operators to find novel ways to evaluate streaming performance at the application layer. Machine learning (ML) has been extensively used to develop solutions that infer application-level Key Performance Indicators (KPI) and/or Quality of Experience (QoE) from the patterns in encrypted traffic. Having such insights provides the means for more user-centric traffic management and enables the mitigation of QoE degradations, thus potentially preventing customer churn. The ML-based QoE/KPI estimation solutions proposed in literature are typically trained on a limited set of network scenarios and it is often unclear how the obtained models perform if applied in a previously unseen setting (e.g., if the model is applied at the premises of a different network operator). In this paper, we address this gap by cross-evaluating the performance of QoE/KPI estimation models trained on 4 separate datasets generated from streaming 48000 video streaming sessions. The paper evaluates a set of methods for improving the performance of models when applied in a different network. Analyzed methods require no or considerably less application-level ground-truth data collected in the new setting, thus significantly reducing the extensiveness of required data collection.