Nikolas Wehner M. Sc.

Web browsing is one of the key applications of the Internet, if not the most important one. We address the problem of Web Quality of Experience (QoE) monitoring from the ISP perspective, relying on in-network, passive measurements. As a proxy to Web QoE, we focus on the analysis of the well-known SpeedIndex (SI) metric. Given the lack of app-level data visibility introduced by the wide adoption of end-to-end encryption, we resort to machine-learning models to infer the SI and the QoE level of individual web page loading sessions, using as input only packet- and flow-level data. Our study targets the impact of different end-user device types (e.g., smartphone, desktop, tablet) on the performance of such models. Empirical evaluations on a large, multi-device, heterogeneous corpus of Web QoE measurements for top popular websites demonstrate that the proposed solution can infer the SI as well as estimate QoE ranges with high accuracy, using either packet-level or flow-level measurements. In addition, we show that the device type adds a strong bias in the feasibility of these Web QoE models, putting into question the applicability of previously conceived approaches on single-device measurements. To improve the state of affairs, we conceive cross-device generalizable models operating at both packet and flow levels, offering a feasible solution for Web QoE monitoring in operational, multi-device networks. To the best of our knowledge, this is the first study tackling the analysis of Web QoE from encrypted network traffic in multi-device scenarios.

When conducting video QoE studies, participants are usually asked to rate the QoE of prepared test videos. However, participants are given no choice to select content, which they like or in which they are interested. This may cause annoyance or frustration when conducting the QoE study, which eventually might affect the QoE results of the study. The consequent question is whether the content liking has a direct impact on the submitted ratings by the participants and whether the freedom of choosing the video content in QoE studies results in better ratings. To investigate this research question, MCA, an existing framework for crowdsourced video testing, is extended and used in a pilot field study. MCA runs on mobile devices and allows users to watch the tested conditions of a QoE study within video content of their interest. The results of a QoE study with individual and dynamic content selection are compared to a QoE study with pre-selected contents. Moreover, this work includes a comparison to a previous QoE study for validation. As the previous study was conducted on desktop PCs, the MCA study further allows to identify differences in the stalling perception between studies on desktop PCs and mobile devices.

Measuring and monitoring YouTube Quality of Experience is a challenging task, especially when dealing with cellular networks and smartphone users. Using a large-scale database of crowdsourced YouTube-QoE measurements in smartphones, we conceive multiple machine-learning models to infer different YouTube-QoE-relevant metrics and userbehavior- related metrics from network-level measurements, without requiring root access to the smartphone, video-player embedding, or any other reverse-engineering-like approaches. The dataset includes measurements from more than 360 users worldwide, spanning over the last five years. Our preliminary results suggest that QoE-based monitoring of YouTube mobile can be realized through machine learning models with high accuracy, relying only on network-related features and without accessing any higher-layer metric to perform the estimations.