A Measurement Tool for Videoconferencing User Experience

Abstract

The COVID-19 pandemic forced people to work remotely and use videoconferencing software like Zoom in their daily lives. While people are returning to their prepandemic lifestyle, many still depend on videoconferencing software. As a result, application developers need to regularly monitor user experience in terms of video quality, stalls, and network conditions, and identify areas of potential improvement. Companies and academic researchers focus user experience analysis on dual-endpoint, controlled conditions that do not reflect everyday user calls. Gathering data on a large scale without knowing the network structure and getting permission for traffic analysis takes time and effort. Such large-scale experiments often use lengthy procedures to obtain the right permissions and deploy monitoring infrastructure in the middle of the campus network.

In contrast to existing approaches, an ideal measurement application would merely run on users’ devices without cooperation from the other endpoint that they’re conversing with. Such an application enables researchers to collect network statistics across a wide range of Internet conditions at a fine-grained level without significant overheads. This thesis proposes the Single Endpoint Zoom Measurement Application (SEZMA) that computes and logs network and video metrics when a user is on a Zoom call and sends metric logs to a centralized server. In addition to providing insights for users and researchers, the application aims to be explanatory, usable, lightweight, and privacy-preserving.