Improving OTT Video Quality

How to Optimize ABR Video Delivery With Server-side Quality Control - $15

With adaptive bitrate streaming, players evaluate the available bandwidth and perform their own quality selection. This approach has several drawbacks. In a fragmented market, each player has a different behavior, causing uneven end-user experiences. In addition, bandwidth evaluation (based on HTTP) may not work in the context of low-latency video streaming, impacting viewers? experience.

This paper will examine an innovative approach wherein the server performs its own bandwidth evaluation (relying on the underlying congestion control) and responds to a player?s segment request with its own selected quality. Experiments have shown the effectiveness of this approach in both unicast and multicast ABR contexts. As the quality selection is centralized, the behaviors of all players is homogeneous, and the service operator can change the selection strategy at any time, providing an unmatched level of control. Low-latency live streaming supports quality switching, since the server-side bandwidth estimate is not based on HTTP.

Guillaume Bichot | Broadpeak | Cesson-Sevign?, France
Pierre-Jean Gu?ry | Broadpeak | Cesson-Sevign?, France
Nicolas Le Scouarnec | Broadpeak | Cesson-Sevign?, France

Machine Learning for Per-Title Encoding - $15

Video streaming content differs in terms of complexity and requires title-specific encoding settings to achieve a certain visual quality. Classic ?one-fits-all? encoding ladders ignore video-specific characteristics and apply the same encoding settings to all video files. In the worst-case scenario, this approach can lead to quality impairments, encoding artifacts or unnecessary large media files. A per-title encoding solution has the potential to significantly decrease the storage and delivery costs of video streams while improving the perceptual quality of the video. Traditional per-title encoding solutions typically require a large number of test encodes, resulting in high computation times and costs. In this paper, we illustrate a solution that implements the traditional per-title encoding approach and uses the resulting data for machine-learning-based improvements. By applying supervised, multivariate regression algorithms like Random Forest Regression, Multilayer Perceptron and Support Vector Regression we are able to predict mandatory video quality metrics (VMAF). That way, the test encodes are eliminated while preserving the benefits of per-title encoding.

Daniel Silhavy | Fraunhofer FOKUS | Berlin, Germany
Christopher Krauss | Fraunhofer FOKUS | Berlin, Germany
Anita Chen | Fraunhofer FOKUS | Berlin, Germany
Anh-Tu Nguyen | Fraunhofer FOKUS | Berlin, Germany
Stefan Arbanowski | Fraunhofer FOKUS | Berlin, Germany
Stephan Steglich | Fraunhofer FOKUS | Berlin, Germany
Louay Bassbouss | Fraunhofer FOKUS | Berlin, Germany