Project R-11526

Optimized video delivery over next-generation protocols (Research)

Video data is projected to constitute 82% of all Internet traffic by 2022, which implies that even small video traffic optimizations will yield significant economical impact. The inflexible transport protocols used nowadays were never designed with video streaming in mind and lack the ability to keep up with the pace at which the Internet landscape evolves. Our work is centered around the application of revolutionary next-generation transport protocols (e.g., QUIC) to the video streaming use case. These protocols promise operational flexibility whilst keeping compatibility with legacy Internet infrastructure. We argue that leveraging this flexibility allows to improve the performance of existing (adaptive) video streaming solutions. Within this next-generation protocols context, we will specifically look at server and network assistance to optimize video delivery. By exploiting server metrics and intelligence about the state of the network, the efficiency of the streaming process can be increased, not only for traditional but also for upcoming unconventional video formats like light fields. This will in turn lead to optimizations concerning bandwidth consumption and Quality of Experience (QoE), of which the latter is a non-trivial element to evaluate. QoE is a subjective measure that is hard to quantify by objective metrics. In the event those objective metrics are deemed to cause undesirable behaviour (from an end-user perspective), we will incorporate crowdsourcing evaluation techniques to subjectively quantify our implementations.

16 September 2019 - 10 January 2024