Deadline-aware TCP congestion control for HTTP adaptive streaming services

Over the last decades, video streaming services have gained a lot of popularity, representing over 70% of the total consumer Internet traffic worldwide. While originally dedicated streaming protocols have been used to deliver these services, for several years, the video streaming industry is steadily shifting to HTTP adaptive streaming (HAS)-based delivery over TCP. With each video segment, implicit delivery deadlines are associated in order to avoid playout interruptions. However, TCP is known to be far from minimizing the number of deadline-missing streams. In this paper, 2 deadline-aware congestion control mechanisms are proposed, based on a parametrization of the traditional TCP New Reno congestion control strategy. By introducing deadline awareness at the transport layer, the modulation of the congestion window size is dynamically adapted to steer the aggressiveness of the considered stream to optimize the delivery of video streaming services. The proposed approaches are thoroughly evaluated in an HAS video-on-demand scenario over tree-based delivery networks. It was shown that in a realistic scenario, the proposed approaches can outperform traditional congestion control strategies by up to 12% in terms of estimated mean opinion score, mainly by reducing the average video freezing time by more than 85%.