Researchers Honored for Work on Improving Online Video Quality

Image
Kevin Spiteri, left, and Ramesh Sitaraman
Kevin Spiteri, left, and Ramesh Sitaraman

Doctoral student Kevin Spiteri and professor Ramesh Sitaraman from the Laboratory for Internet-Scale Distributed Systems (LIDS) in the College of Information and Computer Sciences were honored with the Excellence in DASH Award at the Association for Computing Machinery Multimedia Systems Conference June 12-15 in Amsterdam.

The researchers were honored for their paper, “From Theory to Practice: Improving Bitrate Adaptation in the DASH Reference Player.”

DASH (Dynamic Adaptive Streaming over HTTP) is the open standard for online video streaming on the Internet.

The Excellence in DASH Award is presented each year by the DASH Industry Forum (DASH-IF), a consortium of major enterprises in the video ecosystem, including Microsoft, Google, Verizon, Netflix, Comcast, Cisco, Akamai, Intel, Samsung, LG, Huawei and others.

According to Sitarman, online video, whether streamed on Netflix, YouTube, or news sites like CNN, constitutes nearly 75 percent of all Internet traffic. As anyone who watches video on these or other services knows, stalled or low-quality video can be annoying and can significantly affect the enjoyment or utility of watching a favorite show, news clip or how-to video.

The award-winning paper proposes a solution to video-quality problems, presenting three algorithms, BOLA-E, DYNAMIC, and FAST SWITCHING, used inside video players. These algorithms enable the video player to play with fewer stalls (rebuffers) and play at a higher quality (bitrates). Further, they reduce the video quality switches during playback and respond more quickly to connectivity changes.

Notably, the three algorithms proposed in the paper are now part of the DASH video reference player standard. “This is important, since video providers often implement their own video players by using the algorithms in the reference standard,” says Sitaraman.

The proposed algorithms are already successfully deployed in production by major video providers such as BBC, CBS, Orange and several others who use the reference standard.

The BOLA-E and DYNAMIC algorithms use Lyapunov control theory to set the bitrate of the video to maximize video quality and minimize re-buffering. The FAST SWITCHING algorithm opportunistically replaces low-quality video segments in the video player's buffer with higher quality ones, before they can be rendered to the user. As its name suggests, FAST SWITCHING allows the video playback to switch to higher quality more quickly when your network connection improves.