Minimal Startup Delay in HAS with Staggered Segments

Abstract:

Nowadays HTTP Adaptive Streaming (HAS) is the most prevalent solution on the Internet for multimedia streaming and is responsible for the majority of global traffic. The multimedia content is encoded at different bit rates and split into segments, so that HAS clients can download the segments at the bit rate that best fits the network conditions. Although content segmentation provides a good trade-off between adaptability and encoding efficiency, it is detrimental to low-latency streaming due to high startup delay. In this paper we introduce staggered segments, a means to achieving startup delays of a few frames, while the video is still efficiently encoded. We also demonstrate that systems using staggered segments achieve better Quality-of- Experience in very low-latency streaming as a result of the clients downloading higher bit rates than in traditional deployments.