Chunk Grouping Method for Low-Latency HTTP-Based Live Streaming
Daichi Hattori, Masaki Bandai · 2022
The common media application format (CMAF) provides low-latency, HTTP-based live streaming. However, conventional bandwidth prediction for adaptive bit rate streaming is inaccurate because download time is affected not only by network bandwidth but also by the idle times between chunks in the same segment. Inaccurate bandwidth estimation decreases the viewer's quality of experience. In this paper, we propose a chunk-grouping method for low-latency, HTTP-based live streaming. The proposed method introduces delayed transmissions of HTTP requests for grouping the first of multiple chunks to accurately estimate bandwidth. From the results of the simulations, we show that, compared to the conventional CMAF, the proposed method could obtain higher video quality without excessive degradation in terms of re-buffering time, bit rate variation, or delay.