Optimizing playback quality of HTTP-based dynamic adaptive streaming on smartphones

Due to the seamlessly adaptation to variable wireless network conditions, Dynamic Adaptive Streaming over HTTP (DASH) has been widely used in today's video streaming applications. Both the playback fluency and the average bitrate are significant performance metrics for mobile streaming users. However, the existing bitrate selection mechanisms based on the inaccurate throughput estimation will lead to a feedback loop, resulting in undesirably variable and low-quality video. In this paper, we first present a basic understanding of DASH system and make a simple analysis of the commercial players. We then model the bitrate selection problem under constrains of the buffer and the data requirement of users. To achieve high quality and smoothness playback, we propose an online dynamic video bitrate selection algorithm (DBS) based on the instant throughput and the buffer state. Parameters (α, β) in DBS can be set to adjust the bitrate switch frequency. The extensive simulation demonstrates that DBS can improve user's experience through video playback in different situations.