SIA: Segment importance based rate adaptation for Dynamic Adaptive Streaming over HTTP

Nowadays, Dynamic Adaptive Streaming over HTTP (DASH) emerges as an increasingly popular video streaming technology on the Internet. And many rate adaptation algorithms have been proposed to decide the best bitrate for DASH video players. In different real applications, not all the video segments are of same importance for users. However, none of these existing rate adaptation algorithms takes segment importance into consideration. In this paper, we propose a segment importance based rate adaptation (SIA) algorithm for DASH, which not only considers the network bandwidth and the client video buffer but also takes segment importance into account. To evaluate the performance of SIA under different bandwidth conditions, experiments are conducted under both different values of fixed bandwidth and available bandwidth variations. And results show that SIA achieves a considerable gain for the average bitrate for important segments (ABIS) under low fixed bandwidth and available bandwidth variations while keeping comparable performance of other objective QoE metrics compared to the buffer-based rate adaptation (BBA) [6] and the segment-aware rate adaptation (SARA) algorithm [11]. We also validate that in a certain range, the increment of the percentage of important segments in SIA will lead to the decrement of ABIS and the increment of the duration of video rebuffering.