Ssim-Based Adaptation for Dash With Svc in Mobile Networks

Abstract

Dynamic Adaptive Streaming over HTTP (DASH) depends on adjustment of the quality of a video stream to the available network conditions. In order to increase Quality of Experience, average video quality should be maximized, while keeping the quality switching frequency at low levels. However, achieving high average quality with low switching frequency in highly fluctuating mobile network conditions is a tricky optimization problem. In order to overcome this problem, dynamic structure of Scalable Video Coding (SVC) is utilized in this paper. Another challenge in the quality adaptation algorithms is to proper assessment of the video quality. Most of the adaptation algorithms takes either bitrate or representation level as the input that is used to evaluate the quality of the video. However, bitrate is not strongly correlated with the quality, as it depends on the content of the video. Likewise, representation quality relationship entirely bound to encoding. In this paper, in order to have a more reliable adaptation input, SSIM is used while representing the quality of the video stream. The proposed adaptation is compared with a successful SVC DASH adaptation algorithm using both subjective and objective tests. As a result, considerably higher scores are achieved in terms of both switching frequency and average quality.