Highlights-based Bitrate Adaptation Scheme for Mobile Video Streaming Service
Minsu Kim, Kwangsue Chung · 2019
Bitrate adaptation for video streaming is the de facto solution to optimize viewing experiences of users under the time-varying network conditions. However, most of the existing bitrate adaptation schemes do not consider user-related information, such as video highlights, in determining a bitrate of video chunk to be requested next. The user is more sensitive to the viewing quality in the video highlights than other parts of a video. To maximize Quality of Experience (QoE) for users, the bitrate adaptation needs to consider the video highlights. In this paper, we propose a highlights-based bitrate adaptation scheme for mobile video streaming service. The proposed scheme extracts the video highlights using deep learning-based method and utilizes it in the QoE optimization to determine the next video bitrate. Through the performance evaluation, we confirm that the proposed scheme can provide different video bitrates according to the video highlights.