Joint Preloading and Bitrate Adaptation for Short Video Streaming

Abstract

Short videos have become one of the most popular content mobile users consume nowadays. However, unlike traditional videos, users watch many short videos each time and frequently skip those not of their interest. Not taking into account this factor, conventional video streaming methods result in a large amount of data wastage. Moreover, short videos are usually streamed over time-varying network conditions, which requires preloading to provide satisfactory Quality of Experience (QoE). In this paper, we formulate the QoE-Wastage trade-off optimization problem in short video streaming as a multi-objective optimization problem. We then develop a lightweight joint preloading and bitrate adaptation algorithm that utilizes cross-user behaviors to decide a suitable amount of preloaded data for each video. In addition, our method dynamically adjusts video bitrate per chunk basis to achieve high user QoE. Experimental results show that the proposed method improves QoE by approximately 12% to 45% and reduces data wastage by up to 58% compared to existing methods.