Quality-driven bitrate adaptation method for HTTP live-streaming

Abstract

In this paper, we focus on the bitrate adaptation issue of HTTP live video streaming. We first present a systematic approach that enables the client to choose the best video quality when deciding video bitrate for next several segments. After that, based on the concept of JND (Just Noticeable Difference), an adaptation method is proposed to meet the tradeoff between the requirements of buffer stability and smooth video quality. The experiment results show that our proposed method can provide smooth videos in terms of perceptual quality even under a small buffer size (i.e. 10s). To the best of our knowledge, this paper presents for the first time the way to use a perceptual quality metric for adaptivity in HTTP streaming.