Anticipatory quality adaptation for mobile streaming: Fluent video by channel prediction

Abstract

This demonstration shows a new type of HTTP adaptive streaming (HAS) for mobile users. We will present an anticipatory HAS policy that adapts video quality based on a prediction of the wireless channel state. By anticipating a poor channel state and by reducing video quality in advance, this policy maximizes the number of seconds in the user's playout buffer before coverage is lost. Even in difficult coverage situations such as tunnels and elevators, our approach reaches outstanding fluency without decreasing the average video bitrate. With this anticipatory buffering technique users require no channel resources at low channel state, which increases spectral efficiency. Our solution is based on an modified player application and will be demonstrated on off-the-shelf Smartphones in a conventional 4G network.