3D video has been a visually appealing and costly affordable technology and more sophisticated multi-view videos have also been demonstrated. Yet their remarkably increased data volume and stringent multi-stream synchronization make segment scheduling more difficult. We formulate the segment scheduling as a BQP problem and optimally solve it using an MIQP solver. However, given the high dynamics and the stringent playback deadline in real-time streaming, the optimal solution is too costly to be obtained. Thus, we develop two efficient algorithms that achieve an analytical guarantee in the worst case performance, in particular, the approximation factor is at most 3 comparing with the optimal. We extend our design to support multi-view video with view diversity and dynamics. Under different end-system and network configurations with both stereo and multi-view streaming, the simulation results demonstrate that our algorithms achieve near-optimal performance, and outperform others in terms of streaming quality, stream synchronization/smoothness and scalability.
Copyright is held by the author.
The author granted permission for the file to be printed and for the text to be copied and pasted.
Supervisor or Senior Supervisor
Thesis advisor: Liu, Jiangchuan
Member of collection