Abstract / Description of output
The paper studies the benefits of multi-path content delivery from a rate-distortion efficiency perspective. We develop an optimization framework for computing transmission schedules for streaming media packets over multiple network paths that maximize the end-to-end video quality, for the given bandwidth resources. We comprehensively address the two prospective scenarios of content delivery with packet path diversity. In the context of sender-driven systems, our framework enables the sender to compute at every transmission instance the mapping of packets to network paths that meets a rate constraint while minimizing the end-to-end distortion. In receiver-driven multi-path streaming, our framework enables the client to dynamically decide which packets, if any, to request for transmission and from which media servers, such that the end-to-end distortion is minimized for a given transmission rate constraint. Via simulation experiments, we carefully examine the performance of the scheduling framework in both multi-path delivery scenarios. We demonstrate that the optimization framework closely approaches the performance of an ideal streaming system working at channel capacity with an infinite play-out delay. We also show that the optimization leads to substantial gains in rate-distortion performance over a conventional content-agnostic scheduler. Through the concept of error-cost performance for streaming a single packet, we provide another useful insight into the operation of the optimization framework and the conventional scheduling system.
Original language | English |
---|---|
Pages (from-to) | 1189-1198 |
Number of pages | 10 |
Journal | Journal of visual communication and image representation |
Volume | 23 |
Issue number | 8 |
DOIs | |
Publication status | Published - Nov 2012 |
Keywords / Materials (for Non-textual outputs)
- Video streaming
- Multi-path delivery
- Rate-distortion optimization
- Packet scheduling
- Rate allocation
- Server diversity
- Joint source-channel coding
- Markov decision processes
- VIDEO COMMUNICATION
- DIVERSITY
- INTERNET
- NETWORKS
- MEDIA