Abstract

DVS (Distributed Video Streaming) is an edge streaming architecture for providing smooth video delivery. It is to divide video among multiple streaming senders in order to effectively provide the required throughput. In P2P-based streaming, if some peers are connected with high speed physical link, it is highly probable that they receive the best quality streaming stably from an original source. On these peers, we can impose the role of DVS gateway, i.e., let them translate the original streaming to distributive form and relay to the other peers who cannot connect to the original source with enough bandwidth. In dedicated infrastructure-based streaming, DVS gateway may be recruited by a content delivery company as an infrastructure node, that is to say, this infrastructure node has enough bandwidth to receive the best quality streaming from the original source, and it thus able to relay the original streaming using the distributed streaming to usual subscribers.

In this paper, we consider a minimum DVS gateway deployment problem in DVS-based overlay streaming with satisfying bandwidth requirement of every peer or subscriber. We also consider the mandatory diversity and limitations of DVS gateways’ streaming capacity. The mandatory diversity is beneficial to mitigate the degradation of quality when an available bandwidth decreases suddenly, and the streaming capacity of each DVS gateway may be limited and heterogeneous. Based on recent advances in modeling techniques in flow networks, we provide a mixed integer programming formulation of the minimum DVS gateway deployment problem, and show that this problem is NP-hard. Thus we propose Lagrangian Relaxation with reduction heuristic to obtain an approximated solution. Our theoretical simulation studies show that the results obtained by our method are close to the lower bound obtained using LP relaxation.

Keywords: Distributed Video Streaming; Path diversity; Overlay streaming; P2P streaming; Lagrangian Relaxation

Article Outline

1. Introduction

2. The DVS-based overlay streaming model

3. System model and computational complexity

3.1. Mixed integer programming formulation

3.2. Time complexity of the minimum DG problems

4. Lagrangian Relaxation and reduction heuristic

4.1. Lagrangian Relaxation

4.2. Heuristic for DG reduction

5. Evaluation results

6. Conclusions

Acknowledgements

References

Vitae