Proportional Response based Bandwidth Allocation for Layered P2P Live Streaming

This paper presents the analysis and design of proportional response incentive mechanism for layered P2P live streaming systems. It is shown that, under the proportional response mechanism, high contribution peers are guaranteed to get a better streaming quality. Furthermore, the sensitivity parameter of the mechanism can be used to achieve the trade-off between fairness and efficiency. It is also shown that, to avoid the content bottleneck problem, the sensitivity parameter must be within some specific ranges to ensure a high throughout of the P2P network. The theoretical results are then applied to the design of a layered P2P streaming system, where peers use a guided layer allocation algorithm to subscribe to the most suitable video layers and an efficient packet scheduling algorithm to exchange data with each other. Packet-level simulation experiments show that the optimized system outperforms other previous systems in both performance and fairness.