Optimal data scheduling for P2P video-on-demand streaming systems

Abstract

Peer-to-peer (P2P) overlay-based streaming services have became more and more attractive. However, it is still challenging to provide scalable streaming services over large-scale Internet environment beacause of the stringent quality of service requirements as well as the dynamic nature of P2P overlay network. In this study, the authors focus on the optimisation of streaming data scheduling in P2P video-on-demand (VoD) system, with the objective of minimising the server stress and maximising the playback continuity. The authors first model the data scheduling problem to the maximum network flow problem where the schedule scheme is transformed to find an optimal supplier-consumer relationship assigment among peers with minimal server strees, and then present two max-flow-based streaming data scheduling algorithms by combining the upload capacity of peers as well as the path capacity between peers. The authors prove that the computing complexing of the proposed scheduling algorithms is polynomial. The practicability of the proposal is evaluated via simulations. Simulation results indicate that the proposed scheduling scheme could distribute bandwidth load among peers well while keeping the node degree low. Simulation results also show that the novel proposal has better performance than previous work in term of the server stress and the playback continuity.