Design And Large Scale Simulation Of Scheduling Algorithm Based On P2P Live Streaming System

Nowadays, Live media streaming based on Peer-to-Peer Network, the most typicalmultimedia application in Internet, attracts widely interest and study. AlthoughPeer-to-peer live media streaming technology has almost approached the best, somespecial aspect is still facing the challenge seriously. Compared with the file-sharingsystem, the live streaming system based on peer-to-peer network demands more QoE.The idea system in future should be higher scalability, and well accommodated for thepeers’ dynamic and heterogeneous.Aim at what we have been confronted, I’ve designed a new scheduling algorithmafter a few research about some classical algorithm. We choose the tracker-based systemfor the overlay network, so the peer type contain the tracker、the source and the normalpeers. In chunk-scheduling, we proposed a prior algorithm combined with sequence、rarest and random first. In addition, the demarcation between urgency and urgency-lessregion is not a constant any more, which is self-adapted by the network condition. And inthe chosen of upload peers, we choose the method both consider the remainder bandwithand the transfer delay.We put the algorithm designed before in simulation on the platform ofParallel/Distributed Network Simulator PDNS, and described the most importantmodules like chunk-scheduling in details. In the process of simulation, we first analysethe performance change in different peer population, then we simulated the algorithm wedesigned in this paper and the original algorithm both in high heterogeneous and dynamicnetwork at10thousand peer’ scale, at last, we had a deep analysis with the data we’d getin the simulation. It proved that our design is perfect accommodated to the scalability、dynamic and heterogeneous.