Research Into Scheduling In Bittorrent Streaming System Based On Improved Particle Swarm Optimization Algorithm

In P2 P streaming system, proper scheduling policy not only achieves strong service capabilities, high throughput, short startup delay and good streaming continuity, but also reduces the waste of network bandwidth. Most studies suggest that mesh model reduces fluctuation in the network brought by node failure, which is more suitable for streaming media distribution. In order to provide media streaming service with higher quality, a new data scheduling mechanism based on improved particle swarm optimization algorithm is proposed.First, adaptive dual window piece selection algorithm is designed to improve the original piece selection mechanism. The core of piece selection algorithm of Bit Torrent is rare block priority, which is not suitable for media. In this paper, pieces of media file will be divided to an emergency window and a normal window according to the video-on-demand. Different window adapts different piece selection algorithm to ensure the quality of video play.Secondly, Neighbor Peer Selection based on Improved Particle Swarm Optimization(IPSO-NPS) algorithm is proposed. Improved particle swarm optimization(IPSO) algorithm is developed to overcome the shortcoming of falling easily into local optimum. Particle encoding and decoding are defined in the context of streaming media data scheduling, and fitness function based on the service ability of a peer is proposed. IPSO-based Neighbor Peer Selection(IPSO-NPS) algorithm is designed.Finally, implement experimental program to validate the effectiveness of IPSO algorithm and IPSO-NPS algorithm. Convergence speed and accuracy of evaluation of IPSO algorithm are verified by the six benchmark functions. IPSO-NPS algorithm is compared with random strategy and LRP algorithm from start delay, continuous video playback and network load.