一种高效的对等网络流媒体数据调度算法

An Efficient Data Scheduling Algorithm for Peer-to-Peer Streaming System

在基于无结构对等网络环境的P2P流媒体系统中,针对现有数据调度算法不能充分利用节点带宽资源导致服务器负载较重的问题,提出了一种基于数据块优先级的数据调度算法.该算法根据邻居节点滑动窗口中数据需求信息和数据稀有性计算数据块优先级,调度时优先请求高优先级的数据,以提高节点间数据协作性.当多个源节点可以同时提供所需数据时,根据节点已上传下载数据量计算节点的贡献率,向贡献率最低的源节点请求数据,实现新加入节点上行带宽资源的快速利用.仿真实验与实际系统测试表明,该算法可以更充分地利用节点带宽资源,较传统算法降低服务器负载20%以上,提高了系统可扩展性.

A data-priority based scheduling algorithm is proposed to solve the problem that existing data scheduling algorithm cannot make full use of peers＇ bandwidth resources and causes heavy server load in unstructured overlay based peer-to-peer（P2P）streaming system. The algorithm calculates data block＇ s priority based on data demand information and data scarcity of neighbors. Data blocks with high priority are requested first in scheduling. Moreover, when there are multiple source peers that can provide the desired data simultaneously, the peer with the lowest contribution ratio is selected to request data so that rapid use of newly joined peers＇ upstream bandwidth can be achieved. Peer＇s contribution ratio is calculated based on the amount of data that peer has already downloaded and uploaded. Simulation and experiments in real P2P streaming system show that the proposed algorithm can reduce server＇s load by 20 percent and improve system scalability effectively.