加权负载求导预测的流媒体接纳控制研究

Research on Admission Control for Streaming Media Services of Weight Load Derivation Forecasting

随着流媒体服务要求的不断提高,网络应用对计算、存储资源的需求不断增长。未来一段时间服务器负载的变化情况,是保证QoS要考虑的一个重要问题。因此,对当前网络应用的系统负载分析预测十分有必要。文中以有典型软实时特点的流媒体应用为研究对象,在原有三种资源的基础上,加入了硬盘资源作为第四种考虑对象,提出了加权的负载求导预测法,并以之为基础实现了流媒体的实时接纳控制方法。实验结果表明,该方法预测误差较小,能在高负载的情况下确保较高的服务接纳率和较低的丢包率。

As the increasing requirements of streaming media service, network applications need more and more computing and storage re sources. It is an important issue for ensuring the QoS for the next period of time server load variation. So, it is necessary to analyze the system load of the dedicated network applications. Focus on the typical streaming media applications with soft real-time. In the original three kinds of resources,and on the basis of the hard disk resource as a fourth to consider object,designed a load-measurement method to gather the system load. The experimental results show that this method has smaller error, and ensures the higher acceptance rate of the service and the lower packet loss rate in the high-load conditions.