基于动态预测算法的网络多媒体应用的改进

Improvement of multimedia network application based on dynamic prediction algorithm

提出了一种特定于网络多媒体应用的改进的QoS控制机制,在确定的带宽资源条件下,该控制机制可以以有限的网络多媒体业务服务质量为代价满足更多用户的网络多媒体应用需求.其核心算法——增强型指数平滑预测算法(EEWMA)是以指数平滑预测算法(EWMA)为基础优化而来,可以在预测精度和带宽资源占用量两者之间取舍,适用不同的网络环境和需求.对标准数据库进行测试的实例证明,EEWMA对处于网络应用有效连接建立时间范围内的不同的采样间隔,均可以得到较好的预测效果.

A new QoS mechanism of multimedia network application was proposed.Under the given bandwidth resource,the new mechanism can support more multimedia network application users at the expense of multimedia network application quality.The core algorithm EEWMA（evolution to exponential weighted moving average）evolved from algorithm EWMA（exponential weighted moving average）,and the EEWMA is a network status prediction algorithm.An appropriate prediction result for special network environment and demands,balanced between prediction precision and resource consumption,can be got by EEWMA.EEWMA can get a good prediction result to any time interval in the time range of most valid network applications connection,proved by the experiment adopting a standard database.