一种基于罚因子的DASH调度算法

随着移动互联网的普及,基于DASH的流媒体传输协议的应用越来越广泛。如何在带宽波动较大的移动互联网环境中保证用户实现流媒体的流畅点播,提高用户的体验度是DASH调度算法最主要研究的问题。以提高用户体验度为出发点,结合带宽和缓存深度两方面因素,对带宽预测模型的置信度进行评价。在高置信度情况下,大胆地对网络带宽估计模型的模型参量进行调整;在低置信度情况下,以保护缓冲区深度为目的,谨慎地对模型参量进行调整。这种调整势必会对QoE造成相应的影响,该影响作为"罚因子"反馈回模型置信度的评价,以获得模型参数的动态最优解,得到一种较好的DASH调度算法。

DASH scheduling algorithm based on a penalty factor

As the mobile network has become universal in recent years,DASH-based streaming media transmission proto-col is used more and more widely. Thus,how to guarantee the fluent video-on-demand in the mobile network circumstance with large bandwidth fluctuation and improve the quality of experience（QoE）are the major problems for DASH algorithm. In this pa-per,in order to improve the users’QoE,an evaluation on the confidence coefficient of the bandwidth prediction model is adopted in combination with the bandwidth and buffer depth. A bold adjustment of network bandwidth estimation model parame-ters under the condition of high degree of confidence. A careful adjustment of the model parameters should be conducted to pro-tect the buffer depth under the condition of low degree of confidence. This adjustment will certainly cause corresponding influence on the QoE. The influence will be taken as the ＆quot;penalty factor＆quot; evaluation of feedback back to the model confidence level for dy-namic optimal solution of model parameters and better DASH scheduling algorithm.