基于流行度分类的流媒体分段策略

Segmentation Strategy of Streaming Media Based on Popularity Classifying

对流媒体内部、外部流行度的特点以及流媒体外部流行度与缓存替换次数的关系进行了分析,并在此基础上,针对流媒体的缓存替换通常集中在某一小段流行度范围的特点,提出了一种基于外部流行度分类的流媒体分段方法,令替换频繁的视频采用较小的分段,而其他替换不频繁的视频采用较大的分段,从而提高缓存空间的利用率。实验结果表明,与均匀分段相比,提出的流媒体分段方法可以用更少的分段数目达到相同的命中率。

The characteristics of external and internal popularity of streaming media and the relationship of external popularity and the number of cache replacement were analyzed. Respecting the characteristics that cache replacements are focus on a small rang of popularity, a segmentation method based on popularity classifying of streaming media, which uses smaller segments on videos replaced frequently and bigger segments on the other videos, was proposed. The experimental result shows that the segmentation method can effectively reduce the number of segments compared with the uniform segmentation method under the same conditions of the byte hit ratio.