提出了基于自然数分段的流媒体主动预取算法,代理服务器向用户传送已被缓存的数据,同时,提前预取没被缓存的数据,提高了流媒体传送质量,减少了播放抖动。根据提出的自然数分段方法,理论分析了代理服务器预取点的位置和代理服务器为此所需要的最小缓存空间。仿真实验表明,在缓存空间相同的情况下,自然数分段方法比指数分段方法具有更高的字节命中率和更低的代理服务器抖动率,而与相同分段方法接近。
An active prefetching algorithm for streaming media based on natural number segmentation was proposed. The proxy prefetched actively the uncached data when the proxy delivered the cached data to the users. It improved the streaming media delivery quality and reduced playback jitter. Theoretical analysis determined the position that deter- mined the prefetching starting time and the needed minimum cache size at proxy server. Simulation results show that natural number segmentation strategy can gain more byte hit ratio and less proxy jitter ratio than exponential segmenta- tion strategy and near to those of uniform segmentation strategy under the circumstance of the same cache space.