基于分段的适应性流媒体缓存管理算法

为提高流媒体代理服务器的缓存效率,从而提高整个流媒体服务系统的性能,提出一种基于用户访问模式的适应性媒体分段缓存策略。该策略充分考虑用户行为模式对具体影片访问特性的不同,采取有针对性的分段缓存管理策略,可显著提高代理缓存命中率。为避免由于分段策略所潜在引发的播放抖动问题,对相应的预取算法进行了理论分析。事件驱动试验结果表明:所提出的缓存算法命中率较之传统的定长分段算法及系统现有的间隔缓存策略均有显著的性能提高,系统稳定时缓存命中率分别高出后两者54%和31%。

Segment-based adaptive proxy caching of multimedia streams

A user-access-pattern-based adaptive media segmentation strategy was developed to improve the cache efficiency of multimedia proxy server and to enhance the performance of the whole streaming media system.The strategy takes the user access patterns into consideration and uses different segmentation methods based on the different user access characteristics.An active prefetching mechanism was developed to prevent the potential problem of proxy jitter.Event-driven experiments show that the user-access-pattern-based adaptive segmentation strategy is remarkably more efficient than traditional uniform segmentation and interval caching strategies,with 54% and 31% higher cache hit ratio once the system stabilizes.