基于CDN的流媒体动态调度算法

Dynamic scheduling algorithm for streaming media based on CDN

采用指数分段缓存补丁块方案,根据媒体流行度更新缓存窗口大小,实现了流媒体对象在代理服务器中缓存的数据量和其流行度成正比的原则。仿真结果表明,该算法比MBP(multicast batched patching)算法和OBP(optimized batch patching)+prefix&patch caching算法具有更好的适应性,在最大缓存空间相同的情况下,能显著减少通过补丁通道传输的补丁数据,从而降低了服务器和骨干网络带宽的使用,同时节省了传输成本。

The scheme that the patch bytes were segmented and cached was employed. The cache window size was updated periodically according to the popularity of streaming media object. The principle was obeyed that the data cached for each streaming media object were in proportion to their popularity at the proxy server. Simulation results show that the strategy is more adaptive than MBP（multicast batched patching） algorithm and OBP（optimized batch patching）＋prefix ＆ patch caching algorithm. It can significantly reduce patching data through patching channel under the circumstance of the same maximum cache space. It can effectively reduce the server load and network bandwidth usage on backbone link. It can also save the transmission cost.