Aiming at enormous overhead of data transmission in content distribution networks ( CDN) , a replica placement approach based on cognition is proposed. In which a cognition-based predictive model is built up in regard to content popularity according to the dynamic clients demand information gathered as the feedback from proxy servers. And based on the model the distribution and placement of the contents are heuristically implemented. Simulation results indicate that the approach can noticeably reduce the cost of data transmission in CDN as well as satisfying the dynamic demand of time-varying clients simultaneously. It has the advantages of low latency and low cost. Comparing with normal algorithms, this approach can distinctly improve the cache hit rate and reduce the average latency of clients' requests.%针对内容分发网中数据传输开销巨大的问题,提出一种基于认知的副本放置方法.通过对代理服务器反馈的用户需求动态信息,对内容流行度建立基于认知的预测模型,依据此模型启发式地完成内容的分发和放置.仿真结果表明,该方法能明显降低内容分发网络的传输开销,同时满足时变用户的动态需求,具有低时延、低开销等优点.与一般的算法相比,该方法能显著提高缓存命中率和降低用户请求的平均响应时延.
展开▼
