中文摘要：

针对信息中心网络（ICN）内置缓存系统中的海量内容块流行度获取和存储资源高效利用问题,以最大化节省内容访问总代价为目标,建立针对内容块流行度的缓存收益优化模型,提出了一种基于收益感知的缓存机制。该机制利用缓存对请求流的过滤效应,在最大化单点缓存收益的同时潜在地实现节点间协作和多样化缓存;使用基于布隆过滤器的滑动窗口策略,在检测请求到达间隔时间的同时兼顾从源服务器获取内容的代价,捕获缓存收益高的内容块。分析表明,该方法能够大幅压缩获取内容流行度的存储空间开销;仿真结果表明,该方法能够较为准确地实现基于流行度的缓存收益感知,且在内容流行度动态变化的情况下,在带宽节省和缓存命中率方面更具优势。

英文摘要：

The in-network caching system of information-centric networking had to deal with the popularity of huge number of content chunks and make efficient usage of storage resources. A content popularity based caching gain optimization model aimed to get maximum reduction of content retrieve cost was established, and a gain-aware caching scheme was proposed. By utilizing filtering effect of cache to request flow, this scheme achieves caching cooperation and diversity potentially while maximizing caching gain of every single node. Bloom filter based sliding window strategy captures the content chunks with high caching gain according to request arrival interval and retrieval cost from the source. Analysis shows that the method can drastically reduce memory consumption caused by popularity monitoring. The simulation results illuminate that this scheme is well aware of content popularity based caching gain, and gets better bandwidth saving and cache hit ratio when content popularity is changing dynamically.