命名数据网中基于流行度的网内缓存替换策略

Novel caching replacement strategy based on popularity in named data networking

针对命名数据网络中如何高效地对节点内的数据进行替换的问题,对节点内已经缓存的数据块,根据被请求的频率、请求时间间隔,准确判断数据块在当前时间的流行度,提出了一种基于流行度的替换策略Po-Rep。从命中节点返回的数据决定要存储在相应节点时,把节点内流行度低的数据进行剔除替换。该策略使节点的内容保持最大价值,满足后续的用户请求。仿真结果表明,该策略有效提高了网内节点存储的命中率,降低了服务器的负载,提高了网络的整体性能。

How to effectively solve the cache replacement problem in the NDN（Named Data Networking）nodes, this paper puts forward a strategy＂Po-Rep＂to predict the popularity of all every cache chunk stored in the node. When a cache chunk traverses from the hit nodes, the strategy determines the coming cache chunk which will be stored in the node replacing the one with the least value. According to the request frequency and request time interval of the cache chunk, this strategy accurately predicts the popularity of the content in the node. So that all of the contents in a node can maintain the maximum value, which can satisfy the following user requests. The simulation results show that this strategy can improve the hit rate in the stored nodes of the network, reduce the server load and improve the overall performance of the Named Data Networking.