부산대학교 도서관

To identify a cause of time-consuming per-packet caching and name-based forwarding of Named Data Networking (NDN) routers, we analyze the state-of-the-art NDN implementation. The analysis identifies wasteful cache computation of unpopular content as one of the causes of the time-consuming per-packet caching. To overcome the time-consuming per-packet caching and improve the packet forwarding rate of NDN software routers, this paper proposes a filter, which filters out unpopular content so that such wasteful cache computation is bypassed. We analytically prove that the filter efficiently identifies unpopular content without sacrificing the cache hit rate. Building blocks of the state-of-the-art NDN software and our filter are implemented as a proof of concept prototype. Using the implementation, we prove that the filter reduces the computation time for packet forwarding and thus improves the packet forwarding rate.