PeNCache: Popularity based cooperative caching in Named Data Networks

Abstract

The performance of the caching system in Named Data Networks (NDN) is influenced by the way requests are routed and contents are cached. Due to the limited capacity of caches available at routers, different caching techniques have emerged to effectively use the available space. Cooperative content searching and caching improves performance, and this comes with additional communication overhead. In this paper, we present a popularity-based, lightweight neighborhood cooperative content searching and caching system PeNCache for NDN architecture to improve the overall performance. PeNCache reactively explores all neighborhood routers with the aim of retrieving content from nearby routers while routing requests towards the content source. To aid its caching decisions, PeNCache considers the local and global popularity into account. Global popularity is estimated by a set of designated nodes in the network who periodically exchange local popularity information to help derive global popularity. We present details of how Interest packets and Data packets are processed at each router and also how popularity estimation is done in PeNCache. We perform a simulation study to evaluate the performance of PeNCache on realistic network topologies using a discrete event simulator. Outcomes of simulation demonstrate that it outperforms state-of-the-art caching schemes in terms of cache hit ratio, content access time, average hit distance, and cache diversity.