An Efficient Distributed Content Store-Based Caching Policy for Information-Centric Networking

Abstract

Content store (CS) is one of the main components of information-centric networking (ICN), which enables content objects to be cached and retrieved from any intermediate node in the network. However, in existing ICN designs, CS information is not exploited to coordinate content caching and content retrieval. CS of nodes in the network operates independently while Interest packets forwarding mainly uses forwarding information base (FIB). This paper highlights the importance of CS information for efficient content caching and content retrieval to improve the performance of information-centric networking, especially in resource-constrained environments like the Internet of Things. We propose an efficient caching policy to coordinate the CS of a node with its neighbor nodes in a distributed manner so that more and more popular content objects are cached in the neighborhood of the node. To exploit and coordinate CS information among nodes, we urge to enable CS information in the data plane of the network and design an efficient way for CS information transmission. Each node contributes to the objective of its neighborhood by maximizing its number of unique popular content objects being cached in its CS and not cached in the CS of its neighbors. We implement the proposed policy on top of state-of-the-art popularity-based caching schemes. Through analysis and experiments, we show that the proposed caching policy achieves a significant improvement in terms of cache hit ratio, stretch ratio, content retrieval latency, and energy efficiency significantly compared to state-of-the-art schemes.