Optimizing the Performance of the Interaction-based Caching in CCN

LCE (Leave Copy Everywhere), as the default caching strategy of CCN (Content-Centric Networking), caches demanded contents in all en route routers. To optimally reduce the resulting redundancy, the dynamic features of contents, clients, and network elements must be entangled with those of the network's topology. Inspired by Quantum Field Theory and Newtonian Mechanics, we developed a model to comprehensively describe the clients' interactions with contents, which generate a “gravitational force” on the latter moving them toward their requesters. The “gravity” is then transformed into a topological metric to confine the replication of contents spatially to an optimum subset of caches. This Interaction-based Caching (IC) uses the couplings of contents and clients to regulate the kinematics of contents in the network deterministically. To maximize the IC's performance, the interaction terms must be adapted to the capacity and betweenness centrality degree of caches and the contents' popularity. The proposed algorithms are extensively analyzed and evaluated in various simulation scenarios showing that the IC significantly outperforms comparable caching strategies in terms of hit ratio.