Interaction-based Caching in Content-Centric Networking

CCN improves the hit performance, the content retrieval latency, and the bandwidth consumption of the network by increasing the availability of contents by in-network caching of them. To mitigate the resulting content redundancy issue, besides content replacement policies, deterministic and probabilistic caching strategies are developed to temporarily and spatially limit content replication. However, the entangling of content-related characteristics with topology-related specifics to jointly optimize caching requires further research. Inspired by Quantum Field Theory, we present a caching theory that interprets the interplay between the entities involved in caching as interaction. In the Interaction-based Caching (IC), clients of contents play the role of sources of gravitational fields, which couple to the demanded contents through intermediary force-carrying request quanta. The gravity on contents is then transformed to a graph-theoretical metric that enables the unification of the properties and states of the interacting entities with the network topology in coupling terms that describe the interactions. The multidimensional matrix structure of the coupling terms allows the differentiation between individual clients and contents, making the conduct of the kinematics of contents and requests in the network possible. This control gives rise to various policies for network management and prioritization of content clients and content producers. As a heterogeneous, implicitly cooperative, on-path as well as off-path caching strategy, the IC significantly increases the caching performance, especially concerning cache hit ratio, path length contraction, content retrieval latency, resource utilization, and service portfolio. We evaluated the performance of the IC model in a series of scenarios based on multiple simulation models with different parameterizations.