Information Centric Modeling for Two-tier Cache Enabled Cellular Networks

In this article, we introduce a new metric called `information centric coverage probability' to characterize the performance of a two-tier cache enabled cellular network. The proposed metric unifies the dynamics of in-network caching and heterogeneous networking to provide a unified performance measure. Specifically, it quantifies the probability that a mobile user (MU) is covered at a desired rate when a certain content is requested from a global content library. In other words, it quantifies the percentage of time when an MU can be served locally without paying the traffic penalties at backhaul, fronthaul and core networks. Caching dynamics are modeled by considering that the content which is least recently used (LRU) is evicted while the requested content is stored in the cache. The considered two-tier cellular model leverages coordination between the macro base-station (MBS) and the small cell base-stations (SBSs) to maximize the resource efficiency. More specifically, coordination between macro and small cells enables an arbitrary SBS to exploit the caches at other SBSs in the neighborhood. Thus reducing the requirement for huge and expensive memory modules at individual SBSs. The spatial dynamics of cellular network are modeled by borrowing well established tools from stochastic geometry. Propagation uncertainties are explicitly factored in characterization by considering the small scale Rayleigh fading and the large scale power-law path-loss model. It is shown that the information centric coverage probability is a function of (i) the size of caches at the SBSs and the MBS; (ii) the content eviction strategy; (iii) the underlying popularity law for referenced objects; (iv) the size of the global content library; (v) desired downlink transmission rate; (vi) the amount of spectrum allocated to each tier; (vii) pathloss exponent; and (viii) the deployment density of the SBSs and the MBSs. Our analysis reveals that significant performance gains can be harnessed with appropriate dimensioning of both cache sizes and deployment density. Finally, identification of memory limited vs. QoS limited operational regime for two-tier cellular networks is considered.