A novel prediction-based content update scheme in cache-enabled smallcell networks

The smallcell network (SCN) is restricted by capacity of backhaul, which forces the smallcells (SCs) to prefer acquiring demanded contents from the neighbor SCs rather than data centers far away. In this paper, we investigate the content update problem aiming to maximize the local hit rates in a cache-enabled SCN where a macro BSs (MBS) is overlaid with a tier of SCs with caches. By employing mankind mobility model (MMM) and hidden Markov model (HMM), we describe the movement-in-area nature of human mathematically. Moreover, we derive the transition probability between each movement result and further predict the content demands on SCs. Based on the prediction, we transform the maximum sum hit rate content update problem as a kind of modified multi-commodity max flow problem (MC-MFP). Inspired by the previous works, we design an algorithm to solve the MC-MFP and then propose a Prediction-Based Content Update Scheme (PB-CUS). Simulation and numerical results show the efficiency of PB-CUS.