Caching Optimization in User-Centric Networks: A Stochastic Geometry Perspective

User-centric network (UCN) is regarded as a promising technology to provide users with high network capacity through a cluster of cooperative base stations (BSs). However, to support many-to-one transmission in UCN, the densely deployed BSs need to repeatedly acquire files from the core network via backhaul links, which further lead to tremendous backhaul traffic on existing UCN. In order to alleviate the backhaul traffic pressure in UCN, in this paper, we propose to deploy cache at BSs in UCN and investigate the performance of such cache-enabled UCN from a stochastic geometry perspective. To measure the network backhaul traffic offloading performance, we define a Universal Offloading Probability of Backhaul (UOPB) metric with deriving the coverage probability expression by stochastic geometry tools. Then, a UOPB maximization problem is formulated with respect to file caching probabilities to approach an optimal backhaul traffic offloading performance. By mining the monotonicity of the formulated problem, we develop an optimal polyblock outer approximation-based caching optimization algorithm with exponential complexity and a suboptimal stepwise adjustment caching optimization algorithm with lower polynomial complexity. Numerical results demonstrate that the proposed algorithms could bring up to 22% performance gains in terms of UOPB compared with reference caching schemes.