A Stochastic-Geometry Approach to Coverage in Cellular Networks with Multi-Cell Cooperation

Abstract

Multi-cell cooperation is a promising approach for mitigating inter-cell interference in dense cellular networks. Quantifying the performance of multi-cell cooperation is challenging as it integrates physical-layer techniques and network topologies. For tractability, existing work typically relies on the over-simplified Wyner-type models. In this paper, we propose a new stochastic-geometry model for a cellular network with multi-cell cooperation, which accounts for practical factors including the irregular locations of base stations (BSs) and the resultant path-losses. In particular, the proposed network-topology model has three key features: i) the cells are modeled using a Poisson random tessellation generated by Poisson distributed BSs, ii) multi-antenna BSs are clustered using a hexagonal lattice and BSs in the same cluster mitigate mutual interference by spatial interference avoidance, iii) BSs near cluster edges access a different sub-channel from that by other BSs, shielding cluster-edge mobiles from strong interference. Using this model and assuming sparse scattering, we analyze the shapes of the outage probabilities of mobiles served by cluster-interior BSs as the average number $K$ of BSs per cluster increases. The outage probability of a mobile near a cluster center is shown to be proportional to $e^{-c(2-\sqrt{\nu})^2K}$ where $\nu$ is the fraction of BSs lying in the interior of clusters and $c$ is a constant. Moreover, the outage probability of a typical mobile is proved to scale proportionally with $e^{-c' (1-\sqrt{\nu})^2K}$ where $c'$ is a constant.