基于泊松簇过程的三层异构蜂窝网络部署模型

5G标准下的蜂窝网络正在向异构化、超密集化的方向发展，传统的基于六边形网格模型的研究方法较为理想化且并不精确，越来越不适用于如今的异构网络.针对这个问题，目前常用的方法是使用基于随机几何的泊松点过程来研究异构网络的基站部署，这种方法假设基站的空域分布完全随机，因此得到了覆盖概率的理论下界.但是由于宏蜂窝边缘（盲区）以及热点地区（忙区）等特殊区域中，站点的分布可能形成簇，此时，基于泊松点过程的空域分布将不再准确.针对这个问题，本文使用泊松簇过程研究三层异构蜂窝网络的基站部署与规划.首先，提出基于泊松簇过程的基站部署系统模型，讨论了基于簇分布的基站形成过程；其次，在充分分析用户受到的聚集干扰基础上，采用基于瞬时信干噪比的小区选择机制，推导出了中断概率模型，并讨论了三种特殊条件下的中断概率；最后，通过仿真对比分析了基于泊松簇过程与泊松点过程的中断概率的差异以及信干噪比阈值变化时的中断概率的变化曲线，证明了基于簇的空域基站部署具有更低的中断概率.

Deployment model of three-layer heterogeneous cellular networks based on Poisson clustered process

Cellular networks in 5G are tending to be heterogeneous and ultra-dense. Simulation methods used in a traditional hexagonal grid model are too idealized and inaccurate, and they are no more adaptive to heterogeneous cellular networks today. To settle optimal deployment of base stations, a popular approach for analyzing heterogeneous cellular networks is Poisson point process (PPP) based on stochastic geometry, which assumes that the locations of base stations are completely space random. According to the model of PPP, a lower bound of the probability of coverage is gotten. However, the base stations or users may be clustered at specific areas (e.g. cell edge and hot spot areas) so that the space distribution based on PPP will be inaccurate. In order to settle this problem, Poisson clustered process (PCP) is used to model deployment and planning of the base stations in three-layer heterogeneous cellular networks including macrocells, picocells and femtocells. First of all, a model of the locations of base stations based on PCP was proposed and the formation of clusters of the base stations was discussed. Second, based on the analysis of aggregated interference received by users, the cell-association mechanism of instantaneous signal to interference plus noise ratio (SINR) was adopted, the model of outage probability using SINR was derived and three special cases of the model were given. Finally, according to the model, the difference of outage probability between using PCP and PPP was compared by simulation, and the curve of outage probability is changing as the SINR threshold changes. It shows that using PCP can get a lower outage probability than using PPP.