Efficient Spectrum Spatial Reuse Approach Based on Gibbs Sampling for Ultra Dense Networks

The ultra dense deployment of small cells is consid- ered a key technology to achieve the requested capacity in future cellular networks. However, the interference pattern becomes more unpredictable and challenging in these networks. Therefore, a suitable trade-off between spectrum spatial reuse and interfer- ence level has to be pursued to achieve good performance in terms of provided throughput. This paper proposes a new method to maximize the achievable throughput of an UDN with a suitable level of spatial spectrum reuse. In particular, the focus is on the small cells tier where the available spectrum is divided into sub-bands and each cell can use some of these to communicate with its associated users. The goal is to find the sub-bands allocation among cells that maximizes the system throughput. However, to limit complexity and signaling overhead that could result unaffordable in an UDN, a new metric to approximate the cell throughput to be optimized is defined. Moreover, the newly defined problem is solved using the Gibbs Sampling approach. The method effectiveness is proven by comparing the achieved results with those of the maximization of the effective system throughput and the optimal solution.