Dynamic Cell Sectorization Using Clustering Algorithms

Cell sectorization is a common approach used in cellular systems to increase system capacity. In this paper a novel approach to downlink spatial filtering based on clustering algorithms is proposed for UMTS-FDD cellular networks. In more details, the well known segmentation algorithm K-means is applied for grouping users in order to define the spatial beams. It is assumed that the direction of arrival of received uplink channels known at the base-station. The ability of the proposed automated scheme is evaluated by comparing the computed beams with the optimum ones obtained heuristically in a static spatial configuration of users. Furthermore, simulation results in presence of DOA measurements errors demonstrates the reliability of the presented solution.