Geo-Clustering Algorithm for downlink cellular network planning

Instead of using hexagonal model to form cellular coverage, the clustering method is used in this paper. This method is rarely considered but suitable for cellular network planning. We propose an idea by using clustering method considering geographical elements to obtain better solution called Geo-Clustering Algorithm. Usually the simulated terrain is modeled as regular area such as urban, metropolitan, with general assumption that all areas have the same user´s dissemination. In this paper, we provide a simple but representative as logical thinking of geographical model in a given area. This geographical model represents the possible areas that users can be located. Comparison of cellular planning with traditional K-means cluster and Geo-Clustering model is presented in simulation results. In addition, coverage probability and required SINR are taken as main performance metrics.