Data-driven power control of ultra-dense femtocells: A clustering based approach

Abstract

In this paper we present a data-driven power control (DDPC) approach to improve total cell throughput and energy efficiency of ultra-dense femtocells. Although femtocells can increase the capacity and coverage in an indoor environment, ultra-dense femtocells may consume a lot of energy and generate severe interference. We investigate a data-driven clustering approach to reduce co-tier interference among femtocells in a dense deployment scenario. The proposed DDPC approach periodically collects the operation data of dense femtocells, including reference signal received power (RSRP) from each user, the transmission power and the number of users per femtocell, and so on. The collected data are processed via the affinity propagation (AP) clustering algorithm to determine the cluster centers to perform power control. The AP clustering algorithm can automatically determine appropriate the number of clusters and the corresponding cluster centers for various femtocell densities. Simulation results show that the proposed DDPC approach can increase 41% higher total cell throughput and 64% higher energy efficiency respectively, compared to the approach without power control in the ultra-dense femtocells.