Improving spectrum efficiency in self-organized femtocells using learning automata and fractional frequency reuse

Abstract

Deploying heterogeneous networks (HetNets) and especially femtocell technology improves indoor cell coverage and network capacity. However, since users install femtocells which usually reuse the same frequency band as macrocells, interference management is considered a main challenge. Recently, fractional frequency reuse (FFR) has been considered as a way to mitigate the interference in traditional as well as heterogeneous cellular networks. In conventional FFR methods, radio resources are allocated to macrocell/femtocell users only according to their region of presence ignoring the density of users in defined areas inside a cell. However, regarding the unpredictability of cellular traffic, especially on the femtocell level, smart methods are needed to allocate radio resources to the femtocells not only based on FFR rules, but also traffic load. In order to solve this problem, new distributed resource allocation methods are proposed which are based on learning automata (LA) and consider two levels of resource granularity (subband and mini-subband). Using the proposed methods, femto access points learn to choose appropriate subband and mini-subbands autonomously, regarding their resource requirements and the feedback of their users. The goal of the proposed methods is reduction of interference and improvement of spectral efficiency. Simulation results demonstrate higher spectral efficiency and lower outage probability compared to traditional methods in both fixed and dynamic network environments. © 2017, Institut Mines-Télécom and Springer-Verlag France SAS.