ANN based joint coverage self-optimization for random-deployed cognitive femtocell networks

Abstract

Driven by the emerging demands of enhancing the indoor coverage, the network capacity and users' service experience, small scaled femtocell networks will be implemented extensively in home, office and hotspots, which can provide substantial benefits for both operators and users, such as the capital expenditures (CAPEX) and the operational expenditures (OPEX) reduction, the easy plug-and-play (PnP) deployment. Due to users' random-deployed manner indoors, femto base station (FBS) should have cognitive abilities, such as the position awareness, the automatic parameter tuning and the learning ability, in order to self-optimize its indoor coverage and minimize the unexpected interference outdoors. In contrast to the traditional static power allocation algorithm, the FBS has applied the dynamic power allocation scheme to minimize the interference to macro base station (MBS) and macro user equipment (MUE) by considering the path loss under different wall penetration conditions. Based on the position awareness ability, the optimal coverage radii for FBS indoors are proposed and proved by closed-form formulas at typical positions, which were not considered before. To further enable the intelligent coverage self-optimization for the random-deployed FBS indoors, a joint dynamic power allocation and antenna pattern selection scheme has been proposed based on the Artificial Neural Network (ANN) with training and self-tuning abilities. Numerous results prove both the effectiveness and the accuracy of the proposed coverage self-optimization scheme for the random-deployed femtocell networks. © 2014 Springer International Publishing.