AI empowered Smart User Association in LTE Relays HetNets

Relay nodes (RNs) deployment with wireless back haul in future mobile networks is considered a promising solution to enhance the indoor coverage region of regular base stations, overcome shadowing dips, and provide a very high capacity and Quality of Service (QoS). Relay node cell footprint is limited by low transmission power which may not allow it to carry a significant share of the traffic load, thus undermining the gains of emerging ultra-dense heterogeneous networks (HetNets). Hence, cell association in a relay-enhanced scenario is a key design task, and has become a very interesting research topic over the past few years. This paper develops artificial intelligence empowered self-organizing network (SON) solution to optimally distribute users among relays and macro cells in an automated way such that holistic network performance is optimized. Contrary to existing studies, this work presents a novel idea of traffic service class based cell individual offset (SCIO) to further enhance the gains achieved from deployment of relay nodes. Employing an LTE system level simulator, for relay-enhanced scenario with a traffic mix having distinct QoS requirements, we observe that reinforcement learning based strategy coupled with SCIOs can yield significant gain in terms of users satisfaction scores, cumulative system payload throughput and reduction in signaling overhead when compared with the genetic algorithm based scheme and current industrial practice of fixed CIOs.