Power Efficiency Enhancement in 5G HetNets Using Adaptive BPSO and Classification Trees

The exponential demand for multimedia services is one contributing factor to the significant increase in mobile data traffic. Consequently, the overloaded Macro Cells (MCs) are incapable of guaranteeing high Quality of Service (QoS) for every User Equipment (UE). Heterogeneous Networks (HetNets) comprised of Small Cells (SCs) and MCs is an auspicious solution to overcome this problem. Unfortunately, a critical challenge of the dense deployment of SCs in 5G HetNets is the co-tier interference between them. To enhance the power-efficiency in 5G HetNets, exploiting Machine Learning (ML) approaches is proposed in this paper. Adaptive Binary Particle Swarm Optimization (A-BPSO) algorithm is presented in this paper for turning the SCs on&off to alleviate the power consumption of the system. Additionally, a Soft Frequency Reuse (SFR) scheme is proposed exploiting Classification Trees (CTs) in order to mitigate the interference and ameliorate the overall network throughput. The results demonstrate that the proposed scheme surpasses the other conventional schemes in terms of the overall throughput and the power-efficiency of the network.