Deep Learning based Simultaneous Wireless Information and Power Transfer Enabled Massive MIMO NOMA for Beyond 5G

Abstract

The exponential growth of wireless data services driven by mobile internet and connected devices has triggered the thriving of beyond fifth-generation (B5G) cellular networks. The integration of simultaneous wireless information and power transfer (SWIPT), non-orthogonal multiple access (NOMA), with multiple antenna system is a potential solution to improve spectral efficiency (SE) and energy efficiency (EE). Moreover, it paves the way for ultra-reliable and low-latency communication (URLLC) and massive machine-type communication (mMTC) scenarios. This paper explores an optimal solution for power allocation (PA) and power splitting (PS) control for EE maximization in SWIPT-based multiple input multiple outputs (MIMO) NOMA system. The significant aim is to maximize the EE of the system, maintaining equal fairness among the users in the cluster while satisfying the quality-of-service (QoS) requirements. The optimal solution to fulfill the trade-off between data decoded and energy harvested (EH) at the receiver is achieved through a deep learning (DL) model, viz., deep belief network (DBN). The dataset consisting of 3500 samples is created by varying the power levels from 20 dBm to 40 dBm, and also varying the distance of the users from the base station (BS). The PA and PS efficiency of 94.09 and 91.36 percent respectively, is achieved with DBN which aids for energy and SE in SWIPT MIMO NOMA system for 5GB.