Grant Free NOMA with Deep Learning for Massive IoT Applications

Abstract

Internet of Things (IoT) has become a very hot research topic in recent years because it aims to ensure one of the major goals of 5G and beyond– connecting any physical device without human interaction. This type of data has been defined as Machine-Type Communication (MTC). A major design issue is how to sustain the network performance when a very large number of devices access the network at the same time. To resolve this issue, many studies propose new spectrum efficient methods that aim to increase the throughput, energy efficiency and other network performance parameters. One of these solutions is NonOrthogonal Multiple Access (NOMA), an alternative to wellknown orthogonal multiple access schemes. Even though NOMA was initially suggested for coordinated networks, it may be a feasible solution for the Massive IoT networks without coordinated access. The next generation networks should decide in a smart way. 6G-IoT networks also should be more intelligent with self-coordination. In this study Deep-Q-Network (DQN) based NOMA is proposed for the uncoordinated uplink transmission in IoT networks. According to the results proposed method is outperform the NOMA scheme with random selection in terms of throughput and power consumption.